Biomarker Development for Postoperative Cognitive Impairment in the Elderly

Periodic Report Summary 4 - BIOCOG (Biomarker Development for Postoperative Cognitive Impairment in the Elderly)

Project Context and Objectives:
Cognitive impairment is increasingly prevalent in our aging society as result of different, interacting medical conditions. It is a consequence of age-associated primary brain disorders (depression, Alzheimer’s dementia (AD), cerebrovascular disease etc.) and/or secondary brain disorders due to medical conditions like diabetes, inflammation or treatment interventions and life style factors. Postoperative Cognitive Impairment is a prime example in this regard and is characterized by the progressive deterioration of sensory and cognitive function following surgery with incidences of up to 20-80%* – particularly in but not limited to elderly patients. The condition often has an acute phase of Postoperative Delirium (POD) which then may be followed by a more chronic phase of Postoperative Cognitive Dysfunction (POCD), which tends to persist over time. Increased age as well as a range of medical conditions may predispose patients to POD/POCD. In older patients, POCD resembles chronic dementia and appears to accelerate the cognitive decline in AD. In our aging society the socioeconomic implications of postoperative cognitive impairments are therefore profound as POD/POCD are associated with longer and more costly hospital treatment, increased mortality and dependency on social transfer payments. Understanding POD/POCD thus constitutes an urgent medical need. Yet, there are hardly any treatments available, partly due to a lack of understanding of the pathomechanisms.
Research aimed at understanding the diverse pathomechanisms of POD/POCD requires a cross-cutting “systems medicine” approach, with different medical disciplines working together: the study of different physiological and molecular mechanisms and the identification of a range of clinically valid biomarkers, whose description is limited at present. Integration of neuroimaging biomarkers, which provide direct information on brain structure/function with high sensitivity, combined with complementary knowledge from molecular biomarkers (cholinergic, inflammatory, metabolic) as obtained from body fluids (plasma, cerebrospinal fluid, CSF) is expected to allow accurate patient stratification (subtyping).
Within BioCog, we will establish valid biomarkers panels for risk and clinical outcome prediction of POD/POCD. The study population will be 1200 patients, aged 65 to 85 undergoing major elective surgery. In this respect, BioCog is the most comprehensive study in the field as compared to other studies worldwide.
The resulting expert system is expected:
1) to support clinical decision-making in patient care, e.g. to balance the individual POD/POCD risk against the expected overall clinical outcome of a surgical intervention,
2) to allow the design of more sophisticated and hypothesis-driven clinical studies and drug trials (translational research) in the future to evaluate the role of causality in risk factor associations with POD/POCD.
These challenges can only be met by close collaboration between experienced academic and industrial groups, in a multinational and multidisciplinary network, as realized in the present consortium, BioCog.
Furthermore, a state-of-the art clinical database and biobank will be created to become an integral part of the European Biobanking and Biomolecular Resources Research Infrastructure (BBMRI) and will allow to address specific and hypothesis-driven research questions.
We will further analyse the health-economic impact of biomarker-guided strategy.

In summary, our proposed research project is envisaged to improve the way in which medical care and research is conducted in older pre-surgical and surgical patients in the future.
BioCog represents also an opportunity to strengthen the economic competitiveness of the European biomarker research industry: the involved SMEs Immundiagnostik, ATLAS Biolabs, PharmaImage will be provided with a competitive edge in biomarker research simply by having contributed to this project – in particular with regard to future business with research and development (R&D) departments of the pharmaceutical industry.

Project Results:
Neuroimaging biomarkers are expected to allow the detection of relevant aspects of POD/POCD pathophysiology with high sensitivity whereas molecular biomarkers promise high specificity. Developing an expert system on the basis of neuroimaging and molecular biomarkers requires complementary expertise from the fields of clinical (WP1), neuroimaging biomarker (WP2) and molecular biomarker (WP3). For a meaningful integration and interpretation of data from multiple domains, powerful computational tools are needed. The work package bioinformatics (WP4) is critical, as appropriate integration of sufficiently specific and sensitive biomarkers needs to be based on rigorous statistical procedures.
Efficient project management, in synergy with the scientific and business coordination objectives, is a central task within the project (WP5). Dissemination and exploitation strategies have an important role in the BioCog project. They will ensure proper visibility of the project. Strategies will be implemented to target future stakeholders such as the European Commission, clinicians, the academic community, as well as biotech and pharmaceutical companies (WP6).

