编码药物代谢酶和药物转运体基因中遗传变异的多基因风险评分与发热性中性粒细胞减少的关联

Genetic variants in genes encoding drug-metabolizing enzymes and drug transporters may influence the efficacy and toxicity of chemotherapy. A polygenic risk score (PRS), capturing the aggregated effects of these genetic variants, may help to identify patients at increased risk of treatment toxicity. We conducted a study of 1,710, 1,897, 4,143, 4,398, and 4,779 European ancestry cancer patients who received anthracycline-, alkylating-, taxane-, platinum-, and antimetabolite-based regimens to evaluate the PRS of drug-induced febrile neutropenia (FN). Data utilized in our analysis were from electronic health records and genetic data from ~96,000 Vanderbilt Biobank (BioVU) participants. FN was defined as an absolute neutrophil count <1000/µL, or hospitalization with diagnosis codes of neutropenia, and fever, diagnosis code for bacterial/fungal infection, or antibiotic use. Patients with acute myeloid leukemia, bone marrow transplantation, and use of prophylactic Granulocyte Colony-Stimulating Factor (G-CSF) were not included in our study. Among cancer patients who received anthracycline, alkylating, taxane, platinum, and antimetabolite agents, 21.1%, 16.7%, 8.1%, 9.1%, and 10.5% experienced FN, respectively. We performed logistic regression models with adjustment for sex, age, and the first five principal components to investigate the association of common single-nucleotide polymorphisms (SNPs). We derived PRSs based on two sets of FN-related SNPs: 1) targeted gene approach: SNPs with false discovery rate (FDR)-corrected p<0.2 for metabolizing enzyme and drug transporter genes of study drugs; and 2) global untargeted approach: SNPs with p<5×10 -6 based on genome-wide association study (GWAS) of FN. In the targeted gene approach, we built PRSs based on 6, 4, 15, 10, and 6 SNPs in targeted 17, 11, 14, 13, and 19 genes involved in metabolizing or transporting anthracycline, alkylating, taxane, platinum, and antimetabolite agents. Per SD increase of these PRGs was associated with an increased risk of FN, with Odds ratio (OR) and p-value of 1.87 (1.00x10 -17 ), 1.58 (3.41x10 -22 ), 2.03 (1.07x10 -29 ), 1.78 (3.12x10 -28 ), and 1.60 (5.40 x10 -20 ), respectively. In the global untargeted approach, we built PRSs based on 8, 12, 20, 12, and 6 SNPs that were significantly associated with anthracycline-, alkylating-, taxane-, platinum-, and antimetabolite-induced FN, with a per-SD increase in the PRSs being associated with an increased risk of FN, OR of 2.52 (2.29x10 -32 ), 2.52 (3.92 x10 -33 ), 2.83 (1.01x10 -59 ), 2.27 (1.07 x10 -47 ), and 1.58 (1.70x10 -28 ). Whole-genome sequencing data are now available for 250,000 BioVu participants, which will increase the statistical power of GWAS and enable us to validate the PRSs and to include known functional, but rare, variants in drug metabolism and transport genes when deriving PRSs. Updated results will be presented at the conference.