Model-based clinical utility of pharmacogenetic testing and its potential population impact on the use of essential medicines in Zimbabwe.

Local allele frequencies of actionable genes such as UGT1A1, CYP2B6 and CYP2C9 change the number needed to genotype to avoid one event, and hence the clinical yield of each gene-drug pair.

This cross-sectional analysis combined the Zimbabwe Essential Medicines List, DPWG guidelines, local Zimbabwean genotype and phenotype data, and locally approved Summaries of Product Characteristics. Among 308 medicines screened, 30 (9.7%) contained an actionable pharmacogenomic biomarker, corresponding to 38 gene-drug pairs, all classified as vital or essential medicines. The most favourable numbers needed to genotype (NNG) were for UGT1A1-atazanavir (5), UGT1A1-irinotecan (9), CYP2C9-phenytoin (25) and CYP2B6-efavirenz (26), whereas 61% of gene-drug pairs had NNG above 1,000 versus 47% in the Dutch population. Overall 81.6% of recommendations were concordant across populations while 21.1% were reclassified, mainly because of local genetic profiles and the limited pharmacogenomic content of Zimbabwean SmPCs (42% lacking such information versus 24% in the Dutch dataset).

Synthesis written by Geno'X. For the full original abstract, please refer to the source publication.

The message that matters here is not the catalogue of gene-drug pairs but the demonstration that the utility ranking of a pharmacogenetic programme is reshuffled when the population changes: an NNG of 5 for UGT1A1-atazanavir and an NNG above 1,000 for most pairs do not call for the same strategy. The work remains model-based and cross-sectional, with absolute risk reductions borrowed from DPWG and therefore from European data, so local NNG values inherit the very transferability assumption the paper sets out to challenge. The generalisable lesson is that missing pharmacogenetic information in local regulatory documents is as concrete a barrier as access to the test itself.

Analysis by Dr Thibaut Benquey