The Romand Network of Oncology: bridging the gap of personalised oncology

Olivier Michielin, Petros Tsantoulis, Krisztian Homicsko, Yann Christinat, Laurence De Leval, Bettina Bisig, Edoardo Missiaglia, Vincent Zoete, Fanny Krebs, Aurélie Fortin, Lucien Perey, Volker Kirchner, Sergey Nikolaev, Pierre Chappuis, Laura Rubbia, Thomas McKee, George Coukos, Pierre-Yves Dietrich

 

Impact of omics technologies in oncology

Oncology is being revolutionised by technological breakthroughs that permit unprecedented in-depth analysis of tumour tissues; until now, decision making has been based on the analysis of a few well-known molecular alterations. Recent technologies are now providing complete interrogation of germline and somatic mutations (genomics), gene expression levels (transcriptomics), resulting protein levels (proteomics), metabolic status (metabolomics), antigens presented at the surface of tumour cells (immuno-peptidomics), morphological and spatial cellular features (digital pathology), as well as many additional omics to complete this very rich data set. The price of such omics technologies is dropping rapidly, making them competitive with more standard approaches centred on one or a few molecular alterations at the time. Omics are, therefore, expected to play an increasingly important role in the years to come for the standard of care in oncology.
To standardise the use of omics data to guide treatment decisions, most oncology centres have developed molecular tumour boards that bring together medical oncologists, pathologists and geneticists, as well as bioinformaticians. Currently, most molecular tumour boards rely on the analysis of gene panels, allowing interrogation for the majority of known alterations that are potentially actionable. Our experience so far has shown that reimbursement of such panels has not been an issue, as their price is within the same range of that of the standard analyses limited to known drivers whose status is required for correct clinical management. Within 2 to 3 years, one can expect that full exome and even full genome sequencing will fit within the same price bracket. In addition to tumour sample analyses, genomic analyses can also be performed on “liquid biopsies” obtained directly from the blood of the patient. Genomics information can be obtained easily and more regularly via such technologies, making it an interesting complement to tumour material for therapeutic monitoring.
In addition to genomics, transcriptomics is quickly gaining momentum in defining prognostic or predictive signatures based on gene expression profiles. Progress in RNAseq technologies are making these approaches more broadly available. Several signatures are already in use in the clinics (e.g., Oncotype DX for breast cancer, or the CMS1–4 classification of colon cancer) and more are continually being published and validated. In addition, detailed analysis of the immune infiltrate in the tumour microenvironment is quickly becoming a key component of prognostic and predictive biomarker development, where quantification using recent digital pathology techniques provides new exciting options. For example, although PD-L1 positivity by immunohistochemistry (IHC) has been demonstrated to have value in some indications as a predictive biomarker for PD-1 checkpoint blockade, it is clear that more complex biomarkers are urgently required to guide better patient selection. Multiplexed (IHC) staining is providing very rich information that still needs to be completely explored for its ability to help guide treatment decision in immuno-oncology.
These are just examples of an ongoing big data revolution that offers unprecedented opportunities for personalised oncology and, most importantly, immuno-oncology.

A much more complex therapeutic decision strategy ahead of us

This remarkable evolution comes, however, at the price of very high analytical complexity. The multifaceted clinical decision-making process requires cutting edge expertise encompassing pathology, genomics, bioinformatics and medical oncology. Such competencies are difficult to gather within university hospitals and are even more challenging to assemble within the private sector. An additional difficulty is the fragmentation of the typical known disease entities into a much richer set of molecularly defined subtypes, each bearing a low incidence, implying the need to address very large patient populations to derive interpretable outcome statistics.
The Romand Network of Oncology is an initiative originating from the Lausanne University Hospital (CHUV) and the Geneva University Hospital (HUG) that aims at providing a solution to both the knowledge gap from the complex omics data to the treatment decision as well as the assembly of large databases to provide meaningful statistics on rare molecular subtypes.

