Improved Survival in Randomized Trials: A Meaningful Outcome but Increasingly Difficult to Demonstrate

Maurie Markman, MD

There are several well-recognized and somewhat controversial issues surrounding the clinical trials process employed by regulatory agencies, which is responsible for approving antineoplastic agents for noninvestigative use as single agents or as components of a combination regimen. These issues include the relative paucity of older patients included in such studies compared with their actual representation within the population who develop most malignancies, the general exclusion of individuals with common comorbidities (cardiac disease, diabetes, mild to moderate renal dysfunction, etc), and the frequent limited presence of certain ethnic and socioeconomic groups.

There is probably no more widely debated topic in this arena than that of the appropriate primary outcome end point to declare success for having demonstrated improved clinical benefit in a trial, particularly when considering the study findings to define a new standard of care or even a reasonable therapeutic option in a given clinical setting. Before proceeding with the specific issue to be highlighted in this commentary, it is appropriate to acknowledge a related and similarly unresolved critical question as to the specific roles and ultimate authority of various interested parties (eg, patients and their physicians, academic opinion leaders, regulatory bodies, payers) in the determination of what defines clinical benefit.

However, for the purpose of the current discussion, we will acknowledge the primacy of the long-standing gold standard study end point of an experimental anticancer regimen demonstrating a statistically significant improvement in overall survival (OS) compared with that of a carefully selected control arm. Other potential efficacy outcomes, such as progression-free survival (PFS), are generally considered to be surrogates for meaningful clinical benefit.¹

These debates often begin with the stated assumption that patients with cancer consider statistically significantly longer OS in a randomized clinical trial, no matter how objectively limited in magnitude, along with achieving a measured enhancement in quality of life (or at least no detrimental effect) to be evidence of a clinically beneficial impact of therapy. Furthermore, the argument continues that the demonstrated delay in disease progression (including time to development of cancer-related symptoms) or evidence of objective tumor responses are at a lower level of relevance.

It is certainly appropriate to agree with the general conclusion that improved survival will be highly meaningful for patients and their families. In addition, from the regulatory perspective, the use of OS satisfies the requirement for an objective and easily calculable outcome measure (simply subtract the date of death from the date of study entry). Finally, considering the current reality of rapidly increasing costs associated with new anticancer pharmaceutical agents,² it is reasonable that those responsible for paying for this care (employers, employees, taxpayers, etc) mandate well-defined and clearly understood parameters to document the use of approved drugs.

Therefore, it would not be surprising that individuals (often from the nonclinical academic community) and interested groups have expended considerable effort to push back against suggestions that regulatory agencies (including the FDA) approve the noninvestigative use of anticancer agents based on alternative measures of clinical benefit, despite continuing to be labeled as surrogates. In this regard, one might consider a recent American Association for Cancer Research workshop held with the FDA and the American Statistical Association (“Overall Survival in Oncology Clinical Trials” in July 2023 in Bethesda, Maryland) as an example of this type of strategy to encourage a greater focus on OS as a regulatory agency–mandated study end point.

So what is the concern here with OS as a primary study efficacy outcome, perhaps even the single relevant end point? In the opinion of this commentator, the response to this question is both simple and complex. The simple answer is that advanced cancers, where the focus of documenting improved survival is most relevant today, are increasingly becoming chronic illnesses, where ultimate survival may be measured in years rather than in months (in which they were measured in the not-so-distant past), after a research participant completes therapy in a given clinical trial. Following a participant’s removal from the study, whether due to disease progression or toxicity (or simply deciding to withdraw for personal reasons), an increasing number of potentially beneficial therapeutic options are available in multiple settings that will have the realistic opportunity to favorably affect the individual patient’s clinical course and survival.

The more complex answer to this question is that it is difficult to precisely define the impact of specific therapeutic interventions on the survival outcome for individual patients, except to state they may result in a highly meaningful combined effect over a prolonged period. To emphasize the impact of therapy on ultimate survival following completion of trial-based management, a landmark modeling study was conducted and reported a decade ago by 2 researchers from The University of Texas MD Anderson Cancer Center in Houston.³ Recognizing the theoretical nature of this effort and the assumptions required the results of the analysis should be sobering for those who would mandate trial-associated improvement in OS to declare success of an investigative study arm.

In one poignant example, the authors explored the impact of hypothesized posttrial survival on the ability to demonstrate a statistically significant improvement in OS if a population of 280 patients revealed a 3-month improvement in PFS.³ For an anticipated median posttrial survival of only 2 months, a total of 350 patients would be necessary to achieve this end point. However, for a population median posttrial survival of 24 months, which has been increasingly observed in many solid tumors with conventional noninvestigative therapies in multiple clinical settings, 2440 patients would be required to be participants in the study, surely a highly questionable if not completely unrealistic clinical research strategy.

To be clear, there is no intent in this discussion to suggest that it will be impossible for a new anticancer therapeutic agent or combination drug regimen to demonstrate improved OS in a well-designed and well-conducted randomized trial. Rather, the goal here is to emphasize that the inability to document improved OS with the increasingly effective therapeutic options after a patient has finished therapy in a therapeutic trial should not be considered an indication of the absence of such benefit. Although objectively measuring time to disease progression only provides a limited window into the use of a strategy, which must be considered along with toxicity, impact on quality of life, and cost, employing PFS as the primary efficacy end point will permit investigators and regulatory agencies to isolate and directly evaluate the effect of the individual study regimens.

References

1. Chen EY, Joshi SK, Tran A, Prasad V. Estimation of study time reduction using surrogate end points rather than overall survival in oncology clinical trials. JAMA Intern Med. 2019;179(5):642-647. doi:10.1001/ jamainternmed.2018.8351
2. Ostrer IR, Gross CP. Paying for cancer drugs- getting what we are (prohibited from) bargaining for. JAMA Intern Med. 2022;182(12):1245. doi:10.1001/jamainternmed.2022.4921
3. Broglio KR, Berry DA. Detecting an overall survival benefit that is derived from progression-free survival. J Natl Cancer Inst. 2009;101(23):1642-1649. doi:10.1093/jnci/djp369