DPYD Genotyping of Patients with Fluoropyrimidine Treatment: Results of Protocol Implementation and Outcomes of Patients Carrying Unusual DPYD Variants

The intronic DPYD variant c.1129-5923C>G introduces a cryptic splice site and reduces dihydropyrimidine dehydrogenase activity; the synonymous variant c.1236G>A, in linkage disequilibrium within the HapB3 haplotype, is used as its proxy in many commercial kits.

The most frequent causal DPYD variant, c.1129-5923C>G (rs75017182), lies in intron 10 and introduces a cryptic splice site; many commercial kits substitute the synonymous variant c.1236G>A (rs56038477), assumed to be in complete linkage disequilibrium with it within the HapB3 haplotype. The authors report five years of DPYD genotyping performed under clinical laboratory quality regulations (EN-ISO9001:2015 and EN-ISO15189:2022, with NGS, MLPA and Sanger sequencing for validation) in 2007 patients, with genotype frequencies consistent with European reference data. One patient carrying c.1236G>A without c.1129-5923C>G was identified, supporting recent reports that these two variants are not in perfect linkage disequilibrium and exposing such carriers to suboptimal dosing. Three low-frequency variants absent from DPYD testing recommendations (c.496A>G, c.2194G>A and c.1601G>A) also led to dose adjustment in two patients. The authors conclude that relying on a proxy variant risks misclassifying metabolic capacity and argue for expanding DPYD panels to additional rare variants.

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

The operational message is simple: a laboratory genotyping c.1236G>A as a proxy reports a result it has not measured, and the case described here shows that the approximation eventually hits a patient. The demonstration rests on a single discordant carrier, so it carries no frequency estimate, yet it does not need to be quantitative to change practice: it is enough to target the intronic variant directly, which high-throughput sequencing does at no extra cost. The remaining issue is rare variants, captured only by an untargeted approach — and there, interpretation rather than the list of tested positions becomes the limiting factor.

Analysis by Dr Thibaut Benquey