Rational molecular Assessments and Innovative Drug Selection

Project Information

RAIDS

Grant agreement ID: 304810

Closed project

Start date 1 October 2012

End date 31 March 2017

Funded under

• Specific Programme "Cooperation": Health

Total cost € 7 746 906,48

EU contribution € 5 999 642,40

Coordinated by

INSTITUT CURIE
France

Final Report Summary - RAIDS (Rational molecular Assessments and Innovative Drug Selection)

Executive Summary:
Cervical cancer (CC) is the second most common malignancy in women worldwide. Although CC is a single diagnostic entity and infection of high-risk HPV is recognized as an important initiating event in tumourigenesis, CC exhibits differences in clinical behaviour. Stratification of CC into subclasses for progression and response to targeted treatment remains to be defined.

▪ We are lacking prognostic and predictive biomarkers for CC treatment and there is a growing need for the development of biomarkers to follow up the course of the disease.
▪ More importantly, we need to learn which are the most important targets to address as well as for each target(s) the proportion of tumours which need corresponding innovative therapies.

RAIDs project started October 1st, 2012 and ended March, 30th 2017. It gathered 16 partners from 7 countries to perform a cognitive cohort study (BioRAIDs) intended to define tumour stratification for targeted therapies, as well as precision medicine trials using an HPV directed vaccine in combination with checkpoint inhibition.

The RAIDs consortium aimed at:
• providing a safer and more efficient therapy for the individual patient;
• raising awareness in countries with lesser screening practices;
• improving the quality of life for women with cancer via:
✓ a) the acquisition of defined molecular data for better treatment decisions,
✓ b) targeted pilot trials directed at specific alterations as well as targeted vaccine trials directed against the HPV
✓ c) the continuous evaluation of standards of care by comparing standards and outcome in all the RAIDs centers
• disseminating information on innovative practices in concertation with the help of other international structures, be they clinical trial orientated [EORTC (European Oranisation for Research and Treatment of Cancer) and ENGOT (European Network for Gyneacological Oncology Trials)] centers or International Societies such as [ESGO (European Society for Gynaecological Oncology), ESMO (European Society for Medical Oncology) and IGCS (International Gynaecological Cancer Society)].
• providing information on predisposing conditions for immune rejection or tolerance of this virally transmitted disease rendering immune interventions more effective.
• developing new tools and ideas on future treatments using drug combinations which may be exploited and create economic added value.

The RAIDS consortium focussed on algorithm development for targeted therapies. To assess cancer functional events (CFEs), the RAIDs Network collected a prospective dataset (BioRAIDs: NCT02428842) of consecutive tumour tissues, whole blood and sera from 419 cervical cancer patients. This protocol was conducted between 2013-2016 in 18 centers from 7 European Union countries. At 54 months, whole exome sequencing (WES) is presently available for the first 98 patients, for 20 CC cell lines and Reverse Phase Protein Array (RPPA) data for 154 patients with a common core set of 91 patients. Targeted sequencing is available on additional 100 patients. These preliminary data allowed stratification of patients in different subgroups according to their molecula profile and correlation to clinical outcome is currently ongoing.
Project Context and Objectives:
Cervical cancer (CC) is the second cause of gynaecological cancer-related deaths worldwide[1]. Incidence and mortality rates of CC are up to twelve times higher in Eastern Europe as compared to North/Western Europe due to previously inadequate or absent screening practices (Ferlay et al., 2014; Arbyn et al., 2011).
HPV is commonly accepted as the major etiological cause of CC (Zur Hausen et al., 2009). Preventive vaccination is expected to impact incidence rates in more than 20 years from now when the first vaccinated adolescents will reach the age of peak incidence (35-50) of CC. In the meantime, women are at risk and there is an unmet medical need to improve the diagnosis and treatment of CC.

Although CC is a single diagnostic entity and infection of high-risk HPV is recognized as an important initiating event in tumourigenesis, CC exhibits differences in clinical behaviour. Stratification of CC into subclasses for progression and response to targeted treatment remains to be defined. At present, the dominant targets under scrutiny for innovative CC treatments are the following: EGFR/PI3K pathway, proliferation/DNA checkpoint and angiogenesis inhibition as well as anti-HPV vaccines. There have been recent publications (TCGA et al., 2017; Ojesina, 2014; Wright, 2013) on high resolution genetic investigations in CC. So far, there has been no prospective assessment on patient outcome based on whole genome/exome sequencing or protein profiling of their tumours together with quality control evaluation of patient treatment. Early prospective data is available from a small phase 2 clinical trials by Institut Curie, the coordinators’ center, suggesting that EGFR inhibition at the membrane is ineffective in the presence of a downstream PI3K pathway activation (De la Rochefordière et al., 2015).

RAIDs is a multidisciplinary co-operation between academic clinical centers, SMEs and translational research platforms. It combines Next Generation Sequencing (NGS) and Reverse Phase Protein array (RPPA) in a large patient population prior to standard therapy. It includes :
▪ a cognitive cohort study (BioRAIDs) intended to define tumour stratification for targeted therapies,
▪ as well as precision medicine trials using an HPV directed vaccine in combination with checkpoint inhibition.
In addition, high throughput screening techniques have been performed in CC cell lines, allowing to identify new drugs of relevance for advanced stage multi resistant CC. These molecules are to be validated in preclinical mouse models. Ongoing studies will assess in vitro efficacy of drug targeting according to molecular phenotypes

The main objectives of RAIDs as stated in the DoW are:
▪ To identify prognostic and predictive biomarkers for standard and innovative therapies in cervical cancer patients using both high throughput genomic and proteomic approaches, the final aim being to improve treatment response for the individual patient;
▪ To define a set of stratification criteria for therapy in patients with cervical cancer based on the tumour’s molecular profile;
▪ To identify underlying mechanisms causing immune tolerance of this sexually transmitted viral disease in order to facilitate innovative clinical interventions by vaccination studies (together with micro-environment modulators and/or checkpoint inhibitors);
▪ To improve clinical outcome of patients with cervical cancer by conducting interventional precision medicine trials.

