Poor Sleep Raises Alzheimer’s Brain Changes in Older Women at Genetic Risk

The tau protein had already begun accumulating in the wrong places. Not dramatically, not in ways that would show up on a standard clinic assessment, but enough to register on a PET scanner, enough to correlate, in a study of older women in San Diego, with something as mundane as how well they slept last month. The women who slept poorly and carried a high genetic burden for Alzheimer’s disease were showing more of this sticky, tangled protein in the limbic regions of their brains. The women who slept poorly but carried lower genetic risk were not.

That distinction, subtle as it sounds, may turn out to matter rather a lot for how we think about Alzheimer’s prevention in women, who bear a disproportionate share of the disease.

Researchers at the University of California San Diego have published findings from the Women Inflammation Tau Study, known as WITS, an ongoing prospective project focused on older women at elevated Alzheimer’s risk. The team, led by doctoral student Kitty Lui and associate professor Sarah Banks, recruited 69 women aged 65 and older, all of them showing early signs of mild cognitive impairment. Participants completed the Pittsburgh Sleep Quality Index, a validated questionnaire covering seven dimensions of sleep health over the prior month, and underwent neuropsychological testing for both visual and verbal memory. A subset of 63 women also received tau PET scans using a relatively new tracer, [18F]-MK6240, which has stronger affinity for the neurofibrillary tangles of Alzheimer’s than older generation compounds.

A Gene That Tips the Balance

The study used something more refined than the APOE gene variant so often discussed in Alzheimer’s research. The polygenic hazard score, or PHS, factors in nearly 200,000 genetic variants alongside APOE status, making it, in principle, a more sensitive predictor of when Alzheimer’s might emerge. Women scoring above the 75th percentile on this measure were classified as higher genetic risk; those below it, lower risk. As it happened, every single woman in the higher-risk group was also an APOE epsilon-4 carrier, which complicates any attempt to disentangle the effects of broad polygenic risk from the influence of that single well-studied variant.

What the researchers found was a fairly specific pattern. Among women in the higher genetic risk group, poorer self-reported sleep was associated with worse performance on visual memory tests and with greater tau accumulation in Braak stage III/IV regions, the limbic areas of the brain that typically become affected in the early-to-middle stages of Alzheimer’s pathology. The effect size was medium to large, despite a modest sample. Among women with lower genetic risk, the same association was simply absent.

Verbal memory told a different story, or rather no story at all. Sleep complaints showed no significant relationship to verbal recall in either group. That finding is probably not coincidental. Women, on average, perform better on verbal memory tasks than men, even when they already carry significant Alzheimer’s pathology, and this advantage can effectively mask cognitive decline for years. Visual memory tests, by contrast, carry no such sex bias, and may prove more sensitive to the kind of subtle, early changes that sleep disruption and genetic vulnerability together produce in older women.

A Vicious Cycle, Getting Vicious Earlier

The biology behind a sleep-Alzheimer’s connection has been building for some years now. Tau deposition appears to begin, perhaps decades before any symptoms, in subcortical structures that regulate sleep and wakefulness, including regions of the brainstem and basal forebrain. Once tau starts accumulating there, it seems to degrade sleep architecture, which in turn seems to promote further tau spread. It is a feedback loop that starts running long before a patient notices anything wrong. Animal studies have added a wrinkle specific to APOE epsilon-4: that variant appears to accelerate tau aggregation in precisely those sleep-regulating regions, and when combined with disrupted sleep in mouse models, it synergistically speeds up the deposition of amyloid plaques as well. Good sleep consolidation, on the other hand, has in at least one human study been shown to partially offset the Alzheimer’s risk associated with epsilon-4 carriership, even in people who carry the gene.

One genuinely counterintuitive result in the WITS data: the lower-genetic-risk women actually reported worse subjective sleep quality overall than the higher-risk group. The researchers raise the possibility that women with more advanced underlying pathology may be losing their accurate sense of how poorly they are sleeping, a phenomenon sometimes linked to the anosognosia, or self-awareness deficits, seen in early cognitive decline. Whether that interpretation holds up will require objective sleep measures alongside the questionnaire data.