The unique structure of the Romand Network of Oncology

The general paradigm behind the Romand Network of Oncology is to centralise the complex tumour sample analysis and interpretation, which generates treatment propositions, while decentralising the treatment process whenever possible. This design allows concerted and homogenised treatment plans, while reducing patient travel to the strict minimum.
The overall structure of the Romand Network of Oncology is depicted in figure 1. Peripheral hospitals or private practice-based medical oncologists select patient candidates whose performance status and treatment history make them candidates for inclusion in clinical trials or for the administration of off-label treatment. Such patients typically have advanced stage IV disease and have failed standard therapies. Additional indications are being recognised, including rare cancer types or opportunity windows where a molecularly selected treatment can be delivered safely within the global therapeutic strategy for a given patient.
The referring medical oncologist selects the patient and insures that informed consent is signed before enrolment into the network. Once the Network receives the informed consent, it requests the transfer of the existing tumour material to the reference pathology laboratory (at the CHUV or HUG pathology departments) where the molecular profiling is performed. At the same time, detailed medical information is obtained from the referring oncologist. The resulting clinical and molecular profile is discussed by the Network’s specialised staff, comprising molecular pathologists, bioinformaticians, geneticists and specialised medical oncologists, with the active participation of the referring medical oncologist in charge of the patient (fig. 2). The flexible visio-conferencing system (CISCO System) provides seamless connectivity for professional visio-conferencing solutions and smaller clients that run on personal desktops, tablets or smart phones.
Based on the deep molecular analysis and the clinical aspects brought by the referring medical oncologist, the therapeutic proposals of the molecular tumour board can be quite diverse:

• No treatment can be recommended, based on the molecular analysis and, in some cases, some treatments can even be ruled out based on recognised resistance features.
• An off-label treatment can be proposed, whereby an approved drug for another indication can be considered, based on the discovery of a specific molecular alteration.
  • Such treatments can be obtained from compassionate access programmes or through Article 71 requests.
• A clinical trial can be proposed that matches the patient’s molecular and clinical data.
  • The trial can be an early phase (I/II) within the University Hospitals or any other structure of the Network, as well as a late phase (III) pharma or cooperative group trial.
  • The trial can be regional, Swiss, European or worldwide, depending on the capacity and willingness of the patient to travel.

Figure 1: Overall structure of the Romand Network. Complex tumour analyses and datamining are centralised in the two university hospitals of Lausanne and Geneva, whereas patient care is maintained as much as possible at the regional level by the referring medical oncologist. Only specific phase I programmes require temporary transfer of the patient.

 

Figure 2: Structure of the molecular tumour board. The clinical implications of the molecular profiling performed on the tumour tissue are discussed by a team of specialists including pathologists, biologists, geneticists, bioinformaticians and oncologists. The referring medical oncologist takes an active part in the discussion via a flexible visio-conferencing system.

 

Reaching out to a population of 1.9 million of inhabitants

The decentralised structure of the Network interconnects a large region while minimising unnecessary patient mobilisation (fig. 3). We receive requests from most clinical centres in Romandie, covering a global population pool of 1.9 million inhabitants. Clinical outcomes are captured by data-managers who contact the various referring centres to update the clinical databases. The population pool within reach is comparable in size to several comprehensive cancer centres in the US, enabling personalised oncology strategies to be developed and prospective trials initiated within the Network to be conducted.
The informed consent provided by patient will provide the opportunity to use the molecular and clinical information in research programmes in an encoded/anonymised way (fig. 4). Research projects can originate from the University Centres, regional hospitals or private practice-based medical oncologists. They are reviewed by a Network steering committee that grants access to the data, provided there is approval by an ethics committee. Results obtained by the PI can result into academic publications and/or into clinical trial concepts that can be promoted within the Network for prospective validation.
 

Figure 3: Geographical reach out oncology centres in Romandie that are participating in the Network. Clinical data are maintained and updated with the help of data manager in charge of communicating with the centres.

 

Figure 4: Nominative vs anonymised data within the network. All patient care is handled with nominative data. However, retrospective data analysis requires encoding/anonymisation for each patient. Access to the latter is authorised by the steering committee of the Network and by an ethics committee.