Project Results:
I.3.1 - BioRAIDs study
BIO-RAIDs is a prospective multicentre European study, presently recruiting patients in 7 EU countries. BioRAIDS included more than 400 patients to identify genetic mutations and active pathways in cervical cancer while also supervising standards of therapy. The information accumulated in RAIDS, should serve to stratify future patients into personalized, (precision) cancer treatment regimes. Tumour and liquid biopsies from patients with previously non-treated cervical cancer (stages IB1-IV) are collected at defined time points. Patients receive standard primary treatment according to the stage of their disease. The main objectives are the discovery of -dominant molecular alterations, -signalling pathway activation, and -tumour micro-environment patterns that may predict response or resistance to treatment. An exhaustive molecular analysis is performed using 1° Next generation sequencing, 2° Reverse phase protein arrays and 3° Immuno-histochemistry (Ngo et al., 2015)

Clinicians from Eastern European countries (Serbia, Romania, Moldova) worked in close proximity with clinicians fromFrance, The Netherlands, Germany, Ireland) to implement standardized sampling as well as diagnostic and treatment procedures in the different RAIDs centers.

Together with the sponsor of the clinical study, Institut Curie in Paris, the Hannover Clinical Trial Center (HCTC), an academic clinical research institute on the campus of Hannover Medical School, provided the respective infrastructure for site management and monitoring for Ireland, Romania, and Moldova. ECRIN, hosted by INSERM France, provided monitoring support in Serbia. France and the Netherlands had clinical trial support by Institut Curie.

The initiation of the clinical sites included validation of local ethics requirements, necessary to conduct the clinical study. The process of submission to the ethics committees was uneventful, only minor modifications in wording of the respective patients information and consent forms was necessary according to local ethical standards. In France, approval by the competent national authority was obtained for an Interventional study, in the other participating countries the BioRaids study was classified as a non-interventional study (biomarker study). Furthermore, an adequate patient insurance was obtained covering the procedural part of the study (cervical cancer biopsy). After adequate information and consent, patients were recruited to the project as outlined in the study protocol.

In total, 419 patients were recruited to the study. Between 2013 and 2014 nearly 100 patients were recruited, 319 patients were recruited til September 2016 showing intensification of recruitment in the second half of the study period. With the exception of the sites at Institut Curie and IOV who were the highest patient recruiting centres, most active centers recruited between 20 and 30 patients. It was expected that the site in Timisoara would be able to contribute considerably more patients.
In France 177 patients have been recruited at 12 study sites. The 2 sites in The Netherlands contributed 56 patients. Two sites in Germany recruited 27 patients. In Eastern Europe, the single site in NoviSad, Serbia was able to recruit 101 patients, in Romania (accrual started later) 23 patients were included, in Moldova 34 patients were recruited (4).
Late in the project, it was decided to add Dublin as an investigational site, because the clinical trial group at Trinity College showed a very good general organizational status. The Dublin ethics committee gave notice of its approval of the submitted BioRAIDs protocol at the end of August 2016. Unfortunately, this was much later than usual due to an unforeseen delay for which the ethics committee took responsibility. A few days after the ethical approval was available, HCTC initiated the site in Dublin at the Trinity College. Only one patient could be recruited in the remaining time until end of September 2016.

Clinical data were collected by electronic data capture in an electronic case record form (eCRF). The eCRF was adapted to ensure a practical use for extraction of data. Integration of internal auto-checks for clinical items and biobanking part was implemented and applied. In addition a “checking tool” allowing the traceability of samples during shipment was developed. There was a continuous interaction with the clinical trial teams and the centralized data management at Institut Curie. The data collected during the BioRAIDs clinical study represent one of the largest databases on cervical cancer worldwide. The important feature of this database is that it contains not only clinical data, but also biological data, thus enabling comprehensive analysis of the pathophysiological mechanism in cervical cancer. This will lead to more precise therapy options, to the benefit of future patients with cervical cancer. Discussions on future designs of clinical evaluation are ongoing, the discussion of follow-up projects has been initiated by Institut Curie. Importantly, the BioRaids study resulted in an established network which in the future may specifically focus its scientific interest in conducting clinical trials in cervical (and in related) cancers.

During the BioRAIDs study, the largest European cervical carcinoma imaging data base was created. Evaluating cervical carcinoma imaging characteristics and comparing them to clinical, immunological and genetic data could define imaging parameters and prognostic predictors (radiomics). Defining imaging predictors could stratify cervical cancer patients for the most appropriate treatment options prior to the start of treatment. New optimized MRI protocol and practice guidelines for imaging of cervical cancer patients could be suggested at European level and potentially harmonize imaging techniques. Cervical carcinoma imaging using standard and advanced MRI protocols, developed in the BioRaids study, might become useful to predict response or resistance to treatment.

Online Delineation Workshop
Two Online Delineation Workshops (ODW) included 46 clinicians from 14 centres. Clinicians completed baseline (C1), guided (C2) and final contours (C3) for external beam radiotherapy (EBRT) and brachytherapy (BT) for LACC. Interobserver and intraobserver variability was evaluated quantitatively (using the DICE index) and qualitatively compared to contours by expert radiotherapists. ODW offered training, initial contouring harmonization and allowed assessment of the performance of centres (Rivin del campo et al., 2017).

