Identifying risk variants for dihydropyrimidine dehydrogenase deficiency using an isogenic system of expression. S. M. Offer, S. Shrestha, C. R. Jerde, R. B. Diasio Department of Molecular Pharmacology and Experimental Therapeutics, Mayo Clinic, Rochester, MN.

   Dihydropyrimidine dehydrogenase (DPD) deficiency [MIM 274270] manifests most frequently as severe clinical toxicity to the commonly prescribed anti-cancer drug 5-fluorouracil (5-FU), but has also been linked to various degrees of developmental delay in children, which is likely due to altered uracil catabolism. Genetic variations in the gene encoding DPD, DPYD [MIM 612779], have been suggested as the primary cause of DPD deficiency; however, given the rarity of individual candidate causal alleles within the gene, the functional consequences of many variants has not been evaluated. To address this important topic and identify which DPYD variants contribute to DPD deficiency, we have determined the effect of all 144 reported missense, nonsense, frameshift, and splice variants on DPD enzyme function, DPD protein stability, and expression of correctly-spliced DPYD mRNA. Each amino acid-changing variant was expressed in mammalian cells and the in vitro enzyme activity of expressed DPD measured. Approximately one-third of the reported variants directly impaired 5-FU catabolism to a degree that makes them candidate risk alleles for 5-FU toxicity. Based on publically-available allele frequency data, these variants are expected to be carried by 2-6% of the global population. To gain a broader appreciation for the spectrum of coding variations present within the gene, we performed targeted high throughput sequencing of the region encoding DPYD in a cohort of Somali-American individuals from Southeastern Minnesota, a population for which limited DNA sequence data are available. In addition to 8 previously reported variants, 13 novel missense, frameshift, and splice donor site variants were detected in this population. Collectively, these findings suggest that multiple rare variants contribute to DPD deficiency and that the spectrum of risk variants may vary greatly between racial groups. For that reason, targeted predictive tests developed in a single racial group are likely to incompletely genotype relevant alleles in other racial groups, necessitating a sequence-based approach for detecting risk alleles. Furthermore, the described system of phenotypically evaluating recombinantly-expressed variants should be applicable to additional drug-gene pathways and the study of other nucleotide metabolism disorders.