Final Report Summary - BIO-DRIM (Personalized minimization of immunosuppression after solid organ transplantation by biomarker-driven stratification of patients to improve long-term outcome and health-economic data of transplantation)
Executive Summary:
The central focus of the BIO-DrIM project is the implementation of biomarker-driven strategies for personalizing immunosuppression (IS) in order to improve the long-term outcome and to decrease the adverse effects (graft toxicity, diabetes, cardiovascular events, opportunistic and community acquired infections, bone loss, and malignancies) and costs of chronic IS in solid organ transplant patients. The mission of BIO-DrIM reflects the new strategy in the field of solid organ transplantation: leaving the path of one-size-fits-all weaning strategies and administration of new drugs in favour to a more personalized approach.
The biomarker-guided management of immunosuppressive therapy is an ambitious program that started time ago with other projects (IOT and RISET) in which a set of promising biomarkers was identified and developed, but the methodical validation was missing. Implementation of these biomarkers into the clinical routine requires multiple cross platform tests, high quality standardization, multicenter clinical validation and with this an international network. All these requirements meet up in the BIO-DrIM project, that includes 5 innovative clinical trials (~2.000 patients in screening phase and ~800 patients enrolled in trials).
More in detail, the BIO-DrIM project addresses several highly innovative issues:
- First IS withdrawal study in long-term liver transplant patients based on the presence of a molecular tolerance signature (personalized withdrawal – study 1);
- Two novel trials on systematic partial IS withdrawal in selected highly stable long-term kidney transplant patients for validation of decision making biomarkers to detect operationally tolerant patients (personalized withdrawal - studies 2+3);
- First controlled biomarker-driven perioperative stratification of kidney transplant patients into low/high responders to prevent high-dose standard IS in low responders (personalized minimizing IS – study 4)
- Novel approach to target selectively activated (allospecific) memory/effector T cells (to increase the proportion of low responders after kidney transplantation suitable for low-dose monotherapy – validation by biomarkers – study 5
Safety is a key issue of the clinical trials: all studies are performed according to the international rules and all drugs are used according to their applications.
In addition to the well-validated set-1 biomarkers used for decision-making, all trials are accompanied by a panel of standardized biomarkers (set-2a) and a panel of exploratory biomarkers (set-2b):
i) to further improve the predictive value of the biomarkers for patient stratification
ii) to learn more about the mechanisms behind success/failure of IS minimizing.
Gaining knowledge about the mechanisms behind successful weaning (regulation/effector balance) is also part of the project. Well-defined in vitro and experimental mouse/rat transplant studies are used to address questions regarding the mechanisms of success/failure of IS minimizing. In particular, the scientists within BIO-DrIM address the interaction between regulatory pathways and donor-reactive memory/effector T cells.
The research and trials in BIO-DrIM are strongly translational. The involvement of SME´s and pharmaceutical industry guarantees a fast commercialization of promising product candidates.
Furthermore, health-economic studies push forward the translation into products used at the market as reimbursement strategies can be early developed and discussed with the health insurance companies and the government authorities.
Project Context and Objectives:
FIVE MULTICENTER CLINICAL STUDIES
LIFT Study (WP1a; biomarker-based weaning of liver transplant patients): Biomarker-guided IS weaning strategy in long-term (>3yrs) liver transplant recipients by identification of operationally tolerant patients
Previous weaning studies have demonstrated that about 25-40% of selected liver transplant patients after year 3-6 post transplantation develop operational tolerance and can have their maintenance immunosuppression discontinued. However, in the majority (60-75%), standard immunosuppression weaning fails and they develop rejection. The biomarker being developed and validated by the BIO-DrIM consortium, which is a gene expression test using liver tissue, is key to accurately stratify patients before considering immunosuppression withdrawal.
The LIFT study, a biomarker-driven trial coordinated by KCL, involves long-term stable liver transplant patients who are randomized to immunosuppression weaning without taking into account the result of the biomarker (Arm A) or to biomarker-guided weaning (Arm B). The trial is currently open in 12 centres in 4 EU countries and now, there are 190 consented patients, which are or have undergone screening, and 105 of them have been randomised. The first interim analysis of the trial took place in 2018 and confirmed the clinical utility of the 5-gene transcriptional biomarker. The final analysis of the primary outcome is expected to be completed in 2021.
WEANING trial (WP1b; weaning of long-term stable kidney transplant patients): Targeted partial/complete weaning of standard IS in long-term stable kidney transplant patients characterized as low-responders and identified as putative “operationally tolerant” by recently developed biomarker panel.
The fully blinded WEANING trial, coordinated by ITUN (Nantes), aimed at improving the graft function in clinically selected Highly Stable (HS) patients following complete weaning of calcineurin inhibitor (CNI). In December 2014, the group decided to stop the study due to insufficient patient recruitment. The results of the WEANING Study were published in 2016 (Am J Transplant 2016; 16: 3255–3261).
GAMBIT study (WP1c):
the group of KCL defined and validated a new gene expression signature that is independent of drug effects and also differentiates tolerant patients from healthy controls (cross-validated area under the receiver operating characteristic curve [AUC] = 0.81). In a prospective cohort, it was demonstrated that the new signature remained stable before and after steroid withdrawal. In addition, a validated and highly accurate gene expression signature, that can be reliably used to identify patients suitable for IS reduction (approximately 12% of stable patients), irrespective of the IS drugs they are receiving was reported. The results of the studies were published in December 2016 (Am J Transplant. 2016; 16: 3443–3457, PMID: 27328267).
Data obtained will be merged for ontology analysis to define the best final test format.
CELLIMIN trial (WP2; Prospective donor-specific cellular alloresponse assessment for Immunosuppression minimization in de novo renal transplantationtrial using a biomarker as stratificator for IS).
The CELLIMIN trial has been approved by the Voluntary-Harmonization Process (VHP). CELLIMIN is a non-inferiority trial with enrichment design aiming to demonstrate the utility of the IFN-γ ELISpot marker for the stratification of kidney transplant recipients into “low” and “standard-of-care” IS regimen.
All clinical sites participated in extensive lab and interlab comparision to validate the robustness and correctness of the IFN-γ ELISpot assay, the biomarker used as stratification tool.
184 patients have been screened with pre-transplant donor-specific IFN-γ ELISpot. 102 (55.43%) displayed a negative result and therefore underwent randomization. After randomization, 49 (48%) of patients received the “low” immunosuppressive regimen and 53 (52%) received the “High” immunosuppressive regimen. At 30th November 2018, 94 patients have achieved 12-month of follow up, 74 6-month follow-up and 16, 3-month follow-up. Despite not having achieved the number of patients pre-specified in the trial, the relevant number of transplant recipients recruited in this study is already unique and thus, will allow for a relevant evaluation of the impact of using a novel biomarker to stratify patients in different immunosuppressive regimens. Also, the analysis of patients recruited but not randomized to different immunosuppressive regimens, this is, those showing a positive donor-specific IFN-γ ELISpot will enhance the understanding of the impact of cellular alloimmunity in the context of kidney transplantation.
We will conduct a preliminary analysis of the whole recruited study population by next April 2019, when the last recruited patient will have been randomized. We will analyze the primary end-point with a minimum follow-up of six months after transplantation. Furthermore, we plan to perform the final analysis with the total number of patients randomized by next October 2019, when all patients will have achieved 12 months of follow up.
RIMINI trial (WP3, Tacrolimus after rATG and infliximab induction immunosuppression): Shifting kidney transplant patients to low-responders suitable for early IS minimization
RIMINI is an international multicenter open-label single-arm Simon’s two-stage Phase II clinical trial aiming to estimate confidence interval for the observed efficacy of the induction regimen with rATG and infliximab and a go/no go rule for further clinical development. A total of 75 patients will receive the proposed induction regimen. To introduce IS-based on tacrolimus/steroids as early as possible without losing control of acute/chronic rejections would be of great benefit and could reduce adverse effects and costs. However, this is only possible in a minority of patients yet. Memory/effector T cells are a major challenge. Therefore, increasing the pool of low-responders by selective targeting effector/memory T cells would have a big advantage.
Rationale: T-cell depletion is not sufficient to control allo-memory. ATG/Thymoglobulin treatment leads to a reduction of the clonal size of donor-reactive memory/effector T cells. Antigens, present in vivo, induce a bias of lymphopenia-induced proliferation LIP (preferential expansion of surviving alloreactive, Ag-specific Tem/Teff), which is an undesired effect. Here the combination of Thymoglobulin/Infliximab will help to reduce freshly activated naïve Tem/Teff to enlarge the pool of patients suitable for low IS.
Using tolerance/rejection biomarker monitoring we aim to prove shift from high responders to low responders while using infliximab as co-induction agent to target recently activated memory/effector T cells.
The initiation meeting was held in Berlin, on November 15th 2016. Enrolment into the RIMINI trial started in January 2017. At the end of 2018 56 kidney transplant recipients were enrolled into the RIMINI trial, 33 from Prague, 19 from Berlin and 4 from Barcelona. Thus, the recent enrolment rate is 75%. Not all intended clinical partners are currently enrolling patients for the RIMINI trial, which proved to be the main obstacle to reaching the expected numbers of patient enrolment. The principal reason is the enrolment into concurrent CELLIMIN trial. The study enrolment is still open and 3 participating centres enrol patients and will continue to do so beyond the end of the BIO-DrIM consortium. This is critical as only 25% patients (n=20) need to be enrolled into the trial to reach target population. Full set 12-months data evaluation is thus delayed and expected on June 2020 as we anticipate last patient to be enrolled in the trial in June 2019. Recently 20 patients completed the study with 12 months follow-up. 7 patients prematurely terminated the study. Recent number of efficacy failures (defined primary endpoint of the study) is 20 which anticipates acceptable efficacy and safety of the protocol and suggests non inferiority to the standard of care regimen.
VALIDATING SET-1 BIOMARKER TESTS READY FOR DECISION MAKING FOR THE ON-SITE PATIENT STRATIFICATION WITHIN THE CLINICAL TRIALS DESCRIBED IN WP1-3
Validated biomarkers (set-1) – as decision-making markers applicable for guiding minimizing IS:
- Molecular “tolerance signature” in liver biopsy of liver transplant patients
- Molecular “tolerance signatures” in peripheral blood of kidney transplant patients
- Donor-reactive IFN-γ ELISpot
The markers were selected in our previous studies performed in the EU funded projects IOT, RISET. The tests are well methodical validated now and already applied in the ongoing clinical trials.
Establishment of the ELISpot technique as stratification tool clinical trial.
10 ELISpot Readers were installed by Gen-ID at each local laboratory in the clinical sites of Berlin, Barcelona, London, Amsterdam, Regensburg, Hamburg, Prague, Nantes, plus Santander and Oviedo. Establishment at the local sites took place by extensive technology transfer, combined with training sessions and technical support.
VALIDATING RECENTLY ESTABLISHED SET-2 BIOMARKERS AND IMPLEMENTING NEW BIOMARKER CANDIDATES FOR IMPROVING PERSONALIZED IS WITHIN THE BIO-DRIM CLINICAL TRIALS
Methodically validated biomarkers (set-2A) – applicable for prospective validation concerning their clinical value:
- Molecular signatures in urine of kidney transplant patients
- Gene expression signatures diagnosing balance of innate and adaptive immune responses
- Immune cell subset composition profiling (Multiparameter flow cytometry)
New biomarkers (set-2B) – markers applicable for retrospective validation: e.g. quantification of new immune cell cell subsets (DC subsets, T-cell subsets, Treg subsets) applying our standardized multiparameter flow cytometry. Using our established platform, we have implemented such markers into the trials of BIO-DrIM for exploring their putative value in retrospective analyses.
Set-up and validation of Mulitparameter Flow Cytometry
Technical advice and equipment were provided and installed by Beckman Coulter - Immunotech at the clinical sites of six European countries. More precisely, 10-color Navios cytometers were provided by Beckman Coulter - Immunotech to the clinical sites of Berlin, Nantes, London, Barcelona, Prague, Amsterdam, Regensburg and Hamburg,
In a collaborative effort between Beckman Coulter – Immunotech and the Charite antibody panels in a ready-to-use dry format (DuraClone) were established, tested and validated. These panels are now commercially available.
ANALYSING THE HEALTH-ECONOMIC IMPACT OF BIOMARKER-GUIDED PERSONALIZED IS
Health-economic data demonstrated the usefulness of implementing biomarkers into the management of IS (personalized IS) concerning the cost/benefit ratio, performing health-economic analyses by using Micro Costing. The method used attempts to measure costs and benefits of service as accurately as possible, by including all fixed and variable costs of care at local prices, given the institutional structure within which service and care are being given. It also takes into account country-specific rules, such as fixed prices for diagnostic procedures (point system), drugs, and in/out-hospital service.
STUDYING THE MECHANISMS BEHIND SUCCESSFUL WEANING
The groups from Oxford, Nantes and Berlin worked together to develop a humanized mouse model of « Experimental system to investigate ‘high responder’ recipients who have pre-existing donor reactive memory cells ». This project contains two aims. The first one is to set up and validate in the three centres the human skin transplant model in humanized mice using the protocol developed by the Oxford group. Human immunoregulatory cell types have been administrated to humanized mice in this human skin transplant model to evaluate the efficacy of these cell types to control graft rejection. The second aim, performed by the Oxford group is to set up a new experimental system to investigate ‘high responder’ recipients who have pre-existing donor reactive memory cells in humanized mice. Oxford, Nantes and Berlin teams investigated the potency of their different human immunoregulatory cell types to control graft rejection in ‘high responder’ recipients using this model.
In parallel, Nantes group continued their work on the potential of new therapeutics to control accelerated heart rejection in rats, a model recently developed by this team.
DISSEMINATING THE RESULTS TO SCIENTIFIC, PATIENT AND PUBLIC COMMUNITY (AND DEVELOPING COMMERCIALIZATION STRATEGIES BY PARTNERING WITH SMES/INDUSTRIES
The aim is spreading the knowledge and the results generated by the project to wider community of end users and to the scientific community. The results have been shared with the different communities and commercialization strategies have been put in place (diagnostic products, drug and biomarker diagnostic combinatory products, novel targets of IS drugs). GenID developed and meanwhile market several EliSpot IVD kits e.g. EBVSpot, CMVSpot and TransSpot as outcome of the project. The QC kit GenID established together with the partners is a fixed size in offering standardized controls in routine use. Moreover GenID set up a sample bank with different donors to offer this in clinical trials. These kits are CE IVD marked according European guidelines.
With the experience and the obtained expertice from BIO-DrIM project to run clinical multicenter studies with EliSpot, GenID decided to set up a GLP / CRO lab to offer this service to pharmaceutical companies.
Due to the project, GenID has hired two more scientists and secures the jobs of 25 people employed in production and development. In addition, another position in Regulatory Affairs is planned for 2019.
Sales in the EliSpot area (especially EBV and CMV kit) have increased continuously over the last few years by more than 15 % and (rouphly two million) account for more than half of GenID's sales.
For Milenia Biotec the BIO-DrIM project served as a networking platform, resulting in a Co-Development-Agreement with the big pharma for the development of tests to be used within the area of transplantation.
Regarding Beckman Coulter – Immunotech, the collaboration within the BIO-DrIM consortium has led to the commercialization of the 1st multicolor panels in the DuraClone format. Hence, a model of reagent development at Beckman Coulter has been built following the successful collaboration with the BIO-DrIM. Collaborators are directly involved in the development of reagents which led to the commercialization of nine additional DuraClone products following this approach.
Thus, an effort has been devoted to implementation activities in support of the SMEs and industrial partners involved for understanding the early prospective commercialization strategies for the biomarkers of the Consortium. As consequence a good collaboration between the groups ended up with exploitable results.
Project Results:
WORK PACKAGE 1: Targeted IS withdrawal in long-term transplant patients identified as “operationally tolerant”
Study I: Use of tolerance biomarkers to predict immunosuppression withdrawal in liver transplant recipients
During the EU funded projects “Indices of Tolerance” (IOT) and “RISET” it was demonstrated that intra-graft gene expression markers accurately predict the outcome of immunosuppression withdrawal, which is influenced by the time and age of the transplanted patient.
Based on the insights gained from these two projects, it has been possible to go further and design a study with the overall objective to assess the clinical utility and risk/benefit ratio of employing a liver tissue-based transcriptional test of tolerance to stratify liver recipients prior to immunosuppression withdrawal.
Multi-centre, prospective, open label, non-controlled/non-randomized, interventional cohort phase IV study of immunosuppression withdrawal in stable liver recipients. (personalized withdrawal – LIFT trial)
PI: Alberto Sanchez-Fueyo, King’s College, London, UK
LIFT is worldwide the first randomized, multicenter study in long-term liver transplant patients to assess the clinical utility and safety of biomarker-guided immunosuppression, based on the presence of a molecular tolerance signature. The main goal is to determine the diagnostic accuracy of a liver tissue transcriptional profile of tolerance.
Secondary goals: to determine the correlation between tolerance and clinical parameters, iron status, blood gene expression, immunosenescence markers, microRNA, gut microbiota.
Figure 1: LIFT clinical trial.
Sample size/Patients: 148 patients to be recruited at 12 European clinical sites (King´s College London; Royal Free London; Cambridge; Leeds; Birmingham; Newcastle; Edinburgh; University Hospital Leuven; Hannover Medical Centre; Berlin Charité; Barcelona, St Luc Brussels).
Since the beginning of the trial (October 2015), we consented 191 patients, which are or have undergone screening, and 105 of them have been randomised. Recruitment rate is now within the target, but due to the delay in initiating the trial and opening some of the sites, there are plans for extending the enrolment period until May 2019. We are adding 2 new centres to help with the recruitment (Dublin and Palermo).
The breakdown by the centres is given below:
Site Consent Randomised Ineligible % ineligible Reasons for ineligibility Withdrawn consents Awaiting randomisation Awaiting biopsy or histological evaluation
King’s College Hospital 59 29 23 39 Histological- 23
4 1 2
Royal Free, London 18 7 10 56 Histological-10 1 0 0
Leuven 16 10 5 31 Histological- 5 1 0 0
Royal Infirmary of Edinburgh 9 8 0 0 0 1
Barcelona 23 14 3 13 Histological -3 2 4
Addenbrookes, Cambridge 12 7 5 42 Histological-5
0 0
Freeman Hospital 13 8 5 38 Histological- 5 0 0
Berlin 10 5 5 50 Histological- 5 0 0
Leeds 8 4 3 38 Histological- 3 0 1
University Hospitals Birmingham 2 2 0 0 0 0
Hannover 14 7 5 36 Histological-5 1 0
Brussels St Luc 7 4 3 43 Histological-3 0 0
Total 191 105 67 35% 6 4 9
Furthermore, there have been 27 SAEs for the following reasons: Vasovagal episode following liver biopsy, abdominal pain; pain after liver biopsy; post-biopsy liver bleed (2); jaundice, liver dysfunction / Biliary Anastomotic stricture & severe pancreatitis; Bullosus Pemphigold; tonic clonic seizure; rectal bleeding (2); herpes zoster; graft dysfunction; cholangitis (2); hepatocellular carcinoma; diverticulitis of the sigmoid colon; post-transplantation due to NASH cirrhosis; common bile duct calculi; lung carcinoma (2); non-ST elevation Myocardial Infarction with troponin rise of 1169; small bowel obstruction; infected right foot plantar wound; gastroenteritis; fall at home with brief loss of consciousness; fever + E.coli sepsis; salivary gland calculus and biliary tree dilatation.
As of December 2018, there were 11 fully weaned participants and 38 rejections. 21 participants were weaning off the IS and 31 participants were on the maintenance arm of the trial.
Meetings of the Data Monitoring Ethics Committee (DMEC) are taking place regularly (every 3 months). The second version (V2) of the statistical analysis plan is approved by the DMEC and the DMEC charter is fully signed. There is a draft version of the SAP that is being reviewed by the DMEC (V2.1) to be approved at the next meeting.
An electronic Case Report Form was created and has been functioning well.
A minimisation algorithm is implemented for the treatment group assignment.
The primary endpoint will be the successful discontinuation of immunosuppression with stable liver biopsy and liver tests 12 months after IS withdrawal.
Ancillary studies will be:
• HRQoL and Health Economic analyses
• Mechanistic studies:
Influence of iron status
Changes in regulatory T cell populations
Influence of T cell epigenetic imprinting
Influence of gut microbiome
Role of anti-HLA antibodies
The impact of IS withdrawal on quality of life will be evaluated using disease specific questionnaires. Sequential biological specimens will be collected to conduct ancillary mechanistic studies.
The study design of LIFT, the second biomarker-guided trial of the project (prospective randomised marker-based trial to assess the clinical utility and safety of biomarker-guided immunosuppression withdrawal in liver transplantation), has been subject to a series of modifications, resulting in a push-back in the initiation of the trial. There is a further one-year follow-up for each patient after completion of the withdrawal process. The first interim analysis took place in November 2018, with 37 patients reaching the primary endpoint. We plan to have the second interim analysis in May 2019 (n=56). The secondary outcomes will be evaluated at 12 months post completion of weaning.
Study II: A France prospective randomized double-blind, multicentre parallel controlled study of CNI weaning in selected long-term kidney transplant patients
The clinical group from Nantes conducted for the very first time a placebo controlled-double blinded withdrawal study in long-term kidney transplant patients.
Primary objective is the improvement of renal function (glomerular filtration rate >=7 ml/min). Secondary composite objectives are acute and chronic rejection, death, graft lost, anti HLA antibodies, de novo proteinuria. Ancillary studies are the assessment of low and high immunological risk biomarkers (DNA chip, Phenotypes).
Figure 2: WEANING trial.
The study is divided into two stages: the phase prior to the recommendation of the independent scientific committee and the phase after the recommendation. The committee recommended the enrolment of another 6 patients but with strict adherence to the original inclusion criteria. This was done after the BIO-DrIM annual meeting of January 2014. Following the concept of being “as safe as possible”, the inclusion and follow-up criteria were adjusted as follows:
At inclusion
• DSA <1000
• Anti-HLA <2000
• Age>50y
• 4 to 15y after transplantation
• Prograf and Cellcept or myfortic +/- CS
During the follow-up
• if DSA >2000, control 15 days after
• 15 days after:
• If DSA increase > 3000, biopsy
• If DSA do not increase, no biopsy and control 1 month after
• If proteinuria appears or function declines, biopsy
These new criteria have been submitted to the national ethical committee. They have all been accepted.
In total 16 patients were enrolled (9% of patients with pre-inclusion criteria). 10 were randomised, while 6 patients could not be randomized based on the screening results. 9 patients finished the protocol (4 without CNI weaning and 5 with CNI weaning).
All CNI weaning patients went back on full immunosuppression after having finished the weaning protocol. In two of the five weaning patients, weaning was stopped due to subclinical rejection at two thirds of the protocol.
The two patients have been treated (subclinical rejection).
All the other patients keep a good renal function, with no proteinuria and Abs disappeared.
In December 2014, the group decided to stop the study due to insufficient patient recruitment.
Since January 2014, 254 patients were screened, but with the restriction to rigorously stick on the inclusion criteria and with the change from the Anti-HLA Luminex screening to the more sensitive LABScreenT Single Antigen identification only one patient could be included in the study.
This patient was not randomized at day 30 since he did not tolerate the high doses of cellcept.
The biological studies are ongoing and the group has biological material available for the BIO-DrIM consortium.
It has become clear that tacrolimus withdrawal after years of treatment must be avoided even in highly selective long-term stable kidney recipients. It looks like the long-term tacrolimus treatment prevented any tolerance induction.
The results of the WEANING Study were published:
Dugast E, Soulillou J-P, Foucher Y, Papuchon E, Guerif P, Paul C, Riochet D, Chesneau M, Cesbron A, Renaudin K, Giral M & Brouard S. Failure of Calcineurin Inhibitor (Tacrolimus) Weaning Randomized Trial in Long-Term Stable Kidney Transplant Recipients. Am J Transplant 2016; 16: 3255–3261.
Biological material and data from this important trial are available for the consortium. Other demands for materiel cession have been done outside the BIO-DrIM network. The consortium is aware of these transfers. Transfer agreement are established by ITUN.
Study III: Using Biomarkers of Tolerance to guide immunosuppression weaning in kidney transplant recipients
Following the observation of a serious confounding effect of the immunosuppression regimen on gene selection and the performance of gene-expression signatures based on comparisons between already established untreated tolerant patients and treated stable patients, the team has finalised the development of an immunosuppression-free signature comprising 28 genes. The validation in a Fluidigm platform resulted in the selection (with elastic net) of a parsimonious signature of 9 genes, showing no material difference in performance to the best-performing signature comprising 19 of the original 28 genes. The results of our studies were published in December 2016 (PMID: 27328267). However, there is still an ongoing debate in the kidney transplant community on the safety of the clinical application of biomarkers of tolerance. Therefore, proceeding with the originally planned Study III design, as a safety and feasibility pilot Phase II trial for initiating IS minimization in patients who score ‘tolerant’, would have been unethical at this stage, without extensive further validation. Consequently, the team at KCL proceeded with further validation of the immunosuppression-free signature in RT-qPCR platform, which would be the one applicable to clinical use. The team has also embarked on a detailed exploration of the impact of adjustment for immunosuppressant drug intake on the selection of stable treated patients as potentially “tolerant”. The results of these studies are in preparation for publication.
WORK PACKAGE 2: Biomarker-guided stratification into low/high responder after kidney transplantation
The coordinating centres of the clinical study are Barcelona (O. Bestard / J. Grinyó) and Berlin (P. Reinke).
The Study CELLIMIN (Prospective donor-specific Cellular alloresponse assessment for Immunosuppression Minimization in de novo renal transplantation) is an international multicentre open label randomized non-inferiority Phase IV clinical trial for selection of low anti-donor T-cell responders using the IFN-γ ELISPOT assay as biomarker for patient stratification to safely receive long-term drug minimization based on tacrolimus monotherapy.
Since the start of the project, the design of the clinical trial was a main endeavour of this work-package. The CELLIMIN is a non-inferiority trial with enrichment design aiming to demonstrate the utility and IFN-γ ELISPOT marker for the stratification of kidney transplant recipients into “low” and “standard-of-care” IS regimen. The trial design has undergone major modifications from the original intention in response to serious ethical concerns.
The original intention was for a “biomarker by treatment” interaction design randomised controlled trial with a primary objective to demonstrate the clinical utility of IFN-γ ELISPOT assay pre-transplantation to stratify non d-s T-cell allo-reactive patients. Randomisation to maintenance IS with TAC monotherapy at low trough levels and “standard-of-care” high IS regimen separately in ELISPOT positive and negative patients were intended. The expectation was that ELISPOT negative patients at “low” IS would have 20% lower rate of BPAR at 12 months post-transplant (30% vs 10%) when compared to ELISPOT positive patients receiving low IS (Figure 3).
Figure 3: Original version of CELLIMIN clinical trial.
However, the risk associated with the administration of the “low” IS regimen in ELISPOT positive patients was considered unacceptable, which led to a modification of the trial strategy to a biomarker-based design randomized trial with one control group and the following treatment regimen allocations:
- GROUP A STANDARD OF CARE: the result of ELISPOT test will be kept blind and all patients will be treated following a “High” Immunosuppression regimen (Standard of care immunosuppressive regimen based on TAC (achieving 4-8ng/ml trough levels), MMF (1gr bid) and steroids (according to KDIGO guidelines).
- GROUP B BIOMARKER BASED STRATEGY: the investigator will know the result of the ELISPOT and patients will be treated as follows:
B.1: POSITIVE pre-TX ELISPOT or high responders (>25 spots/300.000 PBMC) will be treated following a “High” Immunosuppression regimen (Standard of care).
B.2: NEGATIVE pre-TX ELISPOT or low responders (<25 spots/300.000 PBMC) will be treated following a “Low” Immunosuppression regimen (based on TAC monotherapy to achieve 8-10 ng/ml trough levels during the first 4 weeks after transplantation and 6-8 ng/ml thereafter, MMF (1g bid) during the first 7 days post-transplant and stopped thereafter) and steroids (tapering until discontinuation on month 2 post- transplant).
All patients will homogenously receive 2 doses of Basiliximab (day 0 and day 4 after transplantation).
In order to test non-inferiority of arm B compared to arm A, assuming 15% BPAR in the latter, with a 15% increase non-inferiority limit, 194 patients would have been needed, randomized at a ratio 1:1, to achieve 90% power and 5% type I error rate (a total of 213 patients would have been enrolled, accounting for a drop-out rate of 10%).
Further concerns from the Voluntary harmonization procedure regarded the 15% limit for non-inferiority margin, which was considered not acceptable as the total expected the BPAR (biopsy-proven acute rejection) rate in the “low” IS group would have been too high. In response, the non-inferiority margin was adjusted to 10%, while having to considerably increase the number of patients to enrol. This would have created limitations of the marker based design, introducing a bias towards non-inferiority due to the large proportion of patients receiving the “standard-of-care” treatment (including the high-responders in the marker-based group B and the entire Control group A), creating a ZERO contrast in a great portion of the trial. Another drawback would have been the limited benefit for the patients of enrolling marker positive patients into the study.
In extensive collaboration with the statisticians from London and with the powerful support of the KKS (CRO), the study design was reviewed and modified for re-submission with a revised hypothesis:
Using the ELISPOT assay to assess whether negative donor-specific T-cell alloreactivity as a biomarker based immunosuppression (IS) will be non-inferior with respect to BPAR (T-cell mediated) rate at 6 months post-transplantation, comparing kidney transplant recipients randomized 1:1 to receive either low (TAC monotherapy) or high IS (Standard of Care triple therapy).
The re-submission of the study in November 2014 was successful and the trial received VHP (201446) approval by the European Commission in February 2015.
Figure 4: New hypothesis of the CELLIMIN Trial
The final strategy is an enrichment design where only ELISPOT negative patients will be randomized in a ration 1:1 to “low” and “high” (standard-of-care) regimens.
The current primary objective is to demonstrate the utility and safety of the INF-γ ELISPOT assay for the stratification of kidney transplant recipients into “low” and “high” IS regimens. This (enrichment) study will test non-inferiority of “low” IS regimen compared to “high” IS regimen, assuming 10% of BPAR (TCMR) at 6 months in the control group, and allowing a non-inferiority limit of maximum 10%.
The secondary objectives will be to investigate differences across treatment arms at 6 and 12 months post-transplantation in:
– eGFR
– Prevalence of biomarkers of tolerance/hyporesponsiveness
– Prevalence, type, severity, treatment and outcome of BPAR
– Prevalence, type, severity, treatment and outcome of 3-mo subclinical rejection
– Prevalence of death and graft loss
– Prevalence of metabolic and cardiovascular co-morbidity
– Prevalence of patients that remain MMF and steroid free
– Prevalence of acute and chronic hystologic lesions (Bannf’11 score) in protocol biopsies at 3 and 12 months post RT
– Prevalence of patients that remain on therapy
– Distribution of patients in distinct chronic kidney disease stages
– Health economics, H-R QoL and treatment cost (cost/benefit) at 1, 3, 6, 12 and 24 months
Subsequently, applications for obtaining approval of the trial by the local Ethics committees and national competent authorities were undertaken in 7 European Kidney transplant centres: Barcelona, Regensburg, Berlin, Prague, Nantes, London and Amsterdam.
After this initial approval, a first amendment was submitted in order to clarify some aspects of the trial and correct typographic errors. This first amendment was approved in Spain on March 2016.
A second amendment was submitted to add new information about MMF toxicity. Approval was obtained in Spain on 19/09/2016.
In December 2015, the first patient was recruited in the trial.
During 2017, two additional kidney transplant centres joined the consortium to participate as recruiting centres. During the first 6 months a thorough “on-side” validation of the standard operating procedures (SOP) of the IFN-γ ELISPOT assay was undertaken.
PATIENT ENROLLMENT:
The recruitment of patients begun on January 2015 and was notified to our competent authority.
Up to now, 184 patients have been screened with a pre-transplant donor-specific IFN-γ ELISpot done, and 102 (55.43%) were negative and thus were randomized. This observation confirms previous retrospective data regarding the prevalence of T-cell alloreactivity as means of preformed circulating donor-reactive memory/effector T cells
PROBLEMS FACED:
The unexpected delay on having the last trial finalized due to the previously mentioned reasons, was a main hurdle that prolonged the initiation of the study. Additional important problems faced that slowed the recruitment of patients in all centres were the significant increase of donors after cardiac death, the high percentage of extended criteria donors, the impossibility to perform the study tests during weekends and the increasing number of sensitized kidney transplant candidates in the waiting lists. These exclusion criterias for patient recruitment significantly impacted on a more dynamic inclusion of patients in the CELLIMIN trial.
WORK PACKAGE 3: Shifting kidney transplant patients to low-responders suitable for early IS minimisation
RIMINI is an International multicenter open-label single-arm Simon’s two-stage Phase II clinical trial aiming to estimate confidence interval for the observed efficacy of the induction regimen with rATG and infliximab and a go/no go rule for further clinical development. A total of 75 patients will receive the proposed induction regimen, accounting for a drop-out rate of 10%. RIMINI was approved by the relevant authorities via VHP process.
Using tolerance/rejection biomarker monitoring we aim to prove shift from high responders to low responders while using infliximab as co-induction agent to target recently activated memory/effector T cells.
Figure 5: RIMINI trial design.
The coordinating centres of the clinical study are Praque (O. Viklicky) and Berlin (P. Reinke). Clinical partners enrolling patients: CHARITE (Berlin), ITUN (Nantes), ICS-Hospital (Barcelona), IKEM (Prague), AMC (Amsterdam), UHREG (Regensburg), UKE (Hamburg)
As presence of donor-specific memory/effector T cells represents a major challenge for -minimizing immunosuppression, the aim of the study is to selectively target alloactivated effector/memory T cells.
Rationale: T-cell depletion is not sufficient to control allo-memory. ATG/Thymoglobuline treatment leads to a reduction of the clonal size of donor-reactive memory/effector T cells.
Antigens, present in vivo, induce a bias of lymphopenia-induced proliferation (LIP, preferential expansion of surviving alloreactive, Ag-specific Tem/Teff), which is an undesired effect, not useful in presensitized patients.
Therefore, targeting of recently activated Teff cells by anti-TNF mAb (Infliximab /Adalimumab) would be desirable. Infliximab/Adalimumab blocks soluble TNF and targets also mTNF on the cell.
Based on these rationales, RIMINI is designed to be a phase II, single-arm trial with 12 months of follow up.
The new induction protocol consists of: Thymoglobuline 1.5 mg/kg (Day 0), Thymoglobuline 1.5 mg/kg (Day 1), Remicade (Infliximab) 5 mg/kg (Day 2). The maintenance immunosuppression is based on Tacrolimus, administered at dose 0.1 mg/kg before surgery and then at dose 0.2 mg/kg (initiated day 1) and prednisone 20 mg (or appropriate dose of metylprednisolone) tapered down to dose 5 mg by day 7.
Figure 6: RIMINI medication scheme.
The following biomarker analyses are implemented in the trial:
- ELISpot/CTLp
- EBV/CMV/BKV load + CMV/EBV T-Ly
- Multi-parameter flow cytometry
- gene expression profiling
- alloantibodies
- urinary IP-10
- HO-1 polymorphisms
- histology (protocol/induced biopsies)
Primary endpoint:
1) Clinical response to the induction determined by the absence of any of the following outcomes up to 12 months post transplantation (start of follow up at transplantation): acute rejection, graft loss or poor graft function defined as eGFR<40 ml/min.
Secondary endpoints:
1) Prevalence of biomarker signatures at 6 and 12 months
2) Incidence of death by 12 months post-transplantation
3) Incidence of graft loss by 12 months post-transplantation
4) Incidence of metabolic and cardiovascular co-morbidity by 12 months post-transplantation (post-transplant diabetes mellitus, dyslipidemia, hypertension)
5) Proportion of subjects who remain on monotherapy at 12 months post-transplantation in the respective study groups
6) Incidence of acute and chronic lesions assessed by the Banff 07 score in protocol biopsy at 12months post-transplantation
7) Incidence of discontinuation of study treatment
8) Overall safety of study therapy immunosuppressive regimen defined as viral and bacterial infections, malignancies and autoimmunity
9) eGFR using CKD-EPI formula at 12 months
10) Health-related quality of life using EQ5D-5L and SF-36v2 questionnaires at baseline, M1, M3, M6, and M12
11) Assessment of patient-specific resource consumption using a trial specific questionnaire at initial discharge, M3, M6, M12, and in cases of repeated hospitalization
The biomarker analyses (e.g. whole blood staining and acquisition for flow cytometry) and sample collections / preparations (e.g. PBMC preparation for ELISpot) is performed in accordance to the same SOPs as filed for CELLIMIN, in order to ensure reproducibility and also comparability of the results between the two clinical trials.
The initiation meeting was held in Berlin, on November 15th 2016. Enrolment into the RIMINI trial started in January 2017. At the end of 2018 56 kidney transplant recipients were enrolled into the RIMINI trial, 33 from Prague, 19 from Berlin and 4 from Barcelona. Thus, the recent enrolment rate is 75%. Not all intended clinical partners are currently enrolling patients for the RIMINI trial, which proved to be the main obstacle to reaching the expected numbers of patient enrolment. The principal reason is the enrolment into concurrent CELLIMIN trial. The study enrolment is still open and 3 participating centres enrol patients and will enrol patients even beyond the end of the BIO-DrIM consortium. This is critical as only 25% patients (n=20) need to be enrolled into the trial to reach target population. Full set 12-months data evaluation is thus delayed and expected on June 2020 as we anticipate last patient to be enrolled in the trial in June 2019. Recently 20 patients completed the study with 12 months follow-up. 7 patients prematurely terminated the study. Recent number of efficacy failures (defined primary endpoint of the study) is 20 which anticipates acceptable efficacy and safety of the protocol and suggests non inferiority to the standard of care regimen.
The final analysis will be performed with per-protocol data set. One-sided upper 95% confidence interval will be calculated for the observed efficacy failure rate. Descriptive statistics will be presented for the individual outcomes comprising the primary outcome and for the secondary outcomes. Secondary binary outcomes will be presented as proportions with two sided 95% confidence intervals. Continuous secondary outcomes following normal distribution will be summarized by mean and standard deviation. Continuous secondary outcomes deviating from the normal distribution will be summarized with a median and interquartile range. Minimum and maximum values will be presented for all continuous outcomes. Survival outcomes will be examined in Kaplan-Meier plots.