I.3.2 - Molecular analyses on samples collected from the BioRAIDs study
I.3.2.1. Biobank
An annotated biobank from the specimens was collected by the clinical centers and the DNA samples isolated from these specimens in order to identify genetic and biochemical biomarkers and develop diagnostic tools that are predictive of the prognosis and might predict the efficacy of different treatment options for cervical cancer. In BioRAIDs only standard treatments (surgery, radio-chemotherapy, chemotherapy first) could be explored. For this purpose, solid and liquid biosies were collected and sent to the central molecular labs for processing and storing. Whole exome sequencing of 20 cervical cancer cell lines and tumor samples from 100 patients with matching blood was to be determined. Based on the sequence, clinical and pharmacological data, the design of a Cervical Cancer Diagnostic Panel targeting a small number (~50) of genes was planned in order to be tested on all further patients. In addition, quantification of total and activated proteins and protein variants that have been suggested of particular relevance in cancer was to be performed, besides also the detection of HPV insertion sites and hotspot (such as PI3K) mutations in serum DNA of the same patients over time.
To achieve the biobanking objectives, sample tracking was improved by inventing a sample checking tool by Quanticsoft and Curie, while Seqomics invested in a barcode-generating software and a printer to create the barcodes for DNA and protein samples from the original barcodes of the biopsies using code128 for encoding. Seqomics sent the labels to the other laboratories.
By the end of the project, Erasmus MC received a total of 6982 biobank samples These samples were checked and processed according to the standard procedure shown in the workflow chart presented in Figure 5

The samples were from 419 patients of whom 177 were from France, 101 were from Serbia, 34 were from Moldavia, 23 were from Romania, 27 were from Germany and 56 were from the Netherlands. These 6982 samples were collected and included as unique patient biobanking materials: 350 fresh frozen tumors and 335 formalin fixed paraffin embedded (FFPE’s), 365 whole bloods and 372 sera samples from 419 patients of which 259, 128, 56, 27, 5 patients had additional non-baseline (mostly) serum samples (at time 1, 2, 3, 4 and 5 respectively). Please note that multiple samples were taken at a defined timepoint.
Of the 419 patients included in the clinical trial, 19 sample sets were excluded by physicians due to patient drop out or error in histology, 8 samples had been sent to the wrong place and melted, 37 had not yet been forwarded to Erasmus MC, on 5 sets there is still missing information and these were not processed for that reason.
Thus 350 sets were biobanked, of which 337 fresh frozen (FF) tumours were eligible for sectioning.
At Erasmus MC the percentage of tumor cells within these 337 FF tissue biopsies was determined by applying cross-contamination-free biopsy cutting according to the guidelines established during the M18 meeting. Moreover, all samples were evaluated by two independent readers for tumor percentages and histology at Erasmus MC, Rotterdam. We also checked the additional non-baseline tissues. However, in 25 non-baseline samples tested, (post treatment) only 2! FF samples showed tumor cells (less than 10%).
The 308 baseline-samples, that contained ≥10% tumor cells, were processed further to isolate DNA for exome sequencing (of these 269 tumor samples had >= 30% cells). For targeted sequencing (tumor % should be >=10%). Hence 29 samples contained <10% tumor cells. Among the samples that were checked by the pathologists 222 were of squamous, 23 of adeno histology.
From those samples that passed the quality control, 5 µm thin sections were sliced for screening before and after preparing ten sections of 30 µm thickness and their tumor content was determined. High molecular weight genomic DNA was isolated from these 10 sections, and those samples that passed QC during the screens were sent to Seqomics. Whole blood samples were forwarded to (or if collected by SeQomics kept at) Seqomics for biobanking and genomic DNA isolation. High quality and large molecular weight (>50kb) genomic DNA from the blood of the patients whose tumor sample passed the quality control was isolated at Seqomics Ltd. The remaining blood samples are stored at Seqomics Ltd, while per March 2017 all tumor specimens, sera, tissue sections, glass slides and isolated DNA samples, that had been previously stored at Erasmus MC, were successfully send off and were received in good condition at Seqomics. The biobank at Seqomics currently contains 6404 tubes of tissue (tumor, blood, sera) samples, 1416 tubes of isolated cfDNA, 326 original and 299 diluted tumor DNA samples.

I.3.2.2. Cervical cancer mutational profile
The sequencing of the whole exome (WES) (all predicted exon of the human genome) from matched tumor and blood from the 100 first BioRAIDs patients allowed us to design a targeted sequencing panel including 600 genes likely to be of interest for CC since these genes were differentially expressed in the first non supervised clustering based on mutational profiles. 100 patients were analyzed using targeted sequencing.

I.3.2.3. Cervical cancer protein activation profiles
To detect protein expression and phosphorylation in cervical cancer, a large effort was made to validate new primary antibodies, with a major focus on metabolism and immune checkpoints/micro-environment. In total, more than 150 new antibodies were tested for their use in reverse phase protein array (RPPA) and around 50 new antibodies could be validated and used in this project.
154 patients enrolled in BioRAIDS were analzed using RPPA. Exploratory analyses identified 3 clusters enriched with EMT, DNA repair and MAPK/PI3K pathways respectively (De Koning et al., 2017).

