There has been much recent discussion regarding the appropriate and acceptable endpoints in evidence-based, randomized trials in ovarian cancer. Prior experience with primary therapy for advanced disease demonstrated the utility of progression-free survival (PFS) in predicting a subsequently documented, statistically significant improvement in overall survival.1-3 However, there has been legitimate concern that, with the availability of an ever-increasing number of biologically active agents in the malignancy, an impact on overall survival may not be observed if the rational use of these drugs after progression on a particular trial affects the ultimate outcome independent of the influence of the study regimen being tested.4
Consequently, many have argued that, although an improvement in overall survival is a most worthy goal of phase 3 trials of new antineoplastic strategies in ovarian cancer, it is acceptable to consider a favorable impact on PFS as a legitimate primary study endpoint. Both the clinical investigative community and regulatory agencies continue to vigorously debate the relative merits and utility of study outcomes, including how best to objectively evaluate/measure the precise time of disease progression.
However, in the opinion of this commentator, a crucial element has been lost in this important research-related discussion and in frequent literature descriptions and interpretations of the clinical relevance of observed changes in measurable tumor masses and markers (eg, cancer antigen 125 [CA 125]) among the women who participate in such trials. In this debate, there has been an apparent assumption by some (or perhaps many) that, if an individual patient meets a study-defined endpoint of progression (eg, an increase in the size of an existing lesion, or the appearance of a new lesion, or a confirmed doubling of the CA 125 antigen level), then this also implies that the current management plan should be changed. It is this very specific “assumption” that must be challenged.
The appropriate conduct of clinical trials requires prospectively delineated criteria to objectively evaluate or measure study endpoints. This statement certainly applies to comparisons of PFS, a common and increasingly used endpoint in ovarian cancer randomized trials. However, in the absence of either newly developed or the worsening of existing malignancy-associated symptoms in a patient with ovarian cancer, one must seriously question the conclusion that treatment must be changed solely because a single mass has increased in size by 25%, a new 1-cm nodule is identified on an imaging study, or the serum CA 125 has doubled from a value of 35 μg/mL to 72 μg/mL. In the opinion of this commentator, justification for such a statement would require scientifically valid evidence that altering therapy based only on a study-defined criterion of “progression” will favorably impact a relevant clinical outcome, such as overall survival or the time to subsequent development cancer-related symptoms.
Recently reported, evidence-based data in a related (but certainly not identical) setting have challenged the notion that the initiation of second-line chemotherapy (rather than changing “current therapy”) based solely on a documented rise in the serum CA 125 level influences survival compared with continued observation for other clinical evidence of disease progression (eg, development of new cancer-related symptoms).5 Yes, the disease has “progressed” (based on objectively reliable CA 125 criteria), but the results of that trial indicated that altering management at this point in time based solely on this measure of “progression” may not favorably impact outcome.
It is certainly appropriate to declare that objective data indicating biologic evidence of “progression” of the disease process can be used in the setting of a randomized trial to indicate a valid element of superiority for 1 particular antineoplastic regimen compared with another.6, 7 However, there truly is nothing inherent in this specific information that either indicates in a particular individual (or in this situation, “study participant”) when cancer-associated symptoms will develop or how long the patient will survive.
From the time of this initial documentation of “progression,” the unique biology of the cancer within an individual patient (its “natural history”) may result in relatively rapid tumor growth, relatively slow tumor growth, or something in between. Reported experiences with observed heterogeneity in the course of changes in the serum CA 125 level after documented, sustained elevations of this important tumor marker serve to emphasize this point.8-10
Of critical relevance to the fundamental theme of this discussion is that, although variability in the subsequent natural history of ovarian cancer does not negate the validity of “progression” for comparing strategies in a phase 3 randomized trial, this single, well defined study endpoint simply cannot be used in the absence of other relevant, patient-specific clinical data with an appreciation of the utility of currently available, setting-specific treatment strategies to accurately determine the future course of the illness, nor can it be used reliably to optimize subsequent management decisions.
In fact, it is highly biologically plausible that, despite the documentation of protocol-defined “progression,” the large majority of the existing, viable malignant cells remain under reasonable control at the clinical level (eg, no symptoms, no change in the size of any other measurable masses). Assuming that the treatment program continues to be tolerated well with minimal or no impact on quality of life or activities of daily living, and in the absence of evidence that switching to a different antineoplastic drug will improve survival in this specific setting, why is a change required?
Yes, the patient has “progressed on treatment” and will be scored as such in the formal analysis of the trial, but this knowledge does not necessary mean that continued treatment lacks clinical utility. Furthermore, although switching to a new agent may have a positive impact, it also is possible realistically that a smaller fraction of the malignant cell population will be responsive biologically to this drug (ie, will exhibit tumor shrinkage or result in a prolonged state of “stable disease”). In addition, it would be inappropriate simply to assume that tolerability to the new program will be as favorable as tolerability to the current regimen.
Therefore, based on these considerations, it is rational to conclude that the decision to alter treatment (unless mandated within a particular research study design) should be made based on the totality of evidence that the disease is worsening, on patient tolerability of the therapy, and on a determination that the current regimen is providing no benefit or unacceptably limited benefit to the specific individual. In this clinical determination, knowledge of an objective measure of “progression” should play an important, but not a singular, role. Finally, it is rational to argue that, for some individuals in specific settings, a reasonable alternative strategy to either continuing existing treatment or changing therapy would be the discontinuation of active antineoplastic drug delivery until there is evidence of symptomatic disease progression.