Genome-wide analysis defines genetic determinants of MPN subtypes and identifies a sex-specific association at CDH22/CD40 Open Access

• Genetic variation at HBS1L-MYB and CDH22/CD40 are the strongest determinant of MPN phenotype, but the latter is only seen in females

• Polygenic risk scores for MPN are improved when optimised for disease subtype

To identify genetic variants that influence myeloproliferative neoplasm (MPN) phenotype, we undertook a two-stage case-only genome-wide association study using cohorts from the UK (including UK Biobank), Spain, Germany and Italy. MPN subtype [essential thrombocythemia (ET); polycythemia vera (PV)] were compared to each other, to healthy controls and stratified analyses was performed based on chromosome 9p aberrations, JAK2 V617F mutation burden and sex. The ET versus PV analysis identified known associations: (i) at HBS1L-MYB that increased ET risk (PMETA=7.93x10-6, OR=1.28) and reduced PV risk (PMETA=9.43x10-5, OR=0.81) and (ii) at GFI1B-GTF3C5 that predisposed to PV only (PMETA=1.43x10-9, OR=1.38). Two further linked intronic SNPs, rs2425786 and rs2425788, at CDH22/CD40 were significant in females only (PMETA=2.67x10-8) with predisposition to PV (PMETA=0.0006, OR=1.3) and reduction of ET risk (PMETA=7.82x10-5, OR=0.75). Associations with JAK2, TERT, ATM, TET2, PINT, GFI1B and SH2B3 were confirmed (PMETA<5x10-8) and nine further loci were replicated (PMETA<0.05). A polygenic risk score consisting of 48 SNPs from 31 loci demonstrated moderate discriminative performance for ET and PV (AUC=0.718) and was improved by optimization for disease subtype (AUCET=0.724 and AUCPV=0.755). Overall, our results reveal that multiple germline variants influence MPN phenotype with HBS1L-MYB and a novel sex-specific association with CDH22/CD40 being the strongest determinants.