Pharmacogenomic (PGx) testing tailors drug selection and dosing to a patient’s genetic profile, reducing adverse drug reactions and improving therapeutic outcomes. The clinical recommendations that drive PGx, published by the Clinical Pharmacogenetics Implementation Consortium (CPIC) and the FDA, are evidence-based and guideline-backed.
However, many of the most clinically important PGx variants occur at very different frequencies across populations. A panel, allele definition, or dosing algorithm optimized for one population can underserve another. For laboratories implementing pharmacogenomics across diverse patient populations, understanding these differences (and being able to customize PGx catalogs to reflect them) is essential to precision prescribing.
Below are four well-characterized examples.
CYP3A5 and tacrolimus
The CYP3A5 *3 loss-of-function allele is near-fixed in European populations: roughly 85–95% of European-ancestry patients are *3/*3 non-expressers, and standard tacrolimus dosing protocols are calibrated to that assumption. Roughly 70% of Latino patients and 85% of patients of African ancestry retain at least one expressing *1 allele, clear tacrolimus more quickly, and require higher per-kilogram doses to reach therapeutic troughs. CPIC has a Tier A recommendation for CYP3A5-guided tacrolimus dosing in solid organ transplant.
HLA-B*15:02 and carbamazepine
HLA-B*15:02 reaches 5–15% allele frequency across Han Chinese, Thai, Malay, and Vietnamese populations and is rare in Europeans. Carriers face a sharply elevated risk of Stevens-Johnson syndrome and toxic epidermal necrolysis on carbamazepine, adverse drug reactions severe enough that the FDA includes a boxed warning. HLA loci also present a technical challenge for NGS-based PGx: CPIC does not currently define variant-level alleles for HLA, which is why the ability to specify diplotypes manually (covered in our earlier post on defining HLA diplotypes for pharmacogenomics) matters for any laboratory reporting on HLA-B.
CYP2C19 and clopidogrel
CYP2C19 metabolizer phenotype distributions differ substantially across populations. The loss-of-function *2 and *3 alleles reach 15–20% or higher in East Asian populations, producing a meaningful share of poor metabolizers who do not activate clopidogrel effectively after a cardiac event and face elevated stroke and reinfarction risk. The ultrarapid *17 allele is most common in European and Middle Eastern populations and shifts metabolism in the opposite direction for proton pump inhibitors and certain antidepressants. The same gene drives different clinical actions depending on the diplotype.
NUDT15 and thiopurines
For years, TPMT was the standard gene-drug pair for predicting thiopurine toxicity, and it explained most cases of azathioprine and mercaptopurine toxicity in European cohorts. NUDT15 c.415C>T was subsequently identified as the dominant predictor of thiopurine toxicity in East Asian and Hispanic patients, and CPIC later added it to its recommendations. The pattern is a familiar one in pharmacogenomics: guidelines evolve as new variants are characterized in new populations, and laboratory PGx catalogs need to keep pace.
VSPGx, the pharmacogenomics module in VarSeq, ships with broad CPIC and FDA gene coverage and automates the steps from variant data to clinical report. CYP3A5, CYP2C19, NUDT15, and HLA-B are all included in the default CPIC + FDA Recommendations track, alongside other pharmacogenes spanning CYP2D6, DPYD, TPMT, SLCO1B1, VKORC1, UGT1A1, G6PD, and others. The CPIC + FDA Recommendations track is regularly updated as new guidelines and FDA label information are released.
For each gene, the PGx Variant Detection and Recommendations algorithm maps variants to standardized star allele nomenclature, assigns diplotypes, classifies metabolizer phenotypes, and joins those results to CPIC and FDA drug recommendations. CYP2D6, which is challenging for conventional variant callers because of high homology with the CYP2D7 pseudogene and frequent copy number and structural variation, can be handled by CypCall, our specialized CYP2D6 caller that feeds results directly into VarSeq.
For laboratories that need to go further (adding a gene CPIC has not yet covered, defining a novel allele, modifying drug interaction strengths, or layering institution-specific recommendations onto the defaults), VSWarehouse’s custom PGx Catalogs make those changes manageable through the catalog interface, without custom curation scripts. Report templates are fully customizable as well, so the clinical reports delivered to clinicians reflect exactly what your laboratory considers actionable.