WP4 Biomarker analyses, Data management & Biostatistics, Health-economic analyses and IVD development
Immune Monitoring Subproject
Immune Monitoring subproject of the BIO-DrIM project, the CHARITE functions as Central Immune Monitoring Lab (CIML) with the following persons involved:
Head CIML: Birgit Sawitzki
Sample material acquisition: Katrin Vogt
ELISpot: Maik Stein (under supervision of Petra Reinke)
Flow cytometry: Mathias Streitz, Stephan Schlickeiser
The main tasks with regard to WP1 (Studies I, II and III) are activities such as the exchange of standard operating procedures (SOPs), the exchange of samples as well as centralized analyses of some assays e.g. Foxp3 demethylation (TSDR). Exchange of SOPs for the analysis of the following biomarker was performed:
1) whole blood flow cytometry of leukocyte subsets (n=6 flow panel)
2) whole blood intracellular Foxp3 staining by flow cytometry
3) IP-10 analysis in urine
4) whole qRT-PCR
In addition, Foxp3 demethylation (TSDR) was quantified in samples from operationally tolerant kidney transplant patients in collaboration with partners from Nantes (Study II; Braza et. al. JASN 2015).
The main task within WP2 and 3 is the coordination and performance of the immune monitoring.
For this purpose, it has been decided on the establishment of the following biomarker/assay portfolio:
Safety marker: viral load (EBV, CMV, BKV) – Research Diagnostics
“Decison making” marker: anti-donor ELISpot (IVD/Companion Diganostics - GenID)
“Efficacy” marker: Flow Cytometry (IVD - Immunotech)
qRT-PCR (IVD – no supporting company yet)
Foxp3 demethylation (TSDR) – Research Diagnostics
IP-10 (IVD – Milenia)
Adjustment and validation of ELISpot and flow cytometry received particular attention. This was done in close collaboration between the Charite as the Central Immune Monitoring Lab (CIML), GenID and Immunotech, respectively (see also detailed description of design, validation process and results further below within the respective paragraphs). The extensive validation process allowed us to extend the age-of-blood time to 12 hours for both assays, which increased the clinical applicability.
The then generated SOPs for standardized assay performance (ELISpot & whole blood flow cytometry) had been sent to the clinical sites and published within the eCRF. Inter-operator and inter-lab comparisons were performed to validate the robustness and correctness of the assays.
Inter-lab and especially inter-operator comparisons for the ELISpot (virus- and allo-specific) resulted in acceptable and low CVs ensuring robustness and applicability as a stratifying biomarker. Because of the test´s impact on the outcome of the clinical trial of WP2, inter-operator and inter-lab comparisons were and will be repeated on a regular basis.
In summary, we were able to implement two well-validated assays at all clinical sites. The technicians and scientists are intensively trained and now ready to perform the assay in strict adherence to the developed and refined standard operation procedure (SOP) for its use as an immune monitoring tool and – more importantly – as a prediction tool in the clinical trial of WP2.
The standardization of the flow cytometry has been a process developed on basis of the experiences from another FP7 project - the project “The ONE Study”. The aim was to obtain accurate and precise data with the least variation among different sites.
The standardization process included the following actions:
Consensual development of flow panels, development of SOPs, panel validation including extensive tests on the influence of “age of blood” and “age of stain”, establishment of reference standards and data analysis (for detailed information see paragraph further below).
Upon refinement of sample processing and staining procedures tests on the influence of “age of blood” revealed a low variability until 12h storage (see figure below), which was now set as a maximal age limit for the clinical trials performed within BIO-DrIM.
Figure 1: Absolute numbers of e.g. CD56+ NK cells, CD8+CD27- cytotoxic effector T cells and CD4+CD25highFoxp3+ regulatory T cells using the DuraClone-based antibody panels in 8 whole blood samples upon 0, 4h, 8h, 12h, 16h and 24h was determined.
Further advice on how to perform and analyse multi-parametric flow cytometry has been published in a book chapter (Schlickeiser S, Streitz M, Sawitzki B. Standardized Multi-Color Flow Cytometry and Computational Biomarker Discovery. Methods Mol Biol. 2016;1371:225-38. doi: 10.1007/978-1-4939-3139-2_15. PubMed PMID: 26530805).
We have also generated and published reference cohort data for healthy controls, which can be used to correct for age- and gender-dependent differences in immune cell subset composition (Kverneland AH, Streitz M, Geissler E, Hutchinson J, Vogt K, Boës D, Niemann N, Pedersen AE, Schlickeiser S, Sawitzki B. Age and gender leucocytes variances and references values generated using the standardized ONE-Study protocol. Cytometry A. 2016 Jun;89(6):543-64. doi: 10.1002/cyto.a.22855.).
In addition to the two tests described above, the consortium has analysed several other parameters either in central core laboratories or in on-site laboratories in order to explore the values of further biomarkers for stratification, safety and monitoring.
For this purpose, additional clinical samples have been collected in the clinical trials of WP2 and WP3. To ensure quality in sample collection, the following procedure was applied:
In agreement with WP leaders WP2- and WP3-specific sample collection schedules were designed (see also WP2 and WP3). All patient material for immune monitoring purpose needs to be labelled accurately. The patient IDs as designed by the CIML team specific for both clinical trials, CELLIMIN and RIMINI, were automatically generated by the eCRF and are always part of the label code.
Almost all samples were collected in pre-labelled sample collection tubes provided by the CIML team in pre-packed sampling bags. The bags are patient-specific and time point/event-specific. As some of the tubes have a short half-life, the bags were provided every 6 months according to the estimated patient enrolment.
ELISPOT analysis
As mentioned above, development and evaluation of ELISpot assays was done in close collaboration between the Charite CIML team and GenID.
The GenID main role within BioDrIM is to develop and evaluate with clinic partners IVD kits for different applications and distribution of these kits worldwide. In detail, GenID supplies established enzymatic EliSpot kits for the detection of IFN gamma (IFN-γ) release after allo- and antigen-specific stimulation. In general, the objective of GenID contribution in the project is the set-up, standardization and validation of EliSpot IVD kit and image analyzer for interpretation in routine use lead to personalized IS. The final aim is the approval of this system in Europe and worldwide (US).
In the following sections, the work and results of GenID within the consortium are described.
WP 2: Prevention of the high-dose standard CNI-based IS in low-responder kidney transplant recipients identified by perioperative patient stratification
This work package focuses on the clinical study on prevention of high-dose standard IS in low T-cell responder kidney transplant recipients by perioperative stratification of patient. GenID offers in this work package kits and antigens.
The kits provided for the CELLIMIN study and the corresponding antigens are shown in the following table:
Table 1: Number of provided EliSpot kits and antigens from 2014-2017
WP 3: Increasing the population of low-responder kidney transplant patients suitable for early minimization (low-dose monotherapy) by the recently explored selective perioperative targeting of alloreactive effector / memory T-cells
Up to now no contribution from GenID`s side.
IVD test development
For the validation, the standardization and implementation of the IFN-γ-EliSpot, it was necessary to monitor the transplant patients according to their pre-transplant frequency of donor-reactive memory / effector T-cells into low / high responder kidney transplant recipients.
For the preoperative identification of low-responder patients, the test had to be performed on-site.
On the basis of the high and stringent requirements on reproducibility and accuracy of the EliSpot technology inside the consortium the main tasks were:
• Designing standard procedure protocols
• Personal (wet) training
• Reader System evaluation
• Implementation of frequent controls to monitor the process at each side
After the workshop of Berlin, held in December 2013, following points had to be cleared:
- 2 generations of AID EliSpot readers in the centers
o Camera sensitivity + resolution
o Light intensity decrease can occur
o Yearly master plate check for light intensity comparison
- Technical questions
o Positive control switched from SEB to pokeweed mitogen
o For functional test only citrate blood
o Blood sample age maximum 8h
o No third party
o One pretested FCS batch
o One count setting for all EliSpot Reader Systems
- Assay performance for pre- and post-transplantation
- Crossvalidation
o SOP update
o Repeat training sessions together with GenID
o Defined counting setting and check with a standard plate for all centers
o Verifying the SOPs and count settings with same samples (n = 5)
▪ Cross check of all raw data at Berlin / Barcelona
Further approaches by GenID were determined and conducted:
1. Set up / calibration
o Definition of a QC-plate (e.g. Masterlot) in Barcelona / Berlin
o QC-plate has to be analyzed in each center
o Adjustment of each system via remote support
2. Definition of settings for counting
o Based on the results of the RISET study and planned cross validation, standard settings for counting had to be fixed
3. Definition of the QC frequency
o By sending sample around like external quality control
o By sending QC-Plate or other plate around
Transversal activities of Charité and GenID on the standardized EliSpot technique
Within the CELLIMIN trial of WP2, the IFN-γ EliSpot assay was used as an enrichment tool in order to select only low-responding, EliSpot negative patients for the study. Thus, proving the robustness of the assay is essential. This required close collaboration between the CHARITE and GenID. In a very meticulous work together with the Charite, all crucial working steps (see Figure 1) were evaluated with regard to the impact of the assay results.
Figure 1: Crucial work steps with possible impact on the test result
The same attention was given to the evaluation of factors with impact on the practical use of the assay.
One example: Since transplantation often takes place at night or during the weekend, it has been important to understand how much time could pass between time of transplantation and start of the assay. For this purpose, it was tested whether the outcome of the assay (spot numbers) changes with the age of the blood. The test results showed that fresh blood can be processed within 12 h (maximum) without changing the outcome of assay. Consequently, it was determined that start of the assay need to occur within 12 h after time of transplantation.
But not only the laboratory assay had to be optimized, the EliSpot reader systems had to be harmonized and standardized as well. The EliSpot reader system consists of an optical unit and specific software to analyze the EliSpot plates and count spots. Although all reader systems are identical in construction, it is not possible to exclude the existence of minor variations that could be the source of differences in the outcome. In order to eliminate such inter-EliSpot reader hardware / software variability, the same plates were analyzed by all centers. Via online remote support every reader system was adjusted in conformity to the results generated by the reader system in Berlin (= reference) and the new settings were saved and locked. A quality control (QC) procedure has been implemented to ensure compliance to these standards.
The EliSpot Reader systems were successfully installed including training at each side (9). To ensure same settings and prevent changing of fixed configurations, different user accounts were created. The Reader Systems have access to the internet for remote support and for upload of data to Berlin and (for data safety reasons) to an individual server at GenID facility by encoded export. There are three levels of users with specific rights:
Figure 2: Screenshots for the different user accounts with individual permission rights
User “AID” is the only supervisor with user management rights and is password-protected. The Password is securely stored at GenID to ensure only the software experts at GenID are able to change other user accounts if necessary.
User “BIO-DrIM” only has the rights to modify results. In software version 7.0b15714 GenID introduced a new button ("review") to ensure count results are confirmed. This feature also locks the plate for further modifications unless plate is "un-reviewed" again; all actions must be performed with user name and are password protected (electronic signature) and saved in the plate history.
All other users are “advanced users” which are allowed to modify settings only in their own profile.
For the BIO-DrIM user account, as first step a layout with specific count settings named BIO-DrIM has been created. Every new document that user BIO-DrIM opens has those settings which are applied for analysis.
Second, stage calibration has been performed and saved as “stage.ini” which is loaded automatically after user log-in. Camera settings were adjusted by remote support from GenID with the same “reference” EliSpot plates for each center. The aim was to calibrate similar well saturations (<3% deviation) leading to reproducible count results; those system-specific camera parameters were saved as AID defaults and are not changeable by the BIO-DriM user account.
Third, to implement a reader / software internal QC System, AID defaults were loaded in a newly established user account named “masterlot” and locked. This masterlot (QC) plate has been read and saved as reference for further QC runs.
Figure 3: Screenshots from the Masterlot user providing the QC system check
All following QC plate readings are compared to the reference plate values and deviations are displayed when performing the system QC test. In case the deviation is greater than 5% for light conditions (and therefore well saturations), the AID remote support should change AID default camera settings to adjust the system to reach same average well saturation as for the QC plate. Then these settings have to be applied to user account BIO-DrIM.
Recognition of the IFN-γ-EliSpot assay as a companion diagnostic tool
Furthermore, several steps have been taken towards implementation of the EAB´s recommendation to open the discussion with the FDA and EU authorities for the recognition of the IFN-γ-EliSpot assay as a companion diagnostic tool.
It was decided to go for two different approvals:
1. License to use the allo-reactive T-cell EliSpot as patient stratification tool in kidney and liver transplantation
2. License to use the EliSpot as monitoring tool (detection of viral infection - CMV, EBV, BKV).
For this purpose, it was decided to benefit from the close collaboration with the industry and the expertise of the members of the EAB and local users (biomarker core facility at the CHARITE / BCRT). All experts have been invited to a meeting, where the technique, its development process and data was presented with the aim to ask for scientific advice and to decide on how to classify the technique and how to present it to the authorities.
AID / GenID has been asked to start promoting the EliSpot as a standardized and marketable test, while defining the intended use and proving the medical need of the assay.
The Documents and forms for FDA submission were prepared for following scenarios:
- For the registration of the test kit “EliSpot for the Detection of Alloreactive Response” or “EliSpot for the Detection of Response of Viral pathogens”, it may be possible to claim a 510(k) regarding the test method used.
- It may be beneficial first to claim a 510(k) for the identical test kit with FDA by AID. Then after approval to amend the 510(k) registration for other antigens and immune response reactions
- The next steps of AID were to determine which pathway to follow and which argumentation is the best for a fast registration.
- In case the FDA does not agree to the 510(k) path, then the PMA (Pre Market Approval) path must be used.
Furthermore, the documents for HDE- pre-submit were prepared:
- Elements of intended use and indication of use,
Clear statement how and by whom the device is to be used
Description for what and for whom the device is to be used
- Define how the device is planned to be used in real-life setting
- Risk Analysis
- Proposed Study Designs
- Specimen information
- Analytical performance
- Method comparison
- Clinical performance
- Statistical analysis plan
For realization of the pre-submit, the following data were collected at GenID / AID:
Assay Performance Characteristics
The minimal detectable unit of response of an EliSpot Assay is one Spot. The limit of detection typically achieved can be up to 1 in 200.000 cells.
Because of natural variations of cell based assays, each control and analyte has to be performed in duplicates or better triplicates (like in this evaluation).
To avoid unspecific assay results the individual immunological constitution of each patient has to be taken into account. Because of that, negative and positive controls have to be performed for every single patient.
The positive control establishes that the assay worked by providing an expected positive result and is used for the documentation of an adequate number of inducible cells. In case of 200.000 cells / well, a minimum of 50 Spots should be achieved. To avoid overdeveloped wells, an additional positive control was performed with 40,000 cells / well.
The negative control is used to establish the unspecific baseline response of the assay. In case of 200.000 cells / well, a maximum of 10 Spots should be achieved. However, routine experience has shown that some patients show slightly higher spot counts dependent on their general individual immunological constitution, disease, medication or even nutrition.
During evaluation of assay results, spontaneous background cytokine secretion has to be taken into account to prevent false positive results. To determine the spontaneous background cytokine secretion, cells are incubated in the presence of cell culture medium without any further stimulating agents like serum or antibiotics. For the background compensation of assay results, the ratio of “analyte triplicate means” to “mean of negative control triplicates” has to be calculated (stimulation index [SI]).
These mean background compensated assay results were grouped into four categories:
• negative 0 – 2 SI
• low response 2 – 5 SI
• medium response 5 – 20 SI
• high response >20 SI
For very low spot counts of the negative control and non-reactive wells, the calculation of the CV% is not very meaningful due to nearly equal values of the mean and standard deviation, which leads to irrational high CV´s. In these cases the observation of the standard deviation is more relevant.
INTRA-Assay
INTRA-Assay precision was analysed from data generated by two operators. To determine INTRA-Assay Variation, each operator performed a total of eleven assays consisting of three replicate wells (triplicates) for all controls and antigens. All cells used for this evaluation were isolated at once from normal human whole blood of one donor and split directly before performing the assay.
Assay performance data was calculated with background compensated assay results of each operator separately. At first, the mean stimulation index (SI) of each of the eleven assays was calculated. After that, corrected sample standard deviation of background compensated assay results was calculated. At last the Coefficient of variation (CV%) was calculated as the ratio of corrected sample standard deviation to mean of background compensated assay results.
The following tables provide an overview of mean response as well as CV% and SD of eleven assays performed in a single run performed by two operators. These results indicate that INTRA-Assay precision is acceptably low and allows the analysis of clinical samples in duplicates.
Operator MS
Antigen Response CV [%] SD
NC negative (xˉ = 1,00)
0 0
PWM high response (xˉ = 26,81)
10,17 2,73
PWM 1:5 high response (xˉ = 29,12)
9,88 2,88
CMV IE1 negative (xˉ = 0,87)
18,23 0,16
CMV pp65 high response (xˉ = 33,81)
10,41 3,52
EBV lyt medium response (xˉ = 11,41)
12,07 1,38
EBV lat medium response (xˉ = 6,22)
11,80 0,73
Operator KS
Antigen Response CV [%] SD
NC negative (xˉ = 1,00)
0 0
PWM high response (xˉ = 29,70)
10,37 3,08
PWM 1:5 high response (xˉ = 32,36)
12,03 3,89
CMV IE1 negative (xˉ = 0,97)
25,60 0,25
CMV pp65 high response (xˉ = 35,23)
12,17 4,29
EBV lyt medium response (xˉ = 12,43)
10,60 1,32
EBV lat medium response (xˉ = 6,41)
15,00 0,96
Table 2: Results for INTRA assay variation for two individual operators (MS and KS). Antigens for specific T-cell stimulation were CMV (Cytomegalovirus) immediate early 1(IE1) and pp65 (tegument protein) and EBV (Epstein-Barr virus) lytic genes (lyt) and latent genes (lat).
INTER-Assay
INTER-Assay precision was analysed from data generated by five different operators using the same four samples, including three buffy coats (501-1 - 501-3) and fresh blood (NE), at three days. Each day a different plate lot was used. All cells used at one test day were thawed and overnight rested or isolated from normal human whole blood at once. All samples were split directly bevor performing the assay. Cells were counted by each operator separately. Assays were performed as three replicate wells (triplicates) for all controls and antigens.
Assay performance data was calculated with background compensated assay results between three test days of each operator separately. At first the mean stimulation index (SI) of each day was calculated. After that corrected sample standard deviation of background compensated assay results between three test days was calculated. At last the Coefficient of variation (CV%) was calculated as the Ratio of corrected sample standard deviation to mean of background compensated assay results.
The following table provides an overview of mean response as well as range of CV% and SD between three different kit lots used at three days obtained by five Operators for two samples (501-2 and NE as example:
Specimen 2 (501-2)
Antigen Response CV [%] SD
NC negative (xˉ = 0,56)
43,30 – 114,56 0,19 – 0,51
PWM high response (xˉ = 409,64)
10,32 – 31,73 47,23 – 110,51
PWM 1:5 high response (xˉ = 83,24)
20,01 – 59,21 16,25 – 42,02
CMV IE1 negative (xˉ = 0,63)
34,64 – 114,56 0,19 – 0,92
CMV pp65 negative (xˉ = 0,57)
20,40 – 91,65 0,08 – 0,51
EBV lyt negative (xˉ = 0,61)
35,25 – 91,65 0,25 – 0,51
EBV lat negative (xˉ = 0,37)
43,30 – 173,21 0,19 – 0,53
Specimen NE
Antigen Response CV [%] SD
NC negative (xˉ = 0,69)
0,00 – 100,00 0,00 – 0,38
PWM high response (xˉ = 276,04)
6,90 – 57,82 28,80 – 143,93
PWM 1:5 high response (xˉ = 206,23)
28,84 – 63,05 54,39 – 97,91
CMV IE1 negative (xˉ = 1,84)
0,00 – 141,95 0,00 – 1,58
CMV pp65 negative (xˉ = 0,93)
30,10 – 86,60 0,29 – 0,96
EBV lyt medium response (xˉ = 7,03)
4,58 – 67,04 0,51 – 3,24
EBV lat high response (xˉ = 20,28)
24,12 – 60,97 4,48 – 9,41
Table 3: Examples for INTER-assay performance from two donors out of five different cell preparations
One possible reason of higher CVs for INTER-Assay precision is the use of different PBMC cryovials and blood drawings on each day. Even though the PBMCs were from the same lot and the fresh blood from the same patient, sample-to-sample variability could explain the findings.
INTER-Operator
INTER-Operator precision was analysed from data generated by five different operators using the same four samples, including three buffy coats (501-1 – 501-3) and fresh blood (NE), at three individual days. Each day a different kit lot was used. To determine INTER-Operator Variation the same dataset used for the calculation of INTER-Assay precision was examined. Cells were counted by each operator separately. Assays were performed as three replicate wells (triplicates) for all controls and antigens.
Assay performance data was calculated with background compensated assay results between fife operators for each test day separately. At first the mean stimulation index (SI) of each operator was calculated. After that corrected sample standard deviation of background compensated assay results between the operators was calculated. At last the Coefficient of variation (CV%) was calculated as the Ratio of corrected sample standard deviation to mean of background compensated assay results.
The range that was obtained for high response (SI > 20) was between 5,75% - 69,15% CV (mean % CV = 39,12), medium response (SI 5 - 20) gave a range of 29,12% - 96,14% CV (mean % CV = 66,33), low response (SI 2 - 5) wasn´t observed and negative results (SI 0 - 2) gave a range of 0,00% - 166,83% CV (mean % CV = 62,05).
The following table provides an overview of mean response as well as range of CV and SD between five operators at three days for two samples (501-2 and NE) as example:
Specimen 2 (501-2)
Antigen Response CV [%] SD
NC negative (xˉ = 0,56)
55,90 – 94,79 0,30 – 0,51
PWM high response (xˉ = 409,64)
16,10 – 24,93 58,66 – 102,49
PWM 1:5 high response (xˉ = 83,24)
14,55 – 44,95 15,42 – 29,47
CMV IE1 negative (xˉ = 0,63)
56,45 – 124,72 0,33 – 0,62
CMV pp65 negative (xˉ = 0,57)
42,59 – 78,20 0,30 – 0,38
EBV lyt negative (xˉ = 0,61)
47,51 – 80,03 0,28 – 0,48
EBV lat negative (xˉ = 0,37)
56,84 – 102,26 0,16 – 0,53
Specimen NE
Antigen Response CV [%] SD
NC negative (xˉ = 0,69)
37,27 – 70,71 0,28 – 0,47
PWM high response (xˉ = 276,04)
14,29 – 65,60 50,86 – 140,51
PWM 1:5 high response (xˉ = 206,23)
13,51 – 46,00 25,36 – 68,15
CMV IE1 negative (xˉ = 1,84)
46,90 – 166,83 0,78 – 3,78
CMV pp65 negative (xˉ = 0,93)
40,75 – 85,67 0,37 – 0,84
EBV lyt medium response (xˉ = 7,03)
29,12 – 72,74 2,19 – 5,21
EBV lat high response (xˉ = 20,28)
5,75 – 47,51 0,95 – 7,78
Table 4: Example for INTER-operator performance for two cell preparations out of five used for the evaluation
Concordance analysis between two EliSpot Reader System
Comparison of Spot counts between two EliSpot reader systems was analysed from data generated by all experiments during assay validation. All Plates were read and counted by two EliSpot reader systems used in the CELLIMIN trial.
The scatter plot is used to display differences in spot counts of the two BioDrIM EliSpot Reader.
Each data point having the value of counts from ELRO7IFL1602236 determining the position on the horizontal axis and the value of counts from ELRIFL061001095 determining the position on the vertical axis.
An identity line of equality (y=x) is added to show 100% conformity. Linear regression of all 1984 pairs of values leads to a linear regression line with the equation y=1.026*x+0,06383 which slightly differs from the line of equality at lower counts.
Additionally Pearson´s correlation coefficient has been calculated. (r=0.994 R2=0.9988 P(two-tailed)=<0.0001)The Bland-Altman plot shows the difference of spot counts between the two measurements of the used EliSpot Reader on the Y axis, and the average of the two measurements on the X axis. Additionally the bias (red dotted line, -3.462) which is the average of the differences, and 95% limits of agreement (black dotted line, from -17.83 to 10.91) are shown.
At low spot counts the discrepancy between the two Reader is low and equally scattered. At high spot counts the scatter around the bias line gets larger and shows the tendency of slightly higher counts for ELRIFL061001095.
Establishing an external QC kit
The last step was the development of an external Quality control kit to monitor the assay procedure as it has been proposed in the requirements:
• GenID offers frequently (each quarter) frozen pretested PBMC´s and frozen antigens to run an EliSpot assay according to the BioDrIM protocol.
• The results of the external QC were recorded in an excel sheet.
• The summary of the results were send to the coordinator in Berlin.
As a first step, for establishing a SOP for thawing of PBMCs, recovery rates were determined to monitor protocol performance
Figure 4: Recovery rates for different donors (502-8, 503-2, 503-4, 503-5, 503-6 and 502-10) and different vials from the same sample. In the right graph additionally Transport of vials overnight on dry ice (TE) is shown regarding recovery rates.
Recovery rates between differences vials from one sample were less than 15%. Although, two PBMC preparations showed significantly lower recovery rates. Transport overnight on dry ice did not influence recovery rates or viability of cells.
Next step was to determine the variation of spot numbers from different vials. We found no significant difference in Spot numbers with standard antigen preparations and overall less than20 % difference in spot numbers between vials from one PBMC lot. Based in this data, QC kit with the established SOPs has been designed and tested with frozen PBMCs as indicated in the beginning.
In two individual runs, all centers tested frozen and fresh samples in parallel and recorded their spot count results to the Charité:
Figure 5: Spot counts for the QC tests from all centers for frozen PBMCs (left) and fresh PBMCs (right).
In conclusion, the external QC kit performance has to be further improved and more runs have to be started for better results and concordance between the centers.
Flow cytometry
Development and evaluation of flow cytometry assays was done in close collaboration between the Charite CIML team and Beckman-coulter Immunotech.
As a manufacturer of reagents and instrument for flow cytometry Beckman Coulter has been previously involved in another FP7 project, i.e “The ONE Study”. Within the ONE Study projects, Beckman coulter has provided flow cytometer instruments, custom designed reagents (6 multicolor panels) and solution for standardization of immunophenotyping. The test analysis showed that the standardization of the flow cytometry was efficient with low variability between the different sites and no clustering by center was observed (Streitz et al. 2013). The collaboration was extended to the BIO-DrIM project where the role of Beckman-coulter Immunotech was to provide 7 panels for immune monitoring by flow cytometry, one flow cytometer per site and solutions for Standardization. Beckman Coulter is committed to provide reagents and support to the BIO-DrIM consortium throughout the study. Hence Beckman Coulter is engaged to manufacture and commercialize the Immune monitoring panels.
1. Method Standardization
With its capability to identify various immune population at the cellular level, Flow cytometry is a suitable technology to reveal immune cell phenotyping and function. However, it is an open technology which make it difficult to reproduce findings and to compare data generate across laboratories. Together with the assay complexity, the pre-analytical factors (1), the reagents (2), the staining protocol (2), the instruments settings (3-4) and the data analysis (5) constitute steps that generates variability (Figure 1).
Figure 1: Flow cytometry experimental workflow
In order to improve standardization, the following has been initially worked out in the “The ONE study”:
- (1) Use Whole blood samples avoiding variability related to sample pre-analytic preparation of the sample (i.e PBMC)
- (2) Cocktail reagents were provided in 3 vials to avoid interaction between tandem dyes conjugates
- (2) SOPs established by the CIML team and specific training was done at each site to ensure handling is done similarly at all sites
- (3) Instruments settings were defined at the central laboratory. Target values using Flow set Pro beads were recorded and transferred to the other site to ensure settings are similar across sites
- (4) Lot match single colors were provided to ensure compensation settings are accurate for the multi-color panel
- (5) All data analyzed in the central laboratory to ensure robustness
In order to further improve the standardization, several settings have been tested and implemented for use in the BIO-DrIM. The changes and addition to the ONE study standardization approach are underlined in the list below:
- (1) Use Whole blood samples avoiding variability related to sample pre-analytic preparation of the sample (i.e PBMC)
- (2) Use stabilized cocktail in the DuraClone Format. Optimized panel design (Clone and Dyes) to improve robustness.
- (2) SOPs established by the CIML team and specific training was done at each site to ensure handling is done similarly at all sites
- (3) Instruments settings according to newly generated standardized SOP. Target values using Flow set Pro beads were recorded and transferred to the other site to ensure settings are similar across sites
- (4) Standard SOP to generate compensation matrix using the Navios Auto Set-up Scheduler. Lot match single colors were provided to ensure compensation settings are accurate for the multi-color panel.
- (5) All data analyzed in the central laboratory to ensure robustness
2. Panel design
The design of the panel has been made in close collaboration with the CIML team. The strategy for the panel design has been to identify the improvement to implement to the ONE study panels by fixing issues and by the addition of new markers of interest. Moreover, the implementation of the Beckman Coulter’s proprietary dry DuraClone format has been done. Each change has been performed as an iterative approach, where Beckman-Coulter has been transferring the newly generated material, i.e liquid antibody conjugates, to the CIML for evaluation. When all antibodies conjugates performances were considered acceptable by the consortium, panel were tested in the liquid format prior the evaluation of the final version in the DuraClone format. A summary of the panel changes is presented in Figure 1. The reasons for changes are summarized in Table 1. When all parameters defined for the various panels, validation has been made at the Central immune monitoring laboratory. All the panels were provided to each site throughout the study (about 10000 tests overall)
Figure 2: Panel design
Panels ID Observations ONE study Panels Implementation in BIO-DrIM panels
#1 to #4 Very bright signal and variable spillover over time of the CD8-APC-A700 conjugate. Requires adjustment of the settings (spillover compensation) during analysis. Time consuming. CD8-AlexaFluor 700 used instead of the CD8-APC-A700. Detection of the CD8 expression remain in the same channel with a bright enough signal, No spillover variability (over-time and inter-lot) and less spillover in other channels
#1 CD64 does not provide additional information versus the CD14 CD64 is not used
#2 CD45RO is not informative in the context of TCR phenotyping.
HLA-DR to replace CD45RO.
Addition of TCRVd1 and TCRVd2 conjugates to have a better phenotyping of the TCRgd subsets.
#3 and #4 Two panels to identify with 5 redundant markers for identification of T cell memory phenotypes and T cell activation status. Additionally, #4 allows identification of the TReg population through CD127 and CD25 expression on CD4+ T Cells. Define 1 panel for identification of the various T cells subsets and their activation status and 1 panel specific for TReg panel allowing a better phentopying of TRegs using FoxP3 (gold standard for detection of TRegs) and additional markers for the identification of the various subsets (Helios, CD39 and CD45RA)
#4 An intracellular approach for TReg phenotyping evaluated in parallel to the ONE study provided good results and allow good discrimination of TReg populations. The detection of TRegs is done using an intracellular approach which allow the detection of two key intracellular markers expressed on TRegs, i.e FoxP3 and Helios
#3 and #4 Interaction between the CCR7 and CD25 antibodies via the complement in a donor dependent fashion. False double positive signal. Requirement of pre-analytic processing prior staining. A CCR7 clones which does not interact with the complement has been used.
#6 Redundancy between BDCA2 and CD123 expression on plasmacytoid dendritic cells Removal of the BDCA2 for identification of plasmacytoid dendritic cells
#6 Literature describes the Clec9A markers to be specific of the BDCA3+ mDC subsets. Moreover, Clec9A expression on myeloid dendritic cells is conserved between human and mouse which allow the translational comparison between animal and human studies.
Only two out of the three subsets of mDCs are identified with the Dendritic cell panel. Replacement of BDCA3 by Clec9A for the detection BDCA3+ mDC subset.
Addition of the CD1c Marker to the panel in order to detect the 3 subsets of mDCs, namely CD1c+ mDCs, Clec9A+mDCs and the CD16+ mDCs.
Table 1: Description of changes in the Immune Monitoring panels
3. DuraCLone Format
Beckman Coulter’s Coulter proprietary DuraClone format allows the stabilization of antibody cocktails in a dry format. DuraClone reagents provide unprecedented standardization and the key features of this stable format are as follow:
• Dry, ready to use, unitized reagent
• Reagent-containing layer coated to sample tube with no loose lyophilized “cakes”.
• No electrostatic charges interfering as with loose lyophilized cakes.
• Storage / logistics at ambient temperature
• Minimal pipetting steps during sample preparation, with preliminary data indicating improved reproducibility of monocyte populations.
• Optimal consistency of results from tube to tube and from batch to batch.
Based on the comparison of tests between dry and liquid reagent, it was found that the staining intensities were completely overlapping demonstrating that the DuraClone format is compatible with all dyes used in the context of the BIO-DrIM (Figure 3A and B). DuraClone sensitivity to temperature variation that could occur during transport and/or storage was assessed by storing a multicolor flow cytometry panel at 37°C and 60°C. The stressed reagents performed as well as their counterpart kept at room temperature (Figure 3C). Moreover, specimen preparation methods including several lyses and cell permeation buffers were investigated (Data not shown). Results showed that the DuraClone dry flow cytometry panels are compatible with routinely used lyse procedures and with intracellular applications.
Figure 3: DuraClone performance evaluation
We also compared the variability between previous ONE-Study liquid-based and DuraClone antibody panels. Interestingly and very assuring, usage of DuraClone-based antibody staining panels reduced the intra-assay CV especially of low-frequent and difficult to stain B cell subpopulations (see figure below).
Figure 4: Intra-Assay CV of liquid versus DuraClone antibody-based staining of B cell subpopulations.
4. Product development
Upon agreement on the design and SOPs, Beckman Coulter has initiated the development of conjugation methodology for the Alexa700 conjugates, the development of the missing single colors conjugates followed by the development and the commercialization of the Immune monitoring panel with a Research use only label.
a. AlexaFluor 700 methodology
The development of the AlexaFluor 700 conjugation methodology consisted on the evaluation and the feasibility of the optimal condition for the conjugation of the AlexaFluor700 dye to purified antibody. The development of the methodology has been done with the CD8-AlexaFluor 700 conjugates required for the BIO-DrIM panels and on two antibody specificities (i.e CD19 and CD20) to ensure the robustness of the process conjugation. As consequence, the three conjugates were commercialized as single color conjugation as indicated in the table below.
Part number Conjugates Launch date
B76283 CD8-AlexaFluor 700 Nov. 2015
B76287 CD19-AlexaFluor700 Nov. 2015
B76279 CD20-AlexaFluor700 Nov. 2015
b. Development of single color reagent
Among the single color conjugates required for the formulation of the panels, nine were not available at Beckman Coulter. Evaluation of the various antibody clones available for each specificity has been done. Ones procurement of the clones has been done, development of the single color conjugates has been initiated. In order to allows customer to use the panel in the liquid, single color conjugates were commercialized as finish good or render available through the custom design service upon request (See summary in table below)
Part Number Conjugates Launch date
B30632 CCR7-PE Nov 2013
B36123 PD-1 PC5.5 Nov. 2013
B46036 CD1c-PC5.5 July 2014
B49309 TCRvd2-PacBlue Sept 2014
B43291 Clec9a-APC Nov 2014
B49310 TCRvd1-PC7 May 2015
B49311 Helios-PacBlue May 2015
B55384 CD39-PC5.5 Aug 2015
CDS Granzyme B-ECD Available through custom design service
c. Development of the BIO-DrIM panels
Upon availability of the conjugates the panel development was performed in sequence. All panels were commercialized with Research Use Only label (See summary in table below).
Part number Panel Launch date
B53309 DuraClone IM Phenotyping BASIC tube Oct. 2014
B53318 DuraClone IM B cell Tube Oct. 2014
B53328 DuraClone IM T cell subsets Tube Dec. 2014
B53351 DuraClone IM Dendritic cell tube Dec. 2014
B53340 DuraClone IM TCR Tube May 2015
B53346 DuraClone IM TReg Tube Sept. 2015
C00162 DuraClone IM Count Dec 2016
The collaboration within the BIO-DrIM consortium has led to the commercialization of the 1st multicolor panels in the DuraClone format. Hence, a model of reagent development at Beckman Coulter has been built following the successful collaboration with the BIO-DrIM. Collaborators are directly involved in the development of reagents which led to the commercialization of nine additional DuraClone products following this approach.
5. Clinical trial data analysis
Finally, we also started analysing flow cytometry staining results generated within CELLIMIN and RIMINI. For CELLIMIN we aimed on revealing leukocyte subsets which in addition to the allo-specific ELISpot are able to predict or diagnose acute rejections and focussed on the pre-transplant and early post-transplant samples. For RIMINI we hypothesize that the combination therapy of ATG and anti-TNF-alpha antibody would induce a stronger reduction of inflammatory subsets and started with a comparative analysis of pre-transplant as well 6 and 12-month post-transplant samples.
In addition, and in collaboration with collaboration partners from the Canadian National Transplant Research Program we have successfully tested algorithms for automated gating procedures. Together we developed a workflow, which includes pre-formatted antibody cocktails, standardized protocols for acquisition, and validated automated analysis pipelines, which can be readily implemented in multicenter clinical trials (Ivison et al. ICI Insight 2018).
Development and manufacturing of a urinary IP-10 point of care test
The specific mission of Milenia Biotec within the BIO-DrIM project is the development of an IP-10 Point-of-Care Test in Urine Samples.
Biological background
IP-10 is a chemokine which is released by immune cells during the course of inflammatory processes and therefore is a candidate to detect infinitesimal immune processes within the graft after kidney transplantation.
Several studies could demonstrate elevated levels of IP-10 in urine during the course of rejection after kidney transplantation both on a mRNA and on a protein level. Furthermore persistant elevated levels of urinary IP-10 within the first weeks of kidney transplantation are predictive of graft function after several months even if an acute rejection is absent.
These findings lead to the conclusion that urinary IP-10 is an “alert signal” of deminished graft function and shows therefore a potential for a screening in kidney transplant patients. However, a very simple to handle test – which ideally should be used as a “home test” by the patient - would allow a better patient care, especially if the data of the tests are directly made available to the patients physician.
