On September 20, 2010, The Engelberg Center for Health Care Reform at Brookings and Friends of Cancer Research convened the third annual Conference on Clinical Cancer Research. Building on discussions from the previous two years, the event aimed to achieve consensus – where possible – on critical questions and issues that affect the effectiveness and efficiency of clinical research. The day’s discussions featured leaders from the National Cancer Institute and the Food and Drug Administration (FDA), along with researchers from academia, the life sciences industry, and patient advocates. The American Society of Clinical Oncology, the American Association for Cancer Research, and Susan G. Komen for the Cure supported the event.
Four panels comprised of 21 thought leaders from the NCI, academia, FDA, patient advocacy groups, and industry used issue briefs developed over months of collaboration to discuss and frame a variety of issues including:
- The use of adaptive clinical trial design in the context of co-development of matched diagnostics and therapeutics.
- Employing systems biology methods to enhance and accelerate safety assessments in the pre-clinical phase of drug development.
- Incorporation of pain metrics in cancer clinical trials, including a discussion of the challenges of measuring these endpoints and ways of overcoming these challenges.
- Patient-initiated study participation as a model for collecting the tissue and data needed to identify biomarkers that predict response to existing cancer therapies.
Dr. Harold Varmus, director of the National Cancer Institute (NCI), delivered a morning keynote address that provided an overview of NCI work and how it relates to clinical cancer research. Emphasizing the heterogeneity of cancer, he spoke of the need to reorient the scientific community away from questions that prioritize funding streams and toward “provocative” questions like the use of combination trials and molecularly targeted agents that could more fully explain the biological consequences of mutations, and as a result accelerate better control of cancer.
However, Dr. Varmus also spoke about the regulatory, scientific, and intellectual property challenges that such efforts face. He gave several examples of new NCI initiatives, such as the Center for Cancer Genomics, that will support this kind of work by incorporating programs like The Cancer Genome Atlas, the Experimental Therapeutics Program (NExT), not sure (CAN), and others under the banner of therapeutic development. He also mentioned the creation of other initiatives like the Lasker Scholarship program – designed to attract clinical researchers often priced out of the market – and the support for collaborative initiatives between institutions like Johns Hopkins University and the Department of Defense through efforts like the NIH Intramural Clinical Program. He also mentioned that there has been a good deal of thinking about how to build appropriate digital databases for clinical cancer – something that he noted could be immensely useful to all major constituencies.
Food and Drug Administration (FDA) Commissioner Dr. Margaret Hamburg delivered the luncheon keynote in which she highlighted the ways in which the focus and work products of this meeting aligned with the FDA’s prioritization of advancing and applying effective regulatory science. Referring to a recently released FDA white paper on regulatory science, Dr. Hamburg observed that cancer is one example of where the better application of regulatory science can make a significant difference in guiding the assessment of subpopulations, and referred to the I-SPY-2 trial as a groundbreaking new clinical trial model and a promising example of regulatory science at work. She discussed a number of new FDA initiatives with the NIH around developing new clinical trial designs, and talked about an upcoming FDA white paper on modernizing safety testing to identify and mitigate drug toxicity.
Panel One: Adaptive Clinical Trials Designs for Simultaneous Testing of Matched Diagnostics and Therapeutics
- Rajeshwari Sridhara, Divisions Director, Office of Biostatistics, Center for Drug Evaluation and Research, US Food and Drug Administration
- Richard Simon, Chief, Biometric Research Branch, National Cancer Institute
- Eric Rubin, Vice President, Oncology Clinical Research, Merck
- Shelley Fuld Nasso, Director, Public and Medical Affairs, Susan G. Komen for the Cure
- Howard Scher, Chief, Genitourinary Oncology Service, Memorial Sloan-Kettering Cancer Center (Moderator)
Cancers of a primary site often represent a heterogeneous group of diverse molecular diseases which vary fundamentally with regard to the oncogenic mutations that cause them or their responsiveness to specific drugs. However, most cancer drugs developed today are designed to inhibit specific cancer pathway targets. Histology-based, “all comers” approaches to developing these drugs have typically led to failure in phase III studies, or demonstration of “success” based on statistically significant, but clinically questionable benefit in an “all comers” population. Moreover, selection of a diagnostic test to identify patients who will benefit from treatment with a drug is difficult in early clinical trials, and when this has been done, it has often been incorrect. Essentially, conventional Phase I and II trials lack sufficient power to identify responsive subgroups, and progress in generating diagnostic tests that can be used to select responsive patients has been slow.
