This article is more than 5 years old. Information may no longer be current.
Are subset analyses an effective way to overcome the lack of data from minorities in clinical trials?
Click here to read the Cover Story, “Effort to recruit more minorities for clinical trials becoming ‘national priority.’”
Yes.
Black individuals experience disparities in health care, such as limited inclusion in clinical research trials. This inadequate enrollment can limit applicability and accuracy of clinical trials, especially for illnesses that disproportionately impact the black community and other ethnic or racial minorities.
Black individuals, for example, have the greatest incidence of developing and dying of the four most common malignancies: lung, breast, colon and prostate cancers. Multiple factors contribute to these disparities, including genetics, lack of access to care, delayed presentation of disease, social determinants of health and lack of participation in research trials by minorities.
Increasingly, it is recognized that black patients metabolize and respond differently to various medications than white patients. For example, black individuals commonly respond to different classes of antihypertensive medications than white individuals.
The medication hydralazine hydrochloride/isosorbide dinitrate (BiDil, Arbor Pharmaceuticals) — used to treat congestive heart failure in black patients — represents the first medication approved by the FDA to be utilized only in self-identified black individuals. Subset analysis showed that BiDil reduced mortality rates in black patients by 43%, and showed no benefit to white patients. In kidney transplantation, it is standard practice that black patients require higher dosages of immunosuppressive medications to prevent rejection than white patients.
Without subset analysis, these differential beneficial effects of certain medications in black patients might never have been identified. By extension, subset analysis of research data in cancer chemotherapy likely will find that black patients respond better to different anticancer medications than white patients.
Charles S. Modlin, MD, MBA, is a kidney transplant surgeon and urologist at Cleveland Clinic, as well as founder and director of Cleveland Clinic’s Minority Men’s Health Center. He can be reached at modlinc@ccf.org. Disclosure: Modlin reports no relevant financial disclosures.
No.
A subset analysis requires asking if the treatment effect in the minority part of the population is different from the overall treatment effect or the treatment effect within its complement. That is dangerous — however, essential — in science and medicine.
Subset analyses are the bane of drug developers who look for subpopulations — sometimes including racial minorities — who benefit from their drug. Empirical evidence shows that subset analyses are almost always wrong, and for well-understood statistical reasons.
A similar example is looking for -omic biomarkers that predict what types of patients will respond to a particular therapy. How many press releases and newspaper articles have we read with titles like, “A 5-gene profile predicts the benefit of ‘X’ drug in prostate cancer”? Historically, most of them are wrong. I stopped reading them long ago.
On the other hand, we have to understand who benefits from our therapies. Not looking is not an answer. Finding the right filters for our looking glasses is the answer, but it is a really tough balance.
Donald A. Berry, PhD, is professor in the department of biostatistics at The University of Texas MD Anderson Cancer Center. He can be reached at don@berryconsultants.com. Disclosure: Berry is co-owner of Berry Consultants LLC.