There are critical and perplexing sex/gender differences in Alzheimer’s disease (AD). Women show a more favorable clinical profile in preclinical AD particularly with verbal memory, but a steeper decline post mild cognitive impairment (MCI) diagnosis and, ultimately, higher rates of AD. Longitudinal studies are needed to understand sex differences across the AD trajectory. Using data from the Alzheimer’s Disease Neuroimaging Initiative, we identified profiles of memory trajectories among those with evidence of preclinical AD or MCI at baseline and how these trajectories differ by sex.

In our sample of 659 participants (age range: 55-90, mean age=72.9 [SD=7.4], 95% non-Hispanic White; mean follow-up=41.2 [SD=32.3] months), 233 were labelled “preclinical” AD (51% women) at baseline based on a cognitively normal status but positivity for either the cerebrospinal fluid p-Tau/Aß42, Amyloid PET or Tau PET biomarkers, and 426 participants (44% women) were MCI at baseline based on Jak/Bondi criteria. We applied latent class growth curve modeling to the heterogeneous change in the Rey Auditory Verbal Learning Test (RAVLT) Immediate and Delayed Recall within preclinical and MCI groups separately. Models in MCI group included Non-Linear Spline to account for differential change rates within subgroups. Models were compared on Bayesian Information Criterion, Entropy, and Class distribution to determine a best-fitting model. Effects of sex on trajectories were the primary outcomes. All models included APOE4 carrier status and age.

Women outperformed men on Immediate and Delayed Recall at baseline in the preclinical and MCI groups (ps<.05). Within the preclinical group, 3-class models representing stable, decline, and accelerated decline provided optimal fit for both Immediate and Delayed Recall. Whereas, on average, preclinical women showed more stable Immediate Recall than men (beta=6.24, SE=.82, p<.0001), they were more likely to be in the Immediate Recall accelerated decline class (23.4% vs. 16.25%; female:male; Chi-square=36.29, p<.00001). On average, preclinical women and men did not differ in Delayed Recall trajectories (beta=.31, SE=.30, p=.28); however, preclinical women were more likely to be in the stable Delayed Recall class (11.04% vs. 6.5%; Chi-Square=19.19, p<.0001). Within the MCI group, 2-class models representing a stable decline group and an accelerated decline group provided optimal fit for both outcomes. Whereas, on average, MCI women showed more stable Immediate Recall than men (beta=3.55, SE=.79, p<.0001), they were more likely to be in the Immediate Recall accelerated decline class, although not significantly. Women and men did not differ, on average, in their Delayed Recall trajectories; however, women were significantly more likely to be in the Delayed Recall accelerated decline class (Chi-square=32.24, p<.0001).

Our findings indicate that sex is an important determinant of the variability observed in early-stage AD trajectories; however, sex differences varied by Immediate versus Delayed Recall likely due, in-part, to psychometric test properties. Our results suggest that, when looking at sex differences in AD trajectories on average, women’s superior stability in verbal learning masks their higher likelihood of rapid decline. Our findings have implications for our ability to optimally diagnose and track disease progression in both sexes.