Genotype-Phenotype Predictive Gap in Pre-Emptive Screening of DPD Deficiency: Improving Sensitivity

Background/Objectives: toxicity related to treatment with the anti-cancer drugs fluoropyrimidines (FP) is mainly driven by a deficiency in the dihydropyrimidine-dehydrogenase (DPD) enzyme activity. Standardized pharmacogenomic markers in the DPYD gene allow pre-emptive prediction and therapy adjustment in patients testing positive, but the rarity of these markers lowers screening sensitivity, leaving a significant part of at-risk individuals undetected. This study aimed to explore the diagnostic performance of genetic screening tests modelled to include unclassified rare DPYD variants and the common SNPs rs1801160 and rs2297595. Methods: this is a retrospective genotype-phenotype association study including 687 Italian subjects who underwent DPD-phenotyping by the 5-fluorouracil degradation assay (5-FUDR) and DPYD genotyping by full exon sequencing. According to their DPD-phenotype, subjects were grouped into a DPD deficiency group (DD, N = 94; >30% reduction in 5-FUDR) and a normal DPD activity group (NDA, N = 593). The phenotype was used as the reference standard to assess genotype-based functional prediction, and discordant genotype-phenotype cases were scored as false negatives or false positives. The phenotype (DD or NDA) prediction power of a genetic, binary (presence/absence) screening test including standardized pharmacogenomic markers (Test 1) was compared to binary tests including missense, nonsense, frameshift or splice-site rare variants (Test 2) or the co-occurrence of rs1801160 and rs2297595 (Test 3). Statistical significance was set at p < 0.05, correcting for multiple testing when appropriate. Results: positivity for all binary tests was significantly associated with the DPD deficiency phenotype (Test1: OR = 6; Test 2: OR = 4.85; Test3: OR = 4.9). Notably, overlap among subjects testing positive in the different tests was minimal. Combining all tests, a 3-fold increase in screening sensitivity was achieved compared to the reference Test 1 (from 12.77% to 35.11%), with a slight decrease in specificity from 97.64% to 91.74%. Conclusions: the proposed approach can refine the genetic predictivity of the DPD-deficiency phenotype in pre-emptive screening, capturing a significantly larger pool of biochemically at-risk patients.