15-Year WHI Trial Finds Hormone Therapy Does Not Alter Alzheimer's Blood Biomarkers

The question of whether hormone therapy (HT) protects against or accelerates Alzheimer's disease (AD) in postmenopausal women has been debated for decades, but hard biological evidence from randomized trials measuring blood-based biomarkers directly has been scarce. This secondary analysis of the Women's Health Initiative Memory Study (WHIMS) represents one of the most rigorous attempts yet to answer that question, leveraging the randomized design of the original WHI trials and newly measured plasma AD biomarkers from archived blood samples collected nearly 15 years apart.

The study enrolled 7,479 cognitively unimpaired postmenopausal women aged 65–79 from 39 U.S. clinical centers between 1996 and 1999. For this biomarker analysis, 2,784 women had baseline plasma measurements, and 967 of those had a second blood draw in 2012–2013, an average of 15.1 years later (SD 0.6). Mean age at enrollment was 69.9 years (SD 3.8) and 84.2 years at the follow-up visit. The sample was racially diverse: 73.1% White, 18.5% Black, 7.3% Hispanic/Latina, 4.7% Asian, and smaller proportions of other groups. Women participated in one of two double-blind randomized trials: oral conjugated equine estrogens (CEE) 0.625 mg/day versus placebo for women with prior hysterectomy, or CEE plus medroxyprogesterone acetate 2.5 mg/day versus placebo for women with an intact uterus.

Five plasma biomarkers were measured at the University of Minnesota Advanced Research and Diagnostics Laboratory in 2024 using state-of-the-art Simoa assays: p-tau217 (ALZpath Simoa v2), p-tau181 (Simoa v2), Aβ42:Aβ40 ratio, GFAP, and NfL. Biomarkers were log₂-transformed and standardized before analysis. Linear mixed effects models with random intercepts and slopes and an unstructured covariance matrix estimated rates of change per decade by treatment arm. Models were weighted for the biomarker sampling design and adjusted for time-varying estimated glomerular filtration rate (eGFR).

All five biomarkers showed significant worsening — indicating accumulating Alzheimer's pathology — across all treatment groups over the 15-year follow-up, consistent with the expected trajectory of aging. However, the critical treatment-by-time interaction terms were non-significant for every biomarker in both trials. Neither CEE alone nor CEE plus medroxyprogesterone acetate accelerated or slowed the progression of any AD biomarker compared to placebo. The 95% confidence intervals for all beta coefficients (representing the differential rate of change per decade in the HT vs. placebo groups) crossed zero, indicating no statistically meaningful difference.

The authors note several important limitations. The WHI HT formulations — oral CEE at relatively high doses — may not generalize to modern regimens using lower doses, transdermal estradiol, or micronized progesterone. Both trials were stopped early (the combined arm in 2002, the estrogen-alone arm in 2004) due to harm signals, and adherence declined over time, potentially diluting any biological effect on biomarkers measured years after discontinuation. Only two blood draw time points were available, preventing analysis of longitudinal trajectories. The biomarker subsample also showed attenuated and non-significant associations of HT with incident MCI/dementia relative to the full WHIMS cohort, suggesting some selection effects. Despite robust randomization, the estrogen-alone trial had minor baseline imbalances in diabetes, hypertension, and HDL cholesterol between groups.