This vision is followed by the BIO-DrIM consortium and the aim of Milenia Biotec was to develop a lateral flow urinary IP-10 test which can be quantified by a smarthphone app. As soon as the data are measured they are stored in the memory of the smartphone and can be transferred to the physicians office.
Assay set up and development process and status
At the start of the BIO-DrIM consortium, the development goals for the urinary IP-10 test have been defined as follows:
- Sample Type: Urine
- Sample Volume: Max. 100 µL
- Time to Result: Max. 20 minutes
- Working Range: 0 – 10.000 pg/mL
- Cut Off 100 pg/mL
- Shelf Life 2 years (at room temperature)
- Technology Lateral Flow Test based on gold particles
- Reader Device Smartphone (App)
- Data Communication e-mail; Cloud
- Reference Method: Quantikine ELISA; R&D Systems, Minneapolis, USA
Figure 15 shows the composition of the test strip:
Figure 15. Composition of the urinary IP-10 test strip
The urinary IP-10 lateral flow test consists of different overlapping membranes. After applying the sample (100 µL urine) it travels from the sample application port to the so called “conjugate release pad”. This is a membrane containing a dried gold particle solution. The gold particles are coated with a monoclonal antibody specific for human IP-10. The sample redilutes the conjugate and if IP-10 is present, it will be captured by the IP-10 antibody on the gold particles. The liquid (urine sample and gold particles) continue to travel through the test strip and enter the “analytical membrane”. This membrane is coated with 2 lines, called the test line and the control line. The test line is coated with a different monoclonal IP-10 antibody (Mouse-Anti-hIP-10). This antibody is directed against a different epitope within the IP-10 molecule than the IP-10 antibody on the gold particles. If IP-10 is present in the sample, the IP-10 antibody coated onto the analytical membrane will bind the IP-10 which was already bound by the other IP-10 antibody coated onto the gold particle. Consequently, the gold particles will be retained at the location of this line, if IP-10 is present in the urine sample. The more IP-10 is in the sample, the higher the intensity of this line.
Excess urine sample and gold particles continue to travel through the strip and cross the control line. At the location of the control line an anti-mouse-antibody is coated which captures all monoclonal mouse antibodies regardless their specificity and for this reason will bind the mouse monoclonal anti hIP-10 antibody on the gold particles as well. This leads to a retention of gold particles at this location and means that this line should occur in any case, regardless whether IP-10 was present in the sample or not. This line is used as a function control of the test strip.
All remaining liquid (urine and gold particles) continued to travel through the strip and are finally completely soaked by a membrane called “Wicking Pad”. The Wicking Pad is a very thick membrane which has the capacity to soak all the liquid of the sample and thus makes sure that the liquid stream through the strip flows only in one direction and that no liquid travels back in the strip.
Figure 16 shows how the urinary IP-10 test strip looks like after running a positive sample.
Figure 16. Urinary IP-10 test strip after running the sample
On the basis of the above described technological approach a prototype of a urinary IP-10 test could be developed.
A typical standard curve of the test is shown in Figure 17.
Figure 17. Typical Standard Curve of the Urinary IP-10 Test
In order to check the performance of the developed urinary IP-10 prototype test, the group of Professor Petra Reinke from Charité, Berlin, Germany supplied 177 urine samples from kidney transplant recipients.
All the samples were run in the Milenia QuickLine urinary IP-10 test and the Quantikine IP-10 ELISA from R&D Systems, Minneapolis, USA. The results of this comparison are shown in Figure 18. The value of 100 pg/mL IP-10 has been chosen as a cut off value for these experiments. This cut off value is the overall accepted value taken from the previously published literature.
Figure 18. Comparison Study of 177 urine samples between Milenia QuickLine IP-10 and R&D Systems Quantikine IP-10 ELISA
The results of Milenia QuickLine urinary IP-10 test in the mentioned comparison study have been measured with a table top reader device, called POCScan. As part of the BIO-DrIM consortium work Milenia has developed a smartphone app working under the iOS-system supplied by Apple. Finally the app, once released, is made available to users via the Apple App Store. The Apple instruments (iPOD or iPhone) need to be placed into a plastic casing supplied by Milenia Biotec. This is shown in Figure 19.
Figure 19. Apple Smartphone in the Reader Casing
The casing contains a drawer where the test units need to be placed in and finally are moved underneath the camera of the smarthphone. It also contains LEDs to illumiate the test unit and a rechargeable battery.
In order to identify the test units they are labeled with a QR code. This QR code encoded the core data of the test, including test name, lot number, lot specific standard curve etc.
The app is ready as a working prototype and is currently used by the R&D group of Milenia Biotec for the first development projects, including the urinary IP-10 test.
The clinical members of the BIO-DrIM project have decided to ship all urine samples collected over the entire time of the project to Charité, Berlin. All samples are kept frozen there until they are run in the Milenia QuickLine urinary IP-10 test.
It has been discussed during the 5th annual BIO-DrIM meeting in Berlin at the beginning of February 2018 that Milenia Biotec manufactures a fresh lot of the IP-10 test and upon availability of this material to run the entire study with the collected urine samples within the month of May 2018 in Berlin.
Close to the end of the BIO-DrIM project Milenia Biotec is close to transfer the developed prototype of the Milenia QuickLine urinary IP-10 test to production and to finally market it. The test will be well documented from the very beginning due to the fact that so many groups helped to supply samples. Of course, it is the hope of us at Milenia Biotec, that if the IP-10 test matches the expectations of the participating clinical study groups, they will continue to use the product in the future for the advantage of their patient care. Furthermore, these groups may be the nucleus of a broader users group in the future, as they are the reference groups in their respective countries and other potential users will follow their recommendations.
Furthermore the BIO-DrIM project served as a networking platform and Milenia Biotec got in touch with big companies from the pharmaceutical industry. Luckily one of these companies was interested in Milenias capability to develop and manufacture point of care test and to read the test results with a smartphone app. With this company Milenia signed a Co-Development-Agreement for the development of tests to be used within the area of transplantation.
BIO-DrIM Statistical Support
Strong statistical support for the BIO-DrIM Consortium was provided by statisticians from King’s College London (KCL). Dr Irene Rebollo-Mesa, a senior lecturer in biostatistics at KCL, oversaw the studies from the establishment of the consortium until December 2015. She was succeeded by Dr Daniel Stahl a reader in biostatistics at KCL. A junior statistician, Dr Sofia Christakoudi was appointed as a postdoctoral researcher with full-time commitment to the BIO-DrIM Consortium for a period of three years, staring from October 2014. The BIO-DrIM statisticians provided crucial support for shaping the design of the consortium trials, protocol development, sample size calculations for all considered trial designs, simulations of potential trial outcomes under different scenarios, establishment of Go/no-Go rules and preparing charters for the Data Monitoring Committees (DMC). They also provided important explanations and responses to queries arising in the process of obtaining approvals from competent authorities in different countries participating in the consortium and implemented the requirements of ethical comities for alterations in the designs of the individual studies, the interim analyses and the establishment of stopping rules. The BIO-DrIM statisticians have collaborated closely with the clinical trials teams, both at KCL and Charité, to develop the electronic case-report forms (eCRFs) which are housing the trial data. They have prepared statistical analysis plans describing in detail the interim and the final analyses of the primary and secondary outcomes and performed detailed data checks and reports for the DMC meetings as required.
However, due to the complex nature of the trials and the involvement of multiple centers and countries, the Consortium has faced several challenges, trial design modifications and delays arising from the process of obtaining approvals from competent authorities in several countries and an initially slow recruitment. This meant that patient recruitment and follow-up could not be completed within the 5 years allocated to BIO-DrIM and, consequently, the final statistical analyses of the trial data could not be performed during this period. Notwithstanding, by the end of 2017 all the consortium trials were well established and actively recruiting participants and the statistical analysis plans determining the final analyses had been prepared. This enabled the transfer of the statistical support during the one year extension to trial statisticians in King’s Clinical Trials Unit (KCTU) for LIFT and the Coordinating Center for Clinical Studies (KKS) at Charité for CELLIMIN and RIMINI. Details of the key engagements of BIO-DrIM statisticians with the individual trials are outlined below:
WPIA: LIFT (Liver Immunosuppression Free Trial) - a prospective randomised biomarker-based trial aiming to assess the clinical utility and safety of biomarker-guided immunosuppression (IS) withdrawal in liver transplantation.
LIFT trial design evolution
The design of the LIFT trial evolved through several stages as a result of interaction and feedback from the funder and the competent authorities. The initial intention to conduct a phase II trial as a prospective, open label, non-controlled/non-randomized, interventional cohort study in which all recruited liver transplant recipients would undergo IS withdrawal would have given the opportunity to validate the diagnostic performance (AUC=0.85) of the previously developed 5-gene liver tissue gene-expression signature (Martinez-Llordella M. et al. J Clin Invest. 2008), but would not have obtained a decisive verdict on the clinical utility of a biomarker-led IS withdrawal when compared with a weaning-all strategy. Consequently, a three-arm randomised controlled biomarker-strategy trial design with two control arms was considered. This aimed to achieve biomarker validation and clinical safety by demonstrating non-inferiority of biomarker-led care with respect to the number of successfully weaned rejection-free patients at 1-year follow-up post IS withdrawal when compared to the weaning-all control arm strategy and a non-inferiority with respect to the increase in fibrosis score at 3 years of follow-up when compared to a weaning-none control arm. This design would have required 150 patients, 50 in each trial arm (for 80% power and 5% probability of type I error), with an expectation of 100% weaning success rate in biomarker-positive patients and including all patients from the biomarker-based strategy arm, including biomarker-negative patients on maintenance IS. However, maintenance IS does not address clinical outcomes related to the establishment of “operational tolerance”, which could only be evaluated following IS withdrawal. This directed the development of the final design of LIFT as a prospective randomized two-arm superiority trial (Figure 20) aiming to demonstrate that the use of a liver tissue transcriptional test of tolerance to stratify liver recipients prior to IS withdrawal accurately identifies operationally tolerant recipients and reduces the incidence of rejection, as compared with a control group in whom IS withdrawal is performed without stratification.
Figure 20. LIFT trial design evolution.
LIFT sample size and Go/no-Go criteria
The study has been designed and powered to test the hypothesis that IS weaning under a "Biomarker-led" strategy (i.e. weaning only biomarker-positive patients, Arm B+) is superior to a “Weaning-All” strategy (irrespective of biomarker status, Arm A), with respect to the proportion of patients who, having started the IS withdrawal protocol, complete it successfully without undergoing allograft rejection by 1 year post withdrawal completion.
Sample size calculations are based on the following expectations:
• Biomarker: 50% biomarker-positive patients – simulation studies showed that if biomarker-positivity is reduced, which was observed in the initial recruitment stages of LIFT, the trial would gain power due to the increased proportion of weaning failures in the weaning-all control Arm A;
• Arm A: 50% successfully weaned patients, based on the weighted average of success rates observed in a previous single-arm weaning-all trial (Benitez C. et al. Hepatology 2013;58:1824) in patients grouped by age and time from transplantation at initiation of withdrawal;
• Arm B+ H0 (null hypothesis) 50% successfully weaned patients (if PPV = 0.50); H1 (alternative hypothesis) 80% or more successfully weaned patients (if PPV ≥ 0.80) more conservative and realistic expectation than the 100% incorporated in the initial trial design;
• Required number of completers: 100 patients for 90% power and 5% type I error rate;
• Allocation ratio: 2:1 between arms A and B+, if 50% of patients are biomarker-positive;
• Target sample size: 148 patients (134) completers, including 34 patients in Arm B- and 14 patients accounting for 10% drop-out rate.
Therefore, with the same number of patients the trial under the final design could achieve higher power than the trial under the initial design and could demonstrate a superiority of biomarker-guided IS withdrawal (Arm B+) compared to a weaning-all strategy (Arm A).
Additional considerations for the biomarker test to be successfully validated for clinical use are a sensitivity no less than 0.70 and a positive predictive value (PPV) no less than 0.80. If the PPV or the sensitivity are 0.5 the prediction of a biomarker would be no better than at random. Consequently, Go/no-Go criteria were set for two planned interim analyses, when 33% and 50% of patients assigned to IS weaning reach 1 year follow-up post completion of IS withdrawal. The trial would progress if the 95% confidence interval for PPV estimated in Arm B+ at the first interim analysis includes 0.80 and excludes 0.50 and the 95% confidence interval for sensitivity estimated within Arm A at the second interim analysis includes 0.70 and excludes 0.50.
LIFT Statistical Analysis Plan (SAP)
The Intention-to-Treat population (determined by treatment assignments) will be used for the primary analysis, which is conventional for superiority trials. Generalised linear mixed-effects models (logistic regression) was selected for the final analyses due to concerns that there would be insufficient power to adjust for all potential covariates, that the results would be conditional on the covariates and that additional clinical centers joined the trial (up to 13), some of them likely to recruit very few patients. The models will include adjustment for the factors determining the minimisation and will consider study centre as a random effect in all analyses and biomarker status as a fixed effect in the secondary analyses. An overall marginal rather than a conditional effect will be estimated. Defining study centre as a random effect, i.e. treating every centre as representative of a larger population of centres, is conceptually appropriate, as all centres are expected to show comparable performance. Mixed-effects regression will also accommodate the analysis of serial measures. Multivariable models for the exploratory analyses of secondary outcomes will include main effects of treatment arm and biomarker status, as well as their interaction term. The influence of potential covariates added individually to the model subgroup analyses for time since transplantation and age at randomization, ethnicity, sex or drug regimen will be examined as part of the sensitivity analyses. The stability of the biomarker signature over the follow-up period will also be examined, both as a binary classification and as a continuous measure of probability of rejection. The effect of potential mechanistic mediators on primary and secondary outcomes will be evaluated using generalized latent mixed models, which will enable the consideration of different models of causation.
The Statistical Analysis Plan (SAP) was approved by the DMC statistician and the protocol was updated and finalized accordingly.
LIFT electronic Case Report Form (eCRF) and data quality checks
A comprehensive eCRF was designed and implemented in collaboration with KCTU. The complicated follow-up and data collection schedule of the LIFT trial was accounted for in some 40 different forms accommodating clinical and laboratory data acquired at the baseline, screening and scheduled transplant centre visits (6-monthly) and details of the protocol liver biopsies, as well as data from monitoring of liver function tests (performed at 3-weekly intervals in the initial phases of IS weaning), information on drug regimens and doses and details of for-cause biopsies. The BIO-DrIM statisticians performed regular data checks at 3-monthly intervals in order to ensure that the baseline information on eligibility, demographics, biochemistry, histology, medical history and clinical events, IS drugs, concomitant medication and quality of life is available and recorded correctly for all consented patients and that follow-up data are timely accrued for all randomized patients. Numerous queries were raised for all identified discrepancies, which were brought to the attention of all individual study sites by the Trial Manager. Efforts were undertaken to identify the patterns of errors and to provide improved guidelines for timely and accurate data recording.
LIFT DMC reports and patient monitoring
Given the potential risk that IS withdrawal poses on patient wellbeing, the Ethics Committee requested a very close monitoring by the DMC. The trial statisticians prepared detailed reports for the DMC meetings at 3-monthly intervals. These included information on consent, randomization and withdrawal figures, baseline demographics and immunosuppressant medication, duration of periods between study procedures and visits in comparison with the target windows set in the protocol, frequency of treatment-related diseases and description of the adverse effects. The data were tabulated by study centre for the open part (accessible to the DMC, Trial Steering Committee) and additionally by treatment (IS weaning or maintenance) for the closed part (confidential to the DMC members). The proportion of biomarker positivity was monitored. In addition, the trial statisticians answered data entry queries raised by the study sites and prepared summaries of the findings in the screening baseline biopsies to assist the trial team to identify which particular histological patters contribute to failure of histological eligibility. The markdown package in R was set up for dynamic reporting. A code was written which generates the data monitoring progress reports exporting directly plots and tables into Word. This removed the possibility of mistakes arising from manual data transfers from the statistical software and ensured the consistent generation of high quality output. The code additionally generates time plots of liver function tests and IS drug doses, individually for each patient and in separate batches per study site. The timings of study visits and liver biopsies are marked on the plots to enables the visualization of the progress of each patient in the trial and to facilitate the identification of missing data and inconsistencies.
WPIC study - using biomarkers of tolerance to guide immunosuppression weaning in kidney transplant recipients
This study, coordinated by KCL, was initially conceived as a safety and feasibility pilot phase II trial for initiating IS minimization in patients identified as ‘tolerant’ on the basis of a 10-gene expression signature in peripheral whole blood developed as part of the preceding Indices of Tolerance study (Sagoo P. et al. J Clin Invest 2010;120:1848). The signature was replicated in samples from patients of the GAMBIT study, achieving high predictive accuracy and good test-retest stability. However, a major statistical point of concern was the possibility that IS drugs influence the levels of expression of the candidate ‘tolerance genes’ and, thus, introduce a potential confounding effect on the gene-expression signature. Statistical analysis revealed that the expression of most IoT genes was, indeed, significantly associated with IS drug intake to the extent that adjustment of the signature genes for drug intake abolished the discrimination between tolerant and non-tolerant patients. Consequently, no clinical trial could go ahead and the project returned to the “drawing board” into the hands of the BIO-DrIM statisticians. A new IS-drug-independent gene-expression signature of tolerance was developed. A set of 28 candidate genes optimal for prediction of tolerance with perfect predictive accuracy was identified from the IoT array using for gene selection drug-adjusted gene-expression levels and penalised logistic regression with elastic net penalty. The candidate genes were then validated in samples from the GAMBIT study using a Fluidigm platform and a parsimonious 9-gene signature was selected with elastic net regression. This new signature (Rebollo-Mesa I. et al. Am J Tranplant 2016;16:3443) is IS-independent and can achieve a very good discrimination of already established tolerant KTRs from treated stable and chronic rejection patients (cross-validated AUC 0.81) identifying approximately 12% of stable patients as potentially “tolerant”.
An important point to consider in stable KTRs selected as “tolerant” whilst receiving IS treatment is whether they would retain such features of “tolerance” once the drugs have been withdrawn. Based on the strong confounding effect of IS drug regiments, revealed by BIO-DrIM statisticians, it has become clear that gene-expression signatures developed without any statistical mechanism to control for IS-drug intake is in danger of being driven by pharmacological and not biological features. Consequently, we have engaged in an active debate with the kidney transplant community with one of the most prominent results an editorial with the explicit title “Lack of adjustment for confounding could lead to misleading conclusions” (Christakoudi S. and Hernandez-Fuentes M. Am J Transplant 2017), prompting an equally explicit response “We agree, Let’s move on” (Asare A. et al. Am J Transplant 2017;17:2505). Before we “move on”, however, there is one more contribution that BIO-DrIM statisticians owe to the transplantation literature and this is the comparison of the clinical characteristics and IS-drug regimens of stable KTRs selected as “tolerant” on the basis of our IS-independent signature and on the basis of signatures developed without adjustment for IS-drug regimens. This analysis will also comprise the validation of our new signature with RT-qPCR, the clinical-grade laboratory technique.
Failure to proceed with a clinical trial of IS minimization in long-term and established stable KTRs within the time-frame of the BIO-DrIM Consortium was determined by major safety concerns of clinical teams and funders in the UK, which made it clear that further detailed studies of “tolerance” signatures are paramount. Consequently, we initiated a ramification of the GAMBIT study aiming to elucidate the role of steroid conversion (HSD11B1, HSD11B2, H6PD) and receptor genes (NR3C1, NR3C2) for “operational tolerance” and chronic rejection in KTRs, as glucocorticoids are major endogenous immunomodulatory and anti-inflammatory factors and synthetic steroids are fundamental for IS drug regimens in KTRs. This work advanced the statistical concept of the importance of adjustment for confounding a step further (Christakoudi S. et al. Mol. Cell. Endocrinol 2018). We examined in detail the impact of exogenous prednisolone and the effects of adjustment for IS drug intake on the expression of steroid conversion and receptor genes, thus demonstrating that drug adjustment enables evaluation of endogenous factors influencing gene expression. Further, steroid conversion and receptor genes, alone, achieved classification of “tolerant” patients and were major contributors to gene-expression signatures of both, tolerance and CR, alongside known tolerance-associated genes, thus revealing a key role of steroid regulation and response in kidney transplantation.
WP2 (CELLIMIN) - prospective, multi-centre, biomarker-strategy enrichment design trial of IS minimization in KTRs.
CELLIMIN trial design evolution
The main objective of the CELLIMIN trial is to demonstrate the utility and safety of the IFN-γ ELISPOT marker for the stratification of kidney transplant recipients. Initially, the trial was intended as a biomarker-stratified design in which KTRs would be first stratified on the basis of their pre-transplantation ELISPOT response in two groups ELISPOT-positive (i.e. at higher risk of rejection) and ELISPOT-negative (i.e. at low risk of rejection) and then patients within each group would be randomized in a ratio 1:1 to a new “Low” or standard-of-care “High” IS-regimen, resulting in a four-arm trial (Figure 2). This trial design, however, would have been appropriate if compelling biological data were lacking that one of the biomarker groups would not benefit from the new treatment. Ethical concerns raised by the competent authorities during the Voluntary Harmonisation Process (VHP) that ELISPOT-positive KTRs have unacceptably high risk of rejection, led to the elimination of the ELISPOT-positive group. The study was modified to an enrichment design, utilising ELISPOT as a screening tool prior to randomisation. Concerns were also raised by the VHP regarding the initially proposed 15% non-inferiority margin. Their requirement to reduce the non-inferiority limit to 10% could not have been met retaining the fairly conservative initial alternative hypothesis of 15% rejection rate in the new-treatment arm (5% higher compared to the control arm), as this would have quadrupled the required sample size to 1205 patients - a financially unsustainable and completely unachievable goal. Figure 21 illustrates how the assumptions under the alternative hypothesis influenced more dramatically sample size for a non-inferiority margin of 10% compared to 15%.
Figure 21. CELLIMIN sample size for different rejection rates under the alternative hypothesis.
BIO-DrIM statisticians, however, were able to retain the original sample size and to accommodate the 10% non-inferiority limit by assuming equivalence in the alternative hypothesis - a more traditional assumption in non-inferiority trial designs and not an unreasonable one, since the prediction for the expected rejection rate in the treatment group was quite uncertain. As a safety net, a stopping boundary for futility, based on conditional power calculated from the observed cumulative proportions, was introduced at the interim analysis step in addition to the pre-existing non-inferiority (efficacy) stopping boundary, based on Lan-De Mets alpha spending function with O’Brian-Fleming parameters. This gives the steering committee the flexibility to consider at the interim analysis whether to stop the trial, if the observed rate of rejection in the treatment arm is more than 15% (judging by its 95% confidence interval) or to increase the sample size adaptively using the actual observed rates of BPAR in the treatment arm, after considering costs and actual accrual rates.
Figure 22. CELLIMIN trial design evolution
CELLIMIN sample size and Go/no-Go criteria
The primary hypothesis is that a “Low” IS regimen is non-inferior with respect to the rate of T-cell mediated BPAR at 6 months post-transplantation when compared to the standard-of-care “High” IS regimen. The final sample size calculations were based on the following expectations:
• “High” IS (standard-of-care) control arm: 10% BPAR at 6 months post-transplantation;
• “Low” IS (new-treatment) arm: H0: 20% BPAR; H1: 10% or less BPAR (reduced from 15%);
• Non-inferiority limit: 10% (reduced from 15%);
• Required number of completers: 271 patients for 80% power and 5% type I error rate;
• Target sample size: 301 ELISPOT-negative patients completers, including 30 patients accounting for 10% drop-out rate;
• Screening: 669 patients to be screened, expecting approximately 45% to be ELISPOT negative;
• Interim analysis: when 122 patients complete 6 months of follow-up, of which 61 in the “new treatment” arm.
The interim analysis involves the calculation of a Z-statistic with a cut-off for the efficacy stopping boundary of -2.698 (on the Z-scale) and for the futility stopping boundary -0.692 on the Z-scale and 0.625 on the conditional power scale. A one-sided 95% CI for a BPAR rate of 15% would be estimated with a precision of +/- 7.5%, with the upper bound of 22.5% defining the cut-off for stopping the trial for futility.
CELLIMIN Statistical Analysis Plan
BIO-DrIM statisticians addressed statistical queries raised in the process of obtaining approval by VHP and the competent authorities of the individual participating countries and provided much appreciated constructive feed back to the eCRF developers at Charité. They also prepared the statistical analysis plan (SAP), which contains a detailed section explicitly defining failure outcome. The problem with primary outcome definitions in CELLIMIN arises from the fact that protocol biopsies are conducted at 3 months post randomisation, to adhere to the conventional clinical management practice, while the primary outcome is evaluated at 6 months and is defined as “biopsy-proven” acute rejection, implying that a for-cause biopsy should be triggered by a clinical indications of rejection. This means that a protocol biopsy at 3 months could uncover subclinical rejection or borderline changes, thus making it impossible to evaluate what would have been the clinical outcome by the time of final evaluation of the primary outcome at 6 months in the absence of this information or a resulting from it treatment. The SAP, therefore, considers failure outcomes with respect to different scenarios of histological results and treatment following the protocol biopsy at 3 months.
The Per-Protocol (PP) population (determined by the administered and not the allocated treatment) will be used for the primary analysis, but an Intention-To-Treat (ITT) analysis will also be performed as part of the sensitivity analyses. The analytical strategy of the primary and secondary outcomes is aligned with the LIFT trial and is based on generalized linear mixed-effects models (logistic regression) with random effects for study centre, as it is a factor used in the randomisation procedure, and a fixed effect for treatment arm. An estimated difference in the proportions of patients with BPAR between the “Low” and the standard-of-care “High” IS regimen arms with a 95% one-sided upper confidence interval will be presented for the population (marginal) effect. Non-inferiority will be evaluated with respect to the upper bound of the confidence interval in relation to the non-inferiority margin as illustrated in Figure 23.
Figure 23. CELLIMIN non-inferiority testing, based on the confidence interval of the difference in BPAR.
WP3 (RIMINI) - Tacrolimus after rATG and infliximab induction immunosuppression
Study design evolution
The three-arm design of the original ICARUS trial was cancelled because it would have required unfeasibly high number of patients and, in addition, the treatment medication was no longer available. The subsequent plan in the renamed RIMINI trial was to proceed with an open-label single-arm Simon’s two-stage pilot Phase II clinical trial in order to provide evidence for efficacy and safety of the new induction regimen and a Go/no-Go rule for further clinical development. However, this design was also abandoned because the sample size, although reduced, was still unfeasibly high, as RIMINI was going to compete with CELLIMIN for patient recruitment. In addition, a Simon’s design would have been more appropriate if the expected proportion of outcome failures under the alternative hypothesis was considerably lower than the expectation for the null hypothesis. In RIMINI a smaller difference or a non-inferiority would also be considered beneficial for the new induction regimen and there was a real risk that under Simon’s design the trial would complete both stages and yet provide inconclusive evidence to proceed into a phase III trial. BIO-DrIM statisticians found a solution in a confidence-interval-based single-arm Phase II design, which reduced the sample size by almost half of the requirement for Simon’s design, without losing power, whilst still addressing the main research question and providing Go/no-Go criteria for a larger parallel-arm-design trial. The confidence-interval-based design would provide an effect estimate and a valuable information for further clinical development with almost the same number of completers that would have been sufficient only for the first stage of a Simon’s design.
Figure 24. RIMINI trial design evolution
Sample size calculations and Go/no-Go criteria
A total of 68 completers will be required to estimate a one-sided 95% confidence interval with a precision (one-sided width) of no more than 10% efficacy failure rate. The target sample size was increased with 10% to account for drop-outs and violations of the protocol. A total of 75 patients will receive the proposed induction regimen - less than half of the 161 patients that would have been required with Simon’s design, which makes completion of recruitment and follow up feasible within a reasonable timeframe. If up to 27 out of the 68 completers experience efficacy failure, a progression into a larger trial will be considered justifiable. If the number of patients experiencing efficacy failure is between 28 and 34 out of 68, the merits of a larger non-inferiority design will be considered depending on the risk/benefit assessment. If more than 34 out of the 68 completers experience efficacy failure, a progression into a larger trial would be considered unjustifiable.
There is no planned interim analysis as the new induction treatment will be applied at enrolment and, due to the 12-month follow-up, the majority of the patients would have already received the new treatment by the time of an interim analysis.
Statistical Analysis Plan
BIO-DrIM statisticians assisted with protocol development and prepared a detailed Statistical Analysis Plan. They raised queries regarding particular details of the definitions of the primary outcomes and the involved in this laboratory measurements of creatinine, which were very much appreciated by the Principal Investigator. These questions were discussed with the members of the consortium during the annual BIO-DrIM meeting and decisions were taken, including corresponding modifications of the eCRF.
For the analysis of the primary outcome, the upper bound of a one-sided 95% confidence interval for the observed efficacy failure rate within the overall group of patients who have received the study treatment will be calculated. Cluster robust standard errors, which are likely to keep or slightly increase the power of the study, will be used to account for study sites.
References
Christakoudi, S. and Hernandez-Fuentes, M.P. 2017. Lack of adjustment for confounding could lead to misleading conclusions, Am J Transplant. 17, 2505-2506.
Christakoudi, S., Runglall, M., Mobillo, P., Rebollo-Mesa, I., Tsui, T.L. Nova-Lamperti, E., Norris, S., Kamra, Y., Hilton, R., Bhandari, S., Baker, R., Berglund, D., Carr, S., Game, D., Griffin, S., Kalra, P.A. Lewis, R., Mark, P.B. Marks, S.D. Macphee, I., McKane, W., Mohaupt, M.G. Pararajasingam, R., Kon, S.P. Seron, D., Sinha, M., Tucker, B., Viklicky, O., Lechler, R.I. Lord, G.M. Stahl, D. and Hernandez-Fuentes, M.P. 2018. Steroid regulation: An overlooked aspect of tolerance and chronic rejection in kidney transplantation, Mol Cell Endocrinol. doi: 10.1016/j.mce.2018.01.021
Nova-Lamperti, E., Romano, M., Christakoudi, S., Runglall, M., McGregor, R., Mobillo, P., Kamra, Y., Tsui, T.L. Norris, S., John, S., Boardman, D.A. Lechler, R.I. Lombardi, G. and Hernandez-Fuentes, M.P. 2018. Reduced TCR Signaling Contributes to Impaired Th17 Responses in Tolerant Kidney Transplant Recipients, Transplantation. 102, e10-e17.
Rebollo-Mesa, I., Nova-Lamperti, E., Mobillo, P., Runglall, M., Christakoudi, S., Norris, S., Smallcombe, N., Kamra, Y., Hilton, R., Indices of Tolerance, E.U.C. Bhandari, S., Baker, R., Berglund, D., Carr, S., Game, D., Griffin, S., Kalra, P.A. Lewis, R., Mark, P.B. Marks, S., Macphee, I., McKane, W., Mohaupt, M.G. Pararajasingam, R., Kon, S.P. Serón, D., Sinha, M.D. Tucker, B., Viklický, O., Lechler, R.I. Lord, G.M. and Hernandez-Fuentes, M.P. 2016. Biomarkers of Tolerance in Kidney Transplantation: Are We Predicting Tolerance or Response to Immunosuppressive Treatment?, Am J Transplant. 16, 3443-3457.
Electronic Case Report Form (eCRF) and the data management (DM) of the trials CELLIMIN and ICARUS 2.0 / RIMINI eCRF/ DM
An extensive effort has been made by the KKS in collaboration with the CHARITE and the statisticians from KCL in setting up an eCRF with an Internet-based Remote Data Entry (RDE) system; a technique that allows electronic data capture in a central study database.
The software platform is SecuTrial® with the following characteristics:
o Compatible to standard Internet browser (Firefox, Internet Explorer, Opera ...)
o No further software installation necessary
o No local storage of data
o Available worldwide 24h (30min. downtime per day [02:00 GMT-1])
o Safety access (Firewalling, SSL encryption, Authentication, Logs)
All the processes are done according to SOPs and Working Procedures and are compliant to:
o GCP
o FDA 21 CFR Part 11
o German Pharmaceuticals Act.
Regarding the eCRF and data management, the following tasks have been accomplished or will be provided:
o Design and creation of database
o Design and programming of eCRF with system and plausibility checks
o Validating and release of eCRF
o Reporting of critical information on weekly basis
o Creation of a Data Validation Plan for data clearing
o Delivery of cleaned data for statistical analysis
o Troubleshooting
For support during the study, the following tools have been implemented:
o Administration: database / eCRF (Logins, roles and rights)
o User support (Reset password, technical problems etc.)
o Query management (Data checks)
o SAS-Reporting
o Staff training
Pharmacovigilance
Regarding the Pharmacovigilance, an adequate SAE (Serious Adverse Event)-Management was implemented to ensure safety surveillance during the study:
o Safety surveillance = application of drug and adverse effects
o Legally defined reporting procedures and timelines
o Central pharmacovigilance
- Central Contact point (all reports KKS)
- Reporting to all concerned authorities, ethics committees and study sites via central pharmacovigilance
All reported SAE are assessed according to their impact on the safety profile of the administrated drugs and treatment procedures. The steps of SAE management, evaluation, and further reporting to CA, EC and investigators in case of a SUSAR are described in the study protocol and study specific SOPs. As a consequence of such safety information all necessary actions to protect patient’s safety are accomplished by the sponsor. All available safety information is continuously monitored and the results are summarized in an annual safety reporting (DSUR) to the CA, EC and investigators.
Monitoring
The objectives of the monitoring procedures are to ensure that the trial subject’s safety and rights as a study participant are respected, that accurate, valid and complete data are collected, and that the trial is conducted in accordance with the trial protocol, the principles of GCP and local legislation. Therefore, to guarantee the quality of the monitoring the KKS provided a general monitoring SOP, a monitoring manual and report templates.
Source data are contained in source documents (original records or certified copies).
The monitoring of CELLIMIN and RIMINI is a combination of on-site visits (initiation, interim and close out) and centralized monitoring based on the eCRF (SecuTrial®).
During the on-site visits the monitor the following data is verified:
• Trial subject existence
• Subject informed consent
• Inclusion / exclusion criteria
• Result of ELISpot randomization (CELLIMIN)
• Correct reduction (group B) of trial medication according to randomisation and trial
protocol.
• Control of drug accountability (only documentation)
• SAEs
• Primary endpoint (ELISpot and Biopsy results BPAR)
• End of study participation/BPAR
The report that KKS generates after each visit, informs the sponsor/PI about:
o Result of SDV
o Protocol violation
o (S)AEs
o Drug account
o Recruitment and problems at the site
o Follow-up activities
Follow-up letter inform the site about:
o Tasks to be completed
o Completion/Updating of study documentation
o Documentation of Drug account
o Others
Health-Economic Studies (Health-Technology Assessment – HTA)
At the outset of the BIO-DrIM consortium, it was necessary to develop three clinical trial protocols, each one including a health-economic data assessment and evaluation strategy that would enable researchers to analyze medium- and long-term healtheconomic outcomes with respect to costs and quality of life.
A central objective of the BIO-DrIM consortium is to prove the clinical feasibility of reducing immunosuppressive (IS) medication and thereby also reducing IS-related medication-costs after solid organ transplantation (kidney and liver). Depending on the trial (LIFT, CELLIMIN, RIMINI), IS reduction is indicated either for selected organ recipients based on pre-transplant biomarker-based patient-stratification procedures (CELLIMIN and LIFT trial) or using an IS weaning protocol for a selected cohort of patients fulfilling specified trial inclusion criteria (RIMINI trial).
As one of the first steps, structured literature reviews were conducted for all indications covered by the three clinical trials in order to identify typical health-economic data assessment strategies, endpoint definitions and respective results. The key findings of the literature reviews were presented at the annual meeting in 2015.
Based on the results of the literature reviews, trial specific data assessment strategies using standardized quality of life questionnaires and study-specific health-economic questionnaires were developed to accurately assess patient-specific resource consumption during the follow-up periods of the trials. The assessment and evaluation strategies were developed based on guidelines for health-economic evaluations alongside clinical trials (Edwards et al., 2008; Petrou & Gray, 2011; Ramsey et al., 2005; Sullivan et al., 2005), they were aligned with the clinical objectives of each clinical trial, and they are described in detail on the following pages.
Health-economic data assessments in the BIO-DrIM trials use questionnaires that are either standardized or were specifically developed for each trial. Precise working instructions on the use of the questionnaires were developed in all cases and made available for the respective study staff. Furthermore, health-economic statistical analysis plans were developed and provided to the study teams. Interim-analyses and final analyses of the health-economic data gathered throughout the trials will be completed as soon as all required data are available.
Contributions to WP1 (LIFT trial)
The health-economic data assessment strategy for the LIFT study was developed in close collaboration with the principal investigator (Prof. Alberto Sanchez-Fueyo) and is based on health-economic guidelines for health-economic evaluations alongside clinical trials (Edwards et al., 2008; Petrou & Gray, 2011; Ramsey et al., 2005; Sullivan et al., 2005). It was presented, discussed and consented within the BioDrIM consortium.
The health-economic data assessment strategy of the LIFT trial is separated into 3 main steps:
1. Assessment of Quality-Adjusted Life Years (QALYs)
2. Assessment of patient-specific resource consumption as a base for cost-estimates
3. Micro-costing approach to evaluate costs of biomarker-stratification process
Quality-Adjusted Life Years are currently being assessed in the LIFT trial using the generic EQ5D-5L questionnaire, which is a standardized and well accepted instrument in the scientific community of health-economics. A working instruction, specifically designed for the LIFT trial, was developed and uploaded to internal BioDrIM website, in order to make the questionnaire as well as the working instruction available for all participating study nurses and other trial staff.
In order to precisely capture other generic aspects of quality of life (qol) which are not covered by the EQ5D-5L instrument, the implementation of the SF36 questionnaire – another well accepted qol-questionnaire – was discussed and agreed. The SF36 questionnaire and a corresponding working instruction were uploaded to the internal BioDrIM website, in order to make them available for all participating study nurses and other trial staff.
As suggested by guidelines on health-economic data assessments in clinical trials (Edwards et al., 2008; Petrou & Gray, 2011; Ramsey et al., 2005; Sullivan et al., 2005), the disease-specific qol-questionnaire NIDDK was proposed and agreed to implement.
