Maurie Markman, MD
All oncology trainees are appropriately taught the established hierarchy of evidence-based medicine. Randomized phase III trials are at the top, although some would strongly argue that a meta-analysis of multiple phase III randomized trials is an even more relevant indicator of the best level of “evidence.” But, of course, if only a single phase III trial is available, that will have to satisfy these academic purists.
These principles have clearly served the oncology community and patients with cancer everywhere quite well during the last half-century of development of the specialty. Unfortunately, for some nothing has changed in this arena over the past 50 years, or apparently should ever change.
The depth of this fundamental belief in randomized trials as essentially the only evidence that ultimately truly matters is highlighted in a rather striking commentary from an oncologist who discusses his worldview of the unquestionable and absolute superiority of such evidence compared with other approaches to evaluate clinical utility1
“When it comes to one’s medical world view, there is no neutral positon. Every clinician makes the choice daily. For most of us, our decisions are capricious, uneven, and arbitrary. For interventions we spurn, we proudly assert ‘there is no evidence.’ For others that we favor, we stress that ‘there are no negative studies.’ For other interventions with promising rational and negative empirical trials, we argue that null data are flawed. Philosophically, this isn’t sure footing.
I propose that the discussion of one’s medical worldview must be made explicit. Students should be taught about the two views. Researchers should formally disclose their camp. I favor evidence first because reasonable belief can easily be hijacked by special interests.”
While there is much to debate and strongly criticize in this rather stunning declaration, the point of this commentary is to note once again just how narrow and increasingly devoid of relevance such assertions have become in the oncology arena. The striking heterogeneity of the entity we call “cancer” has been documented in multiple published experiences, perhaps most notably by the massive Cancer Genome Atlas effort.2
The academic purist who argues that the only acceptable evidence to permit a novel antineoplastic to leave the realm of investigative medicine and become a component of routine oncologic care would presumably demand every possible therapeutic option considered for clinical use—including the massive number of individual molecular abnormalities—be tested in a randomized controlled trial and demonstrate a statistically significant improvement in the only parameter relevant to this group, the gold standard of overall survival.
That such a study is unlikely to be initiated or completed—or, if completed, the results may take a decade or longer to impact the lives of patients with cancer—appears to be of little interest to those who seem to be more concerned with “beliefs being hijacked by special interests” than having more effective treatment available to patients with malignant disease.
Value in Off-Label Prescribing
The FDA, which approves antineoplastic agents for commercial sale, generally relies on data from phase III randomized trials in its decision- making process. Although this agency does not regulate the practice of medicine, many would consider on-label use based on the data required to attain this status to be the highest standard for cancer care. So, in response to the question noted above, how often are the on-label pronouncements by the FDA the standard of care today in the management of malignant disease?
A recent report of an examination of the SEER-Medicare database involving more than 13,347 women with breast cancer who received 16,127 regimens provides provocative insight into the actual relevance of on-label versus off-label antineoplastic drug therapy.3
Only 25% of the regimens utilized were actually on-label, as determined by the FDA, with more than twice that percentage (64%) of selected strategies being supported by sufficient evidence to be included in the National Comprehensive Cancer Network guidelines.
One can only imagine that in less common malignancies or where smaller molecularly defined subsets are recognized that the selected therapeutic strategies would be less likely to follow on-label indications. Further, in these less common settings, therapy would be based on published clinical trial data that fall far short of the declared required evidence enshrined by academic purists.
But shouldn’t the goal of oncologists be to do what is best for their patients, regardless of the currently existing level of evidence and particularly when any timeline for obtaining that “gold standard” evidence will be irrelevant for the patient being cared for today?