Is It Drug Timing or Simply Access? How Clinical Trials Game the System

H. Jack West, MD


February 21, 2019

In my estimation, the improvement in overall survival (OS) with the addition of pembrolizumab to first-line standard doublet chemotherapy for either squamous[1] or nonsquamous[2] advanced non–small cell lung cancer (NSCLC), and the improvement in OS conferred by consolidation durvalumab after patients with unresectable stage III NSCLC have completed chemotherapy/radiation,[3] were the most important clinical developments in NSCLC in 2018. These results defined new standards of care in these settings, making immunotherapy an early, proactive treatment approach for patients.

While this approach improves outcomes, we should recognize that these trials, as well as many others in oncology, beg an important question: Do these approaches benefit patients because it is optimal to give these drugs early, or merely because being on a trial's investigational arm is the only way to reliably receive these active therapies?

Approximately half of patients on the first-line chemotherapy arm received what can only be recognized as substandard care today.

Men with stage I testicular cancers may receive adjuvant therapy following orchiectomy as a reasonable standard of care, but surveillance of these patients has long been recognized as a very appropriate and increasingly favored option. This is because of two key factors. First, deferring adjuvant treatment avoids overtreatment of patients who are already cured and would only experience the short and potentially long-term toxicities, as well as the financial costs, of treatment. Second, patients who undergo "salvage" chemotherapy at relapse can do remarkably well, leading to OS results in the surveillance population that are comparable to those of patients who pursue a proactive treatment approach.

Relapsed testicular cancer is arguably a special case, but the concept that there are several paths to the same destination applies more obviously in a setting of incurable disease. The premise behind focusing on progression-free survival (PFS), objective response rate (ORR), or other time-limited endpoints in clinical trials is based on the presumption that when patients live long enough to benefit from several lines of therapy, it undermines a trial's ability to demonstrate an OS benefit. For example, in many trials, patients assigned to a control arm can receive the investigational treatment subsequently, either as part of a planned crossover to the investigational agent during the trial, or to that therapy or a similar treatment off protocol. It is important to know whether a therapy confers an OS benefit, but we should also want to know whether the benefit with this new treatment is timing-dependent. One approach is to incorporate crossover only after an OS benefit is confirmed, provided that the intervention isn't already a standard of care.

Alyson Haslam, PhD, and Vinay Prasad, MD, recently published a thoughtful editorial[4] that discusses how crossover designs may be ethically appropriate or even mandated for treatments that have a proven benefit, while in the case of testing certain hypotheses, they may be unnecessary or even potentially harmful—for example, if patients cross over to an intervention without established benefit, which may result in deferring use of a therapy with proven value.

In trials like KEYNOTE-407[1] and KEYNOTE-189,[2] which demonstrated highly clinically and statistically significant benefits with the addition of pembrolizumab to first-line chemotherapy in advanced NSCLC, only approximately half of the patients progressing on first-line chemotherapy—who were all eligible for immunotherapy at enrollment—crossed over to receive it later. Of note, in the United States and much of the world, second-line immunotherapy is, and for years has been, a standard of care for advanced NSCLC, based on the consistent and well-established survival benefit it confers. This means that approximately half of the patients on the first-line chemotherapy arm received what can only be recognized as substandard care today. While we may debate whether this standard should be applied going back to the time when these trials were designed, in 2019 we can only conclude that 100% of patients with advanced NSCLC receiving immunotherapy over the course of their illness produces a better OS than half of patients receiving immunotherapy over the course of their illness. These trials didn't demonstrate that first-line chemotherapy/immunotherapy is superior to sequential chemotherapy followed by a reliable transition to immunotherapy, not with an attrition rate of approximately half of patients along the way.

It is incumbent upon us to be judicious and somewhat skeptical consumers of marketed data.

This doesn't happen because 50% of patients are too ill to receive subsequent immunotherapy; immunotherapy requires less physiologic rigor than second-line platinum doublet or docetaxel-based chemotherapy. If half of your patients who were good candidates for triplet chemotherapy/immunotherapy a few months earlier are now dying without receiving the immunotherapy and its associated OS benefit, it is a function of a woefully inadequate healthcare system, inattentive clinical care, or both. The poor quality of cancer care outside of clinical trials in many participating centers should not artificially magnify the difference favoring the investigational arm in clinical trials.

