Does AMH relate to timing of menopause? Results of an individual patient data meta-analysis

Abstract

Context: Anti-Müllerian hormone (AMH)–based age at menopause predictions remain cumbersome due to predictive inaccuracy. Objective: To perform an individual patient data meta-analysis, regarding AMH-based menopause prediction. Design: A systematic literature search was performed using PubMed, Embase, and Cochrane databases. Prospective cohort studies regarding menopause prediction using serum AMH levels were selected by consensus discussion. Individual cases were included if experiencing a regular cycle at baseline. Exclusion criteria were hormone use and gynecological surgery. Results: This meta-analysis included 2596 women, and 1077 experienced menopause. A multivariable Cox regression analysis assessed time to menopause using age and AMH. AMH predicted time to menopause; however, added value on top of age was poor [age alone: C-statistic, 84%; age + AMH: hazard ratio (HR), 0.66; 95% CI, 0.61 to 0.71; C-statistic, 86%). Moreover, the capacity of AMH to predict early (≤45 years) and late menopause (≥55 years) was assessed. An added effect of AMH was demonstrated for early menopause (age alone: C-statistic 52%; age + AMH: HR, 0.33; 95%, CI 0.24 to 0.45; C-statistic, 80%). A Weibull regression model calculating individual age at menopause revealed that predictive inaccuracy remained present and increased with decreasing age at menopause. Lastly, a check of non-proportionality of the predictive effect of AMH demonstrated a reduced predictive effect with increasing age. Conclusion: AMH was a significant predictor of time to menopause and especially of time to early menopause. However, individual predictions of age at menopause demonstrated a limited precision, particularly when concerning early age at menopause, making clinical application troublesome.