Background

Studies examining the cognitive effects of hormone replacement therapy (HRT) for menopausal symptoms have produced mixed results, with some suggesting that HRT increases dementia risk. Our aim was to assess the effect of HRT on brain structure, neurovascular function and cognition using data from the PREVENT-Dementia cohort study.

Methods

Cognitively healthy women (n=388, 40-60 years) were classified using the gold-standard STRAW+10 criteria as: HRT-users (HRT-U,n=60), peri- or post-menopausal non-hormone users (HRT-NU,n=201), and pre-menopausal non-hormone users (PRE,n=127). Episodic memory (EM), affected early in Alzheimer’s dementia and thought to be sensitive to HRT, was derived using principal component analysis of the COGNITO battery. Covariates including vascular risk factors (hypertension, hyperlipidemia, diabetes), depressive symptoms and sleep, were collected using questionnaires (Center for Epidemiological Studies Depression scale, Pittsburgh Sleep Quality Index). Follicle stimulating hormone (FSH), a proxy of circulating sex-hormones, was analysed using fasting blood samples. Total grey matter (GMV), white matter (WMV), white matter hyperintensity (TWMH) volumes, and grey matter cerebral blood flow (GMCBF) were extracted from magnetic resonance imaging sequences (T1-weighted, fluid-attenuated inversion recovery, and arterial spin labelling) using FreeSurfer (v7.1.0), SPM12 and FSL. R Studio was used for imputation of missing questionnaire and FSH data (mice package), and multivariate robust linear regression models.

Results

HRT-U and HRT-NU were older, had higher FSH levels and higher prevalence of sleep disorders than PRE, while HRT-NU had lower education than HRT-U and PRE. HRT-NU had worse EM than HRT-U (B:-0.36[0.14],p=0.01) and PRE (B:-0.52[0.16],p<0.001), correcting for age, education, depressive symptoms and sleep. HRT-U had lower WMV than HRT-NU (B:-13205[4350]mm³,p(FDR)=0.01,n=326) with a similar trend compared to PRE (B:-12645[5650]mm³,p=0.03,p(FDR)=0.10), even correcting for age, education, intracranial volume and vascular risk factors. There were no significant differences in GMV, WMHs or GMCBF. In HRT-U, lower FSH level (suggesting a higher HRT dose) was associated with lower GMV (B:418[108]mm³/unit,p(FDR)=0.001), but not other imaging measures.

Conclusions

Although HRT-U had better episodic memory, there was a hormone dose-dependent reduction in GMV. WMV was also lower in HRT-U, while there was no association with neurovascular function. Ongoing follow-up of these women will reveal whether this effect translates to a higher dementia risk.