Impact of population-specific pharmacogenomic variants on drug dosing in ICU patients

The Pharmacogenomics Journal ARTICLE www.nature.com/tpj OPEN Impact of population-specific pharmacogenomic variants on drug dosing in ICU patients Mahboubeh R. Rostami1,5, Juan Rodriguez-Flores1,5, Ali Ait Hssain2,3, Alya Al Shakaki2, Huda Khan2, Muneera Vakayil ✉ Maha Elhamid3, Lubna Gamal Al Tawil3, Jason G. Mezey1,4, Amal Robay2,6 and Ronald G. Crystal 1,6 2 , Edin Karic3, 1234567890();,: © The Author(s) 2026 Intensive care units (ICU) patients are highly vulnerable to inaccurate drug dosing. Pharmacogenomics (PGx) studies the role of inherited genetic variation in drug metabolism and dose efficacy. To assess the prevalence of PGx variants that may influence therapeutic effect in the ICU, we carried out whole genome sequencing (WGS) of 210 Qataris in ICU care at Hamad Medical Corporation (HMC), Doha, Qatar and assessed the WGS for predicted deleterious variants of genes that metabolize 30 drugs commonly prescribed in the ICU. PGx variation was evaluated using two complementary approaches. First, variants with established functional interpretation were assessed using CPIC guidelines and star-allele haplotypes inferred by PharmCAT to estimate the prevalence of alleles associated with abnormal drug metabolism. Second, a broader exploratory analysis examined computationally predicted deleterious single-nucleotide variants in pharmacogenes that currently lack CPIC guidelines or defined star alleles, with these findings interpreted as descriptive of genomic variation rather than clinical metabolizer phenotypes.Of the ICU patients that received the 5 most commonly prescribed drugs (warfarin, phenytoin, midazolam, vancomycin, levetiracetam), 93% had deleterious metabolism-related variants. Ninety-one % of ICU patients carried at least one variant in a gene with known PGx relevance that could potentially impact the metabolism or activity of at least one medication they received. Most patients had ≥14 deleterious variants of genes that affect the metabolism of administered drugs. Comparison of the deleterious variants related to metabolism of ICU drugs with African/African American and European populations revealed significant population specificity in ICU related PGx variants. Together, these data suggest that population specific, PGx based on the individual’s genome likely plays a significant role in effective, safe dosing in the ICU setting. The Pharmacogenomics Journal (2026)26:23 ; https://doi.org/10.1038/s41397-026-00415-3 INTRODUCTION Accurate drug dosing of patients in the intensive care unit (ICU) presents a significant challenge. ICU patients are critically ill, with complex, acute multiorgan dysfunction superimposed on preexisting chronic diseases. ICU populations have a higher incidence of adverse drug reactions compared to other hospitalized patients [1–3]. These patients are highly vulnerable to inaccurate pharmacologic dosing, resulting in insufficient, ineffective, or excessive dosing, resulting in drug toxicity [4, 5]. Among the many reasons for inaccurate drug dosing in the ICU is PGx, the influence of the patient’s genome on drug metabolism [6–10]. PGx characterizes the effect of inherited genetic variation on drug metabolism, likely playing an important role in the consequences of a given dose of a drug regarding efficacy and toxicity [11–13]. With this background, the focus of this study is to assess the prevalence to PGx-based inaccurate dosing of patients in the ICU. We hypothesize that the typical ICU patient may benefit from PGxguided pharmacotherapy given the complexity of drug dosing in this vulnerable population. To date, there has been no whole genome PGx assessment of actual ICU patients, only limited studies of PGx in the ICU based on analysis of a few genes, analysis of whole genomes of the general population of drugs used in the ICU and have been confined to analysis of only European populations [6–10, 14–17]. In this study, we performed WGS in 210 Qatari ICU patients which enables identification of a wide variety of variants in all protein coding genes. To study both known and novel variants, powerful computational algorithms (CADD & SNPEff) were used to computationally predict variants that are likely to alter protein structure and function. This includes SNVs and indels that are computationally predicted deleterious variants. Star alleles are known for CYP family genes, and in the case of those genes, an analysis of these alleles was conducted. METHODS To assess the prevalence of PGx-based inaccurate drug dosing in an ICU population, we carried out whole genome sequencing (WGS) in 210 patients admitted to the medical ICU of Hamad Medical Corporation (HMC) Hospital, the major in-patient facility in Doha, Qatar and analyzed the 1 Department of Genetic Medicine, Weill Cornell Medicine, New York, NY, USA. 2Department of Genetic Medicine-Qatar, Weill Cornell Medicine-Qatar, Doha, Qatar. 3Medical Intensive Care Unit, Hamad Medical Corporation, Doha, Qatar. 4Department of Biological Statistics and Computational Biology, Cornell University, Ithaca, NY, USA. 5These authors contributed equally: Mahboubeh R. Rostami, Juan Rodriguez-Flores. 6These authors jointly supervised this work: Amal Robay, Ronald G. Crystal. ✉email: Received: 14 July 2025 Revised: 30 March 2026 Accepted: 28 April 2026 M.R. Rostami et al. 2 genomes for predicted deleterious variants of genes known to affect the metabolism of the drugs administered to these patients. For the analysis, we only assessed the genomes of patients with Qatari hereditary, thus eliminating population-related differences in the genomes of PGx-related genes [18–25]. The Qataris are a highly inbred population, and as has been observed in other inbred populations [26], there is high probability of identifying variants at an elevated allele frequency and homozygotes for a variants only found in a heterozygous state in most other populations. To assess whether the ICU drug-related PGx predicted deleterious variants were representative of the Qatari population, We compared the frequency of predicted deleterious variants in ICU PGx-related genes among 210 Qatari ICU patients to those found in 14,669 whole genomes from the general Qatari population, as part of the Qatar Genome Program (QGP) [27–29]. Finally, to evaluate population-specificity of the predicted deleterious variants in the ICU PGx-related genes, we compared the predicted deleterious variants in the ICU PGx-related genes in the Qataris to the predicted deleterious variants in the ICU PGx-related genes in 34,029 European and 20,744 African/African Americans from gnomAD (https://gnomad.broadinstitute.org/ for v.4.1.0) [30]. the full cohort using an established analysis workflow shown to yield highly reproducible variant calls [33]. Sequencing reads were mapped to the latest human reference genome assembly (GRCh38) with BurrowsWheeler Aligner [34] followed by preparation of mapped reads for variant calling using the GATK pipeline [35], SNV and indel va (...truncated)