As a lifelong proponent of well-conducted and properly controlled clinical trials, I am sympathetic to concerns recently raised in a New York Times editorial (“Easier Drug Approval, at What Price,” June 10, 2018) about efforts by the U.S. Food and Drug Administration to “bring new science to the marketplace faster.” As a pragmatic investigator and physician who has witnessed remarkable progress in the treatments we can now offer individuals with cancer, however, I believe the FDA’s support of more rapid and nimble testing and approval of new therapies should be applauded.
The Times editorial targeted the FDA’s approval of drugs based on small, focused trials using improvement in surrogate endpoints rather than quality or length of life – with cancer as their primary example. But in doing so, it overlooked or downplayed critical elements of clinical research and drug approvals as they happen today.
First, FDA’s Accelerated Approval Pathway does not free a company from the responsibility of proving that a new treatment delivers meaningful patient benefits. This initiative permits conditional approval for a new agent based on surrogate endpoints (such as response), with the proviso that a properly controlled and powered trial using clinically meaningful endpoints (such as progression-free or overall survival or quality of life) is underway and likely to be fully accrued. In the oncology community, in fact, there have been moves to raise, not lower, the bar on clinical endpoints.
The Accelerated Approval process may lead to delivery of exciting and promising new drugs earlier than waiting for the more extensive trial to be completed. Several studies have shown that when queried, most patients with cancer would rather face the toxicities of treatment in hopes of benefit than to abandon a possible benefit to avoid side effects. Patients and their physicians can intelligently discuss these pros and cons during the interim between release of the surrogate and final endpoint results.
Second, stated simply, biomarker-driven clinical cancer care is proven to work. Those of us who care for patients with breast cancer have, of course, been doing this for over 40 years – using estrogen receptor (ER) results to select patients who should, or more importantly, should not, receive anti-estrogen, or “endocrine,” therapies. Likewise, identification of HER2 and treatments targeted to it, specifically trastuzumab, provided yet another example of personalizing care so that those most likely to benefit are treated appropriately, and those who are unlikely to avoid the toxicities and costs. There are a growing number of similar examples in cancer care, and in other conditions as well.
Third, biomarker-driven clinical trials have saved time, money, and lives. In the case of HER2, the original phase I, II, and III trials in both the metastatic and adjuvant settings included only patients whose cancers were highly positive for HER2 amplification or over-expression. This strategy led to much more efficient demonstration of activity than had the trials included all patients, since it focused only on those most likely to benefit. With the success of these trials, investigators questioned if, perhaps, the original cutoffs used for HER2 positivity might have been too stringent. Thus, a subsequent trial was conducted in the United States National Clinical Trials Network of the NCI, led by the NRG group, which tested adjuvant trastuzumab in cancers that did not have amplified HER2 and did have only protein expression of 1-2+. This trial was not positive – the treatment did not work. Had these patients been enrolled in the original trial, their results would have diluted the remarkable benefits we have observed in patients with amplified and/or over-expressed HER2. Indeed, these results have raised a newer trial model, in which two or more cohorts are enrolled into exactly the same trial design, but each cohort is properly powered to address whether the new drug/strategy works in that cohort. This design ensures that if the drug works in the biomarker “positive” cohort, it will not be diluted. However, it also obviates the need to wait 15 years in sequential studies (as we have with HER2 and trastuzumab) to determine whether it works in patients whose cancers may not express the presumed target.
Fourth, in an era in which common cancers (such as breast, lung, or colon) are divided into narrow subsets according to their genomics, smaller trials are often the only way forward. These subsets are, in effect, rare cancers. There simply are not enough patients to conduct traditional randomized trials in a timely fashion. To address the feasibility and benefits of a more nimble precision medicine approach, ASCO is sponsoring the Targeting Agent and Profiling Utilization Registry (TAPUR) study and NCI is sponsoring the Molecular Analysis for Therapy Choice (MATCH). Far from lowering our standards of evidence, these studies seek to identify treatments that can deliver major improvements for patients, and then cut a more direct path to meaningful trial results.
Finally, there are better ways to strengthen drug development than to make it more difficult to test drugs in the clinic. In particular, today’s inconsistent and even chaotic regulatory approaches to tumor biomarker tests need to be reformed. These vitally important tests are not valued properly, resulting in lack of truly rigorous clinical testing to determine clinical utility, a sparsity of high levels of evidence to determine proper guidelines, and lack of confidence in whether to use a tumor biomarker test to guide therapy. I believe that a bad tumor marker is as bad as a bad drug.
Meanwhile, there is no question that the rising financial costs of treatments must be addressed. We all agree that the current situation is unsustainable. The problem has no simple solutions, but policymakers have an array of potential approaches before them, and the “fix” should not be to delay or slow the development and testing of new treatments that patients need.
Taken together, the efforts of the FDA to streamline the approval process, coupled with novel and pragmatic trial designs, should result in getting new therapeutic agents to the right patients at the right time faster than we have in the past. Further, cleverly designed trials, with collection of biospecimens for secondary tumor biomarker test analyses, should lead to even more efficient use of novel drugs.
There will still be a role for prospective randomized clinical trials, but they need to be more efficiently designed, and embrace several endpoints. As a community, we need to think about not only innovation in the drugs themselves, but in new ways to bring those drugs to the patients who need them. The FDA continues to pursue efforts to answer that challenge. I, for one, applaud those efforts.