Furthermore, a trial specific health-economic questionnaire was developed to capture individual health-care resource consumption, during the follow-up of each enrolled patient. The development of this questionnaire had to be well balanced in terms of efforts to be undertaken by the trial staff to assess the needed data in patient-interviews and the precision of expected results. This questionnaire as well as a corresponding working instruction was developed in close collaboration between the principal investigator and Cellogic GmbH and both documents were uploaded to the internal BioDrIM website, in order to make them available for all participating study nurses and other trial staff.
Regular checks of the assessed health-economic trial data have been defined in close collaboration between the principal investigator, statisticians at King`s College London, who are responsible for data management and storage, and Cellogic GmbH. Every three months, the complete trial dataset of all health-economic electronic case-report forms (eCRF) of all included patients are submitted to Cellogic GmbH and checked for completeness and consistency. In case of systematically missing data or inconsistent values, adequate measures, e.g. intensified briefing and training of study nurses, are taken to prevent poor data quality and more missing data in future assessments. The results of the last regular completeness-check were presented at the last annual meeting on February 2nd, 2018 and are shown on figure 25.
Figure 25. Results of data completeness-check (LIFT), Data extraction date: January 2nd, 2018
Contributions to WP2 (CELLIMIN trial)
The health-economic data assessment strategy was developed in close collaboration with the principal investigator (PI, Prof. Dr. Josep M. Grinyó) and the sponsor representative (Prof. Dr. Petra Reinke) and is based on health-economic guidelines for health-economic evaluations alongside clinical trials (Edwards et al., 2008; Petrou & Gray, 2011; Ramsey et al., 2005; Sullivan et al., 2005). Advantages and disadvantages of using different assessment strategies were presented and discussed during the BIO-DrIM meeting on November 12th, 2012. The final assessment strategy was consented with the principal investigator shortly after.
The health-economic data assessment strategy of the CELLIMIN trial is separated into three major steps:
1. Assessment of Quality of Life (QoL) at baseline, months 1, 3, 6, 12, 18 and 24 and subsequent computation of Quality-Adjusted Life Years (QALYs)
2. Assessment of patient-specific resource consumption at initial discharge following transplantation, repeated discharges (if occurred) and regularly at months 3, 6, 12, 18 and 24
3. Micro-costing of biomarker-stratification process in order to assess costs related to pre-transplantation patient-stratification
QALYs are being assessed in the CELLIMIN trial using the generic EQ5D-5L questionnaire, which is a standardized and well accepted instrument in the scientific community of health-economics. A working instruction, specifically designed for the CELLIMIN trial, was developed and uploaded to internal BIO-DrIM website, in order to make the questionnaire as well as the working instruction available for all participating study nurses and other trial staff.
In order to precisely capture other generic aspects of quality of life (qol) which are not covered by the EQ5D-5L instrument, the implementation of the SF36 questionnaire – another well accepted qol-questionnaire – was discussed and agreed. The SF36 questionnaire and a corresponding working instruction were uploaded to the internal BIO-DrIM website, in order to make them available for all participating study nurses and other trial staff.
As suggested by guidelines on health-economic data assessments in clinical trials (Edwards et al., 2008; Petrou & Gray, 2011; Ramsey et al., 2005; Sullivan et al., 2005), the disease-specific questionnaire KTQ-25 was proposed to the PI and agreed to implement.
Furthermore, a trial specific health-economic questionnaire was developed to capture individual health-care resource consumption, during the follow-up of each enrolled patient. The development of this questionnaire had to be well balanced in terms of efforts to be undertaken by the trial staff to assess the needed data in patient-interviews and the precision of expected results. This questionnaire as well as a corresponding working instruction was developed in close collaboration between the PI and Cellogic GmbH and both documents were uploaded to the internal BIO-DrIM website, in order to make them available for all participating study nurses and other trial staff.
Regular checks of the assessed health-economic trial data have been defined in close collaboration between the PI, KKS-Charité (CRO), who is responsible for data management and storage, and Cellogic GmbH. Every three months, the complete trial dataset of all health-economic electronic case-report forms (eCRF) of all included patients are submitted to Cellogic GmbH and checked for completeness and consistency. In case of systematically missing data or inconsistent values, adequate measures, e.g. intensified briefing and training of study nurses, are taken to prevent poor data quality and more missing data in future assessments. Additionally, several meetings and reviews with the project coordinator and pilot-center study nurse were carried out in order to clarify and communicate appropriate use of the developed health economic questionnaires since study initiation. The results of a recent completeness check were presented at the last annual meeting on February 2nd, and are shown in figure 26.
Figure 26: Results of completeness-check (CELLIMIN), Data extraction date: November 22nd, 2017
Cost of prospective biomarker-stratification (Micro-Costing)
In order to analyze the health-economic effects of a patient-selection process for a given cohort, e.g. in order to identify those patients who are eligible for reduced IS medication protocols, the costs of that patient-stratification need to be known.
The expected costs for the biomarker- based stratification process were calculated using a micro-costing approach for the ELISPOT-assessment, as it is conducted throughout the CELLIMIN trial. The micro-costing procedure consists of three major steps:
1. Counting of necessary inputs to the patient-stratification process in terms of consumerables as well as labor time
2. The counted inputs are multiplied with the respective market prices to arrive at average provision costs of the patient-stratification process
3. These final costs are discounted (depending on the year of market price identification) with appropriate rates in order to report present costs of patient stratification.
The micro-costing procedure was conducted in close collaboration with AID and laboratory personnel at the Berlin-Brandenburg Center for Regenerative Therapies (BCRT) at a time (in late 2017) when the stratification process was already standardized for the clinical trial and training of the laboratory staff was completed. The timing of the procedure was chosen in order to account for expected learning effects in the beginning of the clinical trial.
Standard operating procedures (SOPs) for the stratification process (in CELLIMIN study) used in participating study labs have been identified, analyzed and evaluated with regards to necessary input factors. All necessary items were counted, and prices were identified either from officially available product-lists or from hospital-internal catalogues, whenever official prices were not available. On these grounds, it was possible to calculate an upper price-ceiling of one stratification process, as it is done during the trial.
This is an important base for further price calculations that are also input to further health-economic calculations, e.g. break-even calculations for the stratification process.
Details of the micro-costing procedure were presented at the annual meeting on February 2nd, 2018 and are shown in figure 27.
Figure 27: Details of micro-costing procedure for patient-stratification in CELLIMIN
Preliminary cost-model to assess differences in IS medication costs between study arms vs. cost of pre-tx stratification (presented at annual meeting on Feburary 2nd, 2018)
The primary objective for the preliminary cost-model was to assess IS-medication costs per patient and per treatment arm, using the medication plan from the CELLIMIN protocol and to combine them with the cost for pre-Tx stratification process (result from micro-costing procedure) to assess whether the reduced costs of IS-medication in one study arm outweigh the cost pre-Tx stratification. This objective can also be expressed in the form of a decision problem, which seeks to find the best treatment strategy (the strategy that is associated with lowest costs). Therefore, two different treatment strategies had to be assessed and compared. They were called: Stratified treatment vs. high IS for all (Standard of Care).
The decision problem is illustrated in figure 28.
Figure 28: Underlying decision problem of preliminary cost-model
The share of ELISpoT-positive and negative patients was assumed to be 55% and 45% respectively for the stratified treatment strategy and the results of the micro-costing and IS-medication costs (based on medication plan and respective dosages in study protocol) served as inputs.
Generalized medications and respective dosages were obtained from study protocol and expert opinion by Prof. Reinke (for Tacrolimus). The costs of IS-medication were calculated for each day of the trial using prices obtained from Rote-Liste (official German market prices of pharmaceuticals). Costs for each day of the trial were calculated for Arm A (high IS) and Arm B (low IS), and total IS medication costs were calculated for the full follow-up period until M24.
Using the outlined model-structure and described inputs, it was possible to calculate generalized IS-medication costs for each study arm, and treatment strategy. Using these results and the given assumptions, a break-even point was calculated in terms of time using an upper price-ceiling for the patient-stratification process. The results were presented at the annual meeting on February 2nd, 2018.
Contributions to WP3 (RIMINI trial)
The development of a health-economic data assessment process for the RIMINI trial was aligned with the CELLIMIN study in order to ensure comparability of results between clinical trials, because both trials evaluate IS medication protocols following kidney transplantation in comparable patient cohorts. Data quality checks were also defined equivalently for RIMINI for the same reason. In contrast to these aspects, the data analyses plan and outcome evaluation of the RIMINI trial will differ from CELLIMIN due to its single-arm character, although all health-economic data and endpoints are identical.
The health-economic data assessment strategy was developed in close collaboration with the principal investigator (PI, Prof. Ondrej Viklicky) and Prof. Dr. Petra Reinke, and is based on health-economic guidelines for health-economic evaluations alongside clinical trials (Edwards et al., 2008; Petrou & Gray, 2011; Ramsey et al., 2005; Sullivan et al., 2005). Advantages and disadvantages of using different assessment strategies were presented and discussed during several BioDrIM meetings, due to several significant changes of the study protocol caused by changes of consortium structure. The final assessment strategy was consented with the principal investigator and the rest of the consortium.
Regular meetings and reviews with the Berlin study nurse were carried out in order to clarify and communicate appropriate use of the developed health-economic questionnaires since study initiation.
The health-economic data assessment strategy of the RIMINI trial is separated into three major steps:
1. Assessment of Quality of Life (QoL) at baseline, months 1, 3, 6, 12, 18 and 24 and subsequent computation of Quality-Adjusted Life Years (QALYs)
2. Assessment of patient-specific resource consumption at initial discharge following transplantation, repeated discharges (if occurred) and regularly at months 3, 6, 12, 18 and 24
QALYs are currently being assessed in the RIMINI trial using the generic EQ5D-5L questionnaire, which is a standardized and well accepted instrument in the scientific community of health-economics. A working instruction, specifically designed for the RIMINI trial, was developed and uploaded to internal BIO-DrIM website, in order to make the questionnaire as well as the working instruction available for all participating study nurses and other trial staff.
In order to precisely capture other generic aspects of quality of life (qol) which are not covered by the EQ5D-5L instrument, the implementation of the SF36 questionnaire – 10 Authors: Simon Weber, Dr. Malte Pietzsch (Cellogic GmbH) another well accepted qol-questionnaire – was discussed and agreed. The SF36 questionnaire and a corresponding working instruction were uploaded to the internal BioDrIM website, in order to make them available for all participating study nurses and other trial staff.
As suggested by guidelines on health-economic data assessments in clinical trials (Edwards et al., 2008; Petrou & Gray, 2011; Ramsey et al., 2005; Sullivan et al., 2005), the disease-specific questionnaire KTQ-25 was proposed to the PI and agreed to implement.
Furthermore, a trial specific health-economic questionnaire was developed to capture individual health-care resource consumption, during the follow-up of each enrolled patient. The development of this questionnaire had to be well balanced in terms of efforts to be undertaken by the trial staff to assess the needed data in patient-interviews and the precision of expected results. This questionnaire as well as a corresponding working instruction was developed in close collaboration between the PI and Cellogic GmbH and both documents were uploaded to the internal BioDrIM website, in order to
make them available for all participating study nurses and other trial staff.
Regular checks of the assessed health-economic trial data have been defined in close collaboration between the PI, KKS-Charité (CRO), who is responsible for data management and storage, and Cellogic GmbH. Every three months, the complete trial dataset of all health-economic electronic case-report forms (eCRF) of all included patients are submitted to Cellogic GmbH and checked for completeness and consistency. In case of systematically missing data or inconsistent values, adequate measures, e.g. intensified briefing and training of study nurses, are taken to prevent poor data quality and more missing data in future assessments. The results of a recent completeness-check were presented at the last annual meeting on February 2nd, and are shown in figure 29.
Figure 29. Results of completeness-check (RIMINI), Data extraction date: November 22nd, 2017
WP5 Mechanisms behind successful minimizing immunosuppression (IS)
Work done by the Oxford group
The aim of this study is to generate a clinically-relevant mouse model that reflects the variability in alloresponses seen in clinical transplantation. We sought to develop a model that would facilitate the assessesment of the differential efficacy of treatments in “high responders” (patients who suffer severe and/or frequent rejection episodes) versus “low responders” (patients at a lower risk of rejection episodes ant potentially liable to over-immunosupression). Ultimately, we hope that the results of these studies would inform optimal treatment strategies for patients stratified into these different groups.
As the basis for this pre-clinical model, we used a humanised mouse model of skin allograft rejection developed in our lab, which has previously informed clinical trials in cell therapy1,2 . This model represents a robust immunological response to allogeneic tissue, most resembling that of “high responder” recipients. Thus, a modification of this model was required in order to adequately represent the “low responder” phenotype.
Historical reports suggest that a strong risk factor for early graft rejection is a high frequency of memory lymphocytes3 . Memory lymphocytes are generated by maturation of effector lymphocytes during the course of antigen encounter and persist for many years even following clearance of the antigen. Upon re-exposure to antigen, memory cells have a lower activation threshold than naïve cells and mediate a more robust immune response4. Transplant recipients, particularly older patients or who have had previous antigen exposure through previous transplantation, blood transfusion, or pregnancy, may therefore have specific anti-donor memory cells. Memory-type responses may also occur as a result of antigen receptor cross-reactivity known as heterologous immunity5,6.
Due to the limited in utero exposure to environmental antigens, umbilical cord blood (CB) contains minimal frequencies of memory T cells, as confirmed by early studies7,8. As such, we assessed CB as a candidate model for “low responder” transplant recipients. Conversely, with increasing age and exposure to environmental antigens, adult peripheral blood (AB) becomes enriched for memory lymphocytes generated during previous antigen encounters. Hence, we chose to use adult blood to model “high responder” recipients.
Using a combination of in vitro functional and molecular analyses, we explored the properties of CB and AB T cells. After confirming that these populations exhibit phenotypes suited to represent differential responses to antigen, we assessed the ability of cells from CB or AB to reject an allograft in a humanised mouse model of skin graft rejection. Finally, we obtained preliminary results from an in vivo experiment assessing the efficacy of anti-TNF9 as a treatment to suppress graft rejection in “low-” and “high- responder” models.
Methods overview
Phenotypic and functional characterisation of cord blood and adult peripheral blood T cells
Phenotypic characterisation of cells after adoptive transfer in vivo
In vivo testing of cord blood and adult peripheral blood graft rejection capacity
Results
Phenotypic characterisation of cell populations
Cord blood T cells retain a naïve phenotype after TCR stimulation
Consistent with previous reports, freshly-isolated CB T cells expressed levels of CD45RO protein comparable to that of adult CD45RA+ T cells, and significantly lower than that of adult T cells sorted as CD45RO+. After polyclonal stimulation in vitro for one week, adult CD45RA+ cells upregulated CD45RO expression to a density comparable to that of adult CD45RO+ cells on day 0. CB T cells retained lower expression of CD45RO Likewise, five days after adoptive transfer into immunodeficient mice, very low frequencies of CD45RO+ T cells were detected among CB T cells, compared to AB T cells. Together, these data suggest that CB T cells be more resistant to TCR-signalling-induced maturation, which supports the application of CB cells as a model of “low responders”.
CB T cells express lower levels of apoptosis-inducing molecule CD95
Freshly-isolated CB T cells (both CD4+ and CD8+) expressed lower levels of the cytolytic effector molecule CD95 (Fas) than AB T cells, suggesting that CB T cells are less able to exert cytolytic activity and may be in an earlier state of differentiation.
CB CD4+ T cells express lower levels of the IL-7/IL-15 receptor subunit CD122
As measured by flow cytometry, the density of cell surface CD122 protein expression was significantly higher in freshly-isolated AB CD4+ T cells than among their CB counterparts. The expression of CD122 among CD8+ T cells was not significantly different between CB and AB. This suggests that CB CD4+ T cells may be less responsive to the T cell maturation factors IL-7 and IL-15, which would further support the use of CB cells to model the alloimmunity of “low responders”.
Figure 32. CD122 expression in CB and AB T cells.
MNCs isolated from cord blood (CB) and from peripheral blood of healthy adult donors (AB) were stained with antibodies against CD122. Median fluorescence intensity by flow cytometry, gated on CD3+CD4+ (a) or CD3+CD8+ (b), is plotted for each donor with mean +/-SD for 3 donors per population.
CB and AB T cells express similar levels of costimulatory molecules
Memory T cells exhibit a lower threshold for antigenic stimulation, and a lesser dependency upon costimulation, than naïve T cells. We measured the expression level of two key costimulatory molecules on CB and AB T cells, to assess whether differential expression of these molecules might indicate any differences in sensitivity to costimulation. CB and AB CD4+ T cells expressed comparable cell surface protein levels of costimulatory molecules CD27 and CD28. Meanwhile, CD8+ T cells from AB expressed lower levels of these costimulatory proteins than those in CB. However, our functional assays (below) suggest that the higher costimulatory molecule expression in CB CD8+ T cells does not confer upon these cells more potent activity.
Regulatory T cells are present at lower frequency in cord blood than adult blood
The proportion of regulatory T cells (Tregs) among CD4+ T cells was lower in cord blood than adult peripheral blood. With this regulatory cell population enriched in adult PBMCs, it is possible that there is enhanced Treg-mediated suppression of effector cell function in AB cells.
Figure 34. Treg frequency in CB and AB T cells.
MNCs isolated from cord blood (CB) and from peripheral blood of healthy adult donors (AB) were stained with antibodies against CD3, CD4, FOXP3 and CD127. Percentgae of FOXP3+CD127lo cells among CD3+CD4+ lymphocytes was enumerated by flow cytometry, and plotted as mean +/-SD for 3 donors per population.
Functional characterisation of cell populations
In vitro proliferative capacity is comparable between cord and adult T cells
After polyclonal stimulation in vitro for three days, the proliferation index of CB T cells and of sorted adult CD45RA+ and CD45RO+ T cells was not significantly different. Meanwhile, the level of transcription of CCND1, a gene encoding the cell cycle initiation protein Cyclin D1, was not significantly different between these three cell populations. There were also no significant differences in CCND1 levels between the CD45RA+ and CD45RO+ subpopulations of CB and AB, although a non-significant trend towards higher expression in CB CD45RO+ cells is apparent. These results indicate that CB T cells are capable of proliferating in response to TCR stimulation.
Figure 35. in vitro proliferation of CB and AB T cells in response to polyclonal TCR stimulation
(A) CD45RA+ and CD45RA-CD45RO+ cells were selected from AB MNCs by magnetic bead-based selection. These cells, alongside total CB MNCs, were stained with violet proliferation dye and cultured for 72hrs with αCD3α CD28-coated beads. Dilution of the proliferation dye was measure by flow cytometry and used to calculate a division index (mean number of cell division cycles per progenitor cell). Mean alues +/-SEM are plotted for three donors with at least three technical replicates. (B) total CD3+ (“T”), CD3+CD45RA+CD45RO- (“A”) and CD3+CD45RO+CD45RA- (“O”) T cells were sorted by FACS from freshly-isolated CB and AB MNCs. RNA isolated from these cell opulations was examined by qRT-PCR for transcription of the cell cycle gene CCND1. The mean value of two replicates is plotted for each cell donor, with lines linking autologous samples.
Upregulation of activation markers upon stimulation is lower in CB T cells
Overall, there is a trend toward lower expression of activation markers in CB T cells relative to their equivalent AB populations, although these differences are not statistically significant between CD45RA+ and CD45RO+ subsets for HLA-DR. These results suggest that CB T cells may have a lower activation potential.
CB T cells secrete minimal levels of cytokines in response to TCR stimulation
Cytokines were generally secreted at much higher concentrations by adult antigen-experienced T cells than by either CB T cells or adult naïve T cells. Whilst adult naïve T cells secreted more after stimulation than in their resting state, CB T cells failed to secrete the majority of cytokines even when stimulated. Of particular interest are the cases where CB T cells fail to produce a cytokine that is secreted by adult naïve T cells, such as IL-2 and TNFα. These results may indicate a functional deficit in CB T cells that prohibits a robust response by these cord blood cells against alloantigen.
Figure 37 cytokine secretion by CB and AB T cells in response to polyclonal TCR stimulation
(A) CD45RA+ and CD45RA-CD45RO+ cells were selected from AB MNCs by magnetic bead-based selection. These cells, alongside total CB MNCs, were cultured for 72hrs with αCD3α CD28-coated beads. Supernatents were analysed using a flow cytometric bead assay to determine the concentration of (left to right, top to bottom) IL-2, IL-4, IL-5, IFNγ, IL-4, IL-9, IL-10, TNFα, IL-13, IL-17A, IL-17F and IL-22. Concentrations of each cytokine are plotted as the mean of three replicates for each donor +/- SEM across three donors.
Establishment of in vivo model
CB MNCs cells reject allogeneic skin with delayed kinetics relative to AB PBMCs
Skin grafts were rejected with a median survival time of 30.5 days in mice reconstituted with AB and 83.0 days in mice reconstituted with CB. The differential rejection kinetics mediated by adult and cord cells demonstrates that these two cell populations can be used to model the more rapid rejection seen in “high responder” recipients and the milder, delayed rejection seen in “low responder” recipients, respectively. In both groups, MNCs were depleted of CD4+CD25+ Tregs pre-injection.
Testing of a putative therapy (anti-TNF) in an in vivo model of high and low responsiveness
Our data indicate that there may be some prolongation of graft survival in the presence of anti-TNF treatment among mice reconstituted with AB MNCs. This experiment also appears to validate the model, given that CB-reconstituted mice have failed to reject their skin grafts with the rapidity of AB-reconstituted mice.
Conclusions
From the data presented here, we conclude that cord blood and adult peripheral blood T cells exhibit differential immunological responses to alloantigen both in vitro and in vivo. These distinctive responses may therefore represent the immune phenotypes of “low responder” and “high responder” transplant recipients, respectively. Indeed, our finding that CB MNCs cells mediate delayed graft rejection in a humanised mouse model of skin allograft rejection confirms that these cells are less responsive to allogeneic stimuli than adult PBMCs and thus provide an adequate model of “low responder” recipients. We suggest that this pair of models could be used in parallel to assess putative therapies in order to examine the differential efficacy or dosage requirements in low versus high responder transplant recipients. Our in vivo experiment assessing anti-TNF treatment as an intervention in the low and high responder models is a useful preliminary investigation of this therapy in a humanised transplantation model.
References
1. Nadig, S. N. et al. in vivo Prevention of Transplant Arteriosclerosis by ex vivo Expanded Human Regulatory T Cells. Nat. Med. 16, 809–813 (2010).
2. Issa, F., Hester, J., Goto, R., Nadig, S. & Goodacre, T. E. Ex vivo -expanded human regulatory T cells prevent the rejection of skin allografts in a humanised mouse model 1. 90, 1321–1327 (2010).
3. Valujskikh, A. & Lakkis, F. G. In remembrance of things past: Memory T cells and transplant rejection. Immunological Reviews 196, 65–74 (2003).
4. Farber, D. L., Yudanin, N. A. & Restifo, N. P. Human memory T cells: Generation, compartmentalization and homeostasis. Nat. Rev. Immunol. 14, 24–35 (2014).
5. Adams, A. B. et al. Heterologous immunity provides a potent barrier to transplantation tolerance. J. Clin. Invest. 111, 1887–1895 (2003).
6. Heuvel, H. Van Den, Heidt, S., Roelen, D. L. & Claas, F. H. J. T-cell alloreactivity and transplantation outcome: A budding role for heterologous immunity? Current Opinion in Organ Transplantation 20, 454–460 (2015).
7. Han, P., Hodge, G., Story, C. & Xu, X. Phenotypic analysis of functional T-lymphocyte subtypes and natural killer cells in human cord blood: relevance to umbilical cord blood transplantation. Br J Haematol 89, 733–740 (1995).
8. Garderet, L. et al. The umbilical cord blood alphabeta T-cell repertoire: characteristics of a polyclonal and naive but completely formed repertoire. Blood 91, 340–346 (1998).
9. Feldmann, M. & Maini, R. N. TNF defined as a therapeutic target for rheumatoid arthritis and other autoimmune diseases. Nat Med 9, 1245–1250 (2003).
Work done by the Nantes group
Global aims of the project
There is a high interest today to use animal models in order to understand mechanisms of tolerance and rejection in transplantation but the major difficulty is to find the”best” model. Indeed, most of studies are performed using naive rodents, breeding in pathogen-free facilities. In order to achieve one of the goal of the BIO-DrIM project, i.e. developing rodent models closer to the “immunity reality”, we wished to develop an in vivo model of acute rejection mediated by memory T cells. The aim was then to evaluate the efficacy of cell therapy or new immunosuppressants in these models. We decided to work in parallel in rats and mice (Part II of this report). Moreover, we wished to identify potential new biomarkers of tolerance and chronic rejection using our own rat models (Part I of this report).
Part I: Definition of new biomarkers
A part of this project consists on the identification of new biomarkers of chronic rejection and tolerance. To achieve this goal, we performed microarrays on heart allografts providing either from tolerant rats (following a 20 days-LF15-0195 treatment, a relevant model that mimics the B cell accumulation observed in blood from operationally-tolerant patients)1 or from chronic rejected rats (following donor blood transfusion). Amongst the 110 genes differentially expressed in allograft from tolerant and chronic rejected rats, we studied more precisely two of them, CLEC-1 and LIME. From blood analysis of these two groups of rats, we suggest that CLEC-1 could be potentially used as a blood diagnostic biomarker of chronic rejection. Conversely, the expression of the LIME biomarker was detected in rats that will develop chronic rejection; we then propose LIME as a blood prognostic biomarker of chronic rejection.
In the future, it will be interested to analyze the expression of these two genes in blood samples from transplanted patients in order to validate or not these results observed in our rat models.
Part II: Rodent models of acute rejection
A. First experiments to set up a model of acute rejection based on the literature:
To define an animal model of acute rejection mediated by memory T cells, we first used two different approaches described in the literature in rodents. The first one consists on an intraperitoneal injection of donor splenocytes two weeks before transplantation2. Using this model, the authors demonstrated an acceleration of the skin graft rejection in mice. In accordance with the project, we injected donor splenic T cells (total splenocytes were used in the article) two weeks before transplantation. By this system, no acceleration of allograft survival was observed either in a mouse skin allograft (Figure 40) or in a rat cardiac allograft models.
Figure 40. IP injection of donor T cells does not contribute to accelerated graft rejection in skin graft models in mice
C57BL6 female mice received IP injection of CD90+ T cells from male mice two weeks before male skin graft. A. Design of the experiment B. Graph depicts skin graft survival curves. Control mice received skin graft in the absence of IP injection of cells.
We then tested a second protocol, based on a sensitization of recipient animals by a first graft3,4. As we would like to focus only on the effects of memory T cells, we could not use the same recipient for the two grafts because the first one would induce memory T cells and memory B cells. So, we decided to do a first skin graft on few animals, isolate T cells from these animals one month post-transplantation and then injected these T cells in new recipient animals by intravenous route one week before transplantation. In mice, by performing skin grafts for sensitization and analysis steps, no graft acceleration was obtained. In the rat model, the first recipients (Lew1W) received skin allograft from Lew1A rats. Recipient rats were then sacrificed and their T cells were injected in naive Lew1W rats in order to sensitize them prior to a cardiac Lew1A allograft a week after T cell injection. Using this protocol, we obtained a slight acceleration of heart rejection in sensitized animals compared to non-sensitized animals (rats receiving only cardiac allograft) (Figure 41).
Figure 41. IV injection of T cells from sensitized animals slightly accelerates graft rejection in heart graft model in rats
Lewis1W rats were sensitized with donor antigen by receiving skin graft from Lewis1A rats. One month later, rats were sacrificed and T cells were collected and IV injected to a new Lewis1W recipient, seven days before Lewis 1A heart transplantation. A. Design of the experiment B. Graph depicts heart graft survival curves. Control rats received heart graft in the absence of injection of cells.
B. Improvement of the rat model of acute rejection and evaluation of Molecule X effect
In order to improve the rat model of heart rejection, we set up a new model of accelerated graft rejection following adoptive transfer of cells obtained from polygrafted animals. In addition to the new skin graft on sensitized rats, several modifications were done in this new model: i) the donor/recipient combination, in order to increase the immune response in sensitized animals, ii) the delays between grafts and sacrifice on first animals and the delay between cell injection and heart transplantation, in order to preserve the cells of interest, and iii) enrichment of T cells were replaced by total immune cells of draining LNs. In this new model, Lewis 1A rats received a first skin graft from Lewis 1W rats. One week later, a second graft from the same donor was performed on the same recipient rats. Rats were sacrificed 7 days later to isolate graft-draining LN cells. Analysis of these draining LN cells showed that cells from polygrafted mice displayed a memory phenotype and were able to induce a stronger alloresponse than cells isolated from naive rats, both in terms of T cell proliferation and T cell activation. To evaluate the in vivo role of these cells, graft-draining LN cells from polygrafted animals were intravenously injected to naive recipient Lewis 1A the day before Lewis 1W cardiac transplantation. Interestingly, the adoptive transfer of draining LN cells from polygrafted animals induced a more reproducible acceleration of allograft rejection (Figure 42).
Figure 42. IV injection of immune cells from highly sensitized animals accelerates graft rejection in heart graft model in rats
Lewis1A rats were sensitized with donor antigen by receiving 2 skin grafts from Lewis1W rats. One week later, rats were sacrificed and draining lymph nodes (DrLN) cells were collected and IV injected to a new Lewis1A recipient, the day before Lewis 1W heart transplantation. A. Design of the experiment B. Graph depicts heart graft survival curves. Control rats received heart graft in absence of injection of cells.
In parallel, works from our center recently reported that a new molecule (Molecule X, requested patent) was able to delay graft rejection in non-immunized rats. Administration of this molecule in our accelerated heart rejection model showed that it was also able to slightly prolong graft survival in this model. In order to study the properties of these memory cells, we next decided to move to a model of immunodeficient rats in order to accentuate their effects.
C. Exacerbation of the acute rejection model using immunodeficient rats
The use of immunodeficient mice is well described in the literature5 and allows to efficiently characterize the role of immune cells. In order to obtain a good extinction of the immune cells, genes coding for Rag2 and for IL2Recepteur γ were knock out in these immunodeficient mice. Based on these models, Rag1-/- IL2Recepteur γ-/- rats (RRG) were generated in the laboratory on Sprague Dawley genetic background. These rats were obtained by the breeding of Rag1-/- rats and IL2-Recepteur γ-/- rats. The absence of T cells, B cells and NK cells were validated in these rats, as observed in mice. We also observed that transplanted RRG animals do not reject an allograft, confirming that the immune system was deficient.
However, our results showed that the adoptive transfer of naïve LN cells from immunocompetent rats induces heart rejection in these immunodeficient rats. More interestingly, the injection of graft-draining LN cells from polygrafted immunocompetent animals (obtained as described in the previous section) allows a significant accelerated graft rejection compared to the transfer of naïve LN cells (Figure 43). Our first analyses performed on RRG animals receiving either sensitized LN cells or naïve LN cells from immunocompetent rats do not allow to explain the mechanisms of action of these sensitized animals.
Figure 43. IV injection of immune cells from highly sensitized animals accelerates graft rejection in heart graft model in RRG rats
Sprague-Dawley rats were sensitized with donor antigen by receiving 2 skin grafts from Lewis1W rats. One week later, rats were sacrificed and draining lymph nodes (DrLN) cells were collected and IV injected to RRG recipient rats, the day before Lewis 1W heart transplantation. In controls, RRG rats received IV injection of LN from naïve (non-immunized) Sprague Dawley rats. A. Design of the experiment. B. Graph depicts heart graft survival curves.
In conclusion of this part, we set up a robust model of accelerated heart rejection in RRG rats. In the future, we want to identify the cells responsible for these accelerated rejection. For that, we will in depth phenotype naive and sensitized LN cells and investigate whether a specific marker of a key population exists to understand this immunization. We will also perform in vitro assays of these two types of LN cells to study their proliferative capacities and their cytokine releases. Following injection of these two types of cells in RRG rats, we will also in-depth screen the engrafted cells in blood (T cell subsets). Lastly, we will evaluate the potential of new molecules and tolerogenic/regulatory cells in this model.
Part III: Experiments in immunodeficient mice in collaboration with the Oxford team
In accordance with the scientific advisory board the aims of WP5 for the last two years of the BioDRIM consortium have been modified. Oxford, Nantes and Berlin propose to work together to develop a humanised mice model of « Experimental system to investigate ‘high responder’ recipients who have pre-existing donor reactive memory cells ».
The first aim of this new project consists in setting up in Nantes the model of human skin graft in humanized mice according to the Oxford protocol.
In this model, NOD/SCID/IL-2Rγ-/- (NSG) mice were grafted with human skin graft (obtained from burns and plastic surgery unit of Nantes Hospital) and then received 5 millions of fresh human PBMCs by IV route (5 weeks later). We observed that these mice reject human skin with a mean rejection time of 30 days. This model was thus successfully set up in Nantes. We then investigated whether this rejection was also possible following the injection of frozen human PBMCs. Our results showed that the injection of 5 millions of frozen PBMC also allows human skin graft rejection (Figure 44A). In the two experiments performed from different donors of skin and PBMCs, the mean time of graft rejection was very similar whereas the engraftments were different between both experiments (Figure 44B). Interestingly, thawed PBMCs were able to efficiently induce both graft rejection and GVHD development. Administration of immunoregulatory cells appears today as an attractive strategy to control graft rejection. For safety reasons, the use of autologous cells appears particularly promising and the use of thawed PBMCs allows to induce graft rejection in humanized mice and also to generate and inject imunoregulatory cells in an autologous way.
Figure 44. Thawed human PBMCs efficiently induce graft rejection in a model of human skin graft in humanized mice
NSG mice received human skin graft and were injected with either fresh or thawed PBMC 5 weeks later. A. Human skin graft survival was monitored in these groups of mice from two different experiments. Four mice (2 in each experiment) were sacrificed before graft rejection due to a weight loss >20% of initial weight (GVHD development) B. The engraftment of human cells was measured in blood of mice receiving skin graft and thawed PBMCs from two different experiments.
In relation with the new results from the Oxford team, we then investigated whether the injection of umbilical cord blood cells induces a less vigorous rejection response in vivo than the injection of PBMCs. For that, umbilical cord blood cells were obtained from the maternity unit of Nantes Hospital and freezed. Following thaw-out, the results of our first experiment show that umbilical cord blood cells induce graft rejection and blood cell engraftment. However, we obtained similar time of graft rejection following administration of PBMCs and umbilical cord blood cells. Consequently, our experiment with umbilical cord blood cells in Nantes does not reproduce Oxford datas. To explain these differences, it has to be considered that we performed only one experiment so far, it then has to be reproduced. Furthermore, we obtained a high mortality in umbilical cord blood cells following thaw-out. A phenotype of these cells has to be done to investigate whether these cells are modified following thaw-out. Lastly, it is important to note that these differences could be also explained by the strains of immunodeficient mice used. Indeed, we used NOD/SCID/IL-2Rγ-/- whereas BALB/c Rag2-/-/IL-2Rγ-/- were used in Oxford.
References
1. Le Texier et al. 2011. Long-term allograft tolerance is characterized by the accumulation of B cells exhibiting an inhibited profile. Am. J. Transplant., 11: 429-438
2. Greenlaw et al. 2008. An antibody combination that targets activated T cells extends graft survival in sensitized recipients. AJT, 8: 2272-2282
3. Wang et al. 2007. Inhibition of terminal complement components in presensitized transplant recipients prevents antibody-mediated rejection leading to long-term graft survival and accommodation. JI, 179: 4451-4463
4. Ramsey et al. 2013. Anti-LFA-1 or rapamycin overcome costimulation blockade-resistant rejection in sensitized bone marrow recipients. Transplant. Int., 26: 206-218
5. Shultz et al. 2007. Humanized mice in translational biomedical research. Nat Rev Immunol. 7, 118130.
Potential Impact:
The focus of BIO-DrIM was to contribute to an improved and cost-effective long-term outcome of solid organ transplantation by implementing decision-making biomarkers into the clinical management of transplant patients (personalized IS) but the results are more broadly relevant also to other ancillary disciplines and have very important exploitable potential. At this aim, three research performing SMEs and two big companies were involved in the project with very deep commitment.
The novelty of this project, that can be considered overall an innovative approach to an important unmet need and a research infrastructure, is the stratification of transplanted patients regarding their individual immunological responsiveness to the allograft and their respective individual need of IS. A second area of novelty is the integrative design of this program, whereby a direct comparison (feasibility, safety, cost and promise of effect) of biomarker driven strategies for personalized therapies is foreseen in 5 innovative investigator-driven biomarker clinical trials designed by the consortium.
The outputs and results that we have generated during this project have a strong potential for further commercial exploitation, with the expectation that personalized therapy can ultimately optimize the need for immunosuppressive drugs in organ transplant recipients.
The most important impact of BIO-DrIM are the first steps and useful results that will drive the scientific community to the personalization of immunosuppressive therapy in organ transplant recipients, thereby significantly improving their survival and quality of life, while at the same time decreasing health care costs.
Secondarily, the results of BIO-DrIM broadly impact the treatment of other diseases related to undesired immune reactions, such as autoimmune diseases and graft-versus-host disease after hematopoetic stem cell transplantation. In addition, it might have also impact on the new field of cell transplantation (e.g. adult differentiated cells, like hepatocytes; but also allogeneic stem-cell derived cell products); in other words diseases with chronic or repetitive IS treatment.
Beside the results that we are confident that will directly impact on the patients in the short-medium time, the BIO-DrIM Consortium generated important outputs that can be summarised as below:
-Fostering of small SMEs, which due to the project were able to gain on visibility, develop new products, increase sales, increase employment numbers and interact with big pharma
-BIO-DrIM project as a prototype of new infrastructure (results and data from previous EU funded project could be exploited in many interacting field and continue to be in use)
-New approach of designing clinical trials (i.e. Simon design)
-Development of a new type of eCRF which includes biomarkers that could be transferred to other clinical trial that were funded by the German Federal Ministry of Education and Research (BMBF)
-Validated and robust biomarker assays established at different platforms & training on biomarker performance
-Implementation of health economic an HPQL assessment
-Experiences with new regulatory processes for EU multicenter trials (Voluntary Hamonization Process VHP)
List of Websites:
www.biodrim.eu
The central focus of the BIO-DrIM project is the implementation of biomarker-driven strategies for personalizing immunosuppression (IS) in order to improve the long-term outcome and to decrease the adverse effects (graft toxicity, diabetes, cardiovascular events, opportunistic and community acquired infections, bone loss, and malignancies) and costs of chronic IS in solid organ transplant patients. The mission of BIO-DrIM reflects the new strategy in the field of solid organ transplantation: leaving the path of one-size-fits-all weaning strategies and administration of new drugs in favour to a more personalized approach.
