Does the Alzheimer’s gene still predict dementia in your nineties?
No. In 805 people aged 90 and older, the APOE e4 gene, the strongest known genetic risk factor for dementia at younger ages, was not significantly linked to dementia risk after 90. Sex and race were, and so was a different version of the same gene.
People who live past 90 are the group we know the least about, since little has been established about dementia incidence and its risk factors after that age. Most of what doctors tell patients about dementia risk comes from studies of people in their sixties and seventies. This study, called LifeAfter90, was built to ask whether those rules still hold in the tenth decade of life.
What the data show
Over an average of 2 years of follow-up, 138 of the 805 participants, or 17%, were diagnosed with dementia. The age-standardised rate came to 116.82 new cases per 1,000 person-years, which works out to roughly 117 new diagnoses each year for every 1,000 people in this age range. The researchers can be fairly confident the true rate sits somewhere between about 94 and 140 per 1,000 person-years (95% CI 93.69 to 139.96). During the same stretch, 295 participants (37%) died, which is why the researchers used a statistical method that treats death as a competing outcome.
The gene findings are the surprise. Carriers of APOE e2 had about 61% lower dementia risk than non-carriers (subdistribution hazard ratio 0.39), very likely somewhere between 12% and 83% lower (95% CI 0.17 to 0.88). APOE e4 carriers, by contrast, showed about 51% higher risk (sHR 1.51), but the range ran from 8% lower to 147% higher (95% CI 0.92 to 2.47), so the study could not confirm any real effect. Dementia rates for each gene group were not published, so these are relative figures only.
Dr. Kumar’s Take
I find this study clarifying rather than alarming. APOE e4 has become shorthand for “you are headed for Alzheimer’s,” and patients bring me consumer gene test results with that fear attached. What this data suggests is that e4 does much of its damage earlier in life. If you reach 90 without dementia, that particular loaded card may already have been played. The protective signal from e2 held up, which is consistent with what we see at younger ages.
I would be careful not to overread it. Only 413 of the 805 participants had gene data, and the follow-up averaged 2 years. A weak effect can hide in a sample that size. The authors themselves note the e4 link might differ by sex, which is a thread worth pulling.
Who faced the highest risk
The clearest divides were not genetic. Women had about 89% higher dementia risk than men (sHR 1.89), very likely between 30% and 176% higher (95% CI 1.30 to 2.76). Black participants had about 75% higher risk than Asian participants (sHR 1.75), likely between 7% and 188% higher (95% CI 1.07 to 2.88). That matters because the same ethnoracial gaps show up in studies of people in their seventies, and there had been reasonable hope that they narrowed among the very oldest. They did not. Education level showed no significant difference in this group.
Limits worth knowing
This is an observational cohort, not a trial, so it can show who develops dementia but not why. Everyone was a Kaiser Permanente member in the San Francisco Bay Area or Sacramento, which makes the group unusually diverse but still regional, and all participants had ongoing health coverage. The team excluded 96 people who already had dementia and 219 who came in for only one visit, so this reflects people healthy enough at 90 to stay in a study. Two years of average follow-up is short for a disease that develops over decades.
Practical Takeaways
- If you carry APOE e4 and have reached your nineties without cognitive symptoms, this data suggests that gene is a weaker predictor for you than it would have been at 65, and it is worth discussing with your doctor rather than assuming the worst.
- Dementia screening should not stop at 85 or 90, since roughly 1 in 6 participants in this study developed dementia across just 2 years of follow-up.
- Families caring for a woman in her nineties should know that risk in this cohort was nearly double that of men the same age, so subtle changes in memory or daily function deserve a real evaluation, not a shrug about age.
Related Studies and Research
- The shingles vaccine may cut dementia risk by a third in older adults
- Lifestyle changes protected the brain, but only in adults under 70
- Coffee reduces mortality risk: prospective cohort evidence
- Biodiversity and drug discovery: why we’re losing one important medicine every two years
FAQs
Should someone over 90 bother getting an APOE gene test?
This study gives little reason to. The e4 variant, which is the reason most people order the test, did not show a confirmed link to dementia in participants past 90. A test result would not change screening, since everyone in this age range warrants attention to memory and daily function regardless of genotype. There is also no treatment that is selected on the basis of APOE status in this age group. If a family wants genetic information for younger relatives, that is a different conversation with a different set of tradeoffs.
How was dementia actually diagnosed in this study?
The researchers did not rely on a single test or a chart code. Diagnosis came from a combination of physician assessment, the Clinical Dementia Rating, and a Functional Activities Questionnaire, which asks an informant how well the person manages tasks like finances and medications. Participants were evaluated every 6 months from July 2018 through November 2024, either in person or remotely. That repeated, structured approach is stronger than the insurance-claims data many large dementia studies depend on, and it is a real strength here.
Why does it matter that so many participants died during the study?
When you follow people in their nineties, death competes with dementia as an outcome. Someone who dies at 93 never gets the chance to develop dementia at 95. In this cohort 295 participants (37%) died over the follow-up period, more than double the number diagnosed with dementia. Ordinary survival analysis can overstate risk in that situation, so the researchers used Fine-Gray competing-risk models that treat death as its own outcome. That is why the numbers here are reported as subdistribution hazard ratios rather than plain hazard ratios.
Bottom Line
Among 805 racially and ethnically diverse adults aged 90 and older, 17% developed dementia over an average of 2 years, an age-standardised rate of about 117 cases per 1,000 person-years. The genetic rules appear to shift at this age: APOE e2 carriers had about 61% lower risk, while the much-discussed e4 variant showed no confirmed association. What did track with risk were sex and race, with women at about 89% higher risk than men and Black participants at about 75% higher risk than Asian participants. Reaching 90 does not put you past the reach of dementia, and screening should continue.
