Incomplete linkage disequilibrium between DPYD HapB3 variants: implications for pharmacogenetic phenotype assignment

Background Reduced dihydropyrimidine dehydrogenase (DPD) activity is a major determinant of severe fluoropyrimidine-associated toxicity. The DPYD HapB3 haplotype is commonly identified through the synonymous variant c.1236G>A (rs56038477), which has traditionally been used as a proxy for the functional intronic variant c.1129–5923C>G (rs75017182). However, emerging evidence suggests that the linkage between these variants may not be absolute, potentially leading to inaccurate haplotype assignment. This study aimed to determine the frequency of clinically relevant DPYD variants in our population and to assess concordance between c.1236G>A and c.1129–5923C>G. Methods We conducted a retrospective observational study including 3276 individuals who underwent routine pharmacogenetic testing. Genotyping was performed using the OpenArray™ platform. Allele and genotype frequencies of clinically relevant DPYD variants were calculated, and concordance between c.1236G>A and c.1129–5923C>G was evaluated. Samples showing discordant or atypical HapB3-associated genotypes were subjected to confirmatory analysis by next-generation sequencing (NGS). Results Allele frequencies were consistent with those reported in European and Spanish populations. One biologically discordant genotype configuration was identified among 3276 individuals, in which c.1236G>A was detected in the absence of c.1129–5923C>G; this finding was confirmed by NGS. A second apparently discordant case, initially classified as homozygous for c.1129–5923C>G but heterozygous for c.1236G>A, was attributed to automated genotype misclassification by the Thermo Fisher Connect Platform. NGS analysis subsequently demonstrated homozygosity for both variants, whereas review of the cluster plot revealed atypical fluorescence patterns likely responsible for the erroneous automated genotype call. These findings indicate that, in our population, incomplete linkage disequilibrium between these variants, as well as technical limitations of genotyping platforms, may compromise accurate HapB3 characterization. Conclusion Concordance between the HapB3-associated variants c.1236G>A and c.1129–5923C>G is extremely high but not absolute. Apparent HapB3 discordance may result from either rare biological dissociation or technical genotype misclassification, both of which can affect DPYD phenotype assignment. These findings support direct genotyping of c.1129–5923C>G whenever feasible and orthogonal confirmation of atypical or discordant genotyping results.