The specific main activities performed so far are:

WP1
The screening of patients started in Charité Berlin, Campus Virchow-Klinikum on 20/10/2014 and in Campus Charité-Mitte on 08/06/2015. In Utrecht screening started on 12/01/2015. As of 31/07/2018, a total of 7568 patients were screened (5294 patients were screened in Berlin and 2274 patients were screened in Utrecht). 1033 patients have been included: 747 patients in Berlin, 286 patients in Utrecht. In addition, 114 control subjects (Subjects with no surgical intervention) have been included (35 subjects in Berlin and 79 subjects in Utrecht).
In total: N = 1147 patients and controls have been included.
Regarding the 3-month follow-up visit, N = 369 patients did not complete all study visits due to drop out, loss to follow-up or death: N = 316 in Berlin and N = 53 patients in Utrecht.
The follow-up visit T3 (i.e. 3 months postoperative) was completed for N = 664 patients (N = 431 patients in Berlin and N = 233 patients in Utrecht).
The recruitment of the POCD control group has been started on 13/06/2015 in Berlin and was finalised. Remark: we have decided to reduce the target N = 1200 surgical patients to be included into the study down to N = 1033 and to investigate instead more control subjects (with no surgical intervention).
All drawn blood samples have been collected under the directive protocol of the MDC (WP3). Every patient has received neuropsychological testing via CANTAB battery, geriatric assessment, delirium screening and further examination (EEG, heart rate variability) per protocol. Transthoracic echocardiography (TTE) has been performed in 43 patients in Berlin according to the availability of certified examiners.
In addition to the previous reporting period, the Standard Operating Procedure (SOP) concerning drop-out criteria and regarding primary endpoint of Delirium was updated.
Applications for ethical approval have been submitted to the relevant ethics committees in Berlin and Utrecht and encompass the work within WP1, 2, 3 (for details, please see “Description of the work performed during the...”).
For further analysis of study data, all data had to be entered into an eCRF, curated and checked for plausibility. This working step was strongly supported by staff from the PharmaImage GmbH (Anna-Leah Hüther, Pola Neuling, Nele Klempnow, Lukas Roediger, Charlotte Braun, Saya Speidel, Bennett Borak, Fieras Nosierat, Zdravka Bosancic, Juliane Dörfler). This working step has been successfully completed during early summer 2018. All curated data have now been transferred to the data administration system XNAT (located at PharmaImage GmbH) for integration with Biomarker/Imaging data. We are working together with a new partner the Institute of Biometry and Clinical Epidemiology of the Charité- Universitätsmedizin Berlin. See Part B Core of the report for the details.
WP2
The MRI protocol has been finalized and harmonized between the Charité Berlin and the UMC Utrecht with support from the Cambridge University Hospital and PharmaImage GmbH (Berlin). The MRI contract/agreement was available on 17/11/2014 (start of neuroimaging data collection). A standard operating procedure (SOP) for data collection as well as a protocol for data handling has been developed with central (raw) data storage in a secured research PACS at the UMC Utrecht. In addition, the Extensible Neuroimaging Archive Toolkit (XNAT) for the management of raw and processed