The biomarker-guided management of immunosuppressive therapy is an ambitious program that started time ago with other projects (IOT and RISET) in which a set of promising biomarkers was identified and developed, but the methodical validation was missing. Implementation of these biomarkers into the clinical routine requires multiple cross platform tests, high quality standardization, multicenter clinical validation and with this an international network. All these requirements meet up in the BIO-DrIM project, that includes 5 innovative clinical trials (~2.000 patients in screening phase and ~800 patients enrolled in trials).
More in detail, the BIO-DrIM project addresses several highly innovative issues:
- First IS withdrawal study in long-term liver transplant patients based on the presence of a molecular tolerance signature (personalized withdrawal – study 1);
- Two novel trials on systematic partial IS withdrawal in selected highly stable long-term kidney transplant patients for validation of decision making biomarkers to detect operationally tolerant patients (personalized withdrawal - studies 2+3);
- First controlled biomarker-driven perioperative stratification of kidney transplant patients into low/high responders to prevent high-dose standard IS in low responders (personalized minimizing IS – study 4)
- Novel approach to target selectively activated (allospecific) memory/effector T cells (to increase the proportion of low responders after kidney transplantation suitable for low-dose monotherapy – validation by biomarkers – study 5
Safety is a key issue of the clinical trials: all studies are performed according to the international rules and all drugs are used according to their applications.
In addition to the well-validated set-1 biomarkers used for decision-making, all trials are accompanied by a panel of standardized biomarkers (set-2a) and a panel of exploratory biomarkers (set-2b):
i) to further improve the predictive value of the biomarkers for patient stratification
ii) to learn more about the mechanisms behind success/failure of IS minimizing.
Gaining knowledge about the mechanisms behind successful weaning (regulation/effector balance) is also part of the project. Well-defined in vitro and experimental mouse/rat transplant studies are used to address questions regarding the mechanisms of success/failure of IS minimizing. In particular, the scientists within BIO-DrIM address the interaction between regulatory pathways and donor-reactive memory/effector T cells.
The research and trials in BIO-DrIM are strongly translational. The involvement of SME´s and pharmaceutical industry guarantees a fast commercialization of promising product candidates.
Furthermore, health-economic studies push forward the translation into products used at the market as reimbursement strategies can be early developed and discussed with the health insurance companies and the government authorities.
Project Context and Objectives:
FIVE MULTICENTER CLINICAL STUDIES
LIFT Study (WP1a; biomarker-based weaning of liver transplant patients): Biomarker-guided IS weaning strategy in long-term (>3yrs) liver transplant recipients by identification of operationally tolerant patients
Previous weaning studies have demonstrated that about 25-40% of selected liver transplant patients after year 3-6 post transplantation develop operational tolerance and can have their maintenance immunosuppression discontinued. However, in the majority (60-75%), standard immunosuppression weaning fails and they develop rejection. The biomarker being developed and validated by the BIO-DrIM consortium, which is a gene expression test using liver tissue, is key to accurately stratify patients before considering immunosuppression withdrawal.
The LIFT study, a biomarker-driven trial coordinated by KCL, involves long-term stable liver transplant patients who are randomized to immunosuppression weaning without taking into account the result of the biomarker (Arm A) or to biomarker-guided weaning (Arm B). The trial is currently open in 12 centres in 4 EU countries and now, there are 190 consented patients, which are or have undergone screening, and 105 of them have been randomised. The first interim analysis of the trial took place in 2018 and confirmed the clinical utility of the 5-gene transcriptional biomarker. The final analysis of the primary outcome is expected to be completed in 2021.
WEANING trial (WP1b; weaning of long-term stable kidney transplant patients): Targeted partial/complete weaning of standard IS in long-term stable kidney transplant patients characterized as low-responders and identified as putative “operationally tolerant” by recently developed biomarker panel.
The fully blinded WEANING trial, coordinated by ITUN (Nantes), aimed at improving the graft function in clinically selected Highly Stable (HS) patients following complete weaning of calcineurin inhibitor (CNI). In December 2014, the group decided to stop the study due to insufficient patient recruitment. The results of the WEANING Study were published in 2016 (Am J Transplant 2016; 16: 3255–3261).
GAMBIT study (WP1c):
the group of KCL defined and validated a new gene expression signature that is independent of drug effects and also differentiates tolerant patients from healthy controls (cross-validated area under the receiver operating characteristic curve [AUC] = 0.81). In a prospective cohort, it was demonstrated that the new signature remained stable before and after steroid withdrawal. In addition, a validated and highly accurate gene expression signature, that can be reliably used to identify patients suitable for IS reduction (approximately 12% of stable patients), irrespective of the IS drugs they are receiving was reported. The results of the studies were published in December 2016 (Am J Transplant. 2016; 16: 3443–3457, PMID: 27328267).
Data obtained will be merged for ontology analysis to define the best final test format.
CELLIMIN trial (WP2; Prospective donor-specific cellular alloresponse assessment for Immunosuppression minimization in de novo renal transplantationtrial using a biomarker as stratificator for IS).
The CELLIMIN trial has been approved by the Voluntary-Harmonization Process (VHP). CELLIMIN is a non-inferiority trial with enrichment design aiming to demonstrate the utility of the IFN-γ ELISpot marker for the stratification of kidney transplant recipients into “low” and “standard-of-care” IS regimen.
All clinical sites participated in extensive lab and interlab comparision to validate the robustness and correctness of the IFN-γ ELISpot assay, the biomarker used as stratification tool.
184 patients have been screened with pre-transplant donor-specific IFN-γ ELISpot. 102 (55.43%) displayed a negative result and therefore underwent randomization. After randomization, 49 (48%) of patients received the “low” immunosuppressive regimen and 53 (52%) received the “High” immunosuppressive regimen. At 30th November 2018, 94 patients have achieved 12-month of follow up, 74 6-month follow-up and 16, 3-month follow-up. Despite not having achieved the number of patients pre-specified in the trial, the relevant number of transplant recipients recruited in this study is already unique and thus, will allow for a relevant evaluation of the impact of using a novel biomarker to stratify patients in different immunosuppressive regimens. Also, the analysis of patients recruited but not randomized to different immunosuppressive regimens, this is, those showing a positive donor-specific IFN-γ ELISpot will enhance the understanding of the impact of cellular alloimmunity in the context of kidney transplantation.
We will conduct a preliminary analysis of the whole recruited study population by next April 2019, when the last recruited patient will have been randomized. We will analyze the primary end-point with a minimum follow-up of six months after transplantation. Furthermore, we plan to perform the final analysis with the total number of patients randomized by next October 2019, when all patients will have achieved 12 months of follow up.
RIMINI trial (WP3, Tacrolimus after rATG and infliximab induction immunosuppression): Shifting kidney transplant patients to low-responders suitable for early IS minimization
RIMINI is an international multicenter open-label single-arm Simon’s two-stage Phase II clinical trial aiming to estimate confidence interval for the observed efficacy of the induction regimen with rATG and infliximab and a go/no go rule for further clinical development. A total of 75 patients will receive the proposed induction regimen. To introduce IS-based on tacrolimus/steroids as early as possible without losing control of acute/chronic rejections would be of great benefit and could reduce adverse effects and costs. However, this is only possible in a minority of patients yet. Memory/effector T cells are a major challenge. Therefore, increasing the pool of low-responders by selective targeting effector/memory T cells would have a big advantage.
Rationale: T-cell depletion is not sufficient to control allo-memory. ATG/Thymoglobulin treatment leads to a reduction of the clonal size of donor-reactive memory/effector T cells. Antigens, present in vivo, induce a bias of lymphopenia-induced proliferation LIP (preferential expansion of surviving alloreactive, Ag-specific Tem/Teff), which is an undesired effect. Here the combination of Thymoglobulin/Infliximab will help to reduce freshly activated naïve Tem/Teff to enlarge the pool of patients suitable for low IS.
Using tolerance/rejection biomarker monitoring we aim to prove shift from high responders to low responders while using infliximab as co-induction agent to target recently activated memory/effector T cells.
The initiation meeting was held in Berlin, on November 15th 2016. Enrolment into the RIMINI trial started in January 2017. At the end of 2018 56 kidney transplant recipients were enrolled into the RIMINI trial, 33 from Prague, 19 from Berlin and 4 from Barcelona. Thus, the recent enrolment rate is 75%. Not all intended clinical partners are currently enrolling patients for the RIMINI trial, which proved to be the main obstacle to reaching the expected numbers of patient enrolment. The principal reason is the enrolment into concurrent CELLIMIN trial. The study enrolment is still open and 3 participating centres enrol patients and will continue to do so beyond the end of the BIO-DrIM consortium. This is critical as only 25% patients (n=20) need to be enrolled into the trial to reach target population. Full set 12-months data evaluation is thus delayed and expected on June 2020 as we anticipate last patient to be enrolled in the trial in June 2019. Recently 20 patients completed the study with 12 months follow-up. 7 patients prematurely terminated the study. Recent number of efficacy failures (defined primary endpoint of the study) is 20 which anticipates acceptable efficacy and safety of the protocol and suggests non inferiority to the standard of care regimen.
VALIDATING SET-1 BIOMARKER TESTS READY FOR DECISION MAKING FOR THE ON-SITE PATIENT STRATIFICATION WITHIN THE CLINICAL TRIALS DESCRIBED IN WP1-3
Validated biomarkers (set-1) – as decision-making markers applicable for guiding minimizing IS:
- Molecular “tolerance signature” in liver biopsy of liver transplant patients
- Molecular “tolerance signatures” in peripheral blood of kidney transplant patients
- Donor-reactive IFN-γ ELISpot
The markers were selected in our previous studies performed in the EU funded projects IOT, RISET. The tests are well methodical validated now and already applied in the ongoing clinical trials.
Establishment of the ELISpot technique as stratification tool clinical trial.
10 ELISpot Readers were installed by Gen-ID at each local laboratory in the clinical sites of Berlin, Barcelona, London, Amsterdam, Regensburg, Hamburg, Prague, Nantes, plus Santander and Oviedo. Establishment at the local sites took place by extensive technology transfer, combined with training sessions and technical support.
VALIDATING RECENTLY ESTABLISHED SET-2 BIOMARKERS AND IMPLEMENTING NEW BIOMARKER CANDIDATES FOR IMPROVING PERSONALIZED IS WITHIN THE BIO-DRIM CLINICAL TRIALS
Methodically validated biomarkers (set-2A) – applicable for prospective validation concerning their clinical value:
- Molecular signatures in urine of kidney transplant patients
- Gene expression signatures diagnosing balance of innate and adaptive immune responses
- Immune cell subset composition profiling (Multiparameter flow cytometry)
New biomarkers (set-2B) – markers applicable for retrospective validation: e.g. quantification of new immune cell cell subsets (DC subsets, T-cell subsets, Treg subsets) applying our standardized multiparameter flow cytometry. Using our established platform, we have implemented such markers into the trials of BIO-DrIM for exploring their putative value in retrospective analyses.
Set-up and validation of Mulitparameter Flow Cytometry
Technical advice and equipment were provided and installed by Beckman Coulter - Immunotech at the clinical sites of six European countries. More precisely, 10-color Navios cytometers were provided by Beckman Coulter - Immunotech to the clinical sites of Berlin, Nantes, London, Barcelona, Prague, Amsterdam, Regensburg and Hamburg,
In a collaborative effort between Beckman Coulter – Immunotech and the Charite antibody panels in a ready-to-use dry format (DuraClone) were established, tested and validated. These panels are now commercially available.
ANALYSING THE HEALTH-ECONOMIC IMPACT OF BIOMARKER-GUIDED PERSONALIZED IS
Health-economic data demonstrated the usefulness of implementing biomarkers into the management of IS (personalized IS) concerning the cost/benefit ratio, performing health-economic analyses by using Micro Costing. The method used attempts to measure costs and benefits of service as accurately as possible, by including all fixed and variable costs of care at local prices, given the institutional structure within which service and care are being given. It also takes into account country-specific rules, such as fixed prices for diagnostic procedures (point system), drugs, and in/out-hospital service.
STUDYING THE MECHANISMS BEHIND SUCCESSFUL WEANING
The groups from Oxford, Nantes and Berlin worked together to develop a humanized mouse model of « Experimental system to investigate ‘high responder’ recipients who have pre-existing donor reactive memory cells ». This project contains two aims. The first one is to set up and validate in the three centres the human skin transplant model in humanized mice using the protocol developed by the Oxford group. Human immunoregulatory cell types have been administrated to humanized mice in this human skin transplant model to evaluate the efficacy of these cell types to control graft rejection. The second aim, performed by the Oxford group is to set up a new experimental system to investigate ‘high responder’ recipients who have pre-existing donor reactive memory cells in humanized mice. Oxford, Nantes and Berlin teams investigated the potency of their different human immunoregulatory cell types to control graft rejection in ‘high responder’ recipients using this model.
In parallel, Nantes group continued their work on the potential of new therapeutics to control accelerated heart rejection in rats, a model recently developed by this team.
DISSEMINATING THE RESULTS TO SCIENTIFIC, PATIENT AND PUBLIC COMMUNITY (AND DEVELOPING COMMERCIALIZATION STRATEGIES BY PARTNERING WITH SMES/INDUSTRIES
The aim is spreading the knowledge and the results generated by the project to wider community of end users and to the scientific community. The results have been shared with the different communities and commercialization strategies have been put in place (diagnostic products, drug and biomarker diagnostic combinatory products, novel targets of IS drugs). GenID developed and meanwhile market several EliSpot IVD kits e.g. EBVSpot, CMVSpot and TransSpot as outcome of the project. The QC kit GenID established together with the partners is a fixed size in offering standardized controls in routine use. Moreover GenID set up a sample bank with different donors to offer this in clinical trials. These kits are CE IVD marked according European guidelines.
With the experience and the obtained expertice from BIO-DrIM project to run clinical multicenter studies with EliSpot, GenID decided to set up a GLP / CRO lab to offer this service to pharmaceutical companies.
Due to the project, GenID has hired two more scientists and secures the jobs of 25 people employed in production and development. In addition, another position in Regulatory Affairs is planned for 2019.
Sales in the EliSpot area (especially EBV and CMV kit) have increased continuously over the last few years by more than 15 % and (rouphly two million) account for more than half of GenID's sales.
For Milenia Biotec the BIO-DrIM project served as a networking platform, resulting in a Co-Development-Agreement with the big pharma for the development of tests to be used within the area of transplantation.
Regarding Beckman Coulter – Immunotech, the collaboration within the BIO-DrIM consortium has led to the commercialization of the 1st multicolor panels in the DuraClone format. Hence, a model of reagent development at Beckman Coulter has been built following the successful collaboration with the BIO-DrIM. Collaborators are directly involved in the development of reagents which led to the commercialization of nine additional DuraClone products following this approach.
Thus, an effort has been devoted to implementation activities in support of the SMEs and industrial partners involved for understanding the early prospective commercialization strategies for the biomarkers of the Consortium. As consequence a good collaboration between the groups ended up with exploitable results.
Project Results:
WORK PACKAGE 1: Targeted IS withdrawal in long-term transplant patients identified as “operationally tolerant”
Study I: Use of tolerance biomarkers to predict immunosuppression withdrawal in liver transplant recipients
During the EU funded projects “Indices of Tolerance” (IOT) and “RISET” it was demonstrated that intra-graft gene expression markers accurately predict the outcome of immunosuppression withdrawal, which is influenced by the time and age of the transplanted patient.
Based on the insights gained from these two projects, it has been possible to go further and design a study with the overall objective to assess the clinical utility and risk/benefit ratio of employing a liver tissue-based transcriptional test of tolerance to stratify liver recipients prior to immunosuppression withdrawal.
Multi-centre, prospective, open label, non-controlled/non-randomized, interventional cohort phase IV study of immunosuppression withdrawal in stable liver recipients. (personalized withdrawal – LIFT trial)
PI: Alberto Sanchez-Fueyo, King’s College, London, UK
LIFT is worldwide the first randomized, multicenter study in long-term liver transplant patients to assess the clinical utility and safety of biomarker-guided immunosuppression, based on the presence of a molecular tolerance signature. The main goal is to determine the diagnostic accuracy of a liver tissue transcriptional profile of tolerance.
Secondary goals: to determine the correlation between tolerance and clinical parameters, iron status, blood gene expression, immunosenescence markers, microRNA, gut microbiota.
Figure 1: LIFT clinical trial.
Sample size/Patients: 148 patients to be recruited at 12 European clinical sites (King´s College London; Royal Free London; Cambridge; Leeds; Birmingham; Newcastle; Edinburgh; University Hospital Leuven; Hannover Medical Centre; Berlin Charité; Barcelona, St Luc Brussels).
Since the beginning of the trial (October 2015), we consented 191 patients, which are or have undergone screening, and 105 of them have been randomised. Recruitment rate is now within the target, but due to the delay in initiating the trial and opening some of the sites, there are plans for extending the enrolment period until May 2019. We are adding 2 new centres to help with the recruitment (Dublin and Palermo).
The breakdown by the centres is given below:
Site Consent Randomised Ineligible % ineligible Reasons for ineligibility Withdrawn consents Awaiting randomisation Awaiting biopsy or histological evaluation
King’s College Hospital 59 29 23 39 Histological- 23
4 1 2
Royal Free, London 18 7 10 56 Histological-10 1 0 0
Leuven 16 10 5 31 Histological- 5 1 0 0
Royal Infirmary of Edinburgh 9 8 0 0 0 1
Barcelona 23 14 3 13 Histological -3 2 4
Addenbrookes, Cambridge 12 7 5 42 Histological-5
0 0
Freeman Hospital 13 8 5 38 Histological- 5 0 0
Berlin 10 5 5 50 Histological- 5 0 0
Leeds 8 4 3 38 Histological- 3 0 1
University Hospitals Birmingham 2 2 0 0 0 0
Hannover 14 7 5 36 Histological-5 1 0
Brussels St Luc 7 4 3 43 Histological-3 0 0
Total 191 105 67 35% 6 4 9
Furthermore, there have been 27 SAEs for the following reasons: Vasovagal episode following liver biopsy, abdominal pain; pain after liver biopsy; post-biopsy liver bleed (2); jaundice, liver dysfunction / Biliary Anastomotic stricture & severe pancreatitis; Bullosus Pemphigold; tonic clonic seizure; rectal bleeding (2); herpes zoster; graft dysfunction; cholangitis (2); hepatocellular carcinoma; diverticulitis of the sigmoid colon; post-transplantation due to NASH cirrhosis; common bile duct calculi; lung carcinoma (2); non-ST elevation Myocardial Infarction with troponin rise of 1169; small bowel obstruction; infected right foot plantar wound; gastroenteritis; fall at home with brief loss of consciousness; fever + E.coli sepsis; salivary gland calculus and biliary tree dilatation.
As of December 2018, there were 11 fully weaned participants and 38 rejections. 21 participants were weaning off the IS and 31 participants were on the maintenance arm of the trial.
Meetings of the Data Monitoring Ethics Committee (DMEC) are taking place regularly (every 3 months). The second version (V2) of the statistical analysis plan is approved by the DMEC and the DMEC charter is fully signed. There is a draft version of the SAP that is being reviewed by the DMEC (V2.1) to be approved at the next meeting.
An electronic Case Report Form was created and has been functioning well.
A minimisation algorithm is implemented for the treatment group assignment.
The primary endpoint will be the successful discontinuation of immunosuppression with stable liver biopsy and liver tests 12 months after IS withdrawal.
Ancillary studies will be:
• HRQoL and Health Economic analyses
• Mechanistic studies:
Influence of iron status
Changes in regulatory T cell populations
Influence of T cell epigenetic imprinting
Influence of gut microbiome
Role of anti-HLA antibodies
The impact of IS withdrawal on quality of life will be evaluated using disease specific questionnaires. Sequential biological specimens will be collected to conduct ancillary mechanistic studies.
The study design of LIFT, the second biomarker-guided trial of the project (prospective randomised marker-based trial to assess the clinical utility and safety of biomarker-guided immunosuppression withdrawal in liver transplantation), has been subject to a series of modifications, resulting in a push-back in the initiation of the trial. There is a further one-year follow-up for each patient after completion of the withdrawal process. The first interim analysis took place in November 2018, with 37 patients reaching the primary endpoint. We plan to have the second interim analysis in May 2019 (n=56). The secondary outcomes will be evaluated at 12 months post completion of weaning.
Study II: A France prospective randomized double-blind, multicentre parallel controlled study of CNI weaning in selected long-term kidney transplant patients
The clinical group from Nantes conducted for the very first time a placebo controlled-double blinded withdrawal study in long-term kidney transplant patients.
Primary objective is the improvement of renal function (glomerular filtration rate >=7 ml/min). Secondary composite objectives are acute and chronic rejection, death, graft lost, anti HLA antibodies, de novo proteinuria. Ancillary studies are the assessment of low and high immunological risk biomarkers (DNA chip, Phenotypes).
Figure 2: WEANING trial.
The study is divided into two stages: the phase prior to the recommendation of the independent scientific committee and the phase after the recommendation. The committee recommended the enrolment of another 6 patients but with strict adherence to the original inclusion criteria. This was done after the BIO-DrIM annual meeting of January 2014. Following the concept of being “as safe as possible”, the inclusion and follow-up criteria were adjusted as follows:
At inclusion
• DSA <1000
• Anti-HLA <2000
• Age>50y
• 4 to 15y after transplantation
• Prograf and Cellcept or myfortic +/- CS
During the follow-up
• if DSA >2000, control 15 days after
• 15 days after:
• If DSA increase > 3000, biopsy
• If DSA do not increase, no biopsy and control 1 month after
• If proteinuria appears or function declines, biopsy
These new criteria have been submitted to the national ethical committee. They have all been accepted.
In total 16 patients were enrolled (9% of patients with pre-inclusion criteria). 10 were randomised, while 6 patients could not be randomized based on the screening results. 9 patients finished the protocol (4 without CNI weaning and 5 with CNI weaning).
All CNI weaning patients went back on full immunosuppression after having finished the weaning protocol. In two of the five weaning patients, weaning was stopped due to subclinical rejection at two thirds of the protocol.
The two patients have been treated (subclinical rejection).
All the other patients keep a good renal function, with no proteinuria and Abs disappeared.
In December 2014, the group decided to stop the study due to insufficient patient recruitment.
Since January 2014, 254 patients were screened, but with the restriction to rigorously stick on the inclusion criteria and with the change from the Anti-HLA Luminex screening to the more sensitive LABScreenT Single Antigen identification only one patient could be included in the study.
This patient was not randomized at day 30 since he did not tolerate the high doses of cellcept.
The biological studies are ongoing and the group has biological material available for the BIO-DrIM consortium.
It has become clear that tacrolimus withdrawal after years of treatment must be avoided even in highly selective long-term stable kidney recipients. It looks like the long-term tacrolimus treatment prevented any tolerance induction.
The results of the WEANING Study were published:
Dugast E, Soulillou J-P, Foucher Y, Papuchon E, Guerif P, Paul C, Riochet D, Chesneau M, Cesbron A, Renaudin K, Giral M & Brouard S. Failure of Calcineurin Inhibitor (Tacrolimus) Weaning Randomized Trial in Long-Term Stable Kidney Transplant Recipients. Am J Transplant 2016; 16: 3255–3261.
Biological material and data from this important trial are available for the consortium. Other demands for materiel cession have been done outside the BIO-DrIM network. The consortium is aware of these transfers. Transfer agreement are established by ITUN.
Study III: Using Biomarkers of Tolerance to guide immunosuppression weaning in kidney transplant recipients
Following the observation of a serious confounding effect of the immunosuppression regimen on gene selection and the performance of gene-expression signatures based on comparisons between already established untreated tolerant patients and treated stable patients, the team has finalised the development of an immunosuppression-free signature comprising 28 genes. The validation in a Fluidigm platform resulted in the selection (with elastic net) of a parsimonious signature of 9 genes, showing no material difference in performance to the best-performing signature comprising 19 of the original 28 genes. The results of our studies were published in December 2016 (PMID: 27328267). However, there is still an ongoing debate in the kidney transplant community on the safety of the clinical application of biomarkers of tolerance. Therefore, proceeding with the originally planned Study III design, as a safety and feasibility pilot Phase II trial for initiating IS minimization in patients who score ‘tolerant’, would have been unethical at this stage, without extensive further validation. Consequently, the team at KCL proceeded with further validation of the immunosuppression-free signature in RT-qPCR platform, which would be the one applicable to clinical use. The team has also embarked on a detailed exploration of the impact of adjustment for immunosuppressant drug intake on the selection of stable treated patients as potentially “tolerant”. The results of these studies are in preparation for publication.
WORK PACKAGE 2: Biomarker-guided stratification into low/high responder after kidney transplantation
The coordinating centres of the clinical study are Barcelona (O. Bestard / J. Grinyó) and Berlin (P. Reinke).
The Study CELLIMIN (Prospective donor-specific Cellular alloresponse assessment for Immunosuppression Minimization in de novo renal transplantation) is an international multicentre open label randomized non-inferiority Phase IV clinical trial for selection of low anti-donor T-cell responders using the IFN-γ ELISPOT assay as biomarker for patient stratification to safely receive long-term drug minimization based on tacrolimus monotherapy.
Since the start of the project, the design of the clinical trial was a main endeavour of this work-package. The CELLIMIN is a non-inferiority trial with enrichment design aiming to demonstrate the utility and IFN-γ ELISPOT marker for the stratification of kidney transplant recipients into “low” and “standard-of-care” IS regimen. The trial design has undergone major modifications from the original intention in response to serious ethical concerns.
The original intention was for a “biomarker by treatment” interaction design randomised controlled trial with a primary objective to demonstrate the clinical utility of IFN-γ ELISPOT assay pre-transplantation to stratify non d-s T-cell allo-reactive patients. Randomisation to maintenance IS with TAC monotherapy at low trough levels and “standard-of-care” high IS regimen separately in ELISPOT positive and negative patients were intended. The expectation was that ELISPOT negative patients at “low” IS would have 20% lower rate of BPAR at 12 months post-transplant (30% vs 10%) when compared to ELISPOT positive patients receiving low IS (Figure 3).
Figure 3: Original version of CELLIMIN clinical trial.
However, the risk associated with the administration of the “low” IS regimen in ELISPOT positive patients was considered unacceptable, which led to a modification of the trial strategy to a biomarker-based design randomized trial with one control group and the following treatment regimen allocations:
- GROUP A STANDARD OF CARE: the result of ELISPOT test will be kept blind and all patients will be treated following a “High” Immunosuppression regimen (Standard of care immunosuppressive regimen based on TAC (achieving 4-8ng/ml trough levels), MMF (1gr bid) and steroids (according to KDIGO guidelines).
- GROUP B BIOMARKER BASED STRATEGY: the investigator will know the result of the ELISPOT and patients will be treated as follows:
B.1: POSITIVE pre-TX ELISPOT or high responders (>25 spots/300.000 PBMC) will be treated following a “High” Immunosuppression regimen (Standard of care).
B.2: NEGATIVE pre-TX ELISPOT or low responders (<25 spots/300.000 PBMC) will be treated following a “Low” Immunosuppression regimen (based on TAC monotherapy to achieve 8-10 ng/ml trough levels during the first 4 weeks after transplantation and 6-8 ng/ml thereafter, MMF (1g bid) during the first 7 days post-transplant and stopped thereafter) and steroids (tapering until discontinuation on month 2 post- transplant).
All patients will homogenously receive 2 doses of Basiliximab (day 0 and day 4 after transplantation).
In order to test non-inferiority of arm B compared to arm A, assuming 15% BPAR in the latter, with a 15% increase non-inferiority limit, 194 patients would have been needed, randomized at a ratio 1:1, to achieve 90% power and 5% type I error rate (a total of 213 patients would have been enrolled, accounting for a drop-out rate of 10%).
Further concerns from the Voluntary harmonization procedure regarded the 15% limit for non-inferiority margin, which was considered not acceptable as the total expected the BPAR (biopsy-proven acute rejection) rate in the “low” IS group would have been too high. In response, the non-inferiority margin was adjusted to 10%, while having to considerably increase the number of patients to enrol. This would have created limitations of the marker based design, introducing a bias towards non-inferiority due to the large proportion of patients receiving the “standard-of-care” treatment (including the high-responders in the marker-based group B and the entire Control group A), creating a ZERO contrast in a great portion of the trial. Another drawback would have been the limited benefit for the patients of enrolling marker positive patients into the study.
In extensive collaboration with the statisticians from London and with the powerful support of the KKS (CRO), the study design was reviewed and modified for re-submission with a revised hypothesis:
Using the ELISPOT assay to assess whether negative donor-specific T-cell alloreactivity as a biomarker based immunosuppression (IS) will be non-inferior with respect to BPAR (T-cell mediated) rate at 6 months post-transplantation, comparing kidney transplant recipients randomized 1:1 to receive either low (TAC monotherapy) or high IS (Standard of Care triple therapy).
The re-submission of the study in November 2014 was successful and the trial received VHP (201446) approval by the European Commission in February 2015.
Figure 4: New hypothesis of the CELLIMIN Trial
The final strategy is an enrichment design where only ELISPOT negative patients will be randomized in a ration 1:1 to “low” and “high” (standard-of-care) regimens.
The current primary objective is to demonstrate the utility and safety of the INF-γ ELISPOT assay for the stratification of kidney transplant recipients into “low” and “high” IS regimens. This (enrichment) study will test non-inferiority of “low” IS regimen compared to “high” IS regimen, assuming 10% of BPAR (TCMR) at 6 months in the control group, and allowing a non-inferiority limit of maximum 10%.
The secondary objectives will be to investigate differences across treatment arms at 6 and 12 months post-transplantation in:
– eGFR
– Prevalence of biomarkers of tolerance/hyporesponsiveness
– Prevalence, type, severity, treatment and outcome of BPAR
– Prevalence, type, severity, treatment and outcome of 3-mo subclinical rejection
– Prevalence of death and graft loss
– Prevalence of metabolic and cardiovascular co-morbidity
– Prevalence of patients that remain MMF and steroid free
– Prevalence of acute and chronic hystologic lesions (Bannf’11 score) in protocol biopsies at 3 and 12 months post RT
– Prevalence of patients that remain on therapy
– Distribution of patients in distinct chronic kidney disease stages
– Health economics, H-R QoL and treatment cost (cost/benefit) at 1, 3, 6, 12 and 24 months
Subsequently, applications for obtaining approval of the trial by the local Ethics committees and national competent authorities were undertaken in 7 European Kidney transplant centres: Barcelona, Regensburg, Berlin, Prague, Nantes, London and Amsterdam.
After this initial approval, a first amendment was submitted in order to clarify some aspects of the trial and correct typographic errors. This first amendment was approved in Spain on March 2016.
A second amendment was submitted to add new information about MMF toxicity. Approval was obtained in Spain on 19/09/2016.
In December 2015, the first patient was recruited in the trial.
During 2017, two additional kidney transplant centres joined the consortium to participate as recruiting centres. During the first 6 months a thorough “on-side” validation of the standard operating procedures (SOP) of the IFN-γ ELISPOT assay was undertaken.
PATIENT ENROLLMENT:
The recruitment of patients begun on January 2015 and was notified to our competent authority.
Up to now, 184 patients have been screened with a pre-transplant donor-specific IFN-γ ELISpot done, and 102 (55.43%) were negative and thus were randomized. This observation confirms previous retrospective data regarding the prevalence of T-cell alloreactivity as means of preformed circulating donor-reactive memory/effector T cells
PROBLEMS FACED:
The unexpected delay on having the last trial finalized due to the previously mentioned reasons, was a main hurdle that prolonged the initiation of the study. Additional important problems faced that slowed the recruitment of patients in all centres were the significant increase of donors after cardiac death, the high percentage of extended criteria donors, the impossibility to perform the study tests during weekends and the increasing number of sensitized kidney transplant candidates in the waiting lists. These exclusion criterias for patient recruitment significantly impacted on a more dynamic inclusion of patients in the CELLIMIN trial.
WORK PACKAGE 3: Shifting kidney transplant patients to low-responders suitable for early IS minimisation
RIMINI is an International multicenter open-label single-arm Simon’s two-stage Phase II clinical trial aiming to estimate confidence interval for the observed efficacy of the induction regimen with rATG and infliximab and a go/no go rule for further clinical development. A total of 75 patients will receive the proposed induction regimen, accounting for a drop-out rate of 10%. RIMINI was approved by the relevant authorities via VHP process.
Using tolerance/rejection biomarker monitoring we aim to prove shift from high responders to low responders while using infliximab as co-induction agent to target recently activated memory/effector T cells.
Figure 5: RIMINI trial design.
The coordinating centres of the clinical study are Praque (O. Viklicky) and Berlin (P. Reinke). Clinical partners enrolling patients: CHARITE (Berlin), ITUN (Nantes), ICS-Hospital (Barcelona), IKEM (Prague), AMC (Amsterdam), UHREG (Regensburg), UKE (Hamburg)
As presence of donor-specific memory/effector T cells represents a major challenge for -minimizing immunosuppression, the aim of the study is to selectively target alloactivated effector/memory T cells.
Rationale: T-cell depletion is not sufficient to control allo-memory. ATG/Thymoglobuline treatment leads to a reduction of the clonal size of donor-reactive memory/effector T cells.
Antigens, present in vivo, induce a bias of lymphopenia-induced proliferation (LIP, preferential expansion of surviving alloreactive, Ag-specific Tem/Teff), which is an undesired effect, not useful in presensitized patients.
Therefore, targeting of recently activated Teff cells by anti-TNF mAb (Infliximab /Adalimumab) would be desirable. Infliximab/Adalimumab blocks soluble TNF and targets also mTNF on the cell.
Based on these rationales, RIMINI is designed to be a phase II, single-arm trial with 12 months of follow up.
The new induction protocol consists of: Thymoglobuline 1.5 mg/kg (Day 0), Thymoglobuline 1.5 mg/kg (Day 1), Remicade (Infliximab) 5 mg/kg (Day 2). The maintenance immunosuppression is based on Tacrolimus, administered at dose 0.1 mg/kg before surgery and then at dose 0.2 mg/kg (initiated day 1) and prednisone 20 mg (or appropriate dose of metylprednisolone) tapered down to dose 5 mg by day 7.
Figure 6: RIMINI medication scheme.
The following biomarker analyses are implemented in the trial:
- ELISpot/CTLp
- EBV/CMV/BKV load + CMV/EBV T-Ly
- Multi-parameter flow cytometry
- gene expression profiling
- alloantibodies
- urinary IP-10
- HO-1 polymorphisms
- histology (protocol/induced biopsies)
Primary endpoint:
1) Clinical response to the induction determined by the absence of any of the following outcomes up to 12 months post transplantation (start of follow up at transplantation): acute rejection, graft loss or poor graft function defined as eGFR<40 ml/min.
Secondary endpoints:
1) Prevalence of biomarker signatures at 6 and 12 months
2) Incidence of death by 12 months post-transplantation
3) Incidence of graft loss by 12 months post-transplantation
4) Incidence of metabolic and cardiovascular co-morbidity by 12 months post-transplantation (post-transplant diabetes mellitus, dyslipidemia, hypertension)
5) Proportion of subjects who remain on monotherapy at 12 months post-transplantation in the respective study groups
6) Incidence of acute and chronic lesions assessed by the Banff 07 score in protocol biopsy at 12months post-transplantation
7) Incidence of discontinuation of study treatment
8) Overall safety of study therapy immunosuppressive regimen defined as viral and bacterial infections, malignancies and autoimmunity
9) eGFR using CKD-EPI formula at 12 months
10) Health-related quality of life using EQ5D-5L and SF-36v2 questionnaires at baseline, M1, M3, M6, and M12
11) Assessment of patient-specific resource consumption using a trial specific questionnaire at initial discharge, M3, M6, M12, and in cases of repeated hospitalization
The biomarker analyses (e.g. whole blood staining and acquisition for flow cytometry) and sample collections / preparations (e.g. PBMC preparation for ELISpot) is performed in accordance to the same SOPs as filed for CELLIMIN, in order to ensure reproducibility and also comparability of the results between the two clinical trials.
The initiation meeting was held in Berlin, on November 15th 2016. Enrolment into the RIMINI trial started in January 2017. At the end of 2018 56 kidney transplant recipients were enrolled into the RIMINI trial, 33 from Prague, 19 from Berlin and 4 from Barcelona. Thus, the recent enrolment rate is 75%. Not all intended clinical partners are currently enrolling patients for the RIMINI trial, which proved to be the main obstacle to reaching the expected numbers of patient enrolment. The principal reason is the enrolment into concurrent CELLIMIN trial. The study enrolment is still open and 3 participating centres enrol patients and will enrol patients even beyond the end of the BIO-DrIM consortium. This is critical as only 25% patients (n=20) need to be enrolled into the trial to reach target population. Full set 12-months data evaluation is thus delayed and expected on June 2020 as we anticipate last patient to be enrolled in the trial in June 2019. Recently 20 patients completed the study with 12 months follow-up. 7 patients prematurely terminated the study. Recent number of efficacy failures (defined primary endpoint of the study) is 20 which anticipates acceptable efficacy and safety of the protocol and suggests non inferiority to the standard of care regimen.
The final analysis will be performed with per-protocol data set. One-sided upper 95% confidence interval will be calculated for the observed efficacy failure rate. Descriptive statistics will be presented for the individual outcomes comprising the primary outcome and for the secondary outcomes. Secondary binary outcomes will be presented as proportions with two sided 95% confidence intervals. Continuous secondary outcomes following normal distribution will be summarized by mean and standard deviation. Continuous secondary outcomes deviating from the normal distribution will be summarized with a median and interquartile range. Minimum and maximum values will be presented for all continuous outcomes. Survival outcomes will be examined in Kaplan-Meier plots.
