Maurie Markman, MD
This commentary poses the following question: Is it time to end or at least vigorously challenge the continued rhetoric and associated claims that seek to deny the clinical benefit of regulatory agency–approved antineoplastic pharmaceutical agents due to the absence of a documented statistically significant improvement in overall survival (OS) in a randomized clinical trial?1-3
Before proceeding, it is important to be clear that academics (and others) can certainly elect to analyze any issue they feel appropriate surrounding the regulatory review process, including multiple components of the conduct of clinical studies as well as the statistical analysis employed to make judgments related to approval for commercial sale. This includes specific decisions regarding drug labeling. Further, they have every right (as do all individual citizens or organizations) to criticize or support the final determinations of the FDA.
Finally, there is no intent in the following discussion to defend in any manner dysfunctional pharmaceutical/ biotech industry antineoplastic drug pricing policies, maneuvers to extend exclusive patents, or reported efforts to delay generic or biosimilar marketplace competition when legally permitted. In fact, while various reasons may be suggested as to why high-impact medical journals persist in the publication of manuscripts that fundamentally reject oncology drug regulatory decisions based on so-called surrogate end points (such as progression-free survival [PFS]), it is the opinion of this commentator that perhaps interest in this topic has more to do with the continued staggering rise in costs associated with newly approved drugs rather than the specific end point employed by the FDA in its review process.
What follows are arguments that specifically deal with the question of antineoplastic drug clinical benefit, again recognizing the increasingly intense societal debate regarding the cost associated with these agents.
Point 1
Although it is appropriately recognized that regulatory agencies require clear definitions, objective metrics, and consistency when making decisions, there is a fundamental question within the oncology sphere as to who ultimately should decide and what process should be employed to determine clinical benefit.
As currently suggested in regulatory guidelines, there is no reason to challenge the opinion that patients would agree that both improving survival and enhancing quality of life (QOL) are meaningful goals of antineoplastic therapeutics. However, what is unclear is why some have concluded that prolonging the time to ultimate disease progression following the delivery of an antineoplastic agent in a therapeutic trial is not evidence of clinical benefit. Stated more directly: Should it be the regulators or academic pundits who decide if improved PFS is a relevant indication for treatment with a particular agent, or should it be patients themselves who make this determination?
Point 2
If one asks the question of objective metrics to permit regulatory approval, OS is surely simple to measure, but what about QOL? Who should define this complex parameter for patients? Is there a single global metric agreeable to all that will declare QOL to be better, worse, or unchanged? Surely, the answer to this question is no. And to emphasize this concern, is it relevant to inquire just how many antineoplastic agents have been approved to date for commercial sale based solely on demonstrating an improvement in QOL in an oncology randomized trial?
Point 3
Returning to the question of OS as a primary study end point, there will certainly be no ambiguity regarding the objectivity of the data (date of study entry to date of death). However, as has been previously noted by this commentator, as cancer becomes a more chronic illness with patients with advanced cancer living for extended periods of time (increasingly measured in years, rather than months) following the finish of a specific therapeutic trial, is it rational to routinely consider the multiple potential life-events as being a result of the agent(s) administered in the completed study?
Unfortunately, the reality of cancer care today is that the numerous compounding factors that might influence a patient’s ultimate survival may seriously interfere with the OS metric as a meaningful approach to evaluating the clinical utility of trial-based antineoplastic therapeutics. For example, randomized studies may permit crossover from the control arm to the experimental regimen at the time of disease progression or the investigated drug (or a comparably active agent) may be able to be administered when a patient goes off-study.
Further, with increasingly extended survival, patients may receive multiple agents, some producing a favorable effect on the disease process while others may have no influence, or possibly even a negative effect due to excessive toxicity. Based on the pattern of disease recurrence or progression, individual patients may receive palliative local radiation or surgical interventions that may significantly prolong their survival.
One must add that in the era of molecularly targeted therapeutics, an increasing percentage of patients may benefit from options based on a next-generation sequencing test that reveals unique findings (eg, < 5% of individuals with a particular tumor type), resulting in the delivery of a novel therapeutic and a favorable survival impact.
All the above factors may influence OS independently of any effect from the outcome of the strategy employed in the prior clinical trial.
Point 4
Into this mix, one must add the impact of common comorbidities (eg, cardiac, renal, hepatic, pulmonary, metabolic, neurologic) on a patient’s ultimate survival unrelated to study drug-related adverse effects, especially with the observed increasing time intervals from the completion of a clinical trial until an individual’s death.
Point 5
The rather profound alteration in the relationship between trial-measured PFS and OS was pointedly highlighted in an important publication that examined outcomes in metastatic malignant melanoma early in the targeted therapeutic era vs a time when such agents were more widely available as an option after a patient had completed a clinical trial. In the earlier era, at the time of disease progression when there were essentially no effective commercially available drugs (or relevant trial crossovers to active agents), the analysis revealed study-measured PFS was highly correlated with OS (correlation coefficient 0.96).4 However, at the later time point, the correlation was substantially reduced (correlation coefficient 0.55), indicating the impact of subsequent lines of active treatment (commercially available drugs or trial crossover).
Finally, it is critical to acknowledge there is no intent here to suggest that it is not possible for a new/novel antineoplastic agent to demonstrate a statistically significant improvement in OS in a well-designed, appropriately powered, randomized phase 3 trial. However, the failure to accomplish this aim in most oncology settings today unequivocally does not justify the conclusion that there is an absence of clinical benefit.
Maurie Markman, MD, is president of Medicine & Science at City of Hope Atlanta, Chicago, and Phoenix.
References
Liu ITT, Kesselheim AS, Scheffer Cliff ER. Clinical benefit and regulatory outcomes of cancer drugs receiving accelerated approval. JAMA. 2024;331(17):1471-1479. doi:10.1001/jama.2024.2396
Mittal A, Kim MS, Dunn S, Wright K, Gyawali B. Frequently asked questions on surrogate endpoints in oncology-opportunities, pitfalls, and the way forward. EClinicalMedicine. 2024;76(5):102824. doi:10.1016/j.eclinm.2024.102824
Prasad V, Kim C, Burotto M, Vandross A. The strength of association between surrogate end points and survival in oncology: a systematic review of trial-level meta-analyses. JAMA Intern Med. 2015;175(8):1389-1398. doi:10.1001/jamainternmed.2015.2829
Flaherty KT, Hennig M, Lee SJ, et al. Surrogate endpoints for overall survival in metastatic melanoma: a meta-analysis of randomised controlled trials. Lancet Oncol. 2014;15(3):297-304. doi:10.1016/S1470- 2045(14)70007-5