Anti Mullerian hormone (AMH) levels predict cardiovascular risk assessment in young women with PCOS

Abstract

Serum levels of anti-mullerian hormone (AMH), a member of the TGFb family, are elevated in women with PCOS. There is conflicting information in women with and without PCOS regarding the association between low AMH levels and increased cardiometabolic risk. Young women with PCOS have an increased risk of metabolic syndrome and we hypothesized that serum AMH levels would predict risk of metabolic syndrome in this population. Retrospective cohort study. Women seen at the Penn PCOS center from 2010-2014 with complete metabolic work up were included in this study (n=469). Metabolic syndrome (Met Syn) was defined by modified NCEP-ATP III criteria (BMI≥30, sBP/dBP≥135/85mmHg or taking anti-hypertensives medications, fasting glucose ≥100mg/ml or taking medications for diabetes, TG≥150mg/ml, HDL-C ≤50mg/ml). AMH was measured using the Gen II ELISA, lipids were measured by standard enzymatic methods and total testosterone was measured by radioimmunoassay. Spearman correlation coefficients were used to determine relationships between continuous variables. Linear regression was used to model associations between AMH and lipids adjusting for confounders. AMH was dichotomized based on tertiles and univariate tests and logistic regression modeling were used to evaluate associations of selected variables with tertiles of AMH. The median AMH level in the entire cohort was 5.06ng/ml (IQR 3.08-7.97) . The mean age of the group was 27.6±6years, 42.1% were non-white and 8.1% were smokers. The overall prevalence of Met Syn was 19%. AMH levels positively correlated with total testosterone (p<0.008). On examination of individual components of Met Syn, AMH levels positively correlated with HDL-C (p<0.001) and negatively correlated with fasting glucose (p<0.004), total insulin (p<0.001), BMI (p<0.001), systolic and diastolic BP (p<0.003, p<0.005 respectively). AMH did not have a significant correlation with smoking, LDL-C, TG, CRP, DHEAS and SHBG. Total testosterone did not correlate with individual components of Met Syn. For every unit increase in AMH the odds of Met Syn decreased by 10% (OR 0.9 (0.83-0.97) in the univariate model (p=0.01) and in the multivariate model (p<0.02). The OR for Met Syn for women in the lowest AMH tertile compared to those in the highest AMH tertile was 2.3 (95% CI 1.1-4.6, p=0.02). Adjusted for age, race and testosterone the OR for Met Syn in the lowest AMH tertile was 2.1 (95% CI 1.01-4.3, p=0.04). Total testosterone did not predict risk of Met Syn (p=0.5). Our findings indicate that AMH levels, independent of age, race and testosterone are a strong predictor of Met Syn risk in young women with PCOS. Although the causal or temporal relationship for our findings is unclear, our study suggests that metabolic risk in young women with PCOS is influenced by the ovarian hormone AMH but not testosterone.