With this as background, experts from the first panel set out to discuss how best to develop new drugs in a manner more consistent with modern tumor biology and obtain reliable information about what regimens work for what kinds of patients. Employing castrate-resistant prostate cancer as a case study, the panel proposed the design of a pivotal phase III clinical trial approach that adaptively identifies a responsive patient population and confirms the effectiveness of a new therapeutic in this population in a rigorous statistical manner – which would allow for an actionable claim. Dr. Rajeshwari Sridhara, director of the Office of Biostatistics in the FDA’s Center for Drug Evaluation and Research (CDER) observed that “change to enrich sub-populations” as the primary definition of “adaptive” is critical. She went on to suggest that this type of targeted drug development could meet regulatory approval if four key factors were included: good pre-clinical data, identification of the target population, a drug that actually hits the target population, and a substantial treatment clinical benefit. She gave the examples of Herceptin and Gleevec as successful cases; and PARP inhibitors, BRAF inhibitors, ALK inhibitors as in-development cases.
Dr. Robert Temple, director of the Office of Medical Policy at FDA CDER, endorsed Panel One’s approach and agreed with Dr. Sridhara’s primary definition of “adaptive”. He emphasized that adaptive design allows clinical researchers – even when they do not know what precise target population they are treating in advance – to increase the number of people in a trial that are likely to benefit. Dr. Temple gave the example of Erlautinib, a drug approved for the general population of non-small cell lung cancer. In the fraction of patients that had EGFR positivity, survival was roughly seven months and response rate was very large. In the EGFR negative population, survival was slightly adverse. Studies like these illustrate the “fantastic possibility” of separating responders from non-responders; however, they also point to one of the major questions for FDA: how much data is desirable to have in the off-population group?
Panel Two – Identification and Elucidation of the Biology of Adverse Events: The Challenges of Safety Assessment and Translational Medicine
- John Leighton, Associate Director for Pharmacology/Toxicology, US Food and Drug Administration
- Myrtle Davis, Chief, Toxicology and Pharmacology Branch, National Cancer Institute
- Leigh Ann Burns-Naas, Senior Director for Drug Safety Research and Development, Pfizer Inc
- Adam Clark, Director, Scientific and Federal Affairs, FasterCures
- Ken Turteltaub, Executive Director, Battelle Center for Fundamental and Applied Systems Technology, Battelle Memorial Institute (Moderator)
“Safety is a frequent cause of preclinical and clinical drug candidate attrition. There has been an explosion of technology-enabled scientific insight into the basic biology of the causes of adverse events. This has been driven, in part, by the development of the various “omics” tools (genomics, proteomics, metabolomics etc.) and associated bioinformatics platforms. Meanwhile, for decades, there has been little to no change in pre-clinical testing protocols and guidelines. Integrating novel in-vitro, in-silico and improved in-vivo methods in the discovery and pre-clinical stages could create actionable understanding of the ‘safety biology’ of drug candidates and help bring products with well-understood safety profiles to market more efficiently.” (from ‘Conference Topics’)
Defining systems biology as “complex interactions of a biological system and how it gives rise to the function or loss of function of a system”, the second panel discussed whether systems biology offers major improvements in the safety evaluation of new chemical entities with indications of oncology, where systems biology adds value, and where guidance is needed. Panelists observed that systems biology could measure molecular effects relevant to adverse events, and focused its discussion on how to implement this on a broader scale. One area where the panel agreed more guidance could be useful was in the need for improved models of toxicity and patient benefit. Currently, animal models of carcinogenicity are not good at extrapolating danger in humans, and illustrate a central problem in drug discovery: the large amount of resources that exist between discovering a promising new drug and demonstrating its effectiveness in humans.