 

Benefits for the patients, the healthcare system and research

The benefits of such a structure are multiple and affect all stakeholders: the patients, the referring oncologist and the Academic Centres. An overview of the benefits for the different stakeholders is given in table 1. In addition, there is also a major potential benefit for the health authorities as such a comprehensive network of expertise could ultimately be a highly valuable tool to determine reimbursement of off-label drugs.

 

Table 1: Benefits for the different stakeholders in the Romand Network of Oncology

Benefits for the patient	Benefits for the referring oncologist	Benefit for the academic centres
• A nearby and experienced medical oncologist • Automatic access to medical innovation • Avoidance of unnecessary/inefficient treatments • Transport reduced to procedures or treatments that are not available onsite	• Strong added value with the direct link to innovation for their patients • Access to cutting-edge clinical decisions (bioinformatics, genomics, …) • Involved discussion in tumour boards • Continuing education • No additional cost	• Recruitment in a large pool of 1.9 million inhabitants • An infrastructure to run clinical trials • Publication policy similar to that of cooperative groups • Continuous education for MDs in training

 

Experience gained during the first year

Since the start of the Molecular Tumour Board in the Network in late 2016, more than 400 cases have been analysed and discussed. Global statistics are shown for 2017 in figure 5. A total of 377 patients benefited in 2017 from the TB molecular expertise. The outcomes of the therapeutic decisions are summarised in table 2. As mentioned in the table, several propositions can be made for each patient and the total percentage is higher than 100%.
Actionable molecular alterations were observed in a third of the patients. For more than half of the patients, relevant clinical trials could be proposed. Finally, off-label treatments were an option for about 46% of the cases. It has to be noticed that several propositions can be made during the Molecular Tumour Board. These statistics are preliminary and subject to change. Also, considerations like actionability or possibility to propose off-label use represent moving targets that will evolve and improve with the availability of a larger arsenal of targeted therapies (fig. 6).

 

Table 2: Outcomes of the therapeutic decisions for the cases analysed between January 2017 and December 2017. Several therapeutic options can be proposed to the same patient.

Proposals from the tumour board	Number of patients	%
Off-label treatment	142	46%
Clinical trial*	157	51%
No molecular options	132	43%
Genetic consulting	23	7%

* Clinical trials were proposed based on molecular alterations or on clinical data matching

 

Figure 5: Rapid adoption by the medical oncology community of the Swiss Romand Network (RRO). Since the opening, we have witnessed a continuing increase in the number of requests during the first months, confirming the recognised need for concerted discussion around the complex decision-making process.

 

Figure 6: Tumour type repartitions for patients presented at the molecular Tumour Board at CHUV and HUG in 2017.

 

Outlook and perspectives

Oncology is at a turning point where a much broader and more detailed understanding of the molecular alterations involved in disease progression is at our fingertips. This opportunity comes with an inherent challenge – to be able to make sense of this vast amount of information and use it optimally to guide treatment decisions. In order to succeed and pave the road to personalised oncology, we need to develop efficient and robust strategies to learn from these rich datasets and use them to point towards the best treatment options. As we move towards that ambitious goal, the gap between raw data and clinical decision must be bridged. Initiatives like the Romand Network of Oncology have the ambition to cover an entire geographic region with a homogeneous care approach, allowing access to sufficient patients to discover and validate the best therapeutic strategies, while providing a common platform of expertise to facilitate the interpretation of the complex data so as to guide optimal treatment decisions for individual patients. We suggest that this exemplary initiative represents significant added value for medical practitioners and, most importantly, for their cancer patients.

 

Acknowledgements

The authors wish to acknowledge the important support of two Foundations, FAMSA and Philantropia, as well as the Geneva Cancer league, for their financial support in this project.

Disclosure statement
Dr Coukos has received grants, research support or is coinvestigator in clinical trials by BMS, Celgene, Boehringer Ingelheim, Roche, Iovance and Kite. Dr Coukos has received honoraria for consultations or presentations by Roche, Genentech, BMS, AstraZeneca, Sanofi-Aventis, Nextcure and GeneosTx. Dr Coukos has patents in the domain of antibodies and vaccines targeting the tumor vasculature as well as technologies related to T-cell expansion and engineering for T-cell therapy. George Coukos receives royalties from the University of Pennsylvania.