Sleep as a Screening Tool

There is a practical argument embedded in these findings that the authors are at pains to make. The Pittsburgh Sleep Quality Index costs nothing much to administer. It is a questionnaire, not a brain scan. If sleep complaints reliably signal elevated Alzheimer’s risk in women who also carry high genetic burden, then something as simple as asking older women about their sleep, routinely, clinically, without necessarily ordering an expensive PET scan, could help identify who warrants closer monitoring. Women already account for roughly two thirds of all Alzheimer’s cases globally, and they are more likely than men to report sleep problems while also more likely to have those problems undertreated or dismissed.

The study has real limitations worth noting. The sample is small, skewed toward White, highly educated, non-Hispanic women, and the PHS itself was developed primarily in European cohorts, which limits how far these results generalise. The sleep data are self-reported, not actigraphy or polysomnography. And because the higher-risk PHS group overlapped almost entirely with APOE epsilon-4 carriership, the study cannot yet cleanly say whether the broad polygenic score adds anything beyond what epsilon-4 status alone would have captured.

Still, the signal is there. Improving sleep is, at least in theory, modifiable in a way that genetic risk is not, and the idea that better sleep hygiene might attenuate some of the neurological damage associated with high Alzheimer’s risk is not implausible. Whether that translates into a genuine prevention strategy for women in their 60s and 70s who carry epsilon-4 and report waking at three in the morning, staring at the ceiling, wondering if it means anything, is exactly the kind of question these researchers are now in a position to pursue.

https://doi.org/10.1016/j.tjpad.2026.100581

Frequently Asked Questions

Why does sleep affect Alzheimer’s risk at all?

Tau, one of the proteins that forms the tangles characteristic of Alzheimer’s disease, begins accumulating decades before symptoms appear in brain regions that regulate sleep and wakefulness. Once tau builds up in those areas, it disrupts sleep architecture, and disrupted sleep appears to promote further tau spread elsewhere in the brain. It is a self-reinforcing cycle, and there is evidence from both animal studies and human imaging that good sleep may partially interrupt it, even in people with elevated genetic risk.

Is the APOE epsilon-4 gene the same as having high genetic risk?

Not exactly. APOE epsilon-4 is the single most studied genetic risk factor for Alzheimer’s and contributes substantially to disease risk, but it is one gene among many. The polygenic hazard score used in this study incorporates nearly 200,000 genetic variants and is designed to predict age of Alzheimer’s onset more precisely than APOE status alone. In practice, in the WITS sample, the two overlapped almost entirely, which means the study cannot yet fully disentangle their separate contributions.

Why were visual and verbal memory affected differently?

Women tend to outperform men on verbal memory tasks, even well into the Alzheimer’s disease process, because of a cognitive advantage that appears to buffer the effects of brain pathology on word recall. Visual memory carries no equivalent sex-linked advantage, which means it may be a more sensitive marker for detecting early Alzheimer’s-related changes in women, including changes associated with poor sleep. The researchers suggest visual memory deserves more attention in studies focused specifically on female patients.

Could improving sleep actually prevent Alzheimer’s in high-risk women?

That remains an open question, but the possibility is not far-fetched. At least one previous study found that people who carried the APOE epsilon-4 variant but maintained good sleep consolidation had Alzheimer’s outcomes more similar to non-carriers than to other epsilon-4 carriers with poor sleep. Whether a deliberate sleep intervention, whether through behavioural therapy, treatment of sleep apnoea, or other means, could meaningfully delay cognitive decline in women at high genetic risk is something researchers have not yet tested directly.

What makes this study different from previous sleep and Alzheimer’s research?

Most prior work has used objective sleep measures like actigraphy or polysomnography, and has not specifically focused on women or used the polygenic hazard score as a measure of genetic risk. This study is unusual in using brain tau PET imaging alongside a simple sleep questionnaire in a sample enriched for high-risk older women, which allowed the researchers to ask whether genetic risk modifies the sleep-tau relationship rather than treating genetic risk as noise to be controlled for.