neuroimaging data (and related clinical and molecular data) was implemented at PharmaImage (Berlin). First analyses based on the MRI data were presented at the 2015 BioCog Annual Meeting in June, the BioCog winter school in 2016 the BioCog Annual Meeting in September 2016 in Sienna, the BioCog in Potsdam, March 2017, Palma de Mallorca September 2017, Winter School March 2018. First results presented included pilot analyses (interim analyses) of local or global brain volume relations with POD and POCD. Furthermore, analyses were presented with the collected resting state functional MRI/EEG data and diffusion tensor imaging data. In addition to POD and POCD the relation between anatomical and functional MRI data and other clinical parameter such as frailty and cognitive performance were analysed. Until 07/2018, a total of pre-operative neuroimaging data sets was successfully collected in 363 patients in Berlin (incl. 17 control participants) and 224 patients in Utrecht (incl. 46 control participants). Follow-up neuroimaging data collection (after three months) was conducted in 391 patients in Berlin and Utrecht together (incl. 40 control participants). For comparison of neuroimaging sites, N = 15 subjects obtained measurements both in Utrecht and Berlin. We are currently also collecting neuroimaging data (2-year post-op). All neuroimaging data have now been analysed on the subject level (structural MRI incl. high-res hippocampus, fMRI, ASL, Flair, DTI).
Hosted by PharmaImage GmbH, the data administration system XNAT has been fully implemented which required extensive programming, data curating and plausibility checks. We have also integrated several (pre-)processing pipelines, e.g. for automated analysis of clinical data incl. imaging data (SPM12, Matlab, FreeSurfer53.0). All neuroimaging data (raw data, pre-processed data, analysed data), clinical data, neuropsychological data and laboratory values as well as associated meta-data and ontologies have now been up-loaded in the system. For data exchange and data integration between XNAT (clinical data, neuroimaging data, laboratory values) and the University of Luxembourg in-house data administration system (with omics data), the tranSMART platform (http://transmartfoundation.org/) was implemented together with University of Luxembourg during the current funding period. On the basis of this integrated platform, we are currently conducting statistical analyses including machine learning algorithms (multivariate biomarker-based prediction of POD/POCD) on the basis of the training (exploratory) data set and the test (validation) data set, i.e. using the entire data set. This multivariate biomarker-based prediction paper is supposed to be the first major paper of the BioCog consortium being likely finalized early 2019.
In order to get a deeper understanding of the data, a first wave of publication projects has already been conducted addressing only pre-operative data (interim data analyses using the training (exploratory) data set with patient 1-400). By now, seven papers have been published. Together these exploratory findings indicate (in conjunction with our not yet published data on pre-operative prediction of POD/POCD, manuscript in preparation) that certain structural brain characteristics both predict pre-operative cognitive impairment/physical frailty but also post-operative POD/POCD development. See Part B Core of the report for the details.

