Abstract
BACKGROUND
Compressing the duration of cognitive impairment is critical to preserve quality of life until the end. To what extent cognitive decline is compressed and cognitive resilience increases with extreme longevity is not well understood.
METHODS
We used data from 13,999 deceased participants from the National Alzheimer's Coordinating Center cohort, including 8,146 with neuropathological data. Cognitive function was assessed annually (median follow‐up: 4.9 years). We evaluated cognitive trajectories before death and cognitive resilience (defined as high neuropathological burden without dementia) across lifespan groups (ages 50–100+ years).
RESULTS
Participants with longer lifespans, particularly centenarians, exhibited slower cognitive decline and shorter periods of cognitive impairment before death, although distinct cognitive trajectories existed among centenarians. Cognitive resilience also increased with longer lifespans, but associated factors varied. Apolipoprotein E ε 2 was associated with higher cognitive resilience only in centenarians.
CONCLUSION
Our findings support a general compression of cognitive decline and increased cognitive resilience in extreme longevity.
Highlights
Individuals with longer lifespans, especially centenarians, generally exhibited better cognitive function and slower cognitive decline toward the end of life, suggesting a compression of cognitive decline in extreme longevity.
Although there was a compression of cognitive decline at the group level among centenarians, heterogeneous cognitive trajectories before death were observed across individuals.
The relationship between neuropathological burden and dementia risk attenuated with longer lifespans, indicating greater cognitive resilience in individuals with extreme longevity.
The associations of both genetic and modifiable factors with cognitive resilience varied by lifespan.
Keywords: centenarians, cognitive decline, cognitive resilience, dementia, longevity, neuropathology
1. BACKGROUND
The “compression of morbidity” hypothesis suggests that the period of disability and morbidity before death can be shortened as lifespan increases, especially as it approaches its natural limit. 1 Cognitive impairment, a key component affecting healthy aging, contributes substantially to disability and poor quality of life in older adults. 2 With an increasing number of older adults living with dementia worldwide, 3 reducing or compressing the period of cognitive impairment and improving cognitive resilience are increasingly important to preserve the quality of life until the very end of life. Centenarians, a small group of individuals who live beyond 100 years, often show a delayed onset of chronic diseases and experience a compressed period of morbidity and functional decline, 4 , 5 making them exemplary models of resilient aging.
Previous studies involving centenarians observed a delayed onset of cognitive impairment and dementia with increasing age. 5 , 6 However, because of the cross‐sectional design and short follow‐up period in these studies, the trajectories of cognitive decline in the last decade before death among individuals with varied lifespans, including those with extreme longevity, have not been well characterized. The underlying neuropathological changes, which are critical to understanding cognitive resilience, also remain largely unknown. Cognitive resilience could be a central mechanism underlying the compression of cognitive decline with aging. Emerging evidence suggests that individuals who reach extreme old age may be more resilient to cognitive decline despite substantial neuropathological burden, such as Alzheimer's disease (AD) or cerebrovascular pathologies. 7 , 8 However, factors associated with this resilience remain poorly understood and may vary between individuals of different lifespans.
In the present study, we used longitudinal data from the National Alzheimer's Coordinating Center (NACC) to investigate cognitive trajectories before death and cognitive resilience to neuropathology among adults with lifespans ranging from 50 to over 100 years. We hypothesized that cognitive decline before death is compressed in individuals with longer lifespans, especially centenarians, possibly due to their greater resilience to neuropathology and certain risk factors.
2. METHODS
2.1. Data source
Data for this study were obtained from NACC, which collects standardized clinical and neuropathological data from Alzheimer's Disease Research Centers (ADRCs) across the United States, funded by the National Institute on Aging (NIA). The NACC database is publicly available upon request (https://naccdata.org/).
2.2. Study population
This study used longitudinal data from the NACC Uniform Data Set (UDS) and the NACC Neuropathology Data Set, collected from 46 current and former ADRCs between June 2005 and December 2024. Detailed descriptions of data collection and harmonization procedures for the UDS have been previously published. 9 , 10
RESEARCH IN CONTEXT
Systematic review: We searched PubMed and reviewed previous studies examining the compression of cognitive decline and cognitive resilience with respect to extreme longevity. Most research used cross‐sectional data and primarily focused on the oldest‐old population. We did not identify published studies assessing cognitive trajectories before death across a broad lifespan spectrum, ranging from premature death in midlife to over 100. We also did not find studies characterizing cognitive resilience stratified by lifespan.
