The Apportionment of Pharmacogenomic Variation: Race, Ethnicity, and Adverse Drug Reactions

Abstract

Fifty years ago, Richard Lewontin found that the vast majority of human genetic variation falls within (~85%) rather than between (~15%) racial groups. This result has been replicated numerous times since and is widely taken to support the notion that genetic differences between racial groups are trivial and thus irrelevant for clinical decision-making. The aim of this study was to consider how the apportionment of pharmacogenomic variation within and between racial and ethnic groups relates to risk disparities for adverse drug reactions. We confirmed that the majority of pharmacogenomic variation falls within (97.3%) rather than between (2.78%) the three largest racial and ethnic groups in the United States: Black, Hispanic, and White. Nevertheless, pharmacogenomic variants showing far greater within than between-group variation can have high predictive value for adverse drug reactions, particularly for minority racial and ethnic groups. We predicted excess adverse drug reactions for minority Black and Hispanic groups, compared to the majority White group, and considered these results in light of the apportionment of genetic variation within and between groups. For 85% within and 15% between group variation, there are 700 excess adverse drug reactions per 1,000 patients predicted for a recessive effect model and 300 for a dominant model. We found high numbers of predicted Black and Hispanic excess adverse drug reactions for widely prescribed platinum chemotherapy compounds, such as cisplatin and oxaliplatin, as well as controlled narcotics, including fentanyl and tramadol. Our results indicate that race and ethnicity, while imprecise proxies for genetic diversity, correlate with patterns of pharmacogenomic variation in a way that is clearly relevant to medical treatment decisions. The effects of this variation is particularly pronounced for Black and Hispanic minority groups, owing to genetic differences from the majority White group. Treatment decisions that are made based on (assumed) White pharmacogenomic variant frequencies can be harmful for minority groups. Ignoring clinically relevant genetic differences among racial and ethnic groups, however well-intentioned, will exacerbate rather than ameliorate health disparities.

Population Pharmacogenomics for Health Equity

“…This is consistent with the notion that race and ethnicity are purely social constructs with little or no biological significance [42]. Here, it must be reiterated that observed patterns of pharmacogenomic variation, particularly as they relate to adverse drug reactions, clearly support the clinical relevance of race and ethnicity [21,22,24,25]. In fact, we previously showed that when pharmacogenomic variation is partitioned exactly in the way that Lewontin and others have described, 85% within-group variation and 15% between-group variation, there can be up to 700 excess adverse drug reactions per 1000 patients predicted for the recessive effect mode and as many as 300 adverse reactions predicted for the dominant mode [24].…”

“…Here, we briefly review examples of pharmacogenomic differences among racial and ethnic groups from our own work in the US, the UK, and Colombia [21][22][23][24][25]. Most recently, we used population biobanks to interrogate the relationship between race, ethnicity, and pharmacogenomic variation in the US and the UK [25].…”

“…As we have shown previously, differences in the frequencies of toxicity-associated pharmacogenomic variants between racial and ethnic groups have important implications for public health. Our approach to studying the implications of race and ethnicity for adverse drug reactions relies on the excess number of predicted adverse reactions per one thousand patients [24]. Given the population frequency of toxicity-associated variants, along with their documented effect mode as either dominant requiring a single copy of the toxicity-associated allele or recessive requiring two copies of the toxicity-associated allele, the number of predicted adverse effects can be calculated.…”

“…Application of this approach to US racial and ethnic groups, using the All of Us and HRS cohorts, yields striking results [24,25]. The Black group shows up to 726 predicted excess adverse drug reactions per 1000 patients compared to the majority White group for the dominant effect mode and up to 635 excess adverse reactions for the recessive effect mode.…”

Psychotropic Medications and Dermatological Side Effects: An In-Depth Review

“…In particular, HLA-A has been observed to be most frequently associated with SJS/TEN reactions induced by the AEDs carbamazepine, lamotrigene, phenytoin, and zonisamide, followed by HLA-B or C subtypes [26]. Furthermore, ethnicity has a role to play in the type of reaction observed [27]. Pharmacogenomic considerations such as HLA subtype can be used to guide dose adjustments or drug selections to avoid any life-threatening ACDRs [28].…”