WP4 Biomarker analyses, Data management & Biostatistics, Health-economic analyses and IVD development
Immune Monitoring Subproject
Immune Monitoring subproject of the BIO-DrIM project, the CHARITE functions as Central Immune Monitoring Lab (CIML) with the following persons involved:
Head CIML: Birgit Sawitzki
Sample material acquisition: Katrin Vogt
ELISpot: Maik Stein (under supervision of Petra Reinke)
Flow cytometry: Mathias Streitz, Stephan Schlickeiser
The main tasks with regard to WP1 (Studies I, II and III) are activities such as the exchange of standard operating procedures (SOPs), the exchange of samples as well as centralized analyses of some assays e.g. Foxp3 demethylation (TSDR). Exchange of SOPs for the analysis of the following biomarker was performed:
1) whole blood flow cytometry of leukocyte subsets (n=6 flow panel)
2) whole blood intracellular Foxp3 staining by flow cytometry
3) IP-10 analysis in urine
4) whole qRT-PCR
In addition, Foxp3 demethylation (TSDR) was quantified in samples from operationally tolerant kidney transplant patients in collaboration with partners from Nantes (Study II; Braza et. al. JASN 2015).
The main task within WP2 and 3 is the coordination and performance of the immune monitoring.
For this purpose, it has been decided on the establishment of the following biomarker/assay portfolio:
Safety marker: viral load (EBV, CMV, BKV) – Research Diagnostics
“Decison making” marker: anti-donor ELISpot (IVD/Companion Diganostics - GenID)
“Efficacy” marker: Flow Cytometry (IVD - Immunotech)
qRT-PCR (IVD – no supporting company yet)
Foxp3 demethylation (TSDR) – Research Diagnostics
IP-10 (IVD – Milenia)
Adjustment and validation of ELISpot and flow cytometry received particular attention. This was done in close collaboration between the Charite as the Central Immune Monitoring Lab (CIML), GenID and Immunotech, respectively (see also detailed description of design, validation process and results further below within the respective paragraphs). The extensive validation process allowed us to extend the age-of-blood time to 12 hours for both assays, which increased the clinical applicability.
The then generated SOPs for standardized assay performance (ELISpot & whole blood flow cytometry) had been sent to the clinical sites and published within the eCRF. Inter-operator and inter-lab comparisons were performed to validate the robustness and correctness of the assays.
Inter-lab and especially inter-operator comparisons for the ELISpot (virus- and allo-specific) resulted in acceptable and low CVs ensuring robustness and applicability as a stratifying biomarker. Because of the test´s impact on the outcome of the clinical trial of WP2, inter-operator and inter-lab comparisons were and will be repeated on a regular basis.
In summary, we were able to implement two well-validated assays at all clinical sites. The technicians and scientists are intensively trained and now ready to perform the assay in strict adherence to the developed and refined standard operation procedure (SOP) for its use as an immune monitoring tool and – more importantly – as a prediction tool in the clinical trial of WP2.
The standardization of the flow cytometry has been a process developed on basis of the experiences from another FP7 project - the project “The ONE Study”. The aim was to obtain accurate and precise data with the least variation among different sites.
The standardization process included the following actions:
Consensual development of flow panels, development of SOPs, panel validation including extensive tests on the influence of “age of blood” and “age of stain”, establishment of reference standards and data analysis (for detailed information see paragraph further below).
Upon refinement of sample processing and staining procedures tests on the influence of “age of blood” revealed a low variability until 12h storage (see figure below), which was now set as a maximal age limit for the clinical trials performed within BIO-DrIM.
Figure 1: Absolute numbers of e.g. CD56+ NK cells, CD8+CD27- cytotoxic effector T cells and CD4+CD25highFoxp3+ regulatory T cells using the DuraClone-based antibody panels in 8 whole blood samples upon 0, 4h, 8h, 12h, 16h and 24h was determined.
Further advice on how to perform and analyse multi-parametric flow cytometry has been published in a book chapter (Schlickeiser S, Streitz M, Sawitzki B. Standardized Multi-Color Flow Cytometry and Computational Biomarker Discovery. Methods Mol Biol. 2016;1371:225-38. doi: 10.1007/978-1-4939-3139-2_15. PubMed PMID: 26530805).
We have also generated and published reference cohort data for healthy controls, which can be used to correct for age- and gender-dependent differences in immune cell subset composition (Kverneland AH, Streitz M, Geissler E, Hutchinson J, Vogt K, Boës D, Niemann N, Pedersen AE, Schlickeiser S, Sawitzki B. Age and gender leucocytes variances and references values generated using the standardized ONE-Study protocol. Cytometry A. 2016 Jun;89(6):543-64. doi: 10.1002/cyto.a.22855.).
In addition to the two tests described above, the consortium has analysed several other parameters either in central core laboratories or in on-site laboratories in order to explore the values of further biomarkers for stratification, safety and monitoring.
For this purpose, additional clinical samples have been collected in the clinical trials of WP2 and WP3. To ensure quality in sample collection, the following procedure was applied:
In agreement with WP leaders WP2- and WP3-specific sample collection schedules were designed (see also WP2 and WP3). All patient material for immune monitoring purpose needs to be labelled accurately. The patient IDs as designed by the CIML team specific for both clinical trials, CELLIMIN and RIMINI, were automatically generated by the eCRF and are always part of the label code.
Almost all samples were collected in pre-labelled sample collection tubes provided by the CIML team in pre-packed sampling bags. The bags are patient-specific and time point/event-specific. As some of the tubes have a short half-life, the bags were provided every 6 months according to the estimated patient enrolment.
ELISPOT analysis
As mentioned above, development and evaluation of ELISpot assays was done in close collaboration between the Charite CIML team and GenID.
The GenID main role within BioDrIM is to develop and evaluate with clinic partners IVD kits for different applications and distribution of these kits worldwide. In detail, GenID supplies established enzymatic EliSpot kits for the detection of IFN gamma (IFN-γ) release after allo- and antigen-specific stimulation. In general, the objective of GenID contribution in the project is the set-up, standardization and validation of EliSpot IVD kit and image analyzer for interpretation in routine use lead to personalized IS. The final aim is the approval of this system in Europe and worldwide (US).
In the following sections, the work and results of GenID within the consortium are described.
WP 2: Prevention of the high-dose standard CNI-based IS in low-responder kidney transplant recipients identified by perioperative patient stratification
This work package focuses on the clinical study on prevention of high-dose standard IS in low T-cell responder kidney transplant recipients by perioperative stratification of patient. GenID offers in this work package kits and antigens.
The kits provided for the CELLIMIN study and the corresponding antigens are shown in the following table:
Table 1: Number of provided EliSpot kits and antigens from 2014-2017
WP 3: Increasing the population of low-responder kidney transplant patients suitable for early minimization (low-dose monotherapy) by the recently explored selective perioperative targeting of alloreactive effector / memory T-cells
Up to now no contribution from GenID`s side.
IVD test development
For the validation, the standardization and implementation of the IFN-γ-EliSpot, it was necessary to monitor the transplant patients according to their pre-transplant frequency of donor-reactive memory / effector T-cells into low / high responder kidney transplant recipients.
For the preoperative identification of low-responder patients, the test had to be performed on-site.
On the basis of the high and stringent requirements on reproducibility and accuracy of the EliSpot technology inside the consortium the main tasks were:
• Designing standard procedure protocols
• Personal (wet) training
• Reader System evaluation
• Implementation of frequent controls to monitor the process at each side
After the workshop of Berlin, held in December 2013, following points had to be cleared:
- 2 generations of AID EliSpot readers in the centers
o Camera sensitivity + resolution
o Light intensity decrease can occur
o Yearly master plate check for light intensity comparison
- Technical questions
o Positive control switched from SEB to pokeweed mitogen
o For functional test only citrate blood
o Blood sample age maximum 8h
o No third party
o One pretested FCS batch
o One count setting for all EliSpot Reader Systems
- Assay performance for pre- and post-transplantation
- Crossvalidation
o SOP update
o Repeat training sessions together with GenID
o Defined counting setting and check with a standard plate for all centers
o Verifying the SOPs and count settings with same samples (n = 5)
▪ Cross check of all raw data at Berlin / Barcelona
Further approaches by GenID were determined and conducted:
1. Set up / calibration
o Definition of a QC-plate (e.g. Masterlot) in Barcelona / Berlin
o QC-plate has to be analyzed in each center
o Adjustment of each system via remote support
2. Definition of settings for counting
o Based on the results of the RISET study and planned cross validation, standard settings for counting had to be fixed
3. Definition of the QC frequency
o By sending sample around like external quality control
o By sending QC-Plate or other plate around
Transversal activities of Charité and GenID on the standardized EliSpot technique
Within the CELLIMIN trial of WP2, the IFN-γ EliSpot assay was used as an enrichment tool in order to select only low-responding, EliSpot negative patients for the study. Thus, proving the robustness of the assay is essential. This required close collaboration between the CHARITE and GenID. In a very meticulous work together with the Charite, all crucial working steps (see Figure 1) were evaluated with regard to the impact of the assay results.
Figure 1: Crucial work steps with possible impact on the test result
The same attention was given to the evaluation of factors with impact on the practical use of the assay.
One example: Since transplantation often takes place at night or during the weekend, it has been important to understand how much time could pass between time of transplantation and start of the assay. For this purpose, it was tested whether the outcome of the assay (spot numbers) changes with the age of the blood. The test results showed that fresh blood can be processed within 12 h (maximum) without changing the outcome of assay. Consequently, it was determined that start of the assay need to occur within 12 h after time of transplantation.
But not only the laboratory assay had to be optimized, the EliSpot reader systems had to be harmonized and standardized as well. The EliSpot reader system consists of an optical unit and specific software to analyze the EliSpot plates and count spots. Although all reader systems are identical in construction, it is not possible to exclude the existence of minor variations that could be the source of differences in the outcome. In order to eliminate such inter-EliSpot reader hardware / software variability, the same plates were analyzed by all centers. Via online remote support every reader system was adjusted in conformity to the results generated by the reader system in Berlin (= reference) and the new settings were saved and locked. A quality control (QC) procedure has been implemented to ensure compliance to these standards.
The EliSpot Reader systems were successfully installed including training at each side (9). To ensure same settings and prevent changing of fixed configurations, different user accounts were created. The Reader Systems have access to the internet for remote support and for upload of data to Berlin and (for data safety reasons) to an individual server at GenID facility by encoded export. There are three levels of users with specific rights:
Figure 2: Screenshots for the different user accounts with individual permission rights
User “AID” is the only supervisor with user management rights and is password-protected. The Password is securely stored at GenID to ensure only the software experts at GenID are able to change other user accounts if necessary.
User “BIO-DrIM” only has the rights to modify results. In software version 7.0b15714 GenID introduced a new button ("review") to ensure count results are confirmed. This feature also locks the plate for further modifications unless plate is "un-reviewed" again; all actions must be performed with user name and are password protected (electronic signature) and saved in the plate history.
All other users are “advanced users” which are allowed to modify settings only in their own profile.
For the BIO-DrIM user account, as first step a layout with specific count settings named BIO-DrIM has been created. Every new document that user BIO-DrIM opens has those settings which are applied for analysis.
Second, stage calibration has been performed and saved as “stage.ini” which is loaded automatically after user log-in. Camera settings were adjusted by remote support from GenID with the same “reference” EliSpot plates for each center. The aim was to calibrate similar well saturations (<3% deviation) leading to reproducible count results; those system-specific camera parameters were saved as AID defaults and are not changeable by the BIO-DriM user account.
Third, to implement a reader / software internal QC System, AID defaults were loaded in a newly established user account named “masterlot” and locked. This masterlot (QC) plate has been read and saved as reference for further QC runs.
Figure 3: Screenshots from the Masterlot user providing the QC system check
All following QC plate readings are compared to the reference plate values and deviations are displayed when performing the system QC test. In case the deviation is greater than 5% for light conditions (and therefore well saturations), the AID remote support should change AID default camera settings to adjust the system to reach same average well saturation as for the QC plate. Then these settings have to be applied to user account BIO-DrIM.
Recognition of the IFN-γ-EliSpot assay as a companion diagnostic tool
Furthermore, several steps have been taken towards implementation of the EAB´s recommendation to open the discussion with the FDA and EU authorities for the recognition of the IFN-γ-EliSpot assay as a companion diagnostic tool.
It was decided to go for two different approvals:
1. License to use the allo-reactive T-cell EliSpot as patient stratification tool in kidney and liver transplantation
2. License to use the EliSpot as monitoring tool (detection of viral infection - CMV, EBV, BKV).
For this purpose, it was decided to benefit from the close collaboration with the industry and the expertise of the members of the EAB and local users (biomarker core facility at the CHARITE / BCRT). All experts have been invited to a meeting, where the technique, its development process and data was presented with the aim to ask for scientific advice and to decide on how to classify the technique and how to present it to the authorities.
AID / GenID has been asked to start promoting the EliSpot as a standardized and marketable test, while defining the intended use and proving the medical need of the assay.
The Documents and forms for FDA submission were prepared for following scenarios:
- For the registration of the test kit “EliSpot for the Detection of Alloreactive Response” or “EliSpot for the Detection of Response of Viral pathogens”, it may be possible to claim a 510(k) regarding the test method used.
- It may be beneficial first to claim a 510(k) for the identical test kit with FDA by AID. Then after approval to amend the 510(k) registration for other antigens and immune response reactions
- The next steps of AID were to determine which pathway to follow and which argumentation is the best for a fast registration.
- In case the FDA does not agree to the 510(k) path, then the PMA (Pre Market Approval) path must be used.
Furthermore, the documents for HDE- pre-submit were prepared:
- Elements of intended use and indication of use,
Clear statement how and by whom the device is to be used
Description for what and for whom the device is to be used
- Define how the device is planned to be used in real-life setting
- Risk Analysis
- Proposed Study Designs
- Specimen information
- Analytical performance
- Method comparison
- Clinical performance
- Statistical analysis plan
For realization of the pre-submit, the following data were collected at GenID / AID:
Assay Performance Characteristics
The minimal detectable unit of response of an EliSpot Assay is one Spot. The limit of detection typically achieved can be up to 1 in 200.000 cells.
Because of natural variations of cell based assays, each control and analyte has to be performed in duplicates or better triplicates (like in this evaluation).
To avoid unspecific assay results the individual immunological constitution of each patient has to be taken into account. Because of that, negative and positive controls have to be performed for every single patient.
The positive control establishes that the assay worked by providing an expected positive result and is used for the documentation of an adequate number of inducible cells. In case of 200.000 cells / well, a minimum of 50 Spots should be achieved. To avoid overdeveloped wells, an additional positive control was performed with 40,000 cells / well.
The negative control is used to establish the unspecific baseline response of the assay. In case of 200.000 cells / well, a maximum of 10 Spots should be achieved. However, routine experience has shown that some patients show slightly higher spot counts dependent on their general individual immunological constitution, disease, medication or even nutrition.
During evaluation of assay results, spontaneous background cytokine secretion has to be taken into account to prevent false positive results. To determine the spontaneous background cytokine secretion, cells are incubated in the presence of cell culture medium without any further stimulating agents like serum or antibiotics. For the background compensation of assay results, the ratio of “analyte triplicate means” to “mean of negative control triplicates” has to be calculated (stimulation index [SI]).
These mean background compensated assay results were grouped into four categories:
• negative 0 – 2 SI
• low response 2 – 5 SI
• medium response 5 – 20 SI
• high response >20 SI
For very low spot counts of the negative control and non-reactive wells, the calculation of the CV% is not very meaningful due to nearly equal values of the mean and standard deviation, which leads to irrational high CV´s. In these cases the observation of the standard deviation is more relevant.
INTRA-Assay
INTRA-Assay precision was analysed from data generated by two operators. To determine INTRA-Assay Variation, each operator performed a total of eleven assays consisting of three replicate wells (triplicates) for all controls and antigens. All cells used for this evaluation were isolated at once from normal human whole blood of one donor and split directly before performing the assay.
Assay performance data was calculated with background compensated assay results of each operator separately. At first, the mean stimulation index (SI) of each of the eleven assays was calculated. After that, corrected sample standard deviation of background compensated assay results was calculated. At last the Coefficient of variation (CV%) was calculated as the ratio of corrected sample standard deviation to mean of background compensated assay results.
The following tables provide an overview of mean response as well as CV% and SD of eleven assays performed in a single run performed by two operators. These results indicate that INTRA-Assay precision is acceptably low and allows the analysis of clinical samples in duplicates.
Operator MS
Antigen Response CV [%] SD
NC negative (xˉ = 1,00)
0 0
PWM high response (xˉ = 26,81)
10,17 2,73
PWM 1:5 high response (xˉ = 29,12)
9,88 2,88
CMV IE1 negative (xˉ = 0,87)
18,23 0,16
CMV pp65 high response (xˉ = 33,81)
10,41 3,52
EBV lyt medium response (xˉ = 11,41)
12,07 1,38
EBV lat medium response (xˉ = 6,22)
11,80 0,73
Operator KS
Antigen Response CV [%] SD
NC negative (xˉ = 1,00)
0 0
PWM high response (xˉ = 29,70)
10,37 3,08
PWM 1:5 high response (xˉ = 32,36)
12,03 3,89
CMV IE1 negative (xˉ = 0,97)
25,60 0,25
CMV pp65 high response (xˉ = 35,23)
12,17 4,29
EBV lyt medium response (xˉ = 12,43)
10,60 1,32
EBV lat medium response (xˉ = 6,41)
15,00 0,96
Table 2: Results for INTRA assay variation for two individual operators (MS and KS). Antigens for specific T-cell stimulation were CMV (Cytomegalovirus) immediate early 1(IE1) and pp65 (tegument protein) and EBV (Epstein-Barr virus) lytic genes (lyt) and latent genes (lat).
INTER-Assay
INTER-Assay precision was analysed from data generated by five different operators using the same four samples, including three buffy coats (501-1 - 501-3) and fresh blood (NE), at three days. Each day a different plate lot was used. All cells used at one test day were thawed and overnight rested or isolated from normal human whole blood at once. All samples were split directly bevor performing the assay. Cells were counted by each operator separately. Assays were performed as three replicate wells (triplicates) for all controls and antigens.
Assay performance data was calculated with background compensated assay results between three test days of each operator separately. At first the mean stimulation index (SI) of each day was calculated. After that corrected sample standard deviation of background compensated assay results between three test days was calculated. At last the Coefficient of variation (CV%) was calculated as the Ratio of corrected sample standard deviation to mean of background compensated assay results.
The following table provides an overview of mean response as well as range of CV% and SD between three different kit lots used at three days obtained by five Operators for two samples (501-2 and NE as example:
Specimen 2 (501-2)
Antigen Response CV [%] SD
NC negative (xˉ = 0,56)
43,30 – 114,56 0,19 – 0,51
PWM high response (xˉ = 409,64)
10,32 – 31,73 47,23 – 110,51
PWM 1:5 high response (xˉ = 83,24)
20,01 – 59,21 16,25 – 42,02
CMV IE1 negative (xˉ = 0,63)
34,64 – 114,56 0,19 – 0,92
CMV pp65 negative (xˉ = 0,57)
20,40 – 91,65 0,08 – 0,51
EBV lyt negative (xˉ = 0,61)
35,25 – 91,65 0,25 – 0,51
EBV lat negative (xˉ = 0,37)
43,30 – 173,21 0,19 – 0,53
Specimen NE
Antigen Response CV [%] SD
NC negative (xˉ = 0,69)
0,00 – 100,00 0,00 – 0,38
PWM high response (xˉ = 276,04)
6,90 – 57,82 28,80 – 143,93
PWM 1:5 high response (xˉ = 206,23)
28,84 – 63,05 54,39 – 97,91
CMV IE1 negative (xˉ = 1,84)
0,00 – 141,95 0,00 – 1,58
CMV pp65 negative (xˉ = 0,93)
30,10 – 86,60 0,29 – 0,96
EBV lyt medium response (xˉ = 7,03)
4,58 – 67,04 0,51 – 3,24
EBV lat high response (xˉ = 20,28)
24,12 – 60,97 4,48 – 9,41
Table 3: Examples for INTER-assay performance from two donors out of five different cell preparations
One possible reason of higher CVs for INTER-Assay precision is the use of different PBMC cryovials and blood drawings on each day. Even though the PBMCs were from the same lot and the fresh blood from the same patient, sample-to-sample variability could explain the findings.
INTER-Operator
INTER-Operator precision was analysed from data generated by five different operators using the same four samples, including three buffy coats (501-1 – 501-3) and fresh blood (NE), at three individual days. Each day a different kit lot was used. To determine INTER-Operator Variation the same dataset used for the calculation of INTER-Assay precision was examined. Cells were counted by each operator separately. Assays were performed as three replicate wells (triplicates) for all controls and antigens.
Assay performance data was calculated with background compensated assay results between fife operators for each test day separately. At first the mean stimulation index (SI) of each operator was calculated. After that corrected sample standard deviation of background compensated assay results between the operators was calculated. At last the Coefficient of variation (CV%) was calculated as the Ratio of corrected sample standard deviation to mean of background compensated assay results.
The range that was obtained for high response (SI > 20) was between 5,75% - 69,15% CV (mean % CV = 39,12), medium response (SI 5 - 20) gave a range of 29,12% - 96,14% CV (mean % CV = 66,33), low response (SI 2 - 5) wasn´t observed and negative results (SI 0 - 2) gave a range of 0,00% - 166,83% CV (mean % CV = 62,05).
The following table provides an overview of mean response as well as range of CV and SD between five operators at three days for two samples (501-2 and NE) as example:
Specimen 2 (501-2)
Antigen Response CV [%] SD
NC negative (xˉ = 0,56)
55,90 – 94,79 0,30 – 0,51
PWM high response (xˉ = 409,64)
16,10 – 24,93 58,66 – 102,49
PWM 1:5 high response (xˉ = 83,24)
14,55 – 44,95 15,42 – 29,47
CMV IE1 negative (xˉ = 0,63)
56,45 – 124,72 0,33 – 0,62
CMV pp65 negative (xˉ = 0,57)
42,59 – 78,20 0,30 – 0,38
EBV lyt negative (xˉ = 0,61)
47,51 – 80,03 0,28 – 0,48
EBV lat negative (xˉ = 0,37)
56,84 – 102,26 0,16 – 0,53
Specimen NE
Antigen Response CV [%] SD
NC negative (xˉ = 0,69)
37,27 – 70,71 0,28 – 0,47
PWM high response (xˉ = 276,04)
14,29 – 65,60 50,86 – 140,51
PWM 1:5 high response (xˉ = 206,23)
13,51 – 46,00 25,36 – 68,15
CMV IE1 negative (xˉ = 1,84)
46,90 – 166,83 0,78 – 3,78
CMV pp65 negative (xˉ = 0,93)
40,75 – 85,67 0,37 – 0,84
EBV lyt medium response (xˉ = 7,03)
29,12 – 72,74 2,19 – 5,21
EBV lat high response (xˉ = 20,28)
5,75 – 47,51 0,95 – 7,78
Table 4: Example for INTER-operator performance for two cell preparations out of five used for the evaluation
Concordance analysis between two EliSpot Reader System
Comparison of Spot counts between two EliSpot reader systems was analysed from data generated by all experiments during assay validation. All Plates were read and counted by two EliSpot reader systems used in the CELLIMIN trial.
The scatter plot is used to display differences in spot counts of the two BioDrIM EliSpot Reader.
Each data point having the value of counts from ELRO7IFL1602236 determining the position on the horizontal axis and the value of counts from ELRIFL061001095 determining the position on the vertical axis.
An identity line of equality (y=x) is added to show 100% conformity. Linear regression of all 1984 pairs of values leads to a linear regression line with the equation y=1.026*x+0,06383 which slightly differs from the line of equality at lower counts.
Additionally Pearson´s correlation coefficient has been calculated. (r=0.994 R2=0.9988 P(two-tailed)=<0.0001)The Bland-Altman plot shows the difference of spot counts between the two measurements of the used EliSpot Reader on the Y axis, and the average of the two measurements on the X axis. Additionally the bias (red dotted line, -3.462) which is the average of the differences, and 95% limits of agreement (black dotted line, from -17.83 to 10.91) are shown.
At low spot counts the discrepancy between the two Reader is low and equally scattered. At high spot counts the scatter around the bias line gets larger and shows the tendency of slightly higher counts for ELRIFL061001095.
Establishing an external QC kit
The last step was the development of an external Quality control kit to monitor the assay procedure as it has been proposed in the requirements:
• GenID offers frequently (each quarter) frozen pretested PBMC´s and frozen antigens to run an EliSpot assay according to the BioDrIM protocol.
• The results of the external QC were recorded in an excel sheet.
• The summary of the results were send to the coordinator in Berlin.
As a first step, for establishing a SOP for thawing of PBMCs, recovery rates were determined to monitor protocol performance
Figure 4: Recovery rates for different donors (502-8, 503-2, 503-4, 503-5, 503-6 and 502-10) and different vials from the same sample. In the right graph additionally Transport of vials overnight on dry ice (TE) is shown regarding recovery rates.
Recovery rates between differences vials from one sample were less than 15%. Although, two PBMC preparations showed significantly lower recovery rates. Transport overnight on dry ice did not influence recovery rates or viability of cells.
Next step was to determine the variation of spot numbers from different vials. We found no significant difference in Spot numbers with standard antigen preparations and overall less than20 % difference in spot numbers between vials from one PBMC lot. Based in this data, QC kit with the established SOPs has been designed and tested with frozen PBMCs as indicated in the beginning.
In two individual runs, all centers tested frozen and fresh samples in parallel and recorded their spot count results to the Charité:
Figure 5: Spot counts for the QC tests from all centers for frozen PBMCs (left) and fresh PBMCs (right).
In conclusion, the external QC kit performance has to be further improved and more runs have to be started for better results and concordance between the centers.
Flow cytometry
Development and evaluation of flow cytometry assays was done in close collaboration between the Charite CIML team and Beckman-coulter Immunotech.
As a manufacturer of reagents and instrument for flow cytometry Beckman Coulter has been previously involved in another FP7 project, i.e “The ONE Study”. Within the ONE Study projects, Beckman coulter has provided flow cytometer instruments, custom designed reagents (6 multicolor panels) and solution for standardization of immunophenotyping. The test analysis showed that the standardization of the flow cytometry was efficient with low variability between the different sites and no clustering by center was observed (Streitz et al. 2013). The collaboration was extended to the BIO-DrIM project where the role of Beckman-coulter Immunotech was to provide 7 panels for immune monitoring by flow cytometry, one flow cytometer per site and solutions for Standardization. Beckman Coulter is committed to provide reagents and support to the BIO-DrIM consortium throughout the study. Hence Beckman Coulter is engaged to manufacture and commercialize the Immune monitoring panels.
1. Method Standardization
With its capability to identify various immune population at the cellular level, Flow cytometry is a suitable technology to reveal immune cell phenotyping and function. However, it is an open technology which make it difficult to reproduce findings and to compare data generate across laboratories. Together with the assay complexity, the pre-analytical factors (1), the reagents (2), the staining protocol (2), the instruments settings (3-4) and the data analysis (5) constitute steps that generates variability (Figure 1).
Figure 1: Flow cytometry experimental workflow
In order to improve standardization, the following has been initially worked out in the “The ONE study”:
- (1) Use Whole blood samples avoiding variability related to sample pre-analytic preparation of the sample (i.e PBMC)
- (2) Cocktail reagents were provided in 3 vials to avoid interaction between tandem dyes conjugates
- (2) SOPs established by the CIML team and specific training was done at each site to ensure handling is done similarly at all sites
- (3) Instruments settings were defined at the central laboratory. Target values using Flow set Pro beads were recorded and transferred to the other site to ensure settings are similar across sites
- (4) Lot match single colors were provided to ensure compensation settings are accurate for the multi-color panel
- (5) All data analyzed in the central laboratory to ensure robustness
In order to further improve the standardization, several settings have been tested and implemented for use in the BIO-DrIM. The changes and addition to the ONE study standardization approach are underlined in the list below:
- (1) Use Whole blood samples avoiding variability related to sample pre-analytic preparation of the sample (i.e PBMC)
- (2) Use stabilized cocktail in the DuraClone Format. Optimized panel design (Clone and Dyes) to improve robustness.
- (2) SOPs established by the CIML team and specific training was done at each site to ensure handling is done similarly at all sites
- (3) Instruments settings according to newly generated standardized SOP. Target values using Flow set Pro beads were recorded and transferred to the other site to ensure settings are similar across sites
- (4) Standard SOP to generate compensation matrix using the Navios Auto Set-up Scheduler. Lot match single colors were provided to ensure compensation settings are accurate for the multi-color panel.
- (5) All data analyzed in the central laboratory to ensure robustness
2. Panel design
The design of the panel has been made in close collaboration with the CIML team. The strategy for the panel design has been to identify the improvement to implement to the ONE study panels by fixing issues and by the addition of new markers of interest. Moreover, the implementation of the Beckman Coulter’s proprietary dry DuraClone format has been done. Each change has been performed as an iterative approach, where Beckman-Coulter has been transferring the newly generated material, i.e liquid antibody conjugates, to the CIML for evaluation. When all antibodies conjugates performances were considered acceptable by the consortium, panel were tested in the liquid format prior the evaluation of the final version in the DuraClone format. A summary of the panel changes is presented in Figure 1. The reasons for changes are summarized in Table 1. When all parameters defined for the various panels, validation has been made at the Central immune monitoring laboratory. All the panels were provided to each site throughout the study (about 10000 tests overall)
Figure 2: Panel design
Panels ID Observations ONE study Panels Implementation in BIO-DrIM panels
#1 to #4 Very bright signal and variable spillover over time of the CD8-APC-A700 conjugate. Requires adjustment of the settings (spillover compensation) during analysis. Time consuming. CD8-AlexaFluor 700 used instead of the CD8-APC-A700. Detection of the CD8 expression remain in the same channel with a bright enough signal, No spillover variability (over-time and inter-lot) and less spillover in other channels
#1 CD64 does not provide additional information versus the CD14 CD64 is not used
#2 CD45RO is not informative in the context of TCR phenotyping.
HLA-DR to replace CD45RO.
Addition of TCRVd1 and TCRVd2 conjugates to have a better phenotyping of the TCRgd subsets.
#3 and #4 Two panels to identify with 5 redundant markers for identification of T cell memory phenotypes and T cell activation status. Additionally, #4 allows identification of the TReg population through CD127 and CD25 expression on CD4+ T Cells. Define 1 panel for identification of the various T cells subsets and their activation status and 1 panel specific for TReg panel allowing a better phentopying of TRegs using FoxP3 (gold standard for detection of TRegs) and additional markers for the identification of the various subsets (Helios, CD39 and CD45RA)
#4 An intracellular approach for TReg phenotyping evaluated in parallel to the ONE study provided good results and allow good discrimination of TReg populations. The detection of TRegs is done using an intracellular approach which allow the detection of two key intracellular markers expressed on TRegs, i.e FoxP3 and Helios
#3 and #4 Interaction between the CCR7 and CD25 antibodies via the complement in a donor dependent fashion. False double positive signal. Requirement of pre-analytic processing prior staining. A CCR7 clones which does not interact with the complement has been used.
#6 Redundancy between BDCA2 and CD123 expression on plasmacytoid dendritic cells Removal of the BDCA2 for identification of plasmacytoid dendritic cells
#6 Literature describes the Clec9A markers to be specific of the BDCA3+ mDC subsets. Moreover, Clec9A expression on myeloid dendritic cells is conserved between human and mouse which allow the translational comparison between animal and human studies.
Only two out of the three subsets of mDCs are identified with the Dendritic cell panel. Replacement of BDCA3 by Clec9A for the detection BDCA3+ mDC subset.
Addition of the CD1c Marker to the panel in order to detect the 3 subsets of mDCs, namely CD1c+ mDCs, Clec9A+mDCs and the CD16+ mDCs.
Table 1: Description of changes in the Immune Monitoring panels
3. DuraCLone Format
Beckman Coulter’s Coulter proprietary DuraClone format allows the stabilization of antibody cocktails in a dry format. DuraClone reagents provide unprecedented standardization and the key features of this stable format are as follow:
• Dry, ready to use, unitized reagent
• Reagent-containing layer coated to sample tube with no loose lyophilized “cakes”.
• No electrostatic charges interfering as with loose lyophilized cakes.
• Storage / logistics at ambient temperature
• Minimal pipetting steps during sample preparation, with preliminary data indicating improved reproducibility of monocyte populations.
• Optimal consistency of results from tube to tube and from batch to batch.
Based on the comparison of tests between dry and liquid reagent, it was found that the staining intensities were completely overlapping demonstrating that the DuraClone format is compatible with all dyes used in the context of the BIO-DrIM (Figure 3A and B). DuraClone sensitivity to temperature variation that could occur during transport and/or storage was assessed by storing a multicolor flow cytometry panel at 37°C and 60°C. The stressed reagents performed as well as their counterpart kept at room temperature (Figure 3C). Moreover, specimen preparation methods including several lyses and cell permeation buffers were investigated (Data not shown). Results showed that the DuraClone dry flow cytometry panels are compatible with routinely used lyse procedures and with intracellular applications.
Figure 3: DuraClone performance evaluation
We also compared the variability between previous ONE-Study liquid-based and DuraClone antibody panels. Interestingly and very assuring, usage of DuraClone-based antibody staining panels reduced the intra-assay CV especially of low-frequent and difficult to stain B cell subpopulations (see figure below).
Figure 4: Intra-Assay CV of liquid versus DuraClone antibody-based staining of B cell subpopulations.
4. Product development
Upon agreement on the design and SOPs, Beckman Coulter has initiated the development of conjugation methodology for the Alexa700 conjugates, the development of the missing single colors conjugates followed by the development and the commercialization of the Immune monitoring panel with a Research use only label.
a. AlexaFluor 700 methodology
The development of the AlexaFluor 700 conjugation methodology consisted on the evaluation and the feasibility of the optimal condition for the conjugation of the AlexaFluor700 dye to purified antibody. The development of the methodology has been done with the CD8-AlexaFluor 700 conjugates required for the BIO-DrIM panels and on two antibody specificities (i.e CD19 and CD20) to ensure the robustness of the process conjugation. As consequence, the three conjugates were commercialized as single color conjugation as indicated in the table below.
Part number Conjugates Launch date
B76283 CD8-AlexaFluor 700 Nov. 2015
B76287 CD19-AlexaFluor700 Nov. 2015
B76279 CD20-AlexaFluor700 Nov. 2015
b. Development of single color reagent
Among the single color conjugates required for the formulation of the panels, nine were not available at Beckman Coulter. Evaluation of the various antibody clones available for each specificity has been done. Ones procurement of the clones has been done, development of the single color conjugates has been initiated. In order to allows customer to use the panel in the liquid, single color conjugates were commercialized as finish good or render available through the custom design service upon request (See summary in table below)
Part Number Conjugates Launch date
B30632 CCR7-PE Nov 2013
B36123 PD-1 PC5.5 Nov. 2013
B46036 CD1c-PC5.5 July 2014
B49309 TCRvd2-PacBlue Sept 2014
B43291 Clec9a-APC Nov 2014
B49310 TCRvd1-PC7 May 2015
B49311 Helios-PacBlue May 2015
B55384 CD39-PC5.5 Aug 2015
CDS Granzyme B-ECD Available through custom design service
c. Development of the BIO-DrIM panels
Upon availability of the conjugates the panel development was performed in sequence. All panels were commercialized with Research Use Only label (See summary in table below).
Part number Panel Launch date
B53309 DuraClone IM Phenotyping BASIC tube Oct. 2014
B53318 DuraClone IM B cell Tube Oct. 2014
B53328 DuraClone IM T cell subsets Tube Dec. 2014
B53351 DuraClone IM Dendritic cell tube Dec. 2014
B53340 DuraClone IM TCR Tube May 2015
B53346 DuraClone IM TReg Tube Sept. 2015
C00162 DuraClone IM Count Dec 2016
The collaboration within the BIO-DrIM consortium has led to the commercialization of the 1st multicolor panels in the DuraClone format. Hence, a model of reagent development at Beckman Coulter has been built following the successful collaboration with the BIO-DrIM. Collaborators are directly involved in the development of reagents which led to the commercialization of nine additional DuraClone products following this approach.
5. Clinical trial data analysis
Finally, we also started analysing flow cytometry staining results generated within CELLIMIN and RIMINI. For CELLIMIN we aimed on revealing leukocyte subsets which in addition to the allo-specific ELISpot are able to predict or diagnose acute rejections and focussed on the pre-transplant and early post-transplant samples. For RIMINI we hypothesize that the combination therapy of ATG and anti-TNF-alpha antibody would induce a stronger reduction of inflammatory subsets and started with a comparative analysis of pre-transplant as well 6 and 12-month post-transplant samples.
In addition, and in collaboration with collaboration partners from the Canadian National Transplant Research Program we have successfully tested algorithms for automated gating procedures. Together we developed a workflow, which includes pre-formatted antibody cocktails, standardized protocols for acquisition, and validated automated analysis pipelines, which can be readily implemented in multicenter clinical trials (Ivison et al. ICI Insight 2018).
Development and manufacturing of a urinary IP-10 point of care test
The specific mission of Milenia Biotec within the BIO-DrIM project is the development of an IP-10 Point-of-Care Test in Urine Samples.
Biological background
IP-10 is a chemokine which is released by immune cells during the course of inflammatory processes and therefore is a candidate to detect infinitesimal immune processes within the graft after kidney transplantation.
Several studies could demonstrate elevated levels of IP-10 in urine during the course of rejection after kidney transplantation both on a mRNA and on a protein level. Furthermore persistant elevated levels of urinary IP-10 within the first weeks of kidney transplantation are predictive of graft function after several months even if an acute rejection is absent.
These findings lead to the conclusion that urinary IP-10 is an “alert signal” of deminished graft function and shows therefore a potential for a screening in kidney transplant patients. However, a very simple to handle test – which ideally should be used as a “home test” by the patient - would allow a better patient care, especially if the data of the tests are directly made available to the patients physician.
