Where Overall Survival Falls Short as a Gauge

Maurie Markman, MD

From the perspective of a patient and that patient’s family, it is completely understandable that the single most important goal of an antineoplastic strategy is to prolong survival and, if possible, produce a cure. Further, it is reasonable to argue that this basic philosophy of the aim of treatment has been quite nicely translated into the regulatory environment of drug development, where overall survival (OS) has traditionally been considered the definitive as well as gold standard outcome and endpoint in cancer therapeutic clinical trials.

An additional advantage of using this strategy is its ease and objectivity. The date of beginning a particular regimen and the date of death can appropriately be defined without time-wasting and senseless debate. In addition, if an audit of these metrics is necessary, it should be easy to do.

A final highly relevant historical argument in support of OS as the gold standard in cancer trials is based on the limited clinical utility of most available oncology therapeutics when the current drug regulatory paradigms were established. In fact, in the early days of the modern chemotherapeutic era, antineoplastic drug therapy was, with a few notable exceptions, marginally effective and objectively associated with only modest short-term survival benefits. For patients with advanced, metastatic, or recurrent cancers following the failure of local-regional therapy, survival was most often measured in months, if not days, and very rarely in years.

Further, considering the toxicity (eg, severe emesis, neutropenia) of most single cytotoxic agents and combination regimens examined and employed in this era, it would be rational for those conducting the trials (industry, academic centers) and the governmental regulators considering approval of an agent for commercial sale to require that the therapy demonstrate an ability to improve OS in a particular setting (eg, first-line treatment of epithelial ovarian cancer) before accepting it as a standard-of-care therapeutic option. In fact, the concept of clinically effective second-line therapy that might favorably affect subsequent OS following a patient’s completion of a given clinical trial would have been considered pure fantasy in the vast majority of circumstances.

The modern world of cancer therapeutics bears strikingly limited resemblance to the era described above—despite some quite distressing exceptions, such as treatment options for advanced/metastatic pancreatic cancer.

Higher Inputs for Reliable Data

Biologically and clinically active antineoplastic drug regimens employed as a component of initial therapy (adjuvant or neoadjuvant strategies, metastatic disease) or as second-line or later approaches in disease management have favorably affected both quantity and quality of life. In an increasing number of settings, patients with an advanced or metastatic cancer, while still unable to be cured, are objectively able to lead a high-quality life for a number of years rather than months or even less time. As a result, it is not unreasonable to label such long-term conditions as very serious but chronic. But what does this discussion have to do with endpoints in cancer clinical trials and, specifically, in epithelial ovarian cancer? There are 2 quite powerful arguments that can be provided in response to this highly relevant question.The first argument is related to how subsequent treatments on posttrial outcomes and survival have affected the way clinical trials are conducted in the current oncology arena and will be conducted in the future. Perhaps the most definitive argument against a disturbingly senseless belief that OS must always remain the gold standard in cancer clinical trials comes from a landmark analysis by a leading group of cancer biostatisticians.

They calculated the sample size required for a particular study to reveal a statistically significant improvement in OS.¹ In their hypothetical scenario, in which a patient population of 280 individuals would be required to show a median 3-month improvement in progression-free survival (PFS), the total population needed to demonstrate a difference in OS for a median post-progression survival of 2 versus 24 months would be 350 versus 2440 patients, respectively. Stated differently, if patients were to live for a median of 2 years following the completion of a trial, versus a median of only 2 months following completion, 7 times as many patients would have to be enrolled in the study for it to reliably demonstrate an improvement in ovarian survival.

This stunning difference in mandated sample size is clearly related to the multiple other therapeutics likely to be employed in individual patients over this extended time period, plus death due to other events, including common comorbidities in the patient populations (eg, cardiac disease). And it is critical to acknowledge that survival in the range of 24 months after first-line therapy for ovarian cancer is an increasingly realistic scenario.

In the case of ovarian cancer, a most poignant example of the significant impact of posttrial therapy on OS associated with advances in routine standard of care and how this factor may dramatically influence outcomes is provided by 2 phase III randomized trials that utilized the combination of carboplatin plus gemcitabine in the management of recurrent platinumsensitive disease^.2,3

These studies, which included the rigorous monitoring and follow-up required for drug regulatory submission, employed identical eligibility criteria and carboplatin-plus-gemcitabine regimens. However, the individual studies were conducted 8 years apart. Although there are dangers associated with any attempt at cross-trial comparisons, it is notable that the studies demonstrated an almost identical median PFS for the 2 individual carboplatin-plus-gemcitabine study arms. However, despite the same trial eligibility criteria and drug regimens (doses, schedules) and PFS results, there was a striking difference in the reported OS outcomes for the 2 groups of patients. For the individuals with platinum-sensitive recurrent ovarian cancer receiving carboplatin plus gemcitabine in the trial activated in 1999, the median OS was 18 months. In sharp contrast, for patients administered this trial-based combination chemotherapy in a study initiated in 2007, the median OS was 35.2 months, almost double this survival outcome.

Importance of Patient Preferences

Why did this occur? While it is never possible to provide a definitive answer to such a question, it is highly likely that study participants in the later time period were provided the opportunity to receive increasingly effective posttrial drug (and possibly other therapeutic) strategies that were simply not available during the earlier period. And with the continued documented advances in clinically beneficial approaches to ovarian cancer management (eg, antiangiogenic agents, poly [ADP-ribose] polymerase inhibitors, etc), one can reasonably speculate that such differences over a given 8-year time span would only be magnified today.The second major argument against OS as the primary endpoint is the observation that patients with ovarian cancer and their families are increasingly less interested in having a third party (eg, regulators) decide for them the relative benefits versus the risks of a given therapeutic strategy.

It is critical to emphasize here that this statement does not negate the absolutely essential role of first-class validated clinical research in defining the impact of a particular therapeutic approach on measurable and meaningful endpoints. However, in the opinion of this commentator and longtime investigator, the locus for the determination of what are “clinically meaningful” outcomes in ovarian cancer clinical trials, specifically as they relate to individual therapeutic decisions, must substantially shift from its current often top-down paradigm, which begins with vastly overly prescribed regulatory approval language (eg, “use only as a third-line or a later drug regimen”), to a partnership between the individual patient and that patient’s advisers and clinical team.

References

1. Broglio KR, Berry DA. Detecting an overall survival benefit that is derived from progression-free survival. J National Cancer Inst. 2009;101(23):1642-1649. www.ncbi.nlm.nih.gov/pmc/articles/ PMC4137232/. Accessed August 12, 2017
2. Pfisterer J, Plante M, Vergote I, et al. Gemcitabine plus carboplatin compared with carboplatin in patients with platinum-sensitive recurrent ovarian cancer: an intergroup trial of the AGO-OVAR, the NCIC CTG, and the EORTC GCG. J Clin Oncol. 2006;24(29):4699- 4707. doi: 10.1200/JCO.2006.06.0913.
3. Aghajanian C, Blank SV, Goff BA, et al. OCEANS: a randomized, double-blind, placebo-controlled phase iii trial of chemotherapy with or without bevacizumab in patients with platinum-sensitive recurrent epithelial ovarian, primary peritoneal, or fallopian tube cancer. J Clin Oncol. 2012;30(17):2039-2045. doi: 10.1200/JCO.2012.42.0505.