I.3.2.4. Cervical cancer biomarkers in liquid biopsies
The identification of HPV integration sites (task for Curie) was initially planned on a subset of 48 HPV16 or HPV18 positive tumour DNA samples for which the whole exome sequence would also be available. In the mean time, serum samples were to be collected from the same patients before treatment (baseline) and every six months for 18 months after the treatment. We subsequently planned to use i) HPV DNA sequences, ii) HPV integration site (when identified) and iii) HPV E7 gene sequences, as tumour markers when screening serum samples for circulating tumour DNA (ctDNA). We aimed to assess the sensitivity of the three markers and their value to estimate minimal residual disease and to predict relapse.
HPV integration sites in tumor DNA, were detected in 42 out of 100 samples (42%), HPV integration led to gene disruption in 25 out of 42 (60%) of the cases.
In the parallel serum analysis, we performed droplet digital PCR (ddPCR) on all serum samples using HPV E7 gene as ctDNA marker and we detected ctDNA in 64 out of 96 (67%) of the baseline serum samples. First results from the analysis of 17 patients with multiple non-baseline serum showed that ctDNA was not detectable anymore in the serum after treatment for most of the cases. However, in 3 patients, a positive non-baseline serum sample for HPV E7 strongly suggested a relapse of the disease, which is confirmed by the clinical data so far for one of the 3 patients.

In parallel at Erasmus MC, a pipeline to isolate circulating (tumor) DNA –ctDNA- from serum, as well as from plasma, was developed in order to perform genomic analyses on the retrieved cfDNA, using the QIAamp kits. The DNA is of high genomic quality with a serum input of 2000 µl. Baseline sera of 372 patients and non-baseline sera (taken at 6 and up to 18 months after treatment) summing up to 475 samples have been processed. As expected, the number of sera samples collected showed an exponential growth.
For genomic analyses, different technologies have been applied to evaluate cfDNA for mutations. With improved sensitivity (by approximately 100 times), we analyzed the PIK3CA hot-spot mutations in all FF tumors and corresponding serum samples using Digital PCR (dPCR) which allows absolute sample quantification: 98 out of 308 tumour-DNA (31.8%) showed a PIK3CA mutation. Of these: 26% tested positive in accompanying BASELINE sera samples.
Interestingly, only few samples tested positive after months of follow-up

I.3.2.5 Tumor microenvironment analyses
Formalin-fixed paraffin-embedded material form the whole exome analysed cohort were used for tumor microenvironment analysis by mulitparameter fluorescent immunohistochemistry. Tumor infiltrating T cell- and myeloid cell comprising CD3, CD8, FoxP3 and CD14, CD163, CD33 markers were enumerated in the tumor fields and the tumor stroma. Moreover, PD-L1 staining was performed to analyse the various PD-L1 expression patterns and PD-L1 positive cells in this cohort. In agreement to previous data, PD-L1 expression was found in tumour cells and in infiltrating myeloid cells. PD-L1 tumour expression was located at the invasive front of the tumour or it was diffuse throughout the whole tumour. Currently, all data on the microenvironment analyses is being pooled with the ultimate goal of recognizing immune-patterns that will guide future immunotherapeutic interventions.

I.3.2.6. Data integration and Bioinformatics analyses
From high-thoughput experiments conducted on biopsies of BioRAIDs patients tumor material, the mutational profile as well as the protein level and activivity were assessed. Complex sequencing and microarray technologies have been used requiring dedicated tools and algorithms to extract the relevant biological and clinical signals that will help the biologist/clinican to unravel molecular mechanisms related to tumor progession allowing to tailor therapy for each patient individually (so called precision medicine). Three essential steps can be identified
• data management and data integration
• further development of new mathematical/statistical algorithms adapted to the complexity of these data
• Bioinformatics and Biostatistical analysis of the data


Management of the data
Precision medicine requires interdisciplinary expertise ranging from medicine and biology to translational research, wet lab technical, as well as biotechnological know-hows.
A seamless information system named KDI (Knowledge and Data Integration) was developed to fully support essential bioinformatics requirements for precision medicine (Servant et al, 2014). The system allows management and analysis of clinical information, classical biological data as well as high-throughput molecular profiles. It can deliver in real-time information to be used by the medical and biological staff for therapeutic decision-making. KDI allows to share information and communicate reports and results across numerous stakeholders. The system relies on state-of-the-art informatic technologies, allowing cross-software interoperability, automatic data extraction, quality control and secure data transfer.
This system integrated:
• Exome sequencing data
• Protein microarray data (Reverse-Phase Protein Array)
• Clinical data
• Tumor Microenvironment data
• HPV (Human Papilloma Virus) integration site

Development of new mathematical/statistical algorithms to take into consideration the compexity of these data
In order to detect tumour specific mutations from exome sequencing data and to weed out patient specific variants, we set up a pipeline to compare blood and tumor samples from the same patient.
While it is well established that mutations in relevant genes such as significant tumor suppressor genes or oncogenes are causal to the development of cancer, it is essential to consider gene-gene interactions when we analyse the data. For this reason, we proposed a new algorithm which benefits from the knowledge of a protein-protein interaction network topology. Indeed, using this methodology allows the influence of each mutation to spread over its network neighbourhood. The mutational profile of each patient is thus a ’network-propagated’ profile in which the state of each gene is no longer binary (mutated vs non-mutated) but reflects its network proximity to other mutated genes in that patient. This method was used to stratify our cohort of patients.

Data analysis
Algorithms to perform the stratification of RAIDs 98 patients + 20 cell lines based on the propagation of the mutated genes binary matrix were used (Atlas of Cancer Signaling Network) taking into account gene interactions. Different clusters (group of patient showing similar mutational patterns) could thus be identified.
Similarly, classification using standard method have been applied on protein data. Protein and gene that explain the differences between the clusters have been identified and their role in the tumoral progression is under investigation.

