When new drugs are approved, FDA limits their on-label use to the treatment of specific diseases. A key question is whether the definition of a disease should be narrower or broader. Make the disease definition too broad, then there may not be strong evidence from the clinical trial to support its use among these types of patients; make the disease definition too narrow, however, patients who could benefit from the treatment may not be able to receive it.
One common way that FDA extrapolates the trial results is that not all of the clinical trial inclusion criteria are applied to the FDA label. For instance, if a clinical trial is undertaken in oncology, the trial may exclude patients with serious comorbidities (e.g., heart failure, COPD). The reason they are excluded is that patients may experience poor health outcomes related to these other diseases. For instance, a person could die from heart failure before one finds out if a cancer drug worked. In order to power the analysis with fewer people, oftentimes researchers will exclude patients with serious comorbidities from the clinical trial. The FDA, however, rarely explicitly excludes patients who have these comorbidities from the FDA-approved indication langauge.
On the other hand, perhaps a trial examines patients with a more severe form of the primary disease of interest. Or maybe a trial only examines patients who have failed prior therapy (i.e., second line patients). Should the FDA assume the treatment also works for patients with mild disease or first line patients?
To determine how often FDA does extrapolate clinical trial results to broader disease indications, a paper by Feldman et al. (2022) examines data from new molecular entity (NME) and biologic drug approvals (BLAs) after pivotal trials approved between 2015 and 2017. The authors compare the clinical trial patient characteristics against the FDA-approved indications across three domains: disease severity, subtypes of disease treated, and concomitant medication use.
Using this approach, the authors find that:
Among the 105 novel FDA drug approvals studied, 23 extrapolations of trial population characteristics to the approved indication were identified in 21 drugs (20%)…Extrapolation of trial findings to patients with greater disease severity was most common (n = 14 drugs), followed by differences in disease subtype (n = 6) and concomitant medication use (n = 3).
It would be of interest to examine real-world data to see if patient outcomes improved among those in “extrapolated” populations after the broader drug approvals. A potential research project for one of my readers!