This vision is followed by the BIO-DrIM consortium and the aim of Milenia Biotec was to develop a lateral flow urinary IP-10 test which can be quantified by a smarthphone app. As soon as the data are measured they are stored in the memory of the smartphone and can be transferred to the physicians office.
Assay set up and development process and status
At the start of the BIO-DrIM consortium, the development goals for the urinary IP-10 test have been defined as follows:
- Sample Type: Urine
- Sample Volume: Max. 100 µL
- Time to Result: Max. 20 minutes
- Working Range: 0 – 10.000 pg/mL
- Cut Off 100 pg/mL
- Shelf Life 2 years (at room temperature)
- Technology Lateral Flow Test based on gold particles
- Reader Device Smartphone (App)
- Data Communication e-mail; Cloud
- Reference Method: Quantikine ELISA; R&D Systems, Minneapolis, USA
Figure 15 shows the composition of the test strip:
Figure 15. Composition of the urinary IP-10 test strip
The urinary IP-10 lateral flow test consists of different overlapping membranes. After applying the sample (100 µL urine) it travels from the sample application port to the so called “conjugate release pad”. This is a membrane containing a dried gold particle solution. The gold particles are coated with a monoclonal antibody specific for human IP-10. The sample redilutes the conjugate and if IP-10 is present, it will be captured by the IP-10 antibody on the gold particles. The liquid (urine sample and gold particles) continue to travel through the test strip and enter the “analytical membrane”. This membrane is coated with 2 lines, called the test line and the control line. The test line is coated with a different monoclonal IP-10 antibody (Mouse-Anti-hIP-10). This antibody is directed against a different epitope within the IP-10 molecule than the IP-10 antibody on the gold particles. If IP-10 is present in the sample, the IP-10 antibody coated onto the analytical membrane will bind the IP-10 which was already bound by the other IP-10 antibody coated onto the gold particle. Consequently, the gold particles will be retained at the location of this line, if IP-10 is present in the urine sample. The more IP-10 is in the sample, the higher the intensity of this line.
Excess urine sample and gold particles continue to travel through the strip and cross the control line. At the location of the control line an anti-mouse-antibody is coated which captures all monoclonal mouse antibodies regardless their specificity and for this reason will bind the mouse monoclonal anti hIP-10 antibody on the gold particles as well. This leads to a retention of gold particles at this location and means that this line should occur in any case, regardless whether IP-10 was present in the sample or not. This line is used as a function control of the test strip.
All remaining liquid (urine and gold particles) continued to travel through the strip and are finally completely soaked by a membrane called “Wicking Pad”. The Wicking Pad is a very thick membrane which has the capacity to soak all the liquid of the sample and thus makes sure that the liquid stream through the strip flows only in one direction and that no liquid travels back in the strip.
Figure 16 shows how the urinary IP-10 test strip looks like after running a positive sample.
Figure 16. Urinary IP-10 test strip after running the sample
On the basis of the above described technological approach a prototype of a urinary IP-10 test could be developed.
A typical standard curve of the test is shown in Figure 17.
Figure 17. Typical Standard Curve of the Urinary IP-10 Test
In order to check the performance of the developed urinary IP-10 prototype test, the group of Professor Petra Reinke from Charité, Berlin, Germany supplied 177 urine samples from kidney transplant recipients.
All the samples were run in the Milenia QuickLine urinary IP-10 test and the Quantikine IP-10 ELISA from R&D Systems, Minneapolis, USA. The results of this comparison are shown in Figure 18. The value of 100 pg/mL IP-10 has been chosen as a cut off value for these experiments. This cut off value is the overall accepted value taken from the previously published literature.
Figure 18. Comparison Study of 177 urine samples between Milenia QuickLine IP-10 and R&D Systems Quantikine IP-10 ELISA
The results of Milenia QuickLine urinary IP-10 test in the mentioned comparison study have been measured with a table top reader device, called POCScan. As part of the BIO-DrIM consortium work Milenia has developed a smartphone app working under the iOS-system supplied by Apple. Finally the app, once released, is made available to users via the Apple App Store. The Apple instruments (iPOD or iPhone) need to be placed into a plastic casing supplied by Milenia Biotec. This is shown in Figure 19.
Figure 19. Apple Smartphone in the Reader Casing
The casing contains a drawer where the test units need to be placed in and finally are moved underneath the camera of the smarthphone. It also contains LEDs to illumiate the test unit and a rechargeable battery.
In order to identify the test units they are labeled with a QR code. This QR code encoded the core data of the test, including test name, lot number, lot specific standard curve etc.
The app is ready as a working prototype and is currently used by the R&D group of Milenia Biotec for the first development projects, including the urinary IP-10 test.
The clinical members of the BIO-DrIM project have decided to ship all urine samples collected over the entire time of the project to Charité, Berlin. All samples are kept frozen there until they are run in the Milenia QuickLine urinary IP-10 test.
It has been discussed during the 5th annual BIO-DrIM meeting in Berlin at the beginning of February 2018 that Milenia Biotec manufactures a fresh lot of the IP-10 test and upon availability of this material to run the entire study with the collected urine samples within the month of May 2018 in Berlin.
Close to the end of the BIO-DrIM project Milenia Biotec is close to transfer the developed prototype of the Milenia QuickLine urinary IP-10 test to production and to finally market it. The test will be well documented from the very beginning due to the fact that so many groups helped to supply samples. Of course, it is the hope of us at Milenia Biotec, that if the IP-10 test matches the expectations of the participating clinical study groups, they will continue to use the product in the future for the advantage of their patient care. Furthermore, these groups may be the nucleus of a broader users group in the future, as they are the reference groups in their respective countries and other potential users will follow their recommendations.
Furthermore the BIO-DrIM project served as a networking platform and Milenia Biotec got in touch with big companies from the pharmaceutical industry. Luckily one of these companies was interested in Milenias capability to develop and manufacture point of care test and to read the test results with a smartphone app. With this company Milenia signed a Co-Development-Agreement for the development of tests to be used within the area of transplantation.
BIO-DrIM Statistical Support
Strong statistical support for the BIO-DrIM Consortium was provided by statisticians from King’s College London (KCL). Dr Irene Rebollo-Mesa, a senior lecturer in biostatistics at KCL, oversaw the studies from the establishment of the consortium until December 2015. She was succeeded by Dr Daniel Stahl a reader in biostatistics at KCL. A junior statistician, Dr Sofia Christakoudi was appointed as a postdoctoral researcher with full-time commitment to the BIO-DrIM Consortium for a period of three years, staring from October 2014. The BIO-DrIM statisticians provided crucial support for shaping the design of the consortium trials, protocol development, sample size calculations for all considered trial designs, simulations of potential trial outcomes under different scenarios, establishment of Go/no-Go rules and preparing charters for the Data Monitoring Committees (DMC). They also provided important explanations and responses to queries arising in the process of obtaining approvals from competent authorities in different countries participating in the consortium and implemented the requirements of ethical comities for alterations in the designs of the individual studies, the interim analyses and the establishment of stopping rules. The BIO-DrIM statisticians have collaborated closely with the clinical trials teams, both at KCL and Charité, to develop the electronic case-report forms (eCRFs) which are housing the trial data. They have prepared statistical analysis plans describing in detail the interim and the final analyses of the primary and secondary outcomes and performed detailed data checks and reports for the DMC meetings as required.
However, due to the complex nature of the trials and the involvement of multiple centers and countries, the Consortium has faced several challenges, trial design modifications and delays arising from the process of obtaining approvals from competent authorities in several countries and an initially slow recruitment. This meant that patient recruitment and follow-up could not be completed within the 5 years allocated to BIO-DrIM and, consequently, the final statistical analyses of the trial data could not be performed during this period. Notwithstanding, by the end of 2017 all the consortium trials were well established and actively recruiting participants and the statistical analysis plans determining the final analyses had been prepared. This enabled the transfer of the statistical support during the one year extension to trial statisticians in King’s Clinical Trials Unit (KCTU) for LIFT and the Coordinating Center for Clinical Studies (KKS) at Charité for CELLIMIN and RIMINI. Details of the key engagements of BIO-DrIM statisticians with the individual trials are outlined below:
WPIA: LIFT (Liver Immunosuppression Free Trial) - a prospective randomised biomarker-based trial aiming to assess the clinical utility and safety of biomarker-guided immunosuppression (IS) withdrawal in liver transplantation.
LIFT trial design evolution
The design of the LIFT trial evolved through several stages as a result of interaction and feedback from the funder and the competent authorities. The initial intention to conduct a phase II trial as a prospective, open label, non-controlled/non-randomized, interventional cohort study in which all recruited liver transplant recipients would undergo IS withdrawal would have given the opportunity to validate the diagnostic performance (AUC=0.85) of the previously developed 5-gene liver tissue gene-expression signature (Martinez-Llordella M. et al. J Clin Invest. 2008), but would not have obtained a decisive verdict on the clinical utility of a biomarker-led IS withdrawal when compared with a weaning-all strategy. Consequently, a three-arm randomised controlled biomarker-strategy trial design with two control arms was considered. This aimed to achieve biomarker validation and clinical safety by demonstrating non-inferiority of biomarker-led care with respect to the number of successfully weaned rejection-free patients at 1-year follow-up post IS withdrawal when compared to the weaning-all control arm strategy and a non-inferiority with respect to the increase in fibrosis score at 3 years of follow-up when compared to a weaning-none control arm. This design would have required 150 patients, 50 in each trial arm (for 80% power and 5% probability of type I error), with an expectation of 100% weaning success rate in biomarker-positive patients and including all patients from the biomarker-based strategy arm, including biomarker-negative patients on maintenance IS. However, maintenance IS does not address clinical outcomes related to the establishment of “operational tolerance”, which could only be evaluated following IS withdrawal. This directed the development of the final design of LIFT as a prospective randomized two-arm superiority trial (Figure 20) aiming to demonstrate that the use of a liver tissue transcriptional test of tolerance to stratify liver recipients prior to IS withdrawal accurately identifies operationally tolerant recipients and reduces the incidence of rejection, as compared with a control group in whom IS withdrawal is performed without stratification.
Figure 20. LIFT trial design evolution.
LIFT sample size and Go/no-Go criteria
The study has been designed and powered to test the hypothesis that IS weaning under a "Biomarker-led" strategy (i.e. weaning only biomarker-positive patients, Arm B+) is superior to a “Weaning-All” strategy (irrespective of biomarker status, Arm A), with respect to the proportion of patients who, having started the IS withdrawal protocol, complete it successfully without undergoing allograft rejection by 1 year post withdrawal completion.
Sample size calculations are based on the following expectations:
• Biomarker: 50% biomarker-positive patients – simulation studies showed that if biomarker-positivity is reduced, which was observed in the initial recruitment stages of LIFT, the trial would gain power due to the increased proportion of weaning failures in the weaning-all control Arm A;
• Arm A: 50% successfully weaned patients, based on the weighted average of success rates observed in a previous single-arm weaning-all trial (Benitez C. et al. Hepatology 2013;58:1824) in patients grouped by age and time from transplantation at initiation of withdrawal;
• Arm B+ H0 (null hypothesis) 50% successfully weaned patients (if PPV = 0.50); H1 (alternative hypothesis) 80% or more successfully weaned patients (if PPV ≥ 0.80) more conservative and realistic expectation than the 100% incorporated in the initial trial design;
• Required number of completers: 100 patients for 90% power and 5% type I error rate;
• Allocation ratio: 2:1 between arms A and B+, if 50% of patients are biomarker-positive;
• Target sample size: 148 patients (134) completers, including 34 patients in Arm B- and 14 patients accounting for 10% drop-out rate.
Therefore, with the same number of patients the trial under the final design could achieve higher power than the trial under the initial design and could demonstrate a superiority of biomarker-guided IS withdrawal (Arm B+) compared to a weaning-all strategy (Arm A).
Additional considerations for the biomarker test to be successfully validated for clinical use are a sensitivity no less than 0.70 and a positive predictive value (PPV) no less than 0.80. If the PPV or the sensitivity are 0.5 the prediction of a biomarker would be no better than at random. Consequently, Go/no-Go criteria were set for two planned interim analyses, when 33% and 50% of patients assigned to IS weaning reach 1 year follow-up post completion of IS withdrawal. The trial would progress if the 95% confidence interval for PPV estimated in Arm B+ at the first interim analysis includes 0.80 and excludes 0.50 and the 95% confidence interval for sensitivity estimated within Arm A at the second interim analysis includes 0.70 and excludes 0.50.
LIFT Statistical Analysis Plan (SAP)
The Intention-to-Treat population (determined by treatment assignments) will be used for the primary analysis, which is conventional for superiority trials. Generalised linear mixed-effects models (logistic regression) was selected for the final analyses due to concerns that there would be insufficient power to adjust for all potential covariates, that the results would be conditional on the covariates and that additional clinical centers joined the trial (up to 13), some of them likely to recruit very few patients. The models will include adjustment for the factors determining the minimisation and will consider study centre as a random effect in all analyses and biomarker status as a fixed effect in the secondary analyses. An overall marginal rather than a conditional effect will be estimated. Defining study centre as a random effect, i.e. treating every centre as representative of a larger population of centres, is conceptually appropriate, as all centres are expected to show comparable performance. Mixed-effects regression will also accommodate the analysis of serial measures. Multivariable models for the exploratory analyses of secondary outcomes will include main effects of treatment arm and biomarker status, as well as their interaction term. The influence of potential covariates added individually to the model subgroup analyses for time since transplantation and age at randomization, ethnicity, sex or drug regimen will be examined as part of the sensitivity analyses. The stability of the biomarker signature over the follow-up period will also be examined, both as a binary classification and as a continuous measure of probability of rejection. The effect of potential mechanistic mediators on primary and secondary outcomes will be evaluated using generalized latent mixed models, which will enable the consideration of different models of causation.
The Statistical Analysis Plan (SAP) was approved by the DMC statistician and the protocol was updated and finalized accordingly.
LIFT electronic Case Report Form (eCRF) and data quality checks
A comprehensive eCRF was designed and implemented in collaboration with KCTU. The complicated follow-up and data collection schedule of the LIFT trial was accounted for in some 40 different forms accommodating clinical and laboratory data acquired at the baseline, screening and scheduled transplant centre visits (6-monthly) and details of the protocol liver biopsies, as well as data from monitoring of liver function tests (performed at 3-weekly intervals in the initial phases of IS weaning), information on drug regimens and doses and details of for-cause biopsies. The BIO-DrIM statisticians performed regular data checks at 3-monthly intervals in order to ensure that the baseline information on eligibility, demographics, biochemistry, histology, medical history and clinical events, IS drugs, concomitant medication and quality of life is available and recorded correctly for all consented patients and that follow-up data are timely accrued for all randomized patients. Numerous queries were raised for all identified discrepancies, which were brought to the attention of all individual study sites by the Trial Manager. Efforts were undertaken to identify the patterns of errors and to provide improved guidelines for timely and accurate data recording.
LIFT DMC reports and patient monitoring
Given the potential risk that IS withdrawal poses on patient wellbeing, the Ethics Committee requested a very close monitoring by the DMC. The trial statisticians prepared detailed reports for the DMC meetings at 3-monthly intervals. These included information on consent, randomization and withdrawal figures, baseline demographics and immunosuppressant medication, duration of periods between study procedures and visits in comparison with the target windows set in the protocol, frequency of treatment-related diseases and description of the adverse effects. The data were tabulated by study centre for the open part (accessible to the DMC, Trial Steering Committee) and additionally by treatment (IS weaning or maintenance) for the closed part (confidential to the DMC members). The proportion of biomarker positivity was monitored. In addition, the trial statisticians answered data entry queries raised by the study sites and prepared summaries of the findings in the screening baseline biopsies to assist the trial team to identify which particular histological patters contribute to failure of histological eligibility. The markdown package in R was set up for dynamic reporting. A code was written which generates the data monitoring progress reports exporting directly plots and tables into Word. This removed the possibility of mistakes arising from manual data transfers from the statistical software and ensured the consistent generation of high quality output. The code additionally generates time plots of liver function tests and IS drug doses, individually for each patient and in separate batches per study site. The timings of study visits and liver biopsies are marked on the plots to enables the visualization of the progress of each patient in the trial and to facilitate the identification of missing data and inconsistencies.
WPIC study - using biomarkers of tolerance to guide immunosuppression weaning in kidney transplant recipients
This study, coordinated by KCL, was initially conceived as a safety and feasibility pilot phase II trial for initiating IS minimization in patients identified as ‘tolerant’ on the basis of a 10-gene expression signature in peripheral whole blood developed as part of the preceding Indices of Tolerance study (Sagoo P. et al. J Clin Invest 2010;120:1848). The signature was replicated in samples from patients of the GAMBIT study, achieving high predictive accuracy and good test-retest stability. However, a major statistical point of concern was the possibility that IS drugs influence the levels of expression of the candidate ‘tolerance genes’ and, thus, introduce a potential confounding effect on the gene-expression signature. Statistical analysis revealed that the expression of most IoT genes was, indeed, significantly associated with IS drug intake to the extent that adjustment of the signature genes for drug intake abolished the discrimination between tolerant and non-tolerant patients. Consequently, no clinical trial could go ahead and the project returned to the “drawing board” into the hands of the BIO-DrIM statisticians. A new IS-drug-independent gene-expression signature of tolerance was developed. A set of 28 candidate genes optimal for prediction of tolerance with perfect predictive accuracy was identified from the IoT array using for gene selection drug-adjusted gene-expression levels and penalised logistic regression with elastic net penalty. The candidate genes were then validated in samples from the GAMBIT study using a Fluidigm platform and a parsimonious 9-gene signature was selected with elastic net regression. This new signature (Rebollo-Mesa I. et al. Am J Tranplant 2016;16:3443) is IS-independent and can achieve a very good discrimination of already established tolerant KTRs from treated stable and chronic rejection patients (cross-validated AUC 0.81) identifying approximately 12% of stable patients as potentially “tolerant”.
An important point to consider in stable KTRs selected as “tolerant” whilst receiving IS treatment is whether they would retain such features of “tolerance” once the drugs have been withdrawn. Based on the strong confounding effect of IS drug regiments, revealed by BIO-DrIM statisticians, it has become clear that gene-expression signatures developed without any statistical mechanism to control for IS-drug intake is in danger of being driven by pharmacological and not biological features. Consequently, we have engaged in an active debate with the kidney transplant community with one of the most prominent results an editorial with the explicit title “Lack of adjustment for confounding could lead to misleading conclusions” (Christakoudi S. and Hernandez-Fuentes M. Am J Transplant 2017), prompting an equally explicit response “We agree, Let’s move on” (Asare A. et al. Am J Transplant 2017;17:2505). Before we “move on”, however, there is one more contribution that BIO-DrIM statisticians owe to the transplantation literature and this is the comparison of the clinical characteristics and IS-drug regimens of stable KTRs selected as “tolerant” on the basis of our IS-independent signature and on the basis of signatures developed without adjustment for IS-drug regimens. This analysis will also comprise the validation of our new signature with RT-qPCR, the clinical-grade laboratory technique.
Failure to proceed with a clinical trial of IS minimization in long-term and established stable KTRs within the time-frame of the BIO-DrIM Consortium was determined by major safety concerns of clinical teams and funders in the UK, which made it clear that further detailed studies of “tolerance” signatures are paramount. Consequently, we initiated a ramification of the GAMBIT study aiming to elucidate the role of steroid conversion (HSD11B1, HSD11B2, H6PD) and receptor genes (NR3C1, NR3C2) for “operational tolerance” and chronic rejection in KTRs, as glucocorticoids are major endogenous immunomodulatory and anti-inflammatory factors and synthetic steroids are fundamental for IS drug regimens in KTRs. This work advanced the statistical concept of the importance of adjustment for confounding a step further (Christakoudi S. et al. Mol. Cell. Endocrinol 2018). We examined in detail the impact of exogenous prednisolone and the effects of adjustment for IS drug intake on the expression of steroid conversion and receptor genes, thus demonstrating that drug adjustment enables evaluation of endogenous factors influencing gene expression. Further, steroid conversion and receptor genes, alone, achieved classification of “tolerant” patients and were major contributors to gene-expression signatures of both, tolerance and CR, alongside known tolerance-associated genes, thus revealing a key role of steroid regulation and response in kidney transplantation.
WP2 (CELLIMIN) - prospective, multi-centre, biomarker-strategy enrichment design trial of IS minimization in KTRs.
CELLIMIN trial design evolution
The main objective of the CELLIMIN trial is to demonstrate the utility and safety of the IFN-γ ELISPOT marker for the stratification of kidney transplant recipients. Initially, the trial was intended as a biomarker-stratified design in which KTRs would be first stratified on the basis of their pre-transplantation ELISPOT response in two groups ELISPOT-positive (i.e. at higher risk of rejection) and ELISPOT-negative (i.e. at low risk of rejection) and then patients within each group would be randomized in a ratio 1:1 to a new “Low” or standard-of-care “High” IS-regimen, resulting in a four-arm trial (Figure 2). This trial design, however, would have been appropriate if compelling biological data were lacking that one of the biomarker groups would not benefit from the new treatment. Ethical concerns raised by the competent authorities during the Voluntary Harmonisation Process (VHP) that ELISPOT-positive KTRs have unacceptably high risk of rejection, led to the elimination of the ELISPOT-positive group. The study was modified to an enrichment design, utilising ELISPOT as a screening tool prior to randomisation. Concerns were also raised by the VHP regarding the initially proposed 15% non-inferiority margin. Their requirement to reduce the non-inferiority limit to 10% could not have been met retaining the fairly conservative initial alternative hypothesis of 15% rejection rate in the new-treatment arm (5% higher compared to the control arm), as this would have quadrupled the required sample size to 1205 patients - a financially unsustainable and completely unachievable goal. Figure 21 illustrates how the assumptions under the alternative hypothesis influenced more dramatically sample size for a non-inferiority margin of 10% compared to 15%.
Figure 21. CELLIMIN sample size for different rejection rates under the alternative hypothesis.
BIO-DrIM statisticians, however, were able to retain the original sample size and to accommodate the 10% non-inferiority limit by assuming equivalence in the alternative hypothesis - a more traditional assumption in non-inferiority trial designs and not an unreasonable one, since the prediction for the expected rejection rate in the treatment group was quite uncertain. As a safety net, a stopping boundary for futility, based on conditional power calculated from the observed cumulative proportions, was introduced at the interim analysis step in addition to the pre-existing non-inferiority (efficacy) stopping boundary, based on Lan-De Mets alpha spending function with O’Brian-Fleming parameters. This gives the steering committee the flexibility to consider at the interim analysis whether to stop the trial, if the observed rate of rejection in the treatment arm is more than 15% (judging by its 95% confidence interval) or to increase the sample size adaptively using the actual observed rates of BPAR in the treatment arm, after considering costs and actual accrual rates.
Figure 22. CELLIMIN trial design evolution
CELLIMIN sample size and Go/no-Go criteria
The primary hypothesis is that a “Low” IS regimen is non-inferior with respect to the rate of T-cell mediated BPAR at 6 months post-transplantation when compared to the standard-of-care “High” IS regimen. The final sample size calculations were based on the following expectations:
• “High” IS (standard-of-care) control arm: 10% BPAR at 6 months post-transplantation;
• “Low” IS (new-treatment) arm: H0: 20% BPAR; H1: 10% or less BPAR (reduced from 15%);
• Non-inferiority limit: 10% (reduced from 15%);
• Required number of completers: 271 patients for 80% power and 5% type I error rate;
• Target sample size: 301 ELISPOT-negative patients completers, including 30 patients accounting for 10% drop-out rate;
• Screening: 669 patients to be screened, expecting approximately 45% to be ELISPOT negative;
• Interim analysis: when 122 patients complete 6 months of follow-up, of which 61 in the “new treatment” arm.
The interim analysis involves the calculation of a Z-statistic with a cut-off for the efficacy stopping boundary of -2.698 (on the Z-scale) and for the futility stopping boundary -0.692 on the Z-scale and 0.625 on the conditional power scale. A one-sided 95% CI for a BPAR rate of 15% would be estimated with a precision of +/- 7.5%, with the upper bound of 22.5% defining the cut-off for stopping the trial for futility.
CELLIMIN Statistical Analysis Plan
BIO-DrIM statisticians addressed statistical queries raised in the process of obtaining approval by VHP and the competent authorities of the individual participating countries and provided much appreciated constructive feed back to the eCRF developers at Charité. They also prepared the statistical analysis plan (SAP), which contains a detailed section explicitly defining failure outcome. The problem with primary outcome definitions in CELLIMIN arises from the fact that protocol biopsies are conducted at 3 months post randomisation, to adhere to the conventional clinical management practice, while the primary outcome is evaluated at 6 months and is defined as “biopsy-proven” acute rejection, implying that a for-cause biopsy should be triggered by a clinical indications of rejection. This means that a protocol biopsy at 3 months could uncover subclinical rejection or borderline changes, thus making it impossible to evaluate what would have been the clinical outcome by the time of final evaluation of the primary outcome at 6 months in the absence of this information or a resulting from it treatment. The SAP, therefore, considers failure outcomes with respect to different scenarios of histological results and treatment following the protocol biopsy at 3 months.
The Per-Protocol (PP) population (determined by the administered and not the allocated treatment) will be used for the primary analysis, but an Intention-To-Treat (ITT) analysis will also be performed as part of the sensitivity analyses. The analytical strategy of the primary and secondary outcomes is aligned with the LIFT trial and is based on generalized linear mixed-effects models (logistic regression) with random effects for study centre, as it is a factor used in the randomisation procedure, and a fixed effect for treatment arm. An estimated difference in the proportions of patients with BPAR between the “Low” and the standard-of-care “High” IS regimen arms with a 95% one-sided upper confidence interval will be presented for the population (marginal) effect. Non-inferiority will be evaluated with respect to the upper bound of the confidence interval in relation to the non-inferiority margin as illustrated in Figure 23.
Figure 23. CELLIMIN non-inferiority testing, based on the confidence interval of the difference in BPAR.
WP3 (RIMINI) - Tacrolimus after rATG and infliximab induction immunosuppression
Study design evolution
The three-arm design of the original ICARUS trial was cancelled because it would have required unfeasibly high number of patients and, in addition, the treatment medication was no longer available. The subsequent plan in the renamed RIMINI trial was to proceed with an open-label single-arm Simon’s two-stage pilot Phase II clinical trial in order to provide evidence for efficacy and safety of the new induction regimen and a Go/no-Go rule for further clinical development. However, this design was also abandoned because the sample size, although reduced, was still unfeasibly high, as RIMINI was going to compete with CELLIMIN for patient recruitment. In addition, a Simon’s design would have been more appropriate if the expected proportion of outcome failures under the alternative hypothesis was considerably lower than the expectation for the null hypothesis. In RIMINI a smaller difference or a non-inferiority would also be considered beneficial for the new induction regimen and there was a real risk that under Simon’s design the trial would complete both stages and yet provide inconclusive evidence to proceed into a phase III trial. BIO-DrIM statisticians found a solution in a confidence-interval-based single-arm Phase II design, which reduced the sample size by almost half of the requirement for Simon’s design, without losing power, whilst still addressing the main research question and providing Go/no-Go criteria for a larger parallel-arm-design trial. The confidence-interval-based design would provide an effect estimate and a valuable information for further clinical development with almost the same number of completers that would have been sufficient only for the first stage of a Simon’s design.
Figure 24. RIMINI trial design evolution
Sample size calculations and Go/no-Go criteria
A total of 68 completers will be required to estimate a one-sided 95% confidence interval with a precision (one-sided width) of no more than 10% efficacy failure rate. The target sample size was increased with 10% to account for drop-outs and violations of the protocol. A total of 75 patients will receive the proposed induction regimen - less than half of the 161 patients that would have been required with Simon’s design, which makes completion of recruitment and follow up feasible within a reasonable timeframe. If up to 27 out of the 68 completers experience efficacy failure, a progression into a larger trial will be considered justifiable. If the number of patients experiencing efficacy failure is between 28 and 34 out of 68, the merits of a larger non-inferiority design will be considered depending on the risk/benefit assessment. If more than 34 out of the 68 completers experience efficacy failure, a progression into a larger trial would be considered unjustifiable.
There is no planned interim analysis as the new induction treatment will be applied at enrolment and, due to the 12-month follow-up, the majority of the patients would have already received the new treatment by the time of an interim analysis.
Statistical Analysis Plan
BIO-DrIM statisticians assisted with protocol development and prepared a detailed Statistical Analysis Plan. They raised queries regarding particular details of the definitions of the primary outcomes and the involved in this laboratory measurements of creatinine, which were very much appreciated by the Principal Investigator. These questions were discussed with the members of the consortium during the annual BIO-DrIM meeting and decisions were taken, including corresponding modifications of the eCRF.
For the analysis of the primary outcome, the upper bound of a one-sided 95% confidence interval for the observed efficacy failure rate within the overall group of patients who have received the study treatment will be calculated. Cluster robust standard errors, which are likely to keep or slightly increase the power of the study, will be used to account for study sites.
References
Christakoudi, S. and Hernandez-Fuentes, M.P. 2017. Lack of adjustment for confounding could lead to misleading conclusions, Am J Transplant. 17, 2505-2506.
Christakoudi, S., Runglall, M., Mobillo, P., Rebollo-Mesa, I., Tsui, T.L. Nova-Lamperti, E., Norris, S., Kamra, Y., Hilton, R., Bhandari, S., Baker, R., Berglund, D., Carr, S., Game, D., Griffin, S., Kalra, P.A. Lewis, R., Mark, P.B. Marks, S.D. Macphee, I., McKane, W., Mohaupt, M.G. Pararajasingam, R., Kon, S.P. Seron, D., Sinha, M., Tucker, B., Viklicky, O., Lechler, R.I. Lord, G.M. Stahl, D. and Hernandez-Fuentes, M.P. 2018. Steroid regulation: An overlooked aspect of tolerance and chronic rejection in kidney transplantation, Mol Cell Endocrinol. doi: 10.1016/j.mce.2018.01.021
Nova-Lamperti, E., Romano, M., Christakoudi, S., Runglall, M., McGregor, R., Mobillo, P., Kamra, Y., Tsui, T.L. Norris, S., John, S., Boardman, D.A. Lechler, R.I. Lombardi, G. and Hernandez-Fuentes, M.P. 2018. Reduced TCR Signaling Contributes to Impaired Th17 Responses in Tolerant Kidney Transplant Recipients, Transplantation. 102, e10-e17.
Rebollo-Mesa, I., Nova-Lamperti, E., Mobillo, P., Runglall, M., Christakoudi, S., Norris, S., Smallcombe, N., Kamra, Y., Hilton, R., Indices of Tolerance, E.U.C. Bhandari, S., Baker, R., Berglund, D., Carr, S., Game, D., Griffin, S., Kalra, P.A. Lewis, R., Mark, P.B. Marks, S., Macphee, I., McKane, W., Mohaupt, M.G. Pararajasingam, R., Kon, S.P. Serón, D., Sinha, M.D. Tucker, B., Viklický, O., Lechler, R.I. Lord, G.M. and Hernandez-Fuentes, M.P. 2016. Biomarkers of Tolerance in Kidney Transplantation: Are We Predicting Tolerance or Response to Immunosuppressive Treatment?, Am J Transplant. 16, 3443-3457.
Electronic Case Report Form (eCRF) and the data management (DM) of the trials CELLIMIN and ICARUS 2.0 / RIMINI eCRF/ DM
An extensive effort has been made by the KKS in collaboration with the CHARITE and the statisticians from KCL in setting up an eCRF with an Internet-based Remote Data Entry (RDE) system; a technique that allows electronic data capture in a central study database.
The software platform is SecuTrial® with the following characteristics:
o Compatible to standard Internet browser (Firefox, Internet Explorer, Opera ...)
o No further software installation necessary
o No local storage of data
o Available worldwide 24h (30min. downtime per day [02:00 GMT-1])
o Safety access (Firewalling, SSL encryption, Authentication, Logs)
All the processes are done according to SOPs and Working Procedures and are compliant to:
o GCP
o FDA 21 CFR Part 11
o German Pharmaceuticals Act.
Regarding the eCRF and data management, the following tasks have been accomplished or will be provided:
o Design and creation of database
o Design and programming of eCRF with system and plausibility checks
o Validating and release of eCRF
o Reporting of critical information on weekly basis
o Creation of a Data Validation Plan for data clearing
o Delivery of cleaned data for statistical analysis
o Troubleshooting
For support during the study, the following tools have been implemented:
o Administration: database / eCRF (Logins, roles and rights)
o User support (Reset password, technical problems etc.)
o Query management (Data checks)
o SAS-Reporting
o Staff training
Pharmacovigilance
Regarding the Pharmacovigilance, an adequate SAE (Serious Adverse Event)-Management was implemented to ensure safety surveillance during the study:
o Safety surveillance = application of drug and adverse effects
o Legally defined reporting procedures and timelines
o Central pharmacovigilance
- Central Contact point (all reports KKS)
- Reporting to all concerned authorities, ethics committees and study sites via central pharmacovigilance
All reported SAE are assessed according to their impact on the safety profile of the administrated drugs and treatment procedures. The steps of SAE management, evaluation, and further reporting to CA, EC and investigators in case of a SUSAR are described in the study protocol and study specific SOPs. As a consequence of such safety information all necessary actions to protect patient’s safety are accomplished by the sponsor. All available safety information is continuously monitored and the results are summarized in an annual safety reporting (DSUR) to the CA, EC and investigators.
Monitoring
The objectives of the monitoring procedures are to ensure that the trial subject’s safety and rights as a study participant are respected, that accurate, valid and complete data are collected, and that the trial is conducted in accordance with the trial protocol, the principles of GCP and local legislation. Therefore, to guarantee the quality of the monitoring the KKS provided a general monitoring SOP, a monitoring manual and report templates.
Source data are contained in source documents (original records or certified copies).
The monitoring of CELLIMIN and RIMINI is a combination of on-site visits (initiation, interim and close out) and centralized monitoring based on the eCRF (SecuTrial®).
During the on-site visits the monitor the following data is verified:
• Trial subject existence
• Subject informed consent
• Inclusion / exclusion criteria
• Result of ELISpot randomization (CELLIMIN)
• Correct reduction (group B) of trial medication according to randomisation and trial
protocol.
• Control of drug accountability (only documentation)
• SAEs
• Primary endpoint (ELISpot and Biopsy results BPAR)
• End of study participation/BPAR
The report that KKS generates after each visit, informs the sponsor/PI about:
o Result of SDV
o Protocol violation
o (S)AEs
o Drug account
o Recruitment and problems at the site
o Follow-up activities
Follow-up letter inform the site about:
o Tasks to be completed
o Completion/Updating of study documentation
o Documentation of Drug account
o Others
Health-Economic Studies (Health-Technology Assessment – HTA)
At the outset of the BIO-DrIM consortium, it was necessary to develop three clinical trial protocols, each one including a health-economic data assessment and evaluation strategy that would enable researchers to analyze medium- and long-term healtheconomic outcomes with respect to costs and quality of life.
A central objective of the BIO-DrIM consortium is to prove the clinical feasibility of reducing immunosuppressive (IS) medication and thereby also reducing IS-related medication-costs after solid organ transplantation (kidney and liver). Depending on the trial (LIFT, CELLIMIN, RIMINI), IS reduction is indicated either for selected organ recipients based on pre-transplant biomarker-based patient-stratification procedures (CELLIMIN and LIFT trial) or using an IS weaning protocol for a selected cohort of patients fulfilling specified trial inclusion criteria (RIMINI trial).
As one of the first steps, structured literature reviews were conducted for all indications covered by the three clinical trials in order to identify typical health-economic data assessment strategies, endpoint definitions and respective results. The key findings of the literature reviews were presented at the annual meeting in 2015.
Based on the results of the literature reviews, trial specific data assessment strategies using standardized quality of life questionnaires and study-specific health-economic questionnaires were developed to accurately assess patient-specific resource consumption during the follow-up periods of the trials. The assessment and evaluation strategies were developed based on guidelines for health-economic evaluations alongside clinical trials (Edwards et al., 2008; Petrou & Gray, 2011; Ramsey et al., 2005; Sullivan et al., 2005), they were aligned with the clinical objectives of each clinical trial, and they are described in detail on the following pages.
Health-economic data assessments in the BIO-DrIM trials use questionnaires that are either standardized or were specifically developed for each trial. Precise working instructions on the use of the questionnaires were developed in all cases and made available for the respective study staff. Furthermore, health-economic statistical analysis plans were developed and provided to the study teams. Interim-analyses and final analyses of the health-economic data gathered throughout the trials will be completed as soon as all required data are available.
Contributions to WP1 (LIFT trial)
The health-economic data assessment strategy for the LIFT study was developed in close collaboration with the principal investigator (Prof. Alberto Sanchez-Fueyo) and is based on health-economic guidelines for health-economic evaluations alongside clinical trials (Edwards et al., 2008; Petrou & Gray, 2011; Ramsey et al., 2005; Sullivan et al., 2005). It was presented, discussed and consented within the BioDrIM consortium.
The health-economic data assessment strategy of the LIFT trial is separated into 3 main steps:
1. Assessment of Quality-Adjusted Life Years (QALYs)
2. Assessment of patient-specific resource consumption as a base for cost-estimates
3. Micro-costing approach to evaluate costs of biomarker-stratification process
Quality-Adjusted Life Years are currently being assessed in the LIFT trial using the generic EQ5D-5L questionnaire, which is a standardized and well accepted instrument in the scientific community of health-economics. A working instruction, specifically designed for the LIFT trial, was developed and uploaded to internal BioDrIM website, in order to make the questionnaire as well as the working instruction available for all participating study nurses and other trial staff.
In order to precisely capture other generic aspects of quality of life (qol) which are not covered by the EQ5D-5L instrument, the implementation of the SF36 questionnaire – another well accepted qol-questionnaire – was discussed and agreed. The SF36 questionnaire and a corresponding working instruction were uploaded to the internal BioDrIM website, in order to make them available for all participating study nurses and other trial staff.
As suggested by guidelines on health-economic data assessments in clinical trials (Edwards et al., 2008; Petrou & Gray, 2011; Ramsey et al., 2005; Sullivan et al., 2005), the disease-specific qol-questionnaire NIDDK was proposed and agreed to implement.