I.3.2.7. DNA vaccine trials
Two novel anti-cancer vaccines were produced and clinical tested by NKI-AVL and AmBTU during the RAIDs project.
The general aim of these anti-cancer vaccines is to generate a defense reaction of the patients own body (also called immune response) against the Human Papilloma Virus (HPV) which represents the main cause of cervix cancer. The activated immune cells can then hopefully recognize and kill the cancer cells. This is a type of cancer immune therapy.

The vaccines tested in the RAIDs project are therapeutic vaccines. In contrast to the prophylactic vaccines against cervix cancer, which are successfully used in young adolescents, ideally before the onset of sexual activity to prevent HPV infection, a therapeutic vaccine aims to generate an immune response against an existing cancer (or a pre-stage of cancer). It is directed against epitopes of the virus which are integrated into the human genome while the prophylactic vacciens are directed to an epitopes of the viral envelope.

The vaccines in the RAIDs consortium are applied by a tattoo administration route, and vaccines (in the form of plasmid DNA), are injected into the skin using a conventional tattoo device. This results in local uptake of the vaccine and immune activation directed against the vaccine and the cancer. The DNA tattoo technology has been developed by NKI.

In the first trial, 12 patients with a pre-stage of cervix cancer (usual type vulvar intraepithelial neoplasia (uVIN)) were vaccinated with a DNA vaccine directed against the HPV E7 oncoprotein. This vaccine was well tolerated, however only a low vaccine-induced immune response and no clinical responses were observed.

Therefore the second trial uses a vaccine format that was improved in the laboratory and is directed against the E7 and the E6 oncoprotein of HPV.
The second trial is currently enrolling uVIN patients in the Netherlands and will be finalized after termination of RAIDs.

I.3.1. Preclinical studies
Pharmacological profiling of cell lines has already detected a group of drugs that synergize with the “standard treatment” of advanced cervical cancers. The validation of these data is ongoing and the reformulation/use of old drugs is continued to be investigated according to molecular profiles.

Preclinical mouse models for tumour micro-environmental studies have been developed and are published in journals with a high impact factor. Preclinical mouse data on combining vaccine and radiotherapy are published and report the efficacy of a novel dual targeting approach in a preclinically relevant animal model. A patent has been deposited.

The aim was also to define new biomarkers from the tumor microenvironement (TME) that could help to define the efficacy of standard therapeutic treatments. The final ambition was to identify potential new targets usable in combined therapies in order to improve standard treatment efficacy. A first part of the work proposed to develop in vitro approaches to study the effect of drug combinations on different cervical tumor cell lines. Because studying TME needs to perform in vivo experiments to reconstitute a fully functional immune microenvironement including all the complexity of the cell dynamic, the main part of our work was based on the development of diverses preclinical animal models to study the effect of both chemotherapy and radiotherapy in the immune component of the TME.

Screening of drug combination efficacy on tumor cell lines.
The experimental approach relies on the systematic screening of drugs and on miniaturized cell-based assay. Although the chemical library used was not large (43 drugs), the computing of the dose-effect response of the combined treatment and the comparison of the output of the 23 cell-lines with different growth rates, require numerous assays (more than 70,000 data points). It was chosen to screen only a selection of drugs in combination with Paclitaxel-Cisplatinum treatment to attempt to detect synergizing drugs. Drugs that target other signaling pathways in cancer (as opposed to Paclitaxel-Cisplatinum) were privileged, as well as molecules directed against the energy metabolism such as anti-diabetic biguanides (metformin, phenformin). Finally, classical chemotherapeutic agents (5-FU, temozolomide, vinblastine, etc.) were also included.
This approach led to the indentification of a group of drugs that synergized with the “standard treatment” and these molecules interfere with different signaling circuitries including epigenetic mechanisms (HDAC), the functioning of receptor tyrosine kinases (EGFR, VEGFR/PDGFR) and nuclear hormone receptors (tamoxifen) as well as the energy metabolism (biguanides). Interestingly, these new drug combinations were not obvious to a person skilled in the art and are susceptible to industrial application.

Characterization of TME modification in a mouse model
Standard cancer therapies are supposedly aimed at preferentially targeting neoplastic cells which they hit preferentially based on their high proliferation rate. Inevitable side effects occur on healthy proliferating cells in particular in components of the immune system. Despite the observation of beneficial effects on the adaptive immune response (Zitvogel et al., 2008) complete tumour rejection by standard therapies is a rare event. Numerous immune interference failures have been associated with chemo and radio-resistance.

For instance, we demonstrated T cell infiltration and modification of T cell interaction with tumor associated macrophages following cyclophosphamide treatment using intravital imaging approach (Boissonnas et al Neoplasia 2013). Detailed characterization of the tumor associated macrophage compartment within the tumor showed that TAMs formed a non-static dense network displaying a M2-like phenotype (Mgl1 and IL4R expression). Intravital imaging showed that TILs stably engage TAMs in an antigen-dependent manner leading to a preferential localization of TILs in TAMs rich areas.

We showed that cyclophosphamide treatment induces a transient aplasia of myeloid cells followed by a massive recovery phase which provokes an important monocytosis in peripheral tissues as well as within the tumor. We hypothesized that this rebound effect of monocytes, providing a de novo source of TAMs might impact the efficacy of CTLs. Deciphering the mechanisms by which TAMs rapidly limit CTL-mediated destruction represents a fundamental question which is at the basis to improve the efficacy of therapeutic regimens. We thus further analyzed the impact of cyclophophamide on the mobilisation of medullar monocytes and show the role of chemokine receptors in this process (Jacquelin et al Blood 2013). We confirmed that CCR2 expression was required for the egress of classical monocytes from the bone marrow toward the blood flow during hematopoietic reconstitution induced by chemotherapy. In contrast, we demonstrated that in opposite to CCR2, CX3CR1 contributes to monocyte retention in the bone marrow through its adhesion properties.

