Geriatrics & Gerontology International · Nov 6, 2026
Circulating Insulin‐Like Growth Factor‐1, Dementia, and Cognitive Subdomains: Evidence From the UK Biobank
By Gao T, Li Y, Guo Y et al.
Aim: To examine nonlinear associations of circulating insulin‐like growth factor‐1 (IGF‐1) with dementia risk and longitudinal changes in cognitive subdomains in the UK Biobank. Methods: We analyzed 66 535 participants for all‐cause dementia and dementia subtypes and a nested cohort of 12 582 participants with repeated cognitive assessments. Cox proportional hazards models and restricted cubic splines assessed dementia risk, while linear mixed‐effects models evaluated longitudinal changes in global cognition, fluid intelligence, paired matching, reaction time, and prospective memory. Results: In the fully adjusted linear Cox model, a 1‐SD increase in IGF‐1 was not linearly associated with all‐cause dementia (HR = 1.01, 95% CI: 1.00–1.01; p = 0.290). RCS analysis yielded p = 0.060 for the overall association and p = 0.048 for nonlinearity. AD risk was elevated only in Q3 (HR = 1.52, 95% CI: 1.08–2.13), and VaD risk was higher in Q5 (HR = 1.93, 95% CI: 1.01–3.69); nonlinear tests for AD, VaD, and FTD were not significant. For the other‐dementia category, the nonlinearity test was nominally significant ( p = 0.048), but continuous and quintile analyses were not. Paired matching showed an IGF‐1‐by‐time interaction ( p = 0.008), with a nominally significant test for nonlinearity ( p = 0.049). Other cognitive domains showed no significant nonlinear longitudinal associations. Conclusions: Overall, circulating IGF‐1 showed no clear association with all‐cause dementia risk, while exploratory analyses suggested possible nonlinear patterns and a domain‐specific longitudinal signal in paired matching. These observational findings are hypothesis‐generating and do not establish a clinical target range.