Maurie Markman, MD
Much has been discussed regarding the increasingly recognized limitations of phase III randomized trials in defining the future of cancer management and, indeed, there are many reasons to question whether there is genuine value in such efforts.
There’s the often mind-numbing time it takes to initiate, complete, and report the results of phase III studies—assuming, of course, they are ever completed and reported!—and the staggering costs of these complex initiatives. There’s also the questionable “real world” relevance of the results of many trials that either intentionally or unintentionally exclude the elderly or individuals with common comorbidities, and the dawn of the era of precision cancer medicine where targeted therapeutics impact smaller patient populations, making the conduct of randomized trials problematic.
In addition to these issues, several recent reports are providing even more fundamental, serious challenges to the basic statistical foundation of randomized trial study results.1-3
Although not drawn from the oncology arena, these examples highlight important considerations for cancer specialists seeking to translate clinical trial findings into practice.
A particular concern is the increasingly frequent reporting of P
values in the absence of even more critical statistical data upon which to evaluate the relevance of the findings such as confidence intervals and effect sizes.2
One commentary noted: “For example, it would be unreasonable to decide that a new cancer medication was ineffective because the calculated P
value from a phase II trial was .051 and the predetermined level of statistical significance was considered less than .05.”3
Yet how often does such an absurd situation occur when academic purists report the results of a randomized trial?
Another editorial pointedly acknowledged that “a P
value cannot indicate the importance of a finding; for instance, a drug can have a statistically significant effect on patients’ blood glucose levels without having a therapeutic effect.”1
Impact on Practice
More important, however, is another serious limitation of phase III randomized trial results in the oncology arena: how much do they actually influence clinical practice?
First, it is critical to separate out randomized trials that permit a new antineoplastic agent to enter the market, or add a new clinical indication. Oncologists appropriately eagerly await results that will permit them to provide additional strategies to their patients and, if a commercial company is involved, that entity is highly likely to market the study results. But there is nothing inherent in this discussion that requires a randomized trial. If a new drug entered the market or received on-label FDA approval based on phase II study data, would the agent be less likely to be employed?
Thus, the issue to be addressed here is the question of the impact of phase III trial results that do not promote a new product but rather challenge a particular management paradigm.
Two examples of such studies from the gynecologic cancer literature demonstrate just how limited the impact of such randomized trial results can be where preexisting beliefs, training, economic interests, or well-established practices conflict with what the “evidence” demonstrates. Despite the results of 3 phase III randomized trials that revealed a survival benefit associated with primary platinum-based chemotherapy for ovarian cancer, meta-analyses demonstrating the superior utility of this approach compared with intravenous delivery of cisplatin in this setting, and an NCI-Clinical Alert, it remains the case that only a minority of patients who are potentially candidates for this therapeutic strategy are actually treated with regional drug delivery.4,5
This is apparently the situation even at major cancer centers who often contend they are the champions of “evidence-based medicine.”
Despite the results of 2 phase III randomized trials demonstrating that neoadjuvant chemotherapy in advanced ovarian cancer is equivalent in progression-free and overall survival and is associated with reduced surgical morbidity and mortality compared with primary surgical cytoreduction6,7
, the use of this approach remains limited in the United States and reports of nonrandomized single institutional experiences continue to claim the superiority of primary surgery.8,9
Now, it is not my intent here to advocate that the use of either intraperitoneal therapy or neoadjuvant therapy should be mandated or to suggest that there is anything wrong with challenging the results of phase III randomized trials.