Overall our results define a novel implication of CX3CR1 and CCR2 in the balance between retention and egress of classical monocytes during hematopoietic reconstitution following chemotherapy.

Following these primary results we made additional efforts characterizing CCR2 as an immune marker linked to TME modifications, appearing in the wake of cancer treatment. We identified that the chemokine receptor axis CX3CR1/CCR2 impacts not only monocyte-differentiation but also Treg migration. We showed that the CCR2 + Treg subset is the main IL10-producing cell subset and has an increased sensitivity to low dose cyclophosphamide. However we observed a very rapid rebound of this subset following chemotherapy which could explain the mechanism of tumor resistance suggesting that CCR2 expression on Treg could represent a potential biomarker of chemotherapy efficacy (Loyher et al Can Res 2016).
Our recent results show that the CCR2 axis also regulates to the co-recruitement of Treg and monocytes following radiotherapy and thus contributes to the generation of a local immunosuppressive environment.

In conclusion, we propose that targeting the CCR2 axis could have a double impact on both the Myeloid and the regulatory T cell compartments. This double effect must be taken into consideration in future studies to better understand the mechanism of chemoresistance mediated by the main immuno suppressive subsets (Treg and TAMs). Several antagonists of the receptor or antibodies against the CCL2 exists and represent already good molecules to test to address these questions.

Optimization of dual-targeting approach using animal models
Because the final ambition of ther project is to improve standard treatment efficacy we benefit from our different preclinical models to test standard therapy combined with original immunotherapy.
Fort instance, there is currently a growing interest in combination approaches based on radiotherapy and immunotherapy for the treatment of solid tumors, and cancer vaccines could play an important role in this setting. Peptide vaccines coupled to the non-toxic STxB have been successfully tested in several preclinical settings and represent a promising tool for vaccine-based cancer therapy. Nevertheless, their use in association with radiotherapy has not been previously reported. We report a novel, extremely effective combination of STxB-based vaccination (STxB-E7 vaccine) and local irradiation for the treatment of HPV-associated HNSCC. In our murine model, this novel therapeutic approach resulted in the complete clearance of the majority of treated tumors. The new STxB-based vaccination strategy acted synergistically with either single dose or fractionated IR, which is of particular relevance because fractionated radiotherapy is a mainstay for the treatment of HPV-associated tumors. In our study, when the combination treatment resulted in complete tumor clearance, which occurred in 70% of treated mice, no recurrence or metastasis was found, even after an 18-month observation period. This is a consequence of the induction of an effective T cell memory, with a protective effect both locally and systemically. This represents a promising clinical tool, since a patient treated with this combination therapy who undergoes tumor remission will likely be protected against any local or distant relapse (which is the main cause of mortality of HPV-associated cancer).

The dual targeting treatment tested represents a potential novel approach for the treatment of HPV-associated cancer (Mondini et al Mol Cancer Ther 2015).

General Conclusion
Our work significantly contributed to the understanding that standard cancer therapies do not only target neoplasic cells but strongly modulate the immune compartement both localy in the tumor microenviroenement but also systemically, by affecting the distant immune reservoirs of inflammatory cells such as the bone marrow. The chemokine receptor network and specifically the CCR2 axis represents a major actor in the strong modifications of the TME observed after both chemotherapy and radiotherapy in particular in the constitution of the immunosuppressive environment mediated by Treg and TAM. This axis might represent an interesting target to modulate in order to improve the efficacy of the standard chemotherapies when combined with specific immunotherapies

Potential Impact:
I.4.1. Potential impact
RAIDs aims to define a set of stratification criteria based on molecular profiling. Its results are likely to contribute insight into relevant dominant genomic and protein signalling pathway alterations, in enabling the identification of prognostic and predictive biomarkers for 1° standard and 2° targeted therapy in CC.

Routine molecular diagnostic tests have yet to be introduced to guide personalized cervical cancer treatment in contrast to other solid cancers. To assess ‘cancer functional events’ (CFEs), the RAIDs consortium accrued consecutive tumour tissues, whole blood and sera from 419 cervical cancer patients in 18 centers from 7 European Union countries, between 2013-2016. Whole exome sequencing (WES) is presently available for the first 98 patients, for 20 CC cell lines and RPPA data for 154 patients with a common core set of 91 patients. The RAIDS consortium focussed on algorithm development for new therapeutic targets. Emphasis is on the characterization of 1° mutational load; 2° significantly mutated (actionable) genes (SMG) including epigenetic targets and 3° potential actionable pathways identified in RPPA clusters. Ultimately, curated clinical outcome data in RAIDs will be integrated with mutations and copy number alterations (CNA), protein and gene expression as well as with epigenetics, tumour microenvironment (TME) and Human Papilloma Virus (HPV) parameters to give predictive sets for response and outcome.

While combination of radiotherapy and vaccine in a mouse model was giving unequivocally improved results, DNA vaccine results by tattoo in early development in preclinical lesions were not yet conclusive. An invasive tumour model is likely to have a higher mutational load and might therefore be better suited.
The RAIDs consortium’s future objectives are to better define actionable molecular events in relation to outcome and patient’s quality of life, allowing to focus on relevant CFEs for patients at high risk. The study of combined drug treatments in CC cell lines may rapidly accelerate our knowledge of unique vulnerabilities in cancer treatment. The RAIDs consortium will attempt to validate such strategies both preeclinically on cell lines and clinically in a future prospective trial (RAIDs 2 CURE) designing adjuvant therapies based on the biological stratification of tumours (RAIDS 2 CURE), taking into account actionable molecular alterations while also focusing on altered epigenetic states.

