Ventricular anatomy is a major source of DTI-ALPS variability in aging females

Abstract Background Diffusion tensor image analysis along the perivascular space (DTI-ALPS) is increasingly used as an indirect MRI marker of glymphatic and perivascular fluid transport, yet its dependence on local anatomy remains uncertain. Because ALPS is sampled in periventricular white matter, ventricular enlargement may influence the metric independently of, or alongside, brain-fluid physiology. Endocrine ageing provides a biological test case as reproductive hormone changes occur alongside age-related ventricular enlargement. We examined whether associations between reproductive hormones and DTI-ALPS were independent of lateral ventricular (LV) volume. Methods This cross-sectional cohort study used baseline data from female participants in the Human Connectome Project Aging and Aging Adult Brain Connectome studies. The primary outcome was the DTI-ALPS index; exposures were circulating estradiol, follicle-stimulating hormone (FSH), and luteinising hormone (LH). LV volume was derived from FreeSurfer segmentations. Four sequential linear regression models were fitted: unadjusted; adjusted for age; adjusted for age plus LV volume; and fully adjusted for age, LV volume, body mass index, study site, and hormone replacement therapy status. Potential effects of sleep quality, APOE ε4 status, and choroid plexus volume were explored as secondary analyses. Results The analysis included 784 females (mean age 60.0 years, SD 15.1). In unadjusted models, higher estradiol was associated with higher ALPS (B = 0.044, 95% CI 0.033 to 0.056), whereas higher FSH (B = − 0.046, 95% CI − 0.058 to − 0.033) and LH (B = − 0.046, 95% CI − 0.063 to − 0.029) were associated with lower ALPS. These associations changed after age adjustment and were no longer evident after inclusion of LV volume. LV volume was inversely associated with ALPS ( r = − 0.71, 95% CI − 0.74 to − 0.67). Inclusion of LV volume attenuated estradiol effect estimates by 94%–100% across multiple parameterisations; similar attenuation was observed for FSH and LH. Conclusions LV volume was strongly associated with DTI-ALPS and statistically accounted for most endocrine-related variation in the index. These findings demonstrate substantial anatomical dependence of DTI-ALPS but do not establish whether this reflects measurement geometry, shared age-related biology, or both. Ventricular morphology should therefore be considered when interpreting the index, and inferences about brain-fluid transport require independent validation.