Furthermore, a trial specific health-economic questionnaire was developed to capture individual health-care resource consumption, during the follow-up of each enrolled patient. The development of this questionnaire had to be well balanced in terms of efforts to be undertaken by the trial staff to assess the needed data in patient-interviews and the precision of expected results. This questionnaire as well as a corresponding working instruction was developed in close collaboration between the principal investigator and Cellogic GmbH and both documents were uploaded to the internal BioDrIM website, in order to make them available for all participating study nurses and other trial staff.
Regular checks of the assessed health-economic trial data have been defined in close collaboration between the principal investigator, statisticians at King`s College London, who are responsible for data management and storage, and Cellogic GmbH. Every three months, the complete trial dataset of all health-economic electronic case-report forms (eCRF) of all included patients are submitted to Cellogic GmbH and checked for completeness and consistency. In case of systematically missing data or inconsistent values, adequate measures, e.g. intensified briefing and training of study nurses, are taken to prevent poor data quality and more missing data in future assessments. The results of the last regular completeness-check were presented at the last annual meeting on February 2nd, 2018 and are shown on figure 25.
Figure 25. Results of data completeness-check (LIFT), Data extraction date: January 2nd, 2018
Contributions to WP2 (CELLIMIN trial)
The health-economic data assessment strategy was developed in close collaboration with the principal investigator (PI, Prof. Dr. Josep M. Grinyó) and the sponsor representative (Prof. Dr. Petra Reinke) and is based on health-economic guidelines for health-economic evaluations alongside clinical trials (Edwards et al., 2008; Petrou & Gray, 2011; Ramsey et al., 2005; Sullivan et al., 2005). Advantages and disadvantages of using different assessment strategies were presented and discussed during the BIO-DrIM meeting on November 12th, 2012. The final assessment strategy was consented with the principal investigator shortly after.
The health-economic data assessment strategy of the CELLIMIN trial is separated into three major steps:
1. Assessment of Quality of Life (QoL) at baseline, months 1, 3, 6, 12, 18 and 24 and subsequent computation of Quality-Adjusted Life Years (QALYs)
2. Assessment of patient-specific resource consumption at initial discharge following transplantation, repeated discharges (if occurred) and regularly at months 3, 6, 12, 18 and 24
3. Micro-costing of biomarker-stratification process in order to assess costs related to pre-transplantation patient-stratification
QALYs are being assessed in the CELLIMIN trial using the generic EQ5D-5L questionnaire, which is a standardized and well accepted instrument in the scientific community of health-economics. A working instruction, specifically designed for the CELLIMIN trial, was developed and uploaded to internal BIO-DrIM website, in order to make the questionnaire as well as the working instruction available for all participating study nurses and other trial staff.
In order to precisely capture other generic aspects of quality of life (qol) which are not covered by the EQ5D-5L instrument, the implementation of the SF36 questionnaire – another well accepted qol-questionnaire – was discussed and agreed. The SF36 questionnaire and a corresponding working instruction were uploaded to the internal BIO-DrIM website, in order to make them available for all participating study nurses and other trial staff.
As suggested by guidelines on health-economic data assessments in clinical trials (Edwards et al., 2008; Petrou & Gray, 2011; Ramsey et al., 2005; Sullivan et al., 2005), the disease-specific questionnaire KTQ-25 was proposed to the PI and agreed to implement.
Furthermore, a trial specific health-economic questionnaire was developed to capture individual health-care resource consumption, during the follow-up of each enrolled patient. The development of this questionnaire had to be well balanced in terms of efforts to be undertaken by the trial staff to assess the needed data in patient-interviews and the precision of expected results. This questionnaire as well as a corresponding working instruction was developed in close collaboration between the PI and Cellogic GmbH and both documents were uploaded to the internal BIO-DrIM website, in order to make them available for all participating study nurses and other trial staff.
Regular checks of the assessed health-economic trial data have been defined in close collaboration between the PI, KKS-Charité (CRO), who is responsible for data management and storage, and Cellogic GmbH. Every three months, the complete trial dataset of all health-economic electronic case-report forms (eCRF) of all included patients are submitted to Cellogic GmbH and checked for completeness and consistency. In case of systematically missing data or inconsistent values, adequate measures, e.g. intensified briefing and training of study nurses, are taken to prevent poor data quality and more missing data in future assessments. Additionally, several meetings and reviews with the project coordinator and pilot-center study nurse were carried out in order to clarify and communicate appropriate use of the developed health economic questionnaires since study initiation. The results of a recent completeness check were presented at the last annual meeting on February 2nd, and are shown in figure 26.
Figure 26: Results of completeness-check (CELLIMIN), Data extraction date: November 22nd, 2017
Cost of prospective biomarker-stratification (Micro-Costing)
In order to analyze the health-economic effects of a patient-selection process for a given cohort, e.g. in order to identify those patients who are eligible for reduced IS medication protocols, the costs of that patient-stratification need to be known.
The expected costs for the biomarker- based stratification process were calculated using a micro-costing approach for the ELISPOT-assessment, as it is conducted throughout the CELLIMIN trial. The micro-costing procedure consists of three major steps:
1. Counting of necessary inputs to the patient-stratification process in terms of consumerables as well as labor time
2. The counted inputs are multiplied with the respective market prices to arrive at average provision costs of the patient-stratification process
3. These final costs are discounted (depending on the year of market price identification) with appropriate rates in order to report present costs of patient stratification.
The micro-costing procedure was conducted in close collaboration with AID and laboratory personnel at the Berlin-Brandenburg Center for Regenerative Therapies (BCRT) at a time (in late 2017) when the stratification process was already standardized for the clinical trial and training of the laboratory staff was completed. The timing of the procedure was chosen in order to account for expected learning effects in the beginning of the clinical trial.
Standard operating procedures (SOPs) for the stratification process (in CELLIMIN study) used in participating study labs have been identified, analyzed and evaluated with regards to necessary input factors. All necessary items were counted, and prices were identified either from officially available product-lists or from hospital-internal catalogues, whenever official prices were not available. On these grounds, it was possible to calculate an upper price-ceiling of one stratification process, as it is done during the trial.
This is an important base for further price calculations that are also input to further health-economic calculations, e.g. break-even calculations for the stratification process.
Details of the micro-costing procedure were presented at the annual meeting on February 2nd, 2018 and are shown in figure 27.
Figure 27: Details of micro-costing procedure for patient-stratification in CELLIMIN
Preliminary cost-model to assess differences in IS medication costs between study arms vs. cost of pre-tx stratification (presented at annual meeting on Feburary 2nd, 2018)
The primary objective for the preliminary cost-model was to assess IS-medication costs per patient and per treatment arm, using the medication plan from the CELLIMIN protocol and to combine them with the cost for pre-Tx stratification process (result from micro-costing procedure) to assess whether the reduced costs of IS-medication in one study arm outweigh the cost pre-Tx stratification. This objective can also be expressed in the form of a decision problem, which seeks to find the best treatment strategy (the strategy that is associated with lowest costs). Therefore, two different treatment strategies had to be assessed and compared. They were called: Stratified treatment vs. high IS for all (Standard of Care).
The decision problem is illustrated in figure 28.
Figure 28: Underlying decision problem of preliminary cost-model
The share of ELISpoT-positive and negative patients was assumed to be 55% and 45% respectively for the stratified treatment strategy and the results of the micro-costing and IS-medication costs (based on medication plan and respective dosages in study protocol) served as inputs.
Generalized medications and respective dosages were obtained from study protocol and expert opinion by Prof. Reinke (for Tacrolimus). The costs of IS-medication were calculated for each day of the trial using prices obtained from Rote-Liste (official German market prices of pharmaceuticals). Costs for each day of the trial were calculated for Arm A (high IS) and Arm B (low IS), and total IS medication costs were calculated for the full follow-up period until M24.
Using the outlined model-structure and described inputs, it was possible to calculate generalized IS-medication costs for each study arm, and treatment strategy. Using these results and the given assumptions, a break-even point was calculated in terms of time using an upper price-ceiling for the patient-stratification process. The results were presented at the annual meeting on February 2nd, 2018.
Contributions to WP3 (RIMINI trial)
The development of a health-economic data assessment process for the RIMINI trial was aligned with the CELLIMIN study in order to ensure comparability of results between clinical trials, because both trials evaluate IS medication protocols following kidney transplantation in comparable patient cohorts. Data quality checks were also defined equivalently for RIMINI for the same reason. In contrast to these aspects, the data analyses plan and outcome evaluation of the RIMINI trial will differ from CELLIMIN due to its single-arm character, although all health-economic data and endpoints are identical.
The health-economic data assessment strategy was developed in close collaboration with the principal investigator (PI, Prof. Ondrej Viklicky) and Prof. Dr. Petra Reinke, and is based on health-economic guidelines for health-economic evaluations alongside clinical trials (Edwards et al., 2008; Petrou & Gray, 2011; Ramsey et al., 2005; Sullivan et al., 2005). Advantages and disadvantages of using different assessment strategies were presented and discussed during several BioDrIM meetings, due to several significant changes of the study protocol caused by changes of consortium structure. The final assessment strategy was consented with the principal investigator and the rest of the consortium.
Regular meetings and reviews with the Berlin study nurse were carried out in order to clarify and communicate appropriate use of the developed health-economic questionnaires since study initiation.
The health-economic data assessment strategy of the RIMINI trial is separated into three major steps:
1. Assessment of Quality of Life (QoL) at baseline, months 1, 3, 6, 12, 18 and 24 and subsequent computation of Quality-Adjusted Life Years (QALYs)
2. Assessment of patient-specific resource consumption at initial discharge following transplantation, repeated discharges (if occurred) and regularly at months 3, 6, 12, 18 and 24
QALYs are currently being assessed in the RIMINI trial using the generic EQ5D-5L questionnaire, which is a standardized and well accepted instrument in the scientific community of health-economics. A working instruction, specifically designed for the RIMINI trial, was developed and uploaded to internal BIO-DrIM website, in order to make the questionnaire as well as the working instruction available for all participating study nurses and other trial staff.
In order to precisely capture other generic aspects of quality of life (qol) which are not covered by the EQ5D-5L instrument, the implementation of the SF36 questionnaire – 10 Authors: Simon Weber, Dr. Malte Pietzsch (Cellogic GmbH) another well accepted qol-questionnaire – was discussed and agreed. The SF36 questionnaire and a corresponding working instruction were uploaded to the internal BioDrIM website, in order to make them available for all participating study nurses and other trial staff.
As suggested by guidelines on health-economic data assessments in clinical trials (Edwards et al., 2008; Petrou & Gray, 2011; Ramsey et al., 2005; Sullivan et al., 2005), the disease-specific questionnaire KTQ-25 was proposed to the PI and agreed to implement.
Furthermore, a trial specific health-economic questionnaire was developed to capture individual health-care resource consumption, during the follow-up of each enrolled patient. The development of this questionnaire had to be well balanced in terms of efforts to be undertaken by the trial staff to assess the needed data in patient-interviews and the precision of expected results. This questionnaire as well as a corresponding working instruction was developed in close collaboration between the PI and Cellogic GmbH and both documents were uploaded to the internal BioDrIM website, in order to
make them available for all participating study nurses and other trial staff.
Regular checks of the assessed health-economic trial data have been defined in close collaboration between the PI, KKS-Charité (CRO), who is responsible for data management and storage, and Cellogic GmbH. Every three months, the complete trial dataset of all health-economic electronic case-report forms (eCRF) of all included patients are submitted to Cellogic GmbH and checked for completeness and consistency. In case of systematically missing data or inconsistent values, adequate measures, e.g. intensified briefing and training of study nurses, are taken to prevent poor data quality and more missing data in future assessments. The results of a recent completeness-check were presented at the last annual meeting on February 2nd, and are shown in figure 29.
Figure 29. Results of completeness-check (RIMINI), Data extraction date: November 22nd, 2017
WP5 Mechanisms behind successful minimizing immunosuppression (IS)
Work done by the Oxford group
The aim of this study is to generate a clinically-relevant mouse model that reflects the variability in alloresponses seen in clinical transplantation. We sought to develop a model that would facilitate the assessesment of the differential efficacy of treatments in “high responders” (patients who suffer severe and/or frequent rejection episodes) versus “low responders” (patients at a lower risk of rejection episodes ant potentially liable to over-immunosupression). Ultimately, we hope that the results of these studies would inform optimal treatment strategies for patients stratified into these different groups.
As the basis for this pre-clinical model, we used a humanised mouse model of skin allograft rejection developed in our lab, which has previously informed clinical trials in cell therapy1,2 . This model represents a robust immunological response to allogeneic tissue, most resembling that of “high responder” recipients. Thus, a modification of this model was required in order to adequately represent the “low responder” phenotype.
Historical reports suggest that a strong risk factor for early graft rejection is a high frequency of memory lymphocytes3 . Memory lymphocytes are generated by maturation of effector lymphocytes during the course of antigen encounter and persist for many years even following clearance of the antigen. Upon re-exposure to antigen, memory cells have a lower activation threshold than naïve cells and mediate a more robust immune response4. Transplant recipients, particularly older patients or who have had previous antigen exposure through previous transplantation, blood transfusion, or pregnancy, may therefore have specific anti-donor memory cells. Memory-type responses may also occur as a result of antigen receptor cross-reactivity known as heterologous immunity5,6.
Due to the limited in utero exposure to environmental antigens, umbilical cord blood (CB) contains minimal frequencies of memory T cells, as confirmed by early studies7,8. As such, we assessed CB as a candidate model for “low responder” transplant recipients. Conversely, with increasing age and exposure to environmental antigens, adult peripheral blood (AB) becomes enriched for memory lymphocytes generated during previous antigen encounters. Hence, we chose to use adult blood to model “high responder” recipients.
Using a combination of in vitro functional and molecular analyses, we explored the properties of CB and AB T cells. After confirming that these populations exhibit phenotypes suited to represent differential responses to antigen, we assessed the ability of cells from CB or AB to reject an allograft in a humanised mouse model of skin graft rejection. Finally, we obtained preliminary results from an in vivo experiment assessing the efficacy of anti-TNF9 as a treatment to suppress graft rejection in “low-” and “high- responder” models.
Methods overview
Phenotypic and functional characterisation of cord blood and adult peripheral blood T cells
Phenotypic characterisation of cells after adoptive transfer in vivo
In vivo testing of cord blood and adult peripheral blood graft rejection capacity
Results
Phenotypic characterisation of cell populations
Cord blood T cells retain a naïve phenotype after TCR stimulation
Consistent with previous reports, freshly-isolated CB T cells expressed levels of CD45RO protein comparable to that of adult CD45RA+ T cells, and significantly lower than that of adult T cells sorted as CD45RO+. After polyclonal stimulation in vitro for one week, adult CD45RA+ cells upregulated CD45RO expression to a density comparable to that of adult CD45RO+ cells on day 0. CB T cells retained lower expression of CD45RO Likewise, five days after adoptive transfer into immunodeficient mice, very low frequencies of CD45RO+ T cells were detected among CB T cells, compared to AB T cells. Together, these data suggest that CB T cells be more resistant to TCR-signalling-induced maturation, which supports the application of CB cells as a model of “low responders”.
CB T cells express lower levels of apoptosis-inducing molecule CD95
Freshly-isolated CB T cells (both CD4+ and CD8+) expressed lower levels of the cytolytic effector molecule CD95 (Fas) than AB T cells, suggesting that CB T cells are less able to exert cytolytic activity and may be in an earlier state of differentiation.
CB CD4+ T cells express lower levels of the IL-7/IL-15 receptor subunit CD122
As measured by flow cytometry, the density of cell surface CD122 protein expression was significantly higher in freshly-isolated AB CD4+ T cells than among their CB counterparts. The expression of CD122 among CD8+ T cells was not significantly different between CB and AB. This suggests that CB CD4+ T cells may be less responsive to the T cell maturation factors IL-7 and IL-15, which would further support the use of CB cells to model the alloimmunity of “low responders”.
Figure 32. CD122 expression in CB and AB T cells.
MNCs isolated from cord blood (CB) and from peripheral blood of healthy adult donors (AB) were stained with antibodies against CD122. Median fluorescence intensity by flow cytometry, gated on CD3+CD4+ (a) or CD3+CD8+ (b), is plotted for each donor with mean +/-SD for 3 donors per population.
CB and AB T cells express similar levels of costimulatory molecules
Memory T cells exhibit a lower threshold for antigenic stimulation, and a lesser dependency upon costimulation, than naïve T cells. We measured the expression level of two key costimulatory molecules on CB and AB T cells, to assess whether differential expression of these molecules might indicate any differences in sensitivity to costimulation. CB and AB CD4+ T cells expressed comparable cell surface protein levels of costimulatory molecules CD27 and CD28. Meanwhile, CD8+ T cells from AB expressed lower levels of these costimulatory proteins than those in CB. However, our functional assays (below) suggest that the higher costimulatory molecule expression in CB CD8+ T cells does not confer upon these cells more potent activity.
Regulatory T cells are present at lower frequency in cord blood than adult blood
The proportion of regulatory T cells (Tregs) among CD4+ T cells was lower in cord blood than adult peripheral blood. With this regulatory cell population enriched in adult PBMCs, it is possible that there is enhanced Treg-mediated suppression of effector cell function in AB cells.
Figure 34. Treg frequency in CB and AB T cells.
MNCs isolated from cord blood (CB) and from peripheral blood of healthy adult donors (AB) were stained with antibodies against CD3, CD4, FOXP3 and CD127. Percentgae of FOXP3+CD127lo cells among CD3+CD4+ lymphocytes was enumerated by flow cytometry, and plotted as mean +/-SD for 3 donors per population.
Functional characterisation of cell populations
In vitro proliferative capacity is comparable between cord and adult T cells
After polyclonal stimulation in vitro for three days, the proliferation index of CB T cells and of sorted adult CD45RA+ and CD45RO+ T cells was not significantly different. Meanwhile, the level of transcription of CCND1, a gene encoding the cell cycle initiation protein Cyclin D1, was not significantly different between these three cell populations. There were also no significant differences in CCND1 levels between the CD45RA+ and CD45RO+ subpopulations of CB and AB, although a non-significant trend towards higher expression in CB CD45RO+ cells is apparent. These results indicate that CB T cells are capable of proliferating in response to TCR stimulation.
Figure 35. in vitro proliferation of CB and AB T cells in response to polyclonal TCR stimulation
(A) CD45RA+ and CD45RA-CD45RO+ cells were selected from AB MNCs by magnetic bead-based selection. These cells, alongside total CB MNCs, were stained with violet proliferation dye and cultured for 72hrs with αCD3α CD28-coated beads. Dilution of the proliferation dye was measure by flow cytometry and used to calculate a division index (mean number of cell division cycles per progenitor cell). Mean alues +/-SEM are plotted for three donors with at least three technical replicates. (B) total CD3+ (“T”), CD3+CD45RA+CD45RO- (“A”) and CD3+CD45RO+CD45RA- (“O”) T cells were sorted by FACS from freshly-isolated CB and AB MNCs. RNA isolated from these cell opulations was examined by qRT-PCR for transcription of the cell cycle gene CCND1. The mean value of two replicates is plotted for each cell donor, with lines linking autologous samples.
Upregulation of activation markers upon stimulation is lower in CB T cells
Overall, there is a trend toward lower expression of activation markers in CB T cells relative to their equivalent AB populations, although these differences are not statistically significant between CD45RA+ and CD45RO+ subsets for HLA-DR. These results suggest that CB T cells may have a lower activation potential.
CB T cells secrete minimal levels of cytokines in response to TCR stimulation
Cytokines were generally secreted at much higher concentrations by adult antigen-experienced T cells than by either CB T cells or adult naïve T cells. Whilst adult naïve T cells secreted more after stimulation than in their resting state, CB T cells failed to secrete the majority of cytokines even when stimulated. Of particular interest are the cases where CB T cells fail to produce a cytokine that is secreted by adult naïve T cells, such as IL-2 and TNFα. These results may indicate a functional deficit in CB T cells that prohibits a robust response by these cord blood cells against alloantigen.
Figure 37 cytokine secretion by CB and AB T cells in response to polyclonal TCR stimulation
(A) CD45RA+ and CD45RA-CD45RO+ cells were selected from AB MNCs by magnetic bead-based selection. These cells, alongside total CB MNCs, were cultured for 72hrs with αCD3α CD28-coated beads. Supernatents were analysed using a flow cytometric bead assay to determine the concentration of (left to right, top to bottom) IL-2, IL-4, IL-5, IFNγ, IL-4, IL-9, IL-10, TNFα, IL-13, IL-17A, IL-17F and IL-22. Concentrations of each cytokine are plotted as the mean of three replicates for each donor +/- SEM across three donors.
Establishment of in vivo model
CB MNCs cells reject allogeneic skin with delayed kinetics relative to AB PBMCs
Skin grafts were rejected with a median survival time of 30.5 days in mice reconstituted with AB and 83.0 days in mice reconstituted with CB. The differential rejection kinetics mediated by adult and cord cells demonstrates that these two cell populations can be used to model the more rapid rejection seen in “high responder” recipients and the milder, delayed rejection seen in “low responder” recipients, respectively. In both groups, MNCs were depleted of CD4+CD25+ Tregs pre-injection.
Testing of a putative therapy (anti-TNF) in an in vivo model of high and low responsiveness
Our data indicate that there may be some prolongation of graft survival in the presence of anti-TNF treatment among mice reconstituted with AB MNCs. This experiment also appears to validate the model, given that CB-reconstituted mice have failed to reject their skin grafts with the rapidity of AB-reconstituted mice.
Conclusions
From the data presented here, we conclude that cord blood and adult peripheral blood T cells exhibit differential immunological responses to alloantigen both in vitro and in vivo. These distinctive responses may therefore represent the immune phenotypes of “low responder” and “high responder” transplant recipients, respectively. Indeed, our finding that CB MNCs cells mediate delayed graft rejection in a humanised mouse model of skin allograft rejection confirms that these cells are less responsive to allogeneic stimuli than adult PBMCs and thus provide an adequate model of “low responder” recipients. We suggest that this pair of models could be used in parallel to assess putative therapies in order to examine the differential efficacy or dosage requirements in low versus high responder transplant recipients. Our in vivo experiment assessing anti-TNF treatment as an intervention in the low and high responder models is a useful preliminary investigation of this therapy in a humanised transplantation model.
References
1. Nadig, S. N. et al. in vivo Prevention of Transplant Arteriosclerosis by ex vivo Expanded Human Regulatory T Cells. Nat. Med. 16, 809–813 (2010).
2. Issa, F., Hester, J., Goto, R., Nadig, S. & Goodacre, T. E. Ex vivo -expanded human regulatory T cells prevent the rejection of skin allografts in a humanised mouse model 1. 90, 1321–1327 (2010).
3. Valujskikh, A. & Lakkis, F. G. In remembrance of things past: Memory T cells and transplant rejection. Immunological Reviews 196, 65–74 (2003).
4. Farber, D. L., Yudanin, N. A. & Restifo, N. P. Human memory T cells: Generation, compartmentalization and homeostasis. Nat. Rev. Immunol. 14, 24–35 (2014).
5. Adams, A. B. et al. Heterologous immunity provides a potent barrier to transplantation tolerance. J. Clin. Invest. 111, 1887–1895 (2003).
6. Heuvel, H. Van Den, Heidt, S., Roelen, D. L. & Claas, F. H. J. T-cell alloreactivity and transplantation outcome: A budding role for heterologous immunity? Current Opinion in Organ Transplantation 20, 454–460 (2015).
7. Han, P., Hodge, G., Story, C. & Xu, X. Phenotypic analysis of functional T-lymphocyte subtypes and natural killer cells in human cord blood: relevance to umbilical cord blood transplantation. Br J Haematol 89, 733–740 (1995).
8. Garderet, L. et al. The umbilical cord blood alphabeta T-cell repertoire: characteristics of a polyclonal and naive but completely formed repertoire. Blood 91, 340–346 (1998).
9. Feldmann, M. & Maini, R. N. TNF defined as a therapeutic target for rheumatoid arthritis and other autoimmune diseases. Nat Med 9, 1245–1250 (2003).
Work done by the Nantes group
Global aims of the project
There is a high interest today to use animal models in order to understand mechanisms of tolerance and rejection in transplantation but the major difficulty is to find the”best” model. Indeed, most of studies are performed using naive rodents, breeding in pathogen-free facilities. In order to achieve one of the goal of the BIO-DrIM project, i.e. developing rodent models closer to the “immunity reality”, we wished to develop an in vivo model of acute rejection mediated by memory T cells. The aim was then to evaluate the efficacy of cell therapy or new immunosuppressants in these models. We decided to work in parallel in rats and mice (Part II of this report). Moreover, we wished to identify potential new biomarkers of tolerance and chronic rejection using our own rat models (Part I of this report).
Part I: Definition of new biomarkers
A part of this project consists on the identification of new biomarkers of chronic rejection and tolerance. To achieve this goal, we performed microarrays on heart allografts providing either from tolerant rats (following a 20 days-LF15-0195 treatment, a relevant model that mimics the B cell accumulation observed in blood from operationally-tolerant patients)1 or from chronic rejected rats (following donor blood transfusion). Amongst the 110 genes differentially expressed in allograft from tolerant and chronic rejected rats, we studied more precisely two of them, CLEC-1 and LIME. From blood analysis of these two groups of rats, we suggest that CLEC-1 could be potentially used as a blood diagnostic biomarker of chronic rejection. Conversely, the expression of the LIME biomarker was detected in rats that will develop chronic rejection; we then propose LIME as a blood prognostic biomarker of chronic rejection.
In the future, it will be interested to analyze the expression of these two genes in blood samples from transplanted patients in order to validate or not these results observed in our rat models.
Part II: Rodent models of acute rejection
A. First experiments to set up a model of acute rejection based on the literature:
To define an animal model of acute rejection mediated by memory T cells, we first used two different approaches described in the literature in rodents. The first one consists on an intraperitoneal injection of donor splenocytes two weeks before transplantation2. Using this model, the authors demonstrated an acceleration of the skin graft rejection in mice. In accordance with the project, we injected donor splenic T cells (total splenocytes were used in the article) two weeks before transplantation. By this system, no acceleration of allograft survival was observed either in a mouse skin allograft (Figure 40) or in a rat cardiac allograft models.
Figure 40. IP injection of donor T cells does not contribute to accelerated graft rejection in skin graft models in mice
C57BL6 female mice received IP injection of CD90+ T cells from male mice two weeks before male skin graft. A. Design of the experiment B. Graph depicts skin graft survival curves. Control mice received skin graft in the absence of IP injection of cells.
We then tested a second protocol, based on a sensitization of recipient animals by a first graft3,4. As we would like to focus only on the effects of memory T cells, we could not use the same recipient for the two grafts because the first one would induce memory T cells and memory B cells. So, we decided to do a first skin graft on few animals, isolate T cells from these animals one month post-transplantation and then injected these T cells in new recipient animals by intravenous route one week before transplantation. In mice, by performing skin grafts for sensitization and analysis steps, no graft acceleration was obtained. In the rat model, the first recipients (Lew1W) received skin allograft from Lew1A rats. Recipient rats were then sacrificed and their T cells were injected in naive Lew1W rats in order to sensitize them prior to a cardiac Lew1A allograft a week after T cell injection. Using this protocol, we obtained a slight acceleration of heart rejection in sensitized animals compared to non-sensitized animals (rats receiving only cardiac allograft) (Figure 41).
Figure 41. IV injection of T cells from sensitized animals slightly accelerates graft rejection in heart graft model in rats
Lewis1W rats were sensitized with donor antigen by receiving skin graft from Lewis1A rats. One month later, rats were sacrificed and T cells were collected and IV injected to a new Lewis1W recipient, seven days before Lewis 1A heart transplantation. A. Design of the experiment B. Graph depicts heart graft survival curves. Control rats received heart graft in the absence of injection of cells.
B. Improvement of the rat model of acute rejection and evaluation of Molecule X effect
In order to improve the rat model of heart rejection, we set up a new model of accelerated graft rejection following adoptive transfer of cells obtained from polygrafted animals. In addition to the new skin graft on sensitized rats, several modifications were done in this new model: i) the donor/recipient combination, in order to increase the immune response in sensitized animals, ii) the delays between grafts and sacrifice on first animals and the delay between cell injection and heart transplantation, in order to preserve the cells of interest, and iii) enrichment of T cells were replaced by total immune cells of draining LNs. In this new model, Lewis 1A rats received a first skin graft from Lewis 1W rats. One week later, a second graft from the same donor was performed on the same recipient rats. Rats were sacrificed 7 days later to isolate graft-draining LN cells. Analysis of these draining LN cells showed that cells from polygrafted mice displayed a memory phenotype and were able to induce a stronger alloresponse than cells isolated from naive rats, both in terms of T cell proliferation and T cell activation. To evaluate the in vivo role of these cells, graft-draining LN cells from polygrafted animals were intravenously injected to naive recipient Lewis 1A the day before Lewis 1W cardiac transplantation. Interestingly, the adoptive transfer of draining LN cells from polygrafted animals induced a more reproducible acceleration of allograft rejection (Figure 42).
Figure 42. IV injection of immune cells from highly sensitized animals accelerates graft rejection in heart graft model in rats
Lewis1A rats were sensitized with donor antigen by receiving 2 skin grafts from Lewis1W rats. One week later, rats were sacrificed and draining lymph nodes (DrLN) cells were collected and IV injected to a new Lewis1A recipient, the day before Lewis 1W heart transplantation. A. Design of the experiment B. Graph depicts heart graft survival curves. Control rats received heart graft in absence of injection of cells.
In parallel, works from our center recently reported that a new molecule (Molecule X, requested patent) was able to delay graft rejection in non-immunized rats. Administration of this molecule in our accelerated heart rejection model showed that it was also able to slightly prolong graft survival in this model. In order to study the properties of these memory cells, we next decided to move to a model of immunodeficient rats in order to accentuate their effects.
C. Exacerbation of the acute rejection model using immunodeficient rats
The use of immunodeficient mice is well described in the literature5 and allows to efficiently characterize the role of immune cells. In order to obtain a good extinction of the immune cells, genes coding for Rag2 and for IL2Recepteur γ were knock out in these immunodeficient mice. Based on these models, Rag1-/- IL2Recepteur γ-/- rats (RRG) were generated in the laboratory on Sprague Dawley genetic background. These rats were obtained by the breeding of Rag1-/- rats and IL2-Recepteur γ-/- rats. The absence of T cells, B cells and NK cells were validated in these rats, as observed in mice. We also observed that transplanted RRG animals do not reject an allograft, confirming that the immune system was deficient.
However, our results showed that the adoptive transfer of naïve LN cells from immunocompetent rats induces heart rejection in these immunodeficient rats. More interestingly, the injection of graft-draining LN cells from polygrafted immunocompetent animals (obtained as described in the previous section) allows a significant accelerated graft rejection compared to the transfer of naïve LN cells (Figure 43). Our first analyses performed on RRG animals receiving either sensitized LN cells or naïve LN cells from immunocompetent rats do not allow to explain the mechanisms of action of these sensitized animals.
Figure 43. IV injection of immune cells from highly sensitized animals accelerates graft rejection in heart graft model in RRG rats
Sprague-Dawley rats were sensitized with donor antigen by receiving 2 skin grafts from Lewis1W rats. One week later, rats were sacrificed and draining lymph nodes (DrLN) cells were collected and IV injected to RRG recipient rats, the day before Lewis 1W heart transplantation. In controls, RRG rats received IV injection of LN from naïve (non-immunized) Sprague Dawley rats. A. Design of the experiment. B. Graph depicts heart graft survival curves.
In conclusion of this part, we set up a robust model of accelerated heart rejection in RRG rats. In the future, we want to identify the cells responsible for these accelerated rejection. For that, we will in depth phenotype naive and sensitized LN cells and investigate whether a specific marker of a key population exists to understand this immunization. We will also perform in vitro assays of these two types of LN cells to study their proliferative capacities and their cytokine releases. Following injection of these two types of cells in RRG rats, we will also in-depth screen the engrafted cells in blood (T cell subsets). Lastly, we will evaluate the potential of new molecules and tolerogenic/regulatory cells in this model.
Part III: Experiments in immunodeficient mice in collaboration with the Oxford team
In accordance with the scientific advisory board the aims of WP5 for the last two years of the BioDRIM consortium have been modified. Oxford, Nantes and Berlin propose to work together to develop a humanised mice model of « Experimental system to investigate ‘high responder’ recipients who have pre-existing donor reactive memory cells ».
The first aim of this new project consists in setting up in Nantes the model of human skin graft in humanized mice according to the Oxford protocol.
In this model, NOD/SCID/IL-2Rγ-/- (NSG) mice were grafted with human skin graft (obtained from burns and plastic surgery unit of Nantes Hospital) and then received 5 millions of fresh human PBMCs by IV route (5 weeks later). We observed that these mice reject human skin with a mean rejection time of 30 days. This model was thus successfully set up in Nantes. We then investigated whether this rejection was also possible following the injection of frozen human PBMCs. Our results showed that the injection of 5 millions of frozen PBMC also allows human skin graft rejection (Figure 44A). In the two experiments performed from different donors of skin and PBMCs, the mean time of graft rejection was very similar whereas the engraftments were different between both experiments (Figure 44B). Interestingly, thawed PBMCs were able to efficiently induce both graft rejection and GVHD development. Administration of immunoregulatory cells appears today as an attractive strategy to control graft rejection. For safety reasons, the use of autologous cells appears particularly promising and the use of thawed PBMCs allows to induce graft rejection in humanized mice and also to generate and inject imunoregulatory cells in an autologous way.
Figure 44. Thawed human PBMCs efficiently induce graft rejection in a model of human skin graft in humanized mice
NSG mice received human skin graft and were injected with either fresh or thawed PBMC 5 weeks later. A. Human skin graft survival was monitored in these groups of mice from two different experiments. Four mice (2 in each experiment) were sacrificed before graft rejection due to a weight loss >20% of initial weight (GVHD development) B. The engraftment of human cells was measured in blood of mice receiving skin graft and thawed PBMCs from two different experiments.
In relation with the new results from the Oxford team, we then investigated whether the injection of umbilical cord blood cells induces a less vigorous rejection response in vivo than the injection of PBMCs. For that, umbilical cord blood cells were obtained from the maternity unit of Nantes Hospital and freezed. Following thaw-out, the results of our first experiment show that umbilical cord blood cells induce graft rejection and blood cell engraftment. However, we obtained similar time of graft rejection following administration of PBMCs and umbilical cord blood cells. Consequently, our experiment with umbilical cord blood cells in Nantes does not reproduce Oxford datas. To explain these differences, it has to be considered that we performed only one experiment so far, it then has to be reproduced. Furthermore, we obtained a high mortality in umbilical cord blood cells following thaw-out. A phenotype of these cells has to be done to investigate whether these cells are modified following thaw-out. Lastly, it is important to note that these differences could be also explained by the strains of immunodeficient mice used. Indeed, we used NOD/SCID/IL-2Rγ-/- whereas BALB/c Rag2-/-/IL-2Rγ-/- were used in Oxford.
References
1. Le Texier et al. 2011. Long-term allograft tolerance is characterized by the accumulation of B cells exhibiting an inhibited profile. Am. J. Transplant., 11: 429-438
2. Greenlaw et al. 2008. An antibody combination that targets activated T cells extends graft survival in sensitized recipients. AJT, 8: 2272-2282
3. Wang et al. 2007. Inhibition of terminal complement components in presensitized transplant recipients prevents antibody-mediated rejection leading to long-term graft survival and accommodation. JI, 179: 4451-4463
4. Ramsey et al. 2013. Anti-LFA-1 or rapamycin overcome costimulation blockade-resistant rejection in sensitized bone marrow recipients. Transplant. Int., 26: 206-218
5. Shultz et al. 2007. Humanized mice in translational biomedical research. Nat Rev Immunol. 7, 118130.
Potential Impact:
The focus of BIO-DrIM was to contribute to an improved and cost-effective long-term outcome of solid organ transplantation by implementing decision-making biomarkers into the clinical management of transplant patients (personalized IS) but the results are more broadly relevant also to other ancillary disciplines and have very important exploitable potential. At this aim, three research performing SMEs and two big companies were involved in the project with very deep commitment.
The novelty of this project, that can be considered overall an innovative approach to an important unmet need and a research infrastructure, is the stratification of transplanted patients regarding their individual immunological responsiveness to the allograft and their respective individual need of IS. A second area of novelty is the integrative design of this program, whereby a direct comparison (feasibility, safety, cost and promise of effect) of biomarker driven strategies for personalized therapies is foreseen in 5 innovative investigator-driven biomarker clinical trials designed by the consortium.
The outputs and results that we have generated during this project have a strong potential for further commercial exploitation, with the expectation that personalized therapy can ultimately optimize the need for immunosuppressive drugs in organ transplant recipients.
The most important impact of BIO-DrIM are the first steps and useful results that will drive the scientific community to the personalization of immunosuppressive therapy in organ transplant recipients, thereby significantly improving their survival and quality of life, while at the same time decreasing health care costs.
Secondarily, the results of BIO-DrIM broadly impact the treatment of other diseases related to undesired immune reactions, such as autoimmune diseases and graft-versus-host disease after hematopoetic stem cell transplantation. In addition, it might have also impact on the new field of cell transplantation (e.g. adult differentiated cells, like hepatocytes; but also allogeneic stem-cell derived cell products); in other words diseases with chronic or repetitive IS treatment.
Beside the results that we are confident that will directly impact on the patients in the short-medium time, the BIO-DrIM Consortium generated important outputs that can be summarised as below:
-Fostering of small SMEs, which due to the project were able to gain on visibility, develop new products, increase sales, increase employment numbers and interact with big pharma
-BIO-DrIM project as a prototype of new infrastructure (results and data from previous EU funded project could be exploited in many interacting field and continue to be in use)
-New approach of designing clinical trials (i.e. Simon design)
-Development of a new type of eCRF which includes biomarkers that could be transferred to other clinical trial that were funded by the German Federal Ministry of Education and Research (BMBF)
-Validated and robust biomarker assays established at different platforms & training on biomarker performance
-Implementation of health economic an HPQL assessment
-Experiences with new regulatory processes for EU multicenter trials (Voluntary Hamonization Process VHP)
List of Websites:
www.biodrim.eu