I.4.2. Dissemination and exploitation of results
Dissemination
The RAIDs network is known internationnaly [EORTC (European Oranisation for Research and Treatment of Cancer) and ENGOT (European Network for Gyneacological Oncology Trials)] centers or international societies such as [ESGO (European Society for Gynaecological Oncology), ESMO (European Society for Medical Oncology) and IGCS (International Gynaecological Cancer Society)] and the BioRAIDs participating clinical centres as well as new centres are keen to follow us in a RAIDs 2 targeted trial. RAIDs will be part of the international CCRN (cervical cancer research network) under the auspices of the International Gynaecological cancer society). A major pharma is in discussion for the contribution of some of their products for RAIDS2.

A number of publications are out, a submission for Nature was recently sent of (short communication) and several papers are in preparation involving: 1) clinical: response/progression by treatment and by stage across the EU countries), 2) Imaging: central review of MRI and comparative assessment of quality and techniques as well as attempt of radiomics, 3) IHC phenotypes in the tumour in parallel to mutational load and to specific sets of mutations.

Video clips explaining RAIDs have been produced (http://www.raids-fp7.eu/press/videos.html). The most recent video involves many patients from NoviSad Serbia and shows all the actors of the consortium.

Video clips explaining RAIDs have been produced (http://www.raids-fp7.eu/press/videos.html).

A dropbox for patients (RAIDs drop box - http://www.raids-fp7.eu/a-question.html) allowing requests for information in the field of precision medicine has been inserted on the RAIDs website as well as in the cervical cancer factsheet of ESGO. The dropbox allows patients to ask questions in relation to standard treatment and to innovative protocols regarding precision medicine in their native language.

Interaction with patients and patient advocacy groups continues. A patients’ working group was organized at Institut Curie around precision medicine issues:
http://curie.fr/actualites/echange-avec-patients-autour-projet-entreprise-2015-2020-007181
https://www.facebook.com/InstitutCurie/ le 25 avril
Another patient-doctor meeting was organized at Gustave Roussy in December 2016.

Exploitation of results
While sets of genes, predictive for outcome to standard therapy could be defined, these need further validation in several new rounds. After validation with the remaining RAIDs patients’ datasets, a predictor gene set can be integrated in the planned RAIDs2 clinical trial. Gene sets for the prediction of mutational load could equally be defined and these would be of high interest before the implementation of immune checkpoint inhibitors. The well characterized RAIDs CC cell lines will be useful for further evaluation of drug combinations in particular with epigenetic targeting drugs.
Finally multiple biomarkers could be defined by RPPA and can be integrated by IHC techniques in the future intitial work up of patients, prior to starting treatment.
The preclinical data on mouse models suggest interesting approaches to target the micro environment and thereby to assist standard therapies.
Most importantly, based on BioRAIDs data we shall be soon able to predict which patients are likely not to achieve a full response to present day primary standard therapy and who need alternate approaches.

List of Websites:
Coordinator
Dr. Suzy SCHOLL, Institut Curie - [e-mail]
Project manager: Maud Kamal, Ph.D Institut Curie - [e-mail]
www.raids-fp7.eu

Partners
1. Institut Curie, France
2. AmBTU, Amsterdam Biotherapeutics Unit, Netherlands
3. SeqOmics, Seqomics Biotechnologia Korlatolt Felelossegu Tarsasag, Hungary
4. KEOCYT, Keocyt SAS, France : This partner has left the project
5. INSERM, Institut National de la Santé et de la Recherche Médicale, France
6. AVL, Stichting Antoni Van Leeuwenhoek Ziekenhuis, Netherlands This partner merged with partner 16
7. Erasmus MC, Erasmus Universitair Medisch Centrum Rotterdam, Netherlands
8. KEM, Kliniken Essen-Mitte Evang, Huyssens-Stiftung/Knappschaft Gemeinnutzige GmBH, Germany
9. EMAUG, Ernst-Moritz-Arndt-Universität Greifswald, Germany
10. IOV, Institut Za Onkologiju Vojvodine, Oncology Institute of Vojvodina, Serbia
11. TEO HEALT SA, Romania
12. IMPS IO, Institutia Medico-Sanitara Publica Institutul Oncologic, Republic of Moldova
13. IGR, Institut Gustave Roussy, France
14. Quanticsoft SARL, France
15. Ayming, France
16. NKI, Stichting Het Nederlands Kanker Instituut, Netherlands
17. HCTC, Hannover Clinical Trial Center, Germany. Entered the project from M24
18. ECRIN-ERIC, European Clinical Research Infrastructure Network, France