Dr. Leigh Ann Burns-Naas, senior director for Drug Safety Research and Development at Pfizer, provided a drug-induced vascular injury (DIVI) case study on how a systems biology approach can help elucidate better informed toxicology information than traditional approaches. She pointed to 13-14 year cycle times using traditional preclinical modeling and clinical translation, and suggested that systems biology could allow clinical researchers to effectively determine what adverse events could be “dialed out” of a drug through a better understanding the relationship of drugs to key “omic” signatures. Dr. Myrtle Davis, chief of the Toxicology and Pharmacology Branch at NCI, provided a second case study on tyrosine kinase inhibitors that emphasized many of Dr. Burns-Naas’ observations. However, while Dr. Davis agreed that there is a need for greater perspective on how mechanistic insights can be used more effectively in drug development, she wondered if groups are publishing mechanistic data rapidly enough, and if time frames for mechanistic studies and drug development are compatible. Dr. John Leighton, associate director for Pharmacology/Toxicology at the FDA, closed the panel presentation by describing how the FDA currently uses genomic information, and discussed where the field is going.
Dr. Janet Woodcock, director of the Center for Drug Evaluation and Research at FDA, proposed that the NCI – with support from FDA – catalyze the creation of a public-private partnership or consortium on building a database on cancer drug toxicity. Dr. Woodcock observed that one of the main obstacles to building such a database is the lack of an incentive to do this kind of work in a clinic. She said that the FDA will publish guidance on submission and approval for regulatory use in clinical or pre-clinical. Dr. Vicki Seyfert-Margolis, senior advisor within Science Innovation and Policy for the FDA Commissioner’s Office, agreed that a consortium would be a “great idea,” noting, “We haven’t set up standards to compare against – healthy control sets of data that are very clear and well characterized. In order to nail down correlations and markers that define a clinical state, we need a lot of work on standards and thinking through dynamic processes.”
Panel Three – Integrating Pain Metrics into Oncologic Clinical and Regulatory Decision-Making
- Laurie Burke, Associate Director for Study Endpoints and Labeling Development, Office of New Drugs, US Food and Drug Administration
- Ann O’Mara, Head of Palliative Care Research, Community Oncology and Prevention Trials Research Group, National Cancer Institute
- Martin Zagari, Executive Director, Department of Health Economics and Outcomes Research, Amgen
- Carole Baas, Advocate, Physical Sciences in Oncology, NCI
- Charles Cleeland, Department Chair, Department of Symptom Research, MD Anderson Cancer Center (Moderator)
“The primary outcome of pain intensity or pain relief is subjective and has long been recognized to be susceptible to bias in studies that are not double-blinded. Ethical considerations are often cited for the absence of placebo controls in cancer pain trials. The use of randomized, single-blind crossover trials may be the answer to ethically employ placebo controls by allowing patients access to medication if needed (“rescue” medication). Epidemiologic evidence on the incidence and prevalence of cancer-related pain, and on the likelihood of increasing pain intensity with advancing cancer stage indicates that cancer pain adds substantially to the already considerable national disease burden of cancer. Prevalence data imply that the number of patients enrolled in methodologically sound trials of cancer pain relief is a tiny fraction of those receiving care resulting in poor quantity and quality of scientific evidence on cancer pain relief.
Quality of life is another area has not been adequately assessed in trials of oncology drugs and nondrug interventions for cancer pain. Advances in quality-of-life assessment and insights from research into the relationships among pain, disability, and impairment offer the opportunity to understand these interactions in the context of cancer pain.”
The third panel discussed both the challenges and reasons involved in integrating pain metrics into oncologic clinical and regulatory decision-making. Much of this stems from the fact that pain is a disabling consequence not only of cancer, but also of some cancer treatments – this latter type of pain being much more difficult to control. As Dr. Carole Baas, a patient advocate with the Physical Sciences in Oncology at NCI, put it, “In cancer patients, pain is one of the most feared and burdensome symptoms.” Dr. Laurie Burke, associate director for Study Endpoints and Leaving Development at the Office of New Drugs at FDA, provided an overview of the current state and FDA practice of incorporating pain metrics in cancer trials. She emphasized that we need a very clear definition of terms, as poor planning and funding will “always bias study results to the null.”
Ann O’Mara, head of Palliative Care Research at the Community Oncology and Prevention Trials Research Group at NCI, suggested that there is a need to look at other methods of treating patients. She provided a case study on short- versus long-term course treatment for palliation of painful bone metastasis, where, respectively, patients were required to take one trip to the hospital to receive treatment versus many trips. At three months after radiation therapy there was no difference between the two arms, suggesting that we can deliver treatment differently to the same outcome.
