The Friends of Cancer Research (Friends) Annual Meeting 2019 began with welcoming remarks by Ellen Sigal, Chairperson & Founder of Friends, and Jeff Allen, President & CEO of Friends. In her remarks, Sigal emphasized the time and effort spent by working group members on putting together collaborative whitepapers meant to reach a level of consensus and push the conversations forward to achieve progress for cancer patients. She encouraged meeting participants and working group members to continue this forward-thinking and action-oriented dialogue throughout the day. As part of his welcome remarks, Allen announced the launch of the new collaborative ctMoniTR Project, a Friends initiative aimed at characterizing the use of circulating tumor DNA (ctDNA) as a monitoring tool that will enable the early identification of treatment response in cancer patients. This new project builds off work presented at the Friends Annual Meeting 2018.
Panel 1: Characterizing the Use of External Controls for Augmenting Randomized Control Arms and Confirming Benefit
As an extension of work presented at the Friends Annual Meeting 2018, the first panel sought to answer the following question: While not meeting the same gold standard as a randomized controlled trial (RCT), can a well-designed study with an external control still provide adequate evidence of treatment effectiveness? To provide context for the working group’s whitepaper and panel discussion, moderator Gary Rosner of John Hopkins University began the panel with a brief presentation detailing the history of randomized clinical trials and providing examples of situations in which RCTs are not always feasible, such as in the case of patients seeking single-arm trials or trials that allow treatment crossover, rare diseases, or loss of equipoise.
“We’re not always getting the right conclusion from RCTs; they’re failing because patients don’t want to be on the control arm; we should consider if synthetic control arms could help answer the research question” —Ruthie Davi
Following this background presentation, working group member Ruthie Davi of Acorn AI, a Medidata Company, presented a case study in multiple myeloma that explored whether the treatment effect based on a synthetic control arm (SCA) can mimic the treatment effect based on a randomized control arm. This case study also developed and illustrated statistical methods useful for assessing the impact of unobserved confounders on the demonstration of efficacy in the setting of the SCA. Results from this case study showed that the treatment effect when based on a synthetic control arm was highly similar to the treatment effect based on a RCT, suggesting that synthetic controls may be useful in situations when it is not feasible or ethical to randomize patients to the standard of care. Results also indicated that tipping point analyses are an effective way of understanding the possible impact of unobserved confounders on treatment effect analyses and their statistical significance.
During the panel discussion, panelists reiterated the value of randomization and RCTs, but noted that in certain situations randomization is not feasible and alternative approaches, such as the utilization of external data, are needed. Erik Pulkstenis of AbbVie and Ruthie Davi stressed that in the absence of a research paradigm that allows for the use of alternative data sources in situations when randomization is not working for patients, false conclusions may be reached. Panelists also discussed the value of different data sources such as real-world data garnered from electronic health records (EHRs), which has the benefit of being almost completely concurrent, and clinical trial data, which is most likely collected in a more routine manner. Panelists agreed that sponsors in collaboration with other stakeholders such as regulators should take a serious, scientific approach toward validating these methods.
Panel 2: Immuno-Oncology Drug Development for Patients with Disease Progression After Initial anti-PD-L(1) Therapy
The second panel discussed the challenges associated with enrolling patients who have progressed after anti-PD-L(1) therapies onto clinical trials testing combination therapies including a PD-L(1) inhibitor. After an initial background presentation by moderator Ryan Sullivan of Massachusetts General Hospital, a case study of Merck’s Keynote-001 trial was presented by Eric Rubin. Maurizio Voi of Novartis then gave a presentation about the opportunities platform trials present for combination therapy drug development and provided examples of a couple of active platform trials that seek to investigate the safety and efficacy of combination regimens including PD-(L)1inhibitors in patients who experience disease progression after initial PD-(L)1inhibitors.
