A High-risk Haplotype for Premature Menopause in Childhood Cancer Survivors Exposed to Gonadotoxic Therapy

Abstract

Background: Childhood cancer survivors are at increased risk of therapy-related premature menopause (PM), with a cumulative incidence of 8.0%, but the contribution of genetic factors is unknown.

Methods: Genome-wide association analyses were conducted to identify single nucleotide polymorphisms (SNPs) associated with clinically diagnosed PM (menopause < 40 years) among 799 female survivors of childhood cancer participating in the St. Jude Lifetime Cohort Study (SJLIFE). Analyses were adjusted for cyclophosphamide equivalent dose of alkylating agents and ovarian radiotherapy (RT) dose (all P values two-sided). Replication was performed using self-reported PM in 1624 survivors participating in the Childhood Cancer Survivor Study (CCSS).

Results: PM was clinically diagnosed in 30 (3.8%) SJLIFE participants. Thirteen SNPs (70 kb region of chromosome 4q32.1) upstream of the Neuropeptide Receptor 2 gene (NPY2R) were associated with PM prevalence (minimum P = 3.3 × 10-7 for rs9999820, all P < 10-5). Being a homozygous carrier of a haplotype formed by four of the 13 SNPs (seen in one in seven in the general population but more than 50% of SJLIFE clinically diagnosed PM) was associated with markedly elevated PM prevalence among survivors exposed to ovarian RT (odds ratio [OR] = 25.89, 95% confidence interval [CI] = 6.18 to 138.31, P = 8.2 × 10-6); this finding was replicated in an independent second cohort of CCSS in spite of its use of self-reported PM (OR = 3.97, 95% CI = 1.67 to 9.41, P = .002). Evidence from bioinformatics data suggests that the haplotype alters the regulation of NPY2R transcription, possibly affecting PM risk through neuroendocrine pathways.

Conclusions: The haplotype captures the majority of clinically diagnosed PM cases and, with further validation, may have clinical application in identifying the highest-risk survivors for PM for possible intervention by cryopreservation.

Figure 1.

Manhattan plot from a single–single nucleotide polymorphism genome-wide association analysis, which identified 13 SNPs in close proximity on chromosome 4, all with P values of less than 10^-5, with a minimum P value of 3.3 × 10^-7.

Figure 2.

Linkage disequilibrium (LD) matrix (r² for individuals with European ancestry from zero [white] to one [black]), highlighting four LD blocks that contain the 13 single nucleotide polymorphisms on chromosome 4q32.1 with P values of less than 10^-5, with a mean between-LD block r² of .50 and a mean within-LD block r² of .96 (25).

Figure 3.

Visualization of regulatory annotations for the expanded chromosome 4q32.1 genetic signal associated with premature menopause in neuron and ovary cell types, along with haplotype single nucleotide polymorphisms and bound transcription factors' genomic locations. A) Chromatin state annotations (ChromHMM) in H9- derived neuron cells. Colored genomic regions reflect chromHMM annotations for chromain states (enhancer, transcribed, Polycomb-repressed, and promoter) (23). B) ChromHMM annotations in ovary cells. C) ENCODE histone modifications associated with Polycomb-repressed regions (H3K27me3) for H9-derived neurons (23). D) H3K27me3 marks for placenta amnion cells (ovary cell data unavailable) (23). E) ENCODE histone modifications associated with repressed regions (H3K9me3) for H9-derived neurons (23). F) H3K9me3 marks for ovary cells (23).