I don't mean to say that earlier treatment isn't optimal in settings in which timing may directly limit access. In advanced NSCLC, where early progression may lead to a clinical decline that precludes some subsequent treatment options, the best way to ensure that a patient receives treatment may be to give it early, at the time when they can also tolerate it best. In the recently reported NEJ009 trial of gefitinib versus gefitinib combined with carboplatin/pemetrexed as first-line treatment for patients with EGFR mutation–positive advanced NSCLC,[5] only 75% of patients who progressed on first-line gefitinib alone went on to receive the intended chemotherapy later, a gap in treatment related to the more extensive progression and worse performance status at progression of those patients who received gefitinib alone compared with those given gefitinib and chemotherapy. There is also the real possibility that certain therapies may confer greater benefit when administered earlier and/or concurrently with other treatments, and to patients with a lower tumor burden and better performance status.

But what is the potential harm of just "putting it in the water supply" and treating patients as proactively, and as early, as possible to ensure that they get access? An obvious one is cost, both to patients and society, as cancer treatments now regularly exceed $15,000/month, and many of these treatments have become part of a patient's care for many months or even years. In fact, earlier treatment makes it far more likely that these treatments will have a longer duration. This is a bonanza for pharmaceutical companies but a major concern if it forces patients to sell their homes or spend their family's retirement savings to pay for a treatment that may be just as effective when given later, only as needed, for a more limited time. For instance, a recently reported phase 2 trial of adjuvant erlotinib following resection of EGFR mutation–positive NSCLC demonstrated an impressive disease-free survival but without any indication that this approach improves OS.[6] Reviewing these results, many oncologists may be tempted to test for driver mutations in patients with resected NSCLC and then consider pursuing adjuvant targeted therapy without a clear endpoint—and at a remarkable cost and very significant risk for chronic toxicity. Beyond the issue that many of these patients may already be cured without subsequent standard therapy and are thus only incurring the cost and toxicity of these drugs, even those who are destined to relapse may ultimately do no better than patients who start targeted therapy when their disease recurs.

We should be open to the possibility that distributing treatments over time could be the best approach for some patients.

So, what do we do in the face of trials showing that early use of novel therapies improves clinical outcomes but only compared with sequential, and less reliable, treatment using the same approaches? In the absence of competing data supporting a sequential approach, newer strategies may be optimal, and I think it is appropriate to follow them for most patients. But we shouldn't forget that sequential treatments remain a reasonable option for some patients and may be a strong alternative for those with more indolent disease and therefore at low risk of progressing past the point where they are no longer candidates for later treatment, and for those in whom cost or other practical barriers to early treatment may be problematic. Ideally, we would hope to test concurrent versus sequential strategies in ways in which access is not a confounding variable—though the very question runs counter to the marketing incentives of companies looking to justify earlier and longer treatment.

It is incumbent upon us to be judicious and somewhat skeptical consumers of marketed data, particularly when the products we are selecting have costs that deplete retirement accounts for patients and raise healthcare premiums for everyone. Adding immune checkpoint inhibitors, targeted therapies, or other novel treatments early on may be the best strategy in many settings, but we should not allow ourselves to be misdirected: Earlier is not necessarily better unless the alternative includes fair access to treatments with established benefit later. We should be open to the possibility that distributing treatments over time could be the best approach for some patients and settings, even if that isn't the optimal strategy for drug companies, who are best served by running trials that falsely confound data on earlier treatment by providing better access to helpful interventions.

Follow Medscape on Facebook, Twitter, Instagram, and YouTube

Follow Jack West on Twitter: @JackWestMD


Comments on Medscape are moderated and should be professional in tone and on topic. You must declare any conflicts of interest related to your comments and responses. Please see our Commenting Guide for further information. We reserve the right to remove posts at our sole discretion.