“The medical journey is more important than the clinical trial design; how do we capture those different entry points into trials and segment them so they make statistical sense” – TJ Sharpe
After these initial presentations, panelists began the discussion by agreeing that having harmonized definitions for relapsed and refractory disease across companies and clinical trials could be valuable to advancing drug development. Panelists also highlighted the importance of demonstrating the contributions of each component in a combination therapy in order to prevent patients from being exposed to non-efficacious treatment. The panelists commented on different clinical trial strategies and statistical approaches that would allow for an earlier exploration of treatment response and modifications based on interim analyses. Patient advocate T.J. Sharpe emphasized that the statistical considerations in the design of clinical trials must be accounted for while balancing the unique goals of individual patients. He also stated that the ultimate goal is stakeholders collaborating on clinical trial design to get effective therapies to patients like him who need them. Sharpe advocated for clinical research as a care option for patients and stressed the importance of humanizing clinical research and the interactions between physicians and patients in order to create clinical trials that properly balance efficacy assessment and equipoise.
Lunch Keynote: Project Orbis
This year’s lunch keynote focused on the FDA’s Oncology Center of Excellence (OCE) Project Orbis pilot program and featured representatives from the FDA, Health Canada, Eisai, and Merck. Rick Pazdur, Director of the OCE, began the discussion by describing the aim of the program, which is to improve patient access to new drugs in countries experiencing delayed regulatory submissions through international regulatory collaboration. Panel members then described their experience with the program, which recently completed its review of a submission by Eisai and Merck that was approved by each Project Orbis participant (FDA, Health Canada, The Australian Therapeutic Goods Administration). Pazdur noted that there may be additional countries involved in future application reviews under Project Orbis, and iterated that each country still establishes its own opinion whether to ultimately approve the drug and subsequent drug labeling.
Panel 3: Data Generation (and Review Considerations) for Use of a Companion Diagnostic for a Group of Oncology Therapeutic Products
The third panel explored evidentiary standards that could be useful in supporting a determination that an IVD companion diagnostic (CDx) device is appropriate for use with a class of therapeutic products. Moderator Steffan Ho of Pfizer began the panel with a background presentation framing the issue, reviewing relevant FDA draft guidance, and posing the following question: What is the appropriate evidence we would need to enable a companion diagnostic to be used for a broad class of drugs?
Panelists agreed that the current regulatory paradigm, which requires each test to be evaluated and approved for guiding the use of every therapy on its label, can create challenges in terms of patient access and complicates clinical trials. For example, clinical trial samples are in limited supply, making it infeasible to evaluate every test:drug combination. David Hyman of Memorial Sloan Kettering highlighted the negative effects this has on patients seeking enrollment on a clinical trial who must often wait for additional biopsy results to be received from a central laboratory rather than using previous test results or being tested by a local laboratory. Preeti Narayan, FDA, noted that class labeling could potentially reduce the number of invasive biopsies and time-to-treat for patients but, also, noted the need to be cautious from regulatory perspective to ensure that each drug in a drug class is accurately represented by a test/biomarker to support a group label.
“I think we need to start at the patient side and what they’re experiencing and build everything toward that, not the other way around.” – Andrea Ferris
The panel moved on to discuss why there is still a limited number of CDx approvals despite the release of guidance on the development of CDx in 2014. The panel speculated that, in some cases, business agreements between diagnostic and pharmaceutical companies could be an impediment and current incentives should be re-evaluated, particularly due to the impact FDA approval status can have on coverage and reimbursement. However, all agreed that the 2018 proposed guidance from FDA and the additional considerations outlined in the whitepaper are an important step forward. Panelists went on to discuss the framework for group labeling outlined in the whitepaper in more detail. They stressed the need for alignment on the extent of evidence needed to support a group labeling claim and what considerations, including complexity of technology and biology of the test biomarker, would contribute to that determination. Lakshman Ramamurthy, GSK, pointed out the concern of how to approach development and regulation of complex biomarkers such as homologous recombinant deficiency, which is explored in the whitepaper. The panelists acknowledged that some CDx may not be eligible for group labeling. The panel ended with final thoughts from Steffan Ho, who identified reaching out to his colleagues at diagnostic companies where he believes a group label is feasible as a personal next step, and Andrea Ferris of LUNGevity, who cautioned meeting participants not to let perfect be the enemy of the good on this issue.