Midlife cumulative deficit frailty predicts Alzheimer's disease-related plasma biomarkers in older adults

Abstract

Background: The study explores whether frailty at midlife predicts mortality and levels of biomarkers associated with Alzheimer's disease and related dementias (ADRD) and neurodegeneration by early old age. We also examine the heritability of frailty across this age period.

Methods: Participants were 1,286 community-dwelling men from the Vietnam Era Twin Study of Aging at average ages 56, 62 and 68, all without ADRD at baseline. The cumulative deficit frailty index (FI) comprised 37 items assessing multiple physiological systems. Plasma biomarkers at age 68 included beta-amyloid (Aβ40, Aβ42), total tau (t-tau) and neurofilament light chain (NfL).

Results: Being frail doubled the risk of all-cause mortality by age 68 (OR = 2.44). Age 56 FI significantly predicted age 68 NfL (P = 0.014), Aβ40 (P = 0.001) and Aβ42 (P = 0.023), but not t-tau. Age 62 FI predicted all biomarkers at age 68: NfL (P = 0.023), Aβ40 (P = 0.002), Aβ42 (P = 0.001) and t-tau (P = 0.001). Age 68 FI scores were associated with age 68 levels of NfL (P = 0.027), Aβ40 (P < 0.001), Aβ42 (P = 0.001) and t-tau (P = 0.003). Genetic influences accounted for 45-48% of the variance in frailty and significantly contributed to its stability across 11 years.

Conclusions: Frailty during one's 50s doubled the risk of mortality by age 68. A mechanism linking frailty and ADRD may be through its associations with biomarkers related to neurodegeneration. Cumulative deficit frailty increases with age but remains moderately heritable across the age range studied. With environmental factors accounting for about half of its variance, early interventions aimed at reducing frailty may help to reduce risk for ADRD.

Keywords: cumulative deficit frailty; neurofilament light chain; older people; plasma beta amyloid; plasma biomarkers; plasma t-tau.

Figure 1

Autoregressive best-fitting model. Frail = cumulative deficit FI, cube-root transformed. A1–A4 estimate the cumulative additive genetic influences at each age group; E1–E4 estimate the cumulative unshared environmental influences at each age group; A_i1 estimates the additive influences at age 51–55, whereas A_i2–A_i4 represent the magnitude of new genetic information at ages 56–60, 61–65 and 66–73; E_i1 estimates the magnitude of unshared environmental influences at age 51–55; E_i2–E_i4 represent unshared environmental influences at ages 56–60, 61–65 and 66–73; res represents residuals or unexplained variance. Estimates on the directional paths between A1 and A4 and E1 and E4 represent βs for autoregressive relationships. Residuals and betas for all age groups were set to be equal without significant loss in model fit (Ps > 0.05). β (e.g. 0.93) for A and E pathways was set to be equal. Bolded numbers indicate statistical significance at 95% CI.

Figure 2

Cumulative deficit FI at age 56 and AD-related plasma biomarkers at age 68. FI = Frailty Index. FI was cube-root transformed; NfL and t-tau were log-transformed. Values for NfL and t-tau were log-transformed first to address skewness. Biomarkers were then residualized for site and storage time. Finally, both FI and plasma biomarker measures were z-scored. NfL = Neurofilament light chain; t-tau = total tau; Aβ40 = beta amyloid 40; Aβ42 = beta amyloid 42. Biomarker measures are z-scored for comparability; descriptive statistics for biomarker measures provided in Supplementary Appendix 16 and boxplots are shown in Supplementary Appendix 17. Higher FI scores are associated with higher biomarker levels. FI scores at age 56 significantly predicted age 68 plasma NfL (P = 0.014), Aβ40 (P = 0.001) and Aβ42 (P = 0.023), but not t-tau.