From the earliest moments of medical school instruction, through residency and fellowship training, physicians are taught the primacy of the randomized phase 3 trial in the hierarchy of evidence-based medicine.
The fundamental value of the randomized phase 3 trial in the development of credible medical care and public health policy needs no justification. For more than a half-century in the United States and other countries, ineffective and potentially harmful approaches for treating medical conditions have been removed or prevented from entering the clinician’s armamentarium. This happens after failing to achieve a prospectively defined measurable outcome in an appropriately designed and conducted randomized phase 3 trial.
The importance of this methodology has been particularly poignant in the establishment of a robust approach to the use of pharmaceutical agents, including in the oncology space. Today, it would be impossible to envision medical care in this country where the selection of therapeutics was not based on the conduct, completion, and subsequent publication of randomized phase 3 trial findings, built on studies involving comparisons of thousands of novel concepts over the preceding decades for the purpose of optimizing successful care for individual patients in terms of efficacy and toxicity. From the earliest moments of medical school instruction, through residency and fellowship training, physicians are taught the primacy of the randomized phase 3 trial in the hierarchy of evidence-based medicine.
Molecular Profiling Makes a Difference
However, with our rapidly expanding knowledge of the complexity and heterogeneity of basic molecular biology, it is increasingly clear that past assumptions regarding the homogeneity of various populations included within these studies has been in error, and sometimes, seriously so.
For example, 30 years ago when clinical trials in metastatic lung cancer were conducted, populations were simply divided into 2 relatively large subgroups: small cell and non–small cell. Today, it is understood that there are several highly clinically relevant subgroups of patients with advanced/metastatic lung cancer, some of which may represent a very small percentage (<3%-5%) of the entire population. And, although it may be reasonable to conduct a randomized phase 3 trial over an acceptable period of time that involves 20% to 30% of the large population of patients with non–small cell lung cancer, it is increasingly clear that approach is highly questionable if one is attempting to initiate and complete such a study that only involves 2% to 3% of these patients. How long would society and, most importantly, patients and their families with these molecularly defined cancer subgroups be willing to wait for an answer to the question of the clinical utility of a novel therapeutic approach?
It is quite easy to find multiple highly clinically relevant situations in the oncology domain where the dogma regarding the requirement for definitive randomized trials is increasingly problematic. Consider, for example, a recent report of a randomized phase 2 trial examining the novel targeted agent capivasertib, which is designed to target several potentially clinically relevant molecular abnormalities (PIK3CA/AKT1/PTEN) in patients with triple-negative breast cancer.1 Of note, approximately 10% to 15% of breast cancers are classified as triple negative. The study, conducted over 3 years at 42 academic centers in 6 countries, included 140 patients with previously untreated locally advanced or metastatic triple-negative breast cancer. Overall, the study revealed no statistically significant impact on progression-free survival (PFS), the primary trial end point (HR, 0.74; 2-sided P = .11). However, of the approximately 25% of patients in the study population with available samples who were found to have the hypothesized relevant molecular targets, the HR for PFS was 0.30 (P = .01), whereas there was no impact on PFS (HR, 1.13; P = .61) in the subgroup without these abnormalities. Similar results were observed for the impact of therapy on overall survival in the entire patient population and 2 relevant subpopulations.
So how does one move forward to confirm the clinical value of this observation in a molecularly defined group of women with breast cancer such that the FDA might approve this (or any other) novel antineoplastic agent? Should a definitive randomized phase 3 trial be required and, if so, how many years—or perhaps decades—would be necessary to complete such a study? This question is particularly relevant in this instance considering that the FDA already has approved alpelisib (Piqray) for patients with another subtype of PIK3CA-mutated breast cancer based on phase 3 trial findings.2
Authentic Study Populations Are Needed
Further, it is increasingly recognized that the population of individuals who participate in definitive phase 3 randomized trials that establish the standard of care in the cancer arena are strikingly unrepresentative of the real-world population encountered by oncologists. One recently published peer-reviewed article—an example of the many that could be cited—emphasizes this point.3 In an examination of the National Cancer Database from 2004 through 2015 for the proportion of individuals (total sample size, 12,097,681) receiving their initial course of therapy for cancer in the United States, investigators found that 0.1% (1 in 1000 patients) were enrolled in a clinical trial, and this population quite poorly represented the overall population with a malignancy in this country in such areas as age, comorbidities, and race.3
Unfortunately, this status quo for defining optimal therapy continues to be the gold standard demanded by a large sector of academics involved in statistical evaluations of medical care, as well as by many regulators.
And, disturbingly, the pushback against change has perhaps only intensified, despite increasingly overwhelming evidence of serious issues with this status quo.
How else does one explain the recent publication of a commentary in the New England Journal of Medicine that describes the process of randomization as “magic” and real-world evidence as “myth”?4 Fortunately, with the rapidly evolving revolution in our understanding of the fundamental molecular biology of health and disease and the introduction of increasingly beneficial health-related strategies based on that knowledge (eg, wellness, prevention, screening, treatment), we as a society are beginning to appropriately ignore such biased rhetoric.