WP3:
All blood samples had been fully processed prior to the final 6 months of the project. Data for molecular biomarkers had been curated and integrated on an ongoing basis, and had been made available to the consortium whenever new data were received through XNAT. During the final 6 months of the project, this task was completed (with exception of a single biomarker that is currently being analysed as detailed below).
Specifically, during late 2018, data on 7 biomarkers for the ‘validation set’ of the BioCog cohort were received from our partner Immundiagnostik. These 7 biomarkers had been identified as promising candidates in terms of POD/POCD risk prediction in preliminary univariate analyses of the interim sample (a total of 16 biomarkers had been measured by Immundiagnostik in the interim sample during 2017). MDC also completed measurement of IL-6, adiponectin, C-peptide, high molecular weight adiponectin, leptin, soluble leptin receptor and sortilin-related receptor (SORLA) for the full BioCog cohort during this time period.
Analysis of transcription profiles using Clariom S und MicroRNA arrays as well as genotyping using Illumina’s Global Screening Array (GSA), too, were completed by ATLAS Biolabs during the final 6 months of the project. Data resulting from these analyses are currently being processed by our Luxembourg partner in collaboration with WP02 researchers for multivariate modeling.
CNR Italy received whole blood assay (WBA) samples for around 420 patients of the total BioCog cohort during late 2018 (in the previous reporting period, CNR had only received WBA for the interim sample).
During a previous reporting period, CNR had succeeded in analysing plasma from 137 patients from the interim sample, i.e. all those that were complete (3 time points/patient, complete information). Analysis of those received later has been post-poned until new resources become available. Analysis of the results collected in the interim sample suggests that only sIL-1R4 might be worth testing in the plasma samples of the validation sample. CNR has also received additional sets of WBA samples during the final 6 months of the project and has therefore resumed testing of these samples. The test implies the assessment of the immune capacity of blood cells, by examining the reactivity to increasing concentrations of a bacterial challenge. Reactivity is measured as the capacity of producing free active IL-1β, a cytokine with potent immunostimulatory and inflammatory activity. Free active IL-1β is assessed as the concentration of free IL-1β (i.e. the cytokine not bound by its soluble inhibitor sIL-1R2) relative to the concentration of the other IL-1 inhibitor, the receptor antagonist IL-1Ra. Compared to the pre-surgery immune capacity (considered as baseline for each patient), a lower production of free active IL-1β is considered as sign of immunosuppression, whereas an enhanced production of free active IL-1β is considered symptomatic of inflammation. As of February 2019, 80 complete sets of WBA samples had been tested, and patients had been stratified, based on their immunoreactivity, into nine different groups. A large proportion of patients (around 40%) experienced an immediate post-operative immunosuppression followed by normalisation after 3 months. Statistical analysis on associations of immunosuppression vs. inflammation with development of POD and POCD will be performed by members of the consortium during the months to follow. In spite of the project ending, CNR will continue testing the complete sets of WBA samples of the validation set.
Of note, following EU approval for redistribution of financial resources in late 2018, 2 further parameters were added to the list of biomarkers to be analysed from blood during the final 6 months of the project: 1) beta amyloid (Aβ40; Aβ42) and 2) IL-2, IL-8 and IL-10.
For 1), we initially conducted a pilot study for a subset of 200 patients to determine the validity of the analysis method. Funding for that study was obtained independently and included the group of Oliver Peters at Charité thus fostering cross-disciplinary and cross-workpackage collaboration. Following promising results, we decided to go ahead and measure Aβ in the full BioCog cohort. Those analyses are currently ongoing and we expect to receive results from the lab commissioned with the analysis by the end of February 2019.
For 2), we decided to supplement the analyses of IL-6 (which was complete) and IL-18 (which had been planned but not yet been performed) by using a multiplex assay that detects IL-2, IL-8, IL-10 and IL-18 in a single analysis. Results of those analyses were received in January 2019.
A single data file has been created that includes all biomarkers that have been measured in blood. The file – accompanied by a data dictionary – has been entered into XNAT for use by the entire consortium and will be updated once data for the final biomarker (Aβ) become available.
Our objective was also to push ahead with dissemination of our findings during the final part of the project. WP3 staff led by MDC have used BioCog data to investigate associations of the metabolic syndrome with cognitive impairment before surgery in the interim sample. As of February 2019, a manuscript resulting from those analyses was under review.
In preparation of the analyses on the metabolic syndrome and POCD, WP3 has also collaborated with WP1 to investigate metabolic predictors of cognitive impairment and POCD in three cohort studies that have already been completed. Two manuscripts resulted from this collaboration and were published shortly before (August 2018) or during the reporting period (November 2018).

