Disturbed sleep and reduced hippocampal integrity at midlife: Implications for AD risk in women with early surgical menopause

Abstract

AbstractBackgroundWomen bear the greatest burden of Alzheimer’s disease (AD), and early (<45y) ovarian hormone deprivation via surgical menopause further increases risk (Rocca et al., 2007). Given that sleep disturbance is implicated in AD progression (Lim et al., 2013), an important area of investigation is determining whether younger women with early surgical menopause demonstrate disturbed sleep at midlife, and whether this may confer additional risks by contributing to changes in hippocampal integrity. Thus, we investigated whether middle‐aged women with bilateral salpingo‐oophorectomy not taking hormone therapy demonstrate sleep disturbance, and reduced hippocampal volume and memory. Additionally, we assessed whether sleep and hippocampal volume predict memory performance.MethodWomen with an early BSO that were either taking estradiol‐based hormone therapy (BSO+ET: age=43y; n=22), or not taking ET (BSO: age=46y; n=18) were recruited to the study, and were compared to age‐matched premenopausal (AMC: n=24) and spontaneously postmenopausal women who were ∼10y older (SM: 55y; n=20). Sleep was assessed via portable polysomnography (Vitaport‐5/REMbo‐234, Temec Technologies) for 1‐3 nights, and sleep staging was acquired using automated scoring (Z3score, Neurobit technologies). High‐resolution T2‐weighted scans were acquired using a Siemens 3T Prisma scanner, and hippocampal subfields (CA1, Subiculum, combined region: DGCA23), were manually segmented (Olsen et al., 2017) in FSLView. The interference match‐to‐sample (IMTS) task was used to asses and recognition memory for scenes (Watson et al., 2013).ResultBoth hormone‐deprived groups demonstrated increased sleep latency (BSOvsAMC: p<.05, d= 0.74; SMvsAMC: d= 0.69) and decreased sleep efficiency (BSOvsAMC: p<.05, d= 0.79; SMvsAMC: p<.05, d= 0.67). Only the BSO group demonstrated reduced volume in the anterior CA1 and DGCA23 relative to AMC (p<.05, d= 0.96‐1.06) and BSO+ET (p<.05, d= 0.67‐0.84). Additionally, the BSO demonstrated reduced efficiency (accuracy/RTcorrect) on scene recognition memory (p<.05, d=0.70‐0.86). Multiple regression analyses show that DGCA23 volume and group membership predict scene recognition memory performance.ConclusionWhile both hormone‐deprived groups demonstrated disturbed sleep, women with early BSO also showed reduced hippocampal integrity, despite being ∼10 years younger than SM women. These findings are consistent with the elevated AD risk among women with BSO. It remains to be seen whether sleep disturbance at a younger age exacerbates risk.