P-521 Inter-individual variation in gonadotoxicity in female childhood cancer survivors – a genome-wide association study: results from the PanCareLIFE study

Abstract

Abstract Study question To identify new variants associated with alkylating agent (AA) induced gonadotoxicity among adult female childhood cancer survivors (CCS) using a genome-wide association study approach (GWAS). Summary answer Upon meta-analysis of two independent cohorts, the variant rs78861946 was associated with lower AMH levels in CCS, but not separately in the smaller replication cohort. What is known already Female childhood cancer survivors are at risk of therapy-related gonadal impairment. Alkylating agents are well-established risk factors, and the inter-individual variability in gonadotoxicity may be explained by genetic polymorphisms. Study design, size, duration A genome-wide association study (GWAS) approach was used to discover new variants associated with AA-induced gonadotoxicity among adult female CCS. The discovery cohort included adult female CCS from the pan-European PanCareLIFE cohort (n = 743; median age: 25.8 years), excluding those who received bilateral ovarian irradiation, bilateral oophorectomy, CNS, total body irradiation, or stem cell transplantation. Results were replicated in the USA-based St. Jude Lifetime Cohort (n = 391; median age: 31.3 years) and combined in a meta-analysis. Participants/materials, setting, methods Anti-Müllerian hormone (AMH) levels served as a proxy for ovarian function in both the discovery and replication cohort. A GWAS was performed using a linear regression model adjusted for cyclophosphamide equivalent dose of alkylating agents, age at diagnosis, and age at study. Main results and the role of chance Three genome-wide significant (<5.0x10E-8 ) and 16 genome-wide suggestive (<5.0x10E-6 ) loci were associated with log-transformed AMH levels, adjusted for cyclophosphamide equivalent dose of alkylating agents, age at diagnosis, and age at study in the PanCareLIFE cohort. Based on effect allele frequency (EAF) (>0.01 if not genome-wide significant), p-value (<5.0 × 10E−6 ), and biological relevance, 15 SNPs were selected for replication in the independent replication cohort of the St. Jude Lifetime Cohort. None of the SNPs were statistically significantly associated with AMH levels in this smaller cohort. A meta-analysis indicated thatrs78861946 was associated at borderline genome-wide statistical significance (reference/effect allele: C/T;EAF: 0.04, Beta (SE): −0.484 (0.091), p-value= 9.39 × 10E−8). While the sample size of both the discovery and replication cohorts are large for CCS cohorts, they are very small compared to GWASs in the general population. The limited power is likely one of the reasons the identified SNPs were not statistically significant in our replication cohort of SJLIFE. Limitations, reasons for caution While AMH is a valuable surrogate marker for ovarian reserve, it remains a proxy for ovarian reserve and low AMH levels may still coincide with fertility potential and pregnancy. Wider implications of the findings Identification of risk-contributing variants such as rs78861946 may prove useful to better identify patients who are specifically susceptible for alkylating agent-induced gonadal toxicity. Future polygenic risk scores may aid individualized counselling regarding the treatment-related risk of gonadotoxicity and fertility preservation options for female childhood cancer survivors. Trial registration number not applicable