WP4:
Several options for effective communication were evaluated by the Bioinformatics team in Luxembourg Centre for Systems Biomedicine. Together with the other partners of the consortium, it was decided to build the extension into the project web site (internal data exchange platform).
In Luxembourg, the first draft of the knowledge map was realized by text-mining and manual curation and published in the form of review focussing on the systemic effects of POD and POCD. The extensive text mining for the POD-, POCD-associated molecules improved the initial knowledge map.
The second revision included results from the updated in-house textmining tool, that together with the manual curation provided a revised version of the POD and POCD knowledge maps.
The latest work included machine learning analysis of transcriptomics data from the interim cohort, which detected a preliminary molecular signature discriminating POD patients based on preoperative blood measurements. We also detected many genes differentially expressed between patients with and without delirium after surgery.
For data exchange and data integration between the data administration system XNAT, hosted at the PharmaImage GmbH (administration of clinical data, neuroimaging data, laboratory values) and the Luxembourg in-house data administration system (with omics data), the tranSMART platform (http://transmartfoundation.org/) was implemented during the current funding period. On the basis of this integrated platform, we conducted statistical analyses (biomarker-based prediction of POD/POCD).

WP6:
Development of the project visual identity and logo, web site, flyers by Alta together with the coordinator (Charité), Health and Technology Assessment(HTA) by Cellulogic GmbH together with the Depts. of Anesthesiology (Charité/Utrecht). A publication on the health-economic impact of biomarker-guided pre-operative prediction strategy is currently prepared. Based on available internal and public databases, our current cost estimates (for Germany) suggest that annual extra expenses for health insurance carriers are in the range of 0.5-1 Billion EUR per year (POD) and 1-2 Billion EUR per year for POCD.
In addition, we have now started to prepare - together with the Technology Transfer Office of the Charité and under the lead of PharmaImage GmbH – the development of an expert system/App for pre-operative multivariate prediction of POD/POCD risk for medical professionals. The XNAT system, hosted at PharmaImage GmbH, has been set-up such that an expert system can be securely implemented for App development which is currently in the making.

Potential Impact:
The BioCog project will have a deep impact on the development of effective imaging tools for diagnosis, monitoring and management of mental disorders, in particular on POD/POCD. Given the tremendous unmet need for clinical management of POD/POCD, particularly in patients with multimorbidity, the improvement of clinical management would constitute a glimmer of hope for all those who are afflicted, their family members and society as a whole.
The multivariate expert system based on neuroimaging and molecular biomarkers is expected to allow the identification of patients at risk for POD/POCD with high sensitivity. At the same time, molecular biomarkers will allow stratification of patients at risk for POD/POCD with regard to specific molecular pathologies.
As a direct consequence, clinical management could be improved by supporting decision-making before a surgical intervention: i.e. balancing the risks (incl. POD/POCD) against the benefits of a surgical intervention, particularly in older patients. If a surgical intervention is unavoidable, subtyping of risk patients would enable clinicians to minimize risk factors for developing POD/POCD, for instance by avoiding or replacing drug treatments/anesthetics with anticholinergic side-effects (older patients often receive multiple treatments with anticholinergic components) or by moderate perioperative glucose control, should these risk factors be identified as predictive of POD/POCD in BioCog.
In addition, development of new drugs would be facilitated: in clinical drug development, biomarker-based stratification of patients would improve testing of drugs with specific mechanisms of action. Predictive biomarkers could also be translated back to pre-clinical studies and facilitate the development of animal models for drug testing.
Accordingly, a successful expert system for POD/POCD accompanied by a solid reference database could become a software-based “medical device” with great market potential.
The study is also expected to promote our knowledge on the pathophysiology and molecular basis of POD/POCD, in particular on the role of the cholinergic-inflammatory pathway, glucose control and vascular disease in POD/POCD patients. Along this line, we also expect that our project will help clarify the intimate relationship between AD and POD/POCD.
A highly valuable database of POD/POCD patients incl. biobank will be created that will be unique in the world. This database could also be used to address scientifically and clinically relevant questions that are not directly related to but go far beyond the role of the cholinergic-inflammatory system or glucose control in POD/POCD patients.
Additional value is generated for both the involved SMEs and academia: the newly established biomarkers can be used by scientists to address research questions more accurately and future drug development can be greatly supported. This is because the proven capability to conduct high-end, biomarker-enhanced clinical studies is a major asset when it comes to attracting clinical trials with novel compounds from the major players in R&D pharmaceutical industry. The development of an infrastructure for using biomarkers in clinical studies creates a clinical research environment which is truly “translational” and which is therefore attractive to all R&D departments for testing novel drug candidates, in particular when this kind of infrastructure is combined with a (professional) clinical phase-I/IIa set-up. At present, only few places in Europe display this kind of infrastructure. Sophisticated biomarker-enhanced clinical studies should therefore help to provide Europe with a competitive edge in attracting clinical studies/trials and also in the biomarker industry which is one of the most rapidly growing market segments in biopharmaceutical industry: biomarker industry was worth over $13 billion in 2011 and its value is expected to double over the next 5 years.
List of Websites:
www.biocog.eu/

Periodic Report Summary 4 - BIOCOG (Biomarker Development for Postoperative Cognitive Impairment in the Elderly)

Other report summaries

Udostępnij tę stronę

Pobierz