References
• Ojesina AI, Lichtenstein L, Freeman SS, Pedamallu CS, Imaz-Rosshandler I, Pugh TJ, Cherniack AD, Ambrogio L, Cibulskis K, Bertelsen B, Romero-Cordoba S, Treviño V, Vazquez-Santillan K, Guadarrama AS, Wright AA, Rosenberg MW, Duke F, Kaplan B, Wang R, Nickerson E, Walline HM, Lawrence MS, Stewart C, Carter SL, McKenna A, Rodriguez-Sanchez IP, Espinosa-Castilla M, Woie K, Bjorge L, Wik E, Halle MK, Hoivik EA, Krakstad C, Gabiño NB, Gómez-Macías GS, Valdez-Chapa LD, Garza-Rodríguez ML, Maytorena G, Vazquez J, Rodea C, Cravioto A, Cortes ML, Greulich H, Crum CP, Neuberg DS, Hidalgo-Miranda A, Escareno CR, Akslen LA, Carey TE, Vintermyr OK, Gabriel SB, Barrera-Saldaña HA, Melendez-Zajgla J, Getz G, Salvesen HB, Meyerson M. Landscape of genomic alterations in cervical carcinomas. Nature. 2014 Feb 20;506(7488):371-5.
• Wright AA, Howitt BE, Myers AP, Dahlberg SE, Palescandolo E, Van Hummelen P, MacConaill LE, Shoni M, Wagle N, Jones RT, Quick CM, Laury A, Katz IT, Hahn WC, Matulonis UA, Hirsch MS. Oncogenic mutations in cervical cancer: genomic differences between adenocarcinomas and squamous cell carcinomas of the cervix. Cancer. 2013 Nov 1;119(21):3776-83.
• De la Rochefordiere A, Kamal M, Floquet A, Thomas L, Petrow P, Petit T, Pop M, Fabbro M, Kerr C, Joly F, Sevin E, Maillard S, Curé H, Weber B, Brunaud C, Minsat M, Gonzague L, Berton-Rigaud D, Aumont M, Gladieff L, Peigneux K, Bernard V, Leroy Q, Bieche I, Margogne A, Nadan AT, Fourchotte V, Diallo A, Asselain B, Plancher C, Armanet S, Beuzeboc P, Scholl S.PIK3CA pathway mutations predictive of poor response following Standard radio chemotherapy +/- Cetuximab in cervical cancer patients. Clin Cancer Res. 2015 Feb 27
• Ngo C, Samuels S, Bagrintseva K, Slocker A, Hupé P, Kenter G, Popovic M, Samet N, Tresca P, von der Leyen H, Deutsch E, Rouzier R, Belin L, Kamal M, Scholl S; RAIDs consortium http://www.raids-fp7.eu/. From prospective biobanking to precision medicine: BIO-RAIDs - an EU study protocol in cervical cancer. BMC Cancer. 2015 Nov 4;15:842.
• Sanne Samuels a, Balazs Balint, Heiko von der Leyen, Philippe Hupé, Leanne de Koning, Choumouss Kamoun, Windy Luscap-Rondof, Ksenia Bagrintseva, Marina Popovic, Atttila Kereszt, Els Berns, Gemma G. Kenter, Ekaterina S. Jordanova, Maud Kamal and Suzy Scholl. Precision medicine in cancer: challenges and recommendations from an EU-funded cancer biobanking study. Br J Cancer. 2016 Nov 22.
• Sanne Samuels b, MD; A. Marijne Heeren, MSc; Henry J.M.A.A. Zijlmans, MD, PhD; Marij J.P. Welters, PhD; Joost H. van den Berg, PhD; Pia Kvistborg, PhD; Bastiaan Nuijen, PhD; Ton N.M. Schumacher, PhD; Marc van Beurden, MD, PhD; Ekaterina S. Jordanova, PhD; John B.A.G. Haanen, MD, PhD; Sjoerd H. van der Burg, PhD; Gemma G. Kenter, MD, PhD. HPV16 E7 DNA tattooing: safety, immunogenicity and clinical response in HPV-positive vulvar intraepithelial neoplasia. Submitted
• Ferlay J, Soerjomataram I, Dikshit R, Eser S, Mathers C, Rebelo M, Parkin DM, Forman D, Bray F: Cancer incidence and mortality worldwide: Sources, methods and major patterns in GLOBOCAN 2012. Int J Cancer 2014.
• Arbyn M, Antoine J, Magi M, Smailyte G, Stengrevics A, Suteu O, Valerianova Z, Bray F, Weiderpass E: Trends in cervical cancer incidence and mortality in the Baltic countries, Bulgaria and Romania. Int J Cancer 2011, 128(8):1899-1907.
• Zur Hausen H: The search for infectious causes of human cancers: where and why. Virology 2009, 392(1):1-10.
• Cancer Genome Atlas Research Network et al. Integrated genomic and molecular characterization of cervical cancer. Nature 543, 378–384 (2017).
• Leanne de Koning, Bérengère Ouine, Aurélie Cartier, Els M. Berns, Kirsten Ruigrok-Ritstier, Corine Beaufort, Balazs Balint, Attila Kereszt, Gemma Kenter, Sanne Samuels, Ekaterina S. Jordanova, Emmanuelle Jeannot, Heiko von der Leyen, Marina Popovic, Windy Luscap-Rondof, Vonick Sibut, Choumouss Kamoun, Isabel Britto, Claudia Rincon, Philippe Hupé, Maud Kamal and Suzy M Scholl, Rational molecular assessment and innovative drug selection (RAIDs): Paving the way to personalized medicine in cervical cancer, 2017
• Eleonor Rivin del Campo, Sofia Rivera, María Martínez-Paredes, Philippe Hupé, Andrea Slocker Escarpa, Isabelle Borget, Renaud Mazeron, Suzy Scholl, Amalia Palacios Eito, Christine Haie-Meder, Cyrus Chargari, Eric Deutsch, Assessment of the novel online delineation workshop dummy run approach using FALCON within a European multicentre trial in cervical cancer (RAIDs), Radiotherapy and Oncology, In press
• Loyher PL, Rochefort J, Baudesson de Chanville C, Hamon P, Lescaille G, Bertolus C, Guillot-Delost M, Krummel MF, Lemoine FM, Combadière C, Boissonnas A, CCR2 Influences T Regulatory Cell Migration to Tumors and Serves as a Biomarker of Cyclophosphamide Sensitivity. Cancer Res. 2016 Nov 15;76(22):6483-6494

Related documents

• final1-m54-publishable-summary-vf.pdf