Researchers have found that noninvasive measures of retinal health at midlife can identify patients who are at elevated risk for Alzheimer’s disease and related dementias.
They used data from the Dunedin Multidisciplinary Health and Development Study—a longitudinal birth cohort of 1,037 patients who were born in New Zealand between 1972 and 1973—for their analysis in the Journal of Alzheimer’s Disease. Among the 997 surviving participants, 938 (50.5% male) were assessed at age 45 years. Retention from the original cohort was high (94.1%), and retinal health was evaluated using optical coherence tomography (OCT) and digital fundus photography. Neuronal measures included retinal nerve fiber layer (RNFL) and ganglion cell–inner plexiform layer (GC-IPL) thickness. Microvascular measures included central retinal artery and vein equivalents, which indicated arteriole and venule width.
The researchers examined the association between these retinal measures and five validated Alzheimer’s disease and related dementias (ADRD) risk indexes:
- Cardiovascular Risk Factors, Aging, and Incidence of Dementia (CAIDE)
- Lifestyle for Brain Health (LIBRA)
- Modifiable dementia risks (Lancet)
- Australian National University Alzheimer’s Disease Risk Index (ANU-ADRI)
- A comprehensive 48-factor index developed for this cohort (DunedinARB)
They found that microvascular measures were most strongly associated with ADRD risk. Narrower arterioles and wider venules were significantly associated with higher scores across all five ADRD risk indexes. For instance:
- A 10 µm decrease in arteriole width corresponded to a 0.16 to 0.21 standard deviation increase in risk scores (eg, CAIDE beta = −0.21, P < .001).
- A 10 µm increase in venule width was associated with a 0.15 to 0.31 standard deviation increase in risk scores (eg, DunedinARB beta = 0.31, P < .001).
Neuronal measures showed weaker associations. For example, thinner RNFL was modestly associated with higher ADRD risk on some indexes (eg, CAIDE beta = −0.08, P = .02), but GC-IPL thickness was not significantly associated with risk after correction for multiple comparisons.
Retinal microvasculature also captured diverse ADRD risk domains beyond cardiometabolic factors. Notably, wider venules were associated with lifestyle, socioeconomic, psychosomatic, physical/sensory, inflammatory, and subjective health risk domains; whereas narrower arterioles were associated with cardiometabolic, lifestyle, and physical/sensory risks.
Finally, retinal neuronal measures primarily captured cardiometabolic risk, as evidenced by RNFL and GC-IPL which were not associated with other domains such as genetics, harmful events, or psychological risk.
The researchers acknowledged limitations including limited generalizability, measurement constraints, and potential confounders. The sample was predominantly New Zealand European/Pākehā (White), and the findings may not apply to other ethnic or racial populations. They suggested that the results should be replicated in other settings. Because the analysis used validated risk indices in younger participants who did not have dementia diagnoses, actual disease development might not have been captured fully.
“Our risk index outcome measures are highly predictive of the likelihood of dementia decades later but are not themselves direct measures of dementia pathology or endpoint disease,” wrote lead study author Ashleigh Barrett-Young, of the Department of Psychology at the University of Otago in Dunedin, New Zealand, and colleagues. They suggested future research could also focus on patients with specific cardiometabolic comorbidities.
“Digital fundus imagery is an accessible, scalable modality that could provide insights into ADRD disease processes alongside other clinical data, potentially enhancing the ability to predict ADRD risk at the population level long before disease end-points emerge,” the study authors concluded.
No conflicts of interest were disclosed.