Patient Stratification in Clinical Trials Emerges as Theme at MMTC
By Bio-IT World Staff
February 17, 2014 | For the past 21 years, the Molecular Medicine Tri-Conference (MMTC) has convened in San Francisco, CA, to discuss trends in using molecular knowledge to drive drug discovery and patient care. Since the meeting’s first iteration in the early days of the Human Genome Project, the potential of human genotyping to personalize the practice of medicine has always been a hot topic at MMTC. However, failures to deliver on this promise at any serious scale have haunted the health care system since the Human Genome Project’s completion. For over a decade, MMTC discontinued its clinical sequencing track, which had the most direct relevance to personal genetic medicine, but never seemed any closer to realization year after year.
This year, as MMTC was held from February 9-14, the clinical sequencing track was reintroduced, thanks to a changing environment in which CLIA labs regularly sequence patients at least in extraordinary cases, and the FDA has at last approved a next-generation sequencing instrument to be used in clinical diagnostics. Speakers and attendees at the conference, however, converged on another area where genetic profiling of patients can provide some immediate relief in a failing system. Over the six days of the conference, arguably the most urgent question to emerge was when and how clinical trials will adapt to take full advantage of patients’ genotypes, which have a massive bearing on treatment success. Most drug trials, especially in oncology, now try to establish genetic markers as co-diagnostics in finding the most responsive patient groups, but participants seemed unanimous that more fundamental changes are both possible and needed.
David Gandara of the UC Davis Cancer Center, who delivered the opening keynote on Monday, spoke of his experience working on non-small cell lung cancer (NSCLC), one of the most complex and heterogeneous cancers on the genetic level. He made a provocative case that, despite massive investments in lung cancer, the only serious advance in treating NSCLC over the past decade has been finding biomarkers in advanced stages of the disease, which for a handful of mutants can lead to targeted therapies. Even this is not enough to extend patient lives for very long, because of NSCLC’s rapid acquisition of drug resistance. Most discouraging to Gandara was clinicians’ and trial managers’ failure to use genetic markers that have already been discovered, preferring either histological markers or patient profiles, like race or age, that have been associated with particular genotypes. His message here was simple: when it comes to choosing cancer therapies, “genotype trumps phenotype.”
As a way forward, Gandara touted a radical reimagining of clinical trial process that is now underway as a public-private partnership between five pharmaceutical companies, the FDA, Foundation Medicine, and the Friends of Cancer Research. This cooperative program invites lung cancer patients to have their tumors sequenced before being assigned a therapy. Patients’ tumor genotypes will allow them to be assigned to one of four separate phase 2/3 trials, each undertaken by a different pharmaceutical company and targeting a different mutation – or to a fifth arm testing a non-targeted therapy for patients who are not genetic matches for any of the other four. This multiple-arm approach provides relevant patient populations for each drug under consideration, while ensuring that every patient who volunteers is assigned to a promising clinical trial.
Kimberly Blackwell of the Duke Cancer Institute, in her keynote presentation on Tuesday morning, treated attendees to a whirlwind review of how the discovery of the HER2 mutation in certain breast cancers powered the most rapid advance in breast cancer treatment of the past twenty years. From the first drug targeted to HER2 breast cancer, trastuzumab or Herceptin, approved in 1998, to its relatives pertuzumab and T-DM1 over the past three years, therapies that intercept the HER2 pathway have recently made HER2 breast cancer far more treatable, despite the poor prognosis that comes with a HER2 diagnosis. While the story of HER2 treatments is a hopeful one for molecular targeting of therapies, Blackwell warned that researchers must be prepared for much greater challenges to come. “The reality is that the next revolutions will have to come from understanding those pathways we all find so complex and mind-boggling,” she said. “It’s unlikely that we’ll see another success story [like HER2].”
(At a separate talk on Monday, Michael Liebman of Strategic Medicine, Inc. gave a startling statistic that illustrates how, even in a true success story like HER2, clinical practice has lagged behind knowledge. Two out of three women eligible for HER2 testing are still not genotyped, he said, while one out of five women prescribed Herceptin were never tested for HER2 mutations.)
Speakers in a variety of individual tracks also spoke to opportunities for better genetic description of patients in clinical trials. Peter Jones, former director of the USC Norris Comprehensive Cancer Center, stressed the need to include the epigenome in patient stratification, pointing out that “every human cancer that has been looked at has included methylation changes, without exception,” and considering the example of Vidaza, a hypomethylating drug that took 40 years to win FDA approval. Jocelyn Holash of Novartis cited a need to start patient stratification in preclinical studies, and to look beyond the direct molecular targets of drugs in development, showing how massive screening of cell lines revealed promising gene targets of PIM inhibitors although PIM genes themselves are rarely mutated. And Barry Taylor of UCSF described an approach to moving forward from failed clinical trials, in which researchers find one or two outlier patients who responded exceptionally well to the drug in question, and exhaustively sequence their tumors to find the genetic pathways that made these cancers sensitive to treatment, potentially unlocking new markers for stratification or new drug targets altogether.
Facing Up to the Obstacles
Over the last two days of MMTC, a symposium entitled “Leaders in Precision Medicine” became home to some remarkably frank discussions of how all players in the clinical trial ecosystem have thrown up roadblocks to a thorough embrace of genetic patient stratification. Representatives of regulatory agencies, large payors, and major pharmaceutical companies, as well as academic observers, all contributed as both panelists and audience members. There was remarkable agreement on all sides as participants cited regulators’ slowness to accept meaningful new clinical endpoints like length of hospital stays, and payors’ reluctance to fund original research that could better stratify patients for best treatment with existing and legacy drugs, despite having arguably the largest financial incentives in this research.
Adaptive licensing – a proposed model by which drugs are first approved for very small, targeted patient groups, and could later have their approval expanded pending new trials – was given intense scrutiny at the symposium. While all members agreed that this model is already legal in most countries and will probably be necessary as patient stratification grows more powerful, participants on all sides had concerns about putting it into practice. Regulators worried about ensuring that only the right patients receive a drug once it’s been approved for a narrow population, while both pharma and payors struggled to make sense of the payment model: should the cost be deflated on first approval, to encourage expanded trials? If a drug is licensed for one use, but fails future trials for broader use, will the manufacturer be able to go back and negotiate a higher price? And does pharma really have the incentive or infrastructure to develop drugs for non-orphan conditions at orphan-like population sizes?
A moderator at the symposium also raised the issue of divergent standards for drugs versus diagnostics, pointing out that drugs can be approved after working in as few as 10% of patients, but diagnostics are still held to a minimum of 95% sensitivity and specificity. A pharma representative conceded that many drugs on the market simply aren’t powerful enough without better co-diagnostics, which are hugely difficult to develop under those stringent demands. A payor representative, meanwhile, acknowledged that being receptive to less-accurate co-diagnostics could save lives by getting at least a few more patients the best-targeted treatments. While these participants were speaking in a privileged atmosphere, where they could expect a reasonable freedom to speak off the record and an audience that understood the restrictions they face, the consensus between different players in clinical trials suggested that the time may be approaching when an outmoded system will better adapt to differences between patients, and the need to tailor medicine to the individual from clinical trials up.