Amyloid and tau PET-positive cognitively unimpaired individuals are at high risk for future cognitive decline

Abstract

A major unanswered question in the dementia field is whether cognitively unimpaired individuals who harbor both Alzheimer's disease neuropathological hallmarks (that is, amyloid-β plaques and tau neurofibrillary tangles) can preserve their cognition over time or are destined to decline. In this large multicenter amyloid and tau positron emission tomography (PET) study (n = 1,325), we examined the risk for future progression to mild cognitive impairment and the rate of cognitive decline over time among cognitively unimpaired individuals who were amyloid PET-positive (A⁺) and tau PET-positive (T⁺) in the medial temporal lobe (A⁺T_MTL⁺) and/or in the temporal neocortex (A⁺T_NEO-T⁺) and compared them with A⁺T^- and A^-T^- groups. Cox proportional-hazards models showed a substantially increased risk for progression to mild cognitive impairment in the A⁺T_NEO-T⁺ (hazard ratio (HR) = 19.2, 95% confidence interval (CI) = 10.9-33.7), A⁺T_MTL⁺ (HR = 14.6, 95% CI = 8.1-26.4) and A⁺T^- (HR = 2.4, 95% CI = 1.4-4.3) groups versus the A^-T^- (reference) group. Both A⁺T_MTL⁺ (HR = 6.0, 95% CI = 3.4-10.6) and A⁺T_NEO-T⁺ (HR = 7.9, 95% CI = 4.7-13.5) groups also showed faster clinical progression to mild cognitive impairment than the A⁺T^- group. Linear mixed-effect models indicated that the A⁺T_NEO-T⁺ (β = -0.056 ± 0.005, T = -11.55, P < 0.001), A⁺T_MTL⁺ (β = -0.024 ± 0.005, T = -4.72, P < 0.001) and A⁺T^- (β = -0.008 ± 0.002, T = -3.46, P < 0.001) groups showed significantly faster longitudinal global cognitive decline compared to the A^-T^- (reference) group (all P < 0.001). Both A⁺T_NEO-T⁺ (P < 0.001) and A⁺T_MTL⁺ (P = 0.002) groups also progressed faster than the A⁺T^- group. In summary, evidence of advanced Alzheimer's disease pathological changes provided by a combination of abnormal amyloid and tau PET examinations is strongly associated with short-term (that is, 3-5 years) cognitive decline in cognitively unimpaired individuals and is therefore of high clinical relevance.

Fig. 1. NIA-AA versus IWG criteria.

Differences in the nomenclature of cognitively unimpaired individuals with (⁺) or without (−) in vivo biomarker evidence of Aβ (A) and tau (T) pathology in the NIA-AA versus IWG criteria for AD. Note that for the IWG criteria, the presumed ‘risk for progression’ level rises when both A and T biomarkers are positive.

Fig. 2. Progression to MCI or all-cause dementia in the different AT biomarker profiles.

a,d, Survival curves for progression to MCI (a) or all-cause dementia (d) in the different AT biomarker profiles (A⁻T⁺: n = 292; A⁺T_MTL⁺: n = 51; A⁺T_NEO-T⁺: n = 60) with the A⁻T⁻ group (n = 781) as the reference, including a table of total number of participants available at each time point. The dashed line in a indicates the time point at which 50% of a group had progressed to MCI. b,e, Forest plots showing the HRs and 95% CIs derived from the survival analyses shown in a (b) and d(e), from Cox regression models including age, sex, education and cohort as covariates. c,f, Forest plots showing the HRs and 95% CIs derived from Cox regression models including age, sex, education and cohort as covariates but now using the A⁺T⁻ group as the reference with the outcome being progression to MCI (c) and progression to all-cause dementia (f). Statistics from two-sided tests without adjustment for multiple comparisons are reported. *P < 0.01, **P < 0.001.

Fig. 3. Longitudinal cognitive decline in the different AT biomarker profiles.

a–e, Cognitive trajectories of mPACC5 (a), MMSE (b) and subcomponents of the mPACC5 including delayed episodic memory (c), timed executive function (d) and semantic memory (e) over time in the different AT biomarker profiles. The average regression line for each group was plotted from linear mixed-effect models including age, sex, education and cohort as covariates. Data are anchored to the tau PET visit (time 0); cognitive data up to 1 year before PET was included. The error bands correspond to the 95% CI.

Extended Data Fig. 1. Main analyses from Figs. 2 and 3, but now also including the A-T + group.

This figure resembles parts of Figs. 2 and 3 of the main manuscript but now also includes the A-T + group. a, Survival curves in relation to progression to MCI in the different AT categories, with the table of total number of participants available at each time point. b, Survival curves in relation to progression to all-cause dementia in the different AT categories, with the table of total number of participants available at each time point. c, d Forest plots showing the hazard ratios and 95% confidence intervals from the survival analyses shown in a and b, from Cox regression models including age, sex, education, and cohort as covariates. Statistical tests were two-sided without adjustment for multiple comparisons. e, Cognitive trajectories of mPACC5 scores over time in the different AT categories f, Cognitive trajectories of MMSE scores over time in the different AT categories. The average regression line for each group was plotted from linear mixed effect models including age, sex, education, and cohort as covariates. The error bands correspond to the 95% confidence interval. Data are anchored to the tau-PET visit (Time 0), and cognitive data up to 1 year prior to PET was included. *p = 0.01, **p < 0.001.

Extended Data Fig. 2. Progression to AD dementia in the different AT categories.

a, Survival curves in relation to progression to AD dementia in the main AT categories, with the table of total number of participants available at each time point. b, Survival curves in relation to progression to AD dementia when including the A^-T ⁺ group, with the table of total number of participants available at each time point. Given the small number of events, the hazard ratios are difficult to interpret, but the A + T-, A + T_MTL + and A + T_NEO-T + groups had higher HR’s compared to A-T- reference group (all p < 0.001). Pairwise comparisons indicated that both the A + T_NEO-T + (p < 0.001) and A + T_MTL + (p = 0.008) groups differed from the A + T- group, and the A + T_NEO-T + group differed from the A + T_MTL + group (p = 0.01).

Extended Data Fig. 3. Cognitive decline on mPACC5 and MMSE in the individual cohorts.

a, Cognitive trajectories of mPACC5 scores over time in the different AT categories in each individual cohort. b, Cognitive trajectories of MMSE scores over time in the different AT categories in each individual cohort. The average regression line for each group was plotted from linear mixed effect models including age, sex, and education as covariates. The error bands correspond to the 95% confidence interval. Data are anchored to the tau-PET visit (Time 0), and cognitive data up to 1 year prior to PET was included.