The third panel discussion ended with a presentation of the industry perspective by Dr. Martin Zagari, executive director for the Department of Health Economics and Outcomes Research at Amgen. He observed that integrating pain metrics into oncologic clinical and regulatory decision-making is difficult as drug development costs continue to escalate, research productivity is down, and payment for innovative products is increasingly uncertain. Moreover, the integration of pain metrics is particularly problematic because the measurement of pain and symptoms is imperfect, the path to labeling and meaning of results is unclear, and the risk is perceived as high. As a result, “hard” regulatory requirements and trial “must-haves” are prioritized over “softer” factors. Dr. Zagari went on to provide a series of suggestions for stimulating more pain and symptom research interest.
One commenter made the observation that clinical trials for cancer drugs are artificial constructs of the real world. Every other area of the FDA requires two trials for drug approval, and suggested that cancer should have one smaller trial centered on pain palliation.
Panel Four – Using Patient-Initiated Study Participation in the Development of Evidence for Personalized Cancer Therapy
- Sue-Jane Wang, Associate Director, Pharmacogenomics and Adaptive Design, Office of Biostatistics, US Food and Drug Administration
- Ken Buetow, Director, Bioinformatics and Information Technology, National Cancer Institute
- Jamie Freedman, Vice President, R&D, GlaxoSmithKline
- Laurie Fenton Ambrose, President and CEO, Lung Cancer Alliance
- Stephen Friend, President, Sage Bionetworks and Richard Schilsky, Chief-Section of Hematology-Oncology, Deputy Director, University of Chicago Comprehensive Cancer Center (Co-Moderators)
On average, only one quarter of cancer patients receiving an approved cancer drug regimen will gain a significant benefit from that treatment. Despite this lack of effectiveness, these current therapies are considered “standards of care.” The development of innovative medical technologies provides an opportunity to identify patients who are more or less likely to benefit from currently available cancer therapies, however such studies may be difficult to conduct as prospective clinical trials. “The goal of this panel was to assess the feasibility and limitations of directly engaging patients to provide access to their tumor specimens and clinical data to enable population-based studies of molecular determinants of treatment outcome.”
The fourth panel complemented panel one’s discussion on identifying biomarkers mid-trial by providing the perspectives of patients, providers, clinical investigators, drug developers and regulators in designing a post-approval non-responder study – using non-small cell long cancer as a case study – that could assess such questions around the feasibility and limitations of patient-generated biospecimens and clinical data in guiding regulatory decision-making. Dr. Sue-Jane Wang, associate director for Pharmacogenomics and Adaptive Design at the Office of Biostatistics at FDA, outlined a set of principles for developing evidence to modify an existing label, noting that efficacy versus toxicity may bear different evidential criteria. Dr. Stephen Friend, President of Sage Bionetworks, outlined the Sage Non-responders project that would incorporate these principles, and the panel then discussed in detail the need for patient engagement, effective and rigorous data collection, and data management. Ken Buetow, director of Bioinformatics and Information Technology at NCI, provided the Army of Women’s Health of Women study as an example of how structured information can be used to actively and quickly recruit patients: in 10 days, the study was able to collect observation data on 30,000 women. Several panelists observed that this type of study design offers a promising approach to laying a deep foundation for sub-classifications that could significantly increase the database for clinical studies.
Dr. Mark McClellan, director of the Engelberg Center for Health Care Reform at the Brookings Institution, closed the conference by providing a summary of the discussion and next steps on the recommendations put forward by the panels. Dr. Varmus’ emphasis on the heterogeneity of cancer as a disease and Dr. Hamburg’s emphasis on the importance of regulatory science both underscored the need for an increasing number of substantive collaborations moving forward – a goal well represented by the conference panel contributions. The first panel laid out a clear approach for identifying unknown biomarkers, and developed a rough consensus on how their proposed methods could move forward – possibly with a guidance document that would invite collaboration. The second panel on systems biology recommended that work be focused on shortening the timeline to understanding risk, and using systems biology to help accelerate that process. This led to a dialogue on how a consortium, including the FDA and NCI, could share data to improve processes. The third panel on integrating pain metrics led to conversations about developing better guidance and further studies. Finally, the fourth panel was viewed as a well-suited complement to panel one by helping to significantly increase the size of the database for clinical studies, and beginning to merge health care delivery and clinical research.