Maurie Markman, MD
Editor-in-Chief of OncologyLive
Senior vice president for Clinical Affairs and National Director for Medical Oncology
Cancer Treatment Centers of America, Eastern Regional Medical Center
Peer-reviewed publications increasingly feature quite provocative reports of nonrandomized retrospective analyses examining drug effects that have the realistic potential to be of substantial relevance in the management of cancer.
But the issue here is not the frequently discussed—and critically important—question of “off-label” delivery of antineoplastic agents that have been shown in phase II efficacy studies to produce a meaningful degree of clinical activity outside the label indication that a drug regulatory agency has approved.
Rather, the focus of this commentary is the far more controversial argument about whether oncologists should consider the noninvestigational delivery of certain pharmaceuticals for the specific purpose of improving survival outcomes in settings where such theorized benefits have never previously been documented in a prospective clinical trial.
Consider, for example, the multiple nonrandomized retrospective reports suggesting that patients with gynecologic cancers who were taking several classes of pharmaceutical agents such as metformin and beta-blockers for reasons not related to their malignancy experienced superior overall survival compared with women included in the individual studies who did not receive the particular drug.1,2
Of course, it is not difficult, and in fact it is quite easy to level substantial criticism about such noncontrolled experiences, or to simply ignore these reports, since any conclusions are likely based on nothing more than a retrospective review of the medical or even billing records of highly heterogeneous groups.
Yet these studies involve quite real-world patient populations and mounting evidence suggests they should not be ignored. Instead, oncologists should consider a framework for evaluating whether to recommend a medication to a patient based on such reports (Table
Key Questions to Consider
The studies of beta-blockers and metformin again provide an important example. Multiple retrospective analyses conducted by investigators in different cancer settings and published in the peer-reviewed medical literature (ie, beta-blockers1,3-5
) appear to have reached the same favorable conclusions regarding the impact on survival associated with the use of these individual medications.
There are key questions to consider before dismissing such reports: What if a highly rational biological explanation for such observations can be advanced? What if the specific medications in question are widely utilized in routine medical practice, have a well-established acceptable safety profile, and carry a generic drug cost that would be very modest (especially when compared with commercially available nongeneric antineoplastic agents)?
Finally, what if it is highly unlikely that “evidence-based data” regarding the objective effectiveness of the drug for routine use in clinical practice as a strategy to improve cancer-related survival will ever be generated for a given clinical setting?
For example, what are the realistic chances investigators would conduct a randomized phase III study addressing the survival impact associated with the administration of either metformin or a beta-blocking agent specifically
in women with stage III poorly differentiated papillary serous epithelial ovarian cancer who have attained a clinical partial response to a regimen of carboplatin/paclitaxel/bevacizumab (followed by maintenance bevacizumab) with normalization of radiographic imaging studies and disappearance of all cancer-related symptoms, but with a persistent slight elevation of the serum CA-125 above the “normal range”? Or, what if we modify the scenario a little and state that the patient has undergone an identical treatment plan and experienced a similar clinical course, but the histology is a clear cell cancer?
Table. Key Considerations Before Recommending Nontraditional Agents
If a definitive evidence-based phase III randomized trial is initiated, what is the likelihood of its completion and how long would it take for the results to be available?
What would be the mandated versus clinically relevant study outcomes (eg, overall survival, progression-free survival, time to requirement for subsequent therapy>?
Might a randomized trail design be acceptable with a less stringent requirement fora positive outcome (eg, P < 0.1 or P < 0.15), assuming the agent was associated with a highly acceptable side effect profile?
If a phase III trial is "positive" or "negative" in one tumor type, how should these results be interpreted regarding the potential utility of the agent in (a) other tumor types; (b) other clinical settings in the same tumor type (eg, primary, second-line, maintenance therapy)?
How should the dose, schedule, and treatment duration be determined considering the fact the agent has been widely employed outside the oncology arena?
What would be required for the agent to be approved for an oncology use by the FDA, paid for by third-party carriers, and included in cancer management guidelines?
Would all forms of a generic medicine be considered equivalent for use in the oncology domain, if a specific medication had been employed in an evidence-based phase III clinical trial?
If a positive evidence-based phase III clinical trial included few elderly patients or excluded individuals with relevant comorbid conditions (both quite common events in the oncology clinical trials arena), how should clinicians interpret the data when considering efficacy and toxicity outside the confines of the clinical trial?
Simply stated, what is the likelihood that any organization today would be willing and able to financially support the conduct of such a study in any given patient population? We must acknowledge the multiple relevant clinical settings where the delivery of such nonestablished pharmaceuticals might be considered as a potentially rational therapeutic, the large number of competing scientific agendas in a very crowded field looking at the same patient populations to document the clinical utility of novel antineoplastic strategies, the absence of industry sponsorship to define the benefits of any low-cost generic drug, and the complexity of a trial involving hundreds of patients needing to be followed for many years to define a survival endpoint.