Maurie Markman, MD
The major role played by phase III randomized trials in the development of evidence- based cancer medicine is appropriately well acknowledged. Although legitimate questions have been raised over the years regarding the generalizability of clinical study results in the nonresearch setting, including the unfortunate exclusion of elderly patients and individuals with highly relevant comorbidities (eg, pre-existing cardiac, pulmonary, hepatic, and renal disease), patients with cancer have benefited greatly from the appropriate conduct, completion, and ultimate reporting of such trials in the peer-reviewed medical literature.
The question now is whether that established model can carry cancer care forward in an era of fast-paced innovation and long-term survivorship. As cancer becomes a more chronic condition to be managed over extended periods of time, the fundamental limitations of trials examining the utility of individual drugs or combination strategies at a single point in the natural history of a given malignancy will become increasingly—and likely painfully—obvious. Restrictions on Patient Populations First, let us consider the limits of the current clinical trial protocols concerning patient populations. Of necessity, clinical studies examining new antineoplastic strategies require specific eligibility for patient entry. Justification for limitations regarding who may, or may not, be permitted to be treated on a given clinical trial include concerns regarding the toxicity profile of a particular strategy or evidence that a specific approach may only be clinically active in certain conditions such as a particular tumor type or the presence of a validated biomarker.
In general, early-phase studies (ie, phase I) may be more restrictive in the exclusion of potential comorbidities due to many unknowns regarding the agent’s adverse event profile and less restrictive regarding tumor type or number of prior lines of treatment, compared with later stages (phase II, III) in the drug developmental pathway. In fact, it is in the evidence-based studies employed to obtain regulatory approval where one generally encounters the most restrictive eligibility criteria. As previously noted, the reasons for a variety of entry barriers are fairly obvious. These include concerns for excessive toxicity due to the unique adverse event profile of the agent observed in prior trials using the drug and the presence of particular comorbidities that may potentially negatively impact the ability to define the favorable effects (toxicity, efficacy, or both) of the new product.
Of course, such restrictions, while likely permitting a more focused examination of the experimental agent in question, may quite seriously limit the clinical applicability of data relevant to patients treated in the real world where comorbidities are very common. This issue is especially relevant in the more elderly population most likely to develop the majority of malignant disease.
Restrictions on Prior Treatments
Although the substantial concerns over the value of evidence-based regulatory agency– approved trial data that exclude patients with common and clinically meaningful comorbid conditions are increasingly recognized, there is an additional potential impact of specific restrictions associated with the large majority of the study designs employed to gain permission for a drug to be sold commercially.
That is the impact of restricting clinical trial enrollment based upon the nature of the prior therapies the potential participants have received. Knowledge of the compiled evidence supporting potential indications for a new drug as well as the existing standard of care in particular clinical settings will dictate the specific patient populations where the evaluation of efficacy of a new antineoplastic agent or combination regimen should be undertaken. Further, there is a logical desire to restrict as much as possible the heterogeneity of the patient population entered into a given study in an effort to optimize the potential that a favorable signal specifically related to the drug’s effect will be observed, assuming it actually exists.