Placing the results of a single trial in the context of real-world, everyday practice is increasingly difficult because of the number of available options and the absence of studies that directly compare individual strategies.
Maurie Markman, MD
The importance of randomized trials in oncology in the development of solid evidence-based care is well recognized. Data from such studies presented at major oncology meetings and published in high-impact medical journals can be practice changing. Randomized trials may reveal superior efficacy, such as improved progression-free or overall survival, or the study end point may involve noninferior outcomes with reduced clinically relevant toxicities. Of course, it is always hoped that practice-changing trial data will reveal improvement both in efficacy and in treatment-associated adverse effects.
In an era that produced a limited number of meaningful antineoplastic agents or novel therapeutic concepts pending evaluation in a given clinical setting, it was reasonable to suggest that 1 or several randomized studies could be initiated, conducted, and completed, with the results analyzed and reported, to potentially define a new standard of care. Further, the results of those few, often related trials, essentially identical regarding the specific clinical question addressed and the study arms involved, could be employed to critically inform the control and experimental study arms of subsequent randomized efforts.
This simple scenario nicely characterizes the development of frontline (“primary”) cytotoxic chemotherapy for advanced ovarian cancer for almost 50 years, beginning approximately in the 1960s. Single-agent nonplatinum drugs were replaced by combination nonplatinum therapy and then by nontaxane cisplatin-based therapy, followed by nontaxane carboplatin-based therapy, which was followed by cisplatin plus a taxane (replacing cyclophosphamide); finally, cisplatin was replaced by carboplatin when combined with a taxane.1 In fact, a platinum agent (cisplatin or carboplatin) plus paclitaxel has been the standard- of-care primary chemotherapy approach to managing advanced ovarian cancer for more than 20 years and remains the backbone cytotoxic regimen for the illness.
Importantly, the changes outlined above resulted from the conduct of well-designed randomized trials published in peer-reviewed literature over many years and, in general, with 1 generation of trials addressing a single or closely related question, logically followed by the next.
Rapid Changes Add Complexity
Times have changed rather strikingly in cancer medicine. The virtual revolution in our understanding of the molecular biology of malignant disease, as well as efforts of pharmaceutical/biotech companies, governmental agencies, and clinical investigators, has raised remarkably the number of antineoplastic agents introduced into clinical practice over the past decade, along with investigative drugs likely to soon join the oncologist’s standard-ofcare armamentarium.
It is difficult to overstate the transformation in the speed of introducing clearly beneficial anticancer agents and the resulting complexity of interpreting data from individual, well-conducted randomized trials in the realworld practice of clinical cancer medicine.
For example, in ovarian cancer, 4 antineoplastic agents are approved in the United States specifically for maintenance therapy following a second response to a platinum-based cytotoxic chemotherapy regimen: bevacizumab (Avastin), niraparib (Zejula), olaparib (Lynparza), and rucaparib (Rubraca).2 Each approval was based on the results of 1 or more phase III randomized trials that compared the active drug with a placebo control, an appropriate regulatory decision at the time based on the absence of a known safe and effective maintenance regimen in this clinical setting.
Unfortunately, as a result of the meaningful data emanating from these important, solidly evidence-based clinical trials, today clinicians appropriately have multiple relevant questions regarding the optimal care of patients with ovarian cancer who achieve evidence of a secondline or later response to a platinum-based chemotherapy regimen, a situation that is far from uncommon.
For example, consider this series of questions: (1) If a patient with a known germline BRCA mutation achieves a response to a second-line platinum regimen that contains bevacizumab, should the bevacizumab be discontinued and substituted with a PARP inhibitor? (2) In this specific setting, should bevacizumab be administered in combination with a PARP inhibitor, or should the drugs be given individually but sequentially? (3) Should patients who are wild-type BRCA or negative for having homologous recombination deficiency be treated preferentially with bevacizumab rather than with a PARP inhibitor, or should the agents be delivered in combination or in sequence? (4) Is it preferable to specifically employ a PARP inhibitor or bevacizumab at the earliest point in time in the natural history (that is, line of therapy) of ovarian cancer in an individual patient, and does the answer change based on the presence or absence of a BRCA mutation? (5) If a patient experiences a prolonged stable disease state (>1 year, for example) with a maintenance regimen, is it reasonable to discontinue therapy and restart with evidence of disease progression, such as a rising serum cancer antigen 125 level?
Given that multiple clinically active PARP inhibitor agents are available for the management of ovarian cancer, obvious additional questions arise: If a patient experiences excessive toxicity to 1 of the currently approved PARP agents, is it preferable to continue lowering the dose or try an approved alternative drug? If a patient progresses on 1 PARP inhibitor, is it reasonable to try a second or even third (so-called PARP after PARP)?
The intent of this discussion is not to challenge the design of previously conducted randomized trials or suggest that regulatory approval for any of those agents was an error. Rather, the goal is to highlight the clinically relevant problem of success in oncology, where attempting to place the results of a single trial in the context of real-world, everyday practice is increasingly difficult because of the number of available options and the absence of studies that directly compare individual strategies.