Interpretation: Individuals who had a longer lifespan, particularly centenarians, exhibited both higher levels of preserved cognitive function and slower terminal decline, suggesting a compression of cognitive impairment with longevity. The relationship between neuropathological burden and dementia risk attenuated and cognitive resilience to neuropathology increased substantially with longer lifespans, indicating that cognitive resilience may be more common in individuals with extreme longevity. We also observed the associations of both genetic and modifiable factors with cognitive resilience varied by lifespan, which may reflect the depletion of susceptible individuals in extreme longevity.
Future directions: Our findings support the hypothesis of the compression of cognitive morbidity in late life and highlight the importance of cognitive resilience in achieving healthy aging. Intervention strategies that improve cognitive resilience might help extend lifespan with healthy cognitive function.
We included participants who were aged ≥ 50 years at their initial visit and died before December 2024. A total of 13,999 deceased participants met these criteria and were included in our study. Of these, 8,146 underwent brain autopsy and had neuropathological data available. Figure S1 in Supporting Information shows a detailed flowchart of participant selection. Written informed consent was obtained from all participants at each ADRC and approved by the ADRC institutional review boards. Autopsy permission was obtained according to applicable state laws.
2.3. Cognitive assessment
Cognitive function was assessed approximately annually using the Clinical Dementia Rating (CDR) Dementia Staging Instrument, which consists of six cognitive or functional domains: memory, orientation, judgment and problem solving, community affairs, home and hobbies, and personal care. Each domain is rated with a score ranging from 0 (normal) to 3 (severe). The CDR Sum of Boxes (CDR‐SB) score, ranging from 0 to 18, is calculated by summing the rating scores of the six domains and was used as the primary measure of global cognitive function in our study. 11 Cognitive performance was additionally assessed using the Mini‐Mental State Examination (MMSE).
Cognitive status at each visit was classified as normal cognition, impaired but not mild cognitive impairment (MCI), MCI, or dementia by either a consensus team or a physician according to a standardized protocol. 9 Because only a few participants (n = 391, 2.8%) were classified as impaired but not MCI at initial visit, we combined this category with the MCI category for analysis. Dementia severity was further classified using CDR‐SB scores: mild dementia (CDR‐SB ≥ 4.5 and < 9.5) and moderate to severe dementia (CDR‐SB ≥ 9.5). 12
2.4. Neuropathology assessment
Autopsy‐based neuropathology was assessed by trained neuropathologists using the Neuropathology Form and Coding Guidebook, 13 with details described on the NACC website (https://naccdata.org/data‐collection/forms‐documentation/np‐10). In this study, we primarily assessed AD and cerebrovascular pathologies that were harmonized across ADRCs.
For AD pathologies, we assessed amyloid beta (Aβ) plaques using Thal phase (A score), neurofibrillary tau tangles using Braak stage (B score), density of neuritic plaques according to Consortium to Establish a Registry for Alzheimer's Disease (CERAD) score (C score), and density of diffuse plaques according to CERAD semi‐quantitative score. To evaluate the overall AD pathological burden, we first dichotomized each of the AD pathologies based on the National Institute on Aging–Alzheimer's Association guidelines, 14 and their associations with dementia: Thal phases 3 to 5, Braak stages III to VI, frequent neuritic plaques, and frequent diffuse plaques. We then summed the positive dichotomies as a composite AD pathological score (range from 0–4). A composite AD score ≥ 3 (out of 4) was classified as high AD pathology. We also included AD neuropathologic change (ADNC, with 40% missing data) as a composite measure of AD pathologies in secondary analysis.
For cerebrovascular pathologies, we followed the NACC neuropathology guidelines, which recommend the use of a qualitative and subjective grading system for overall severity rather than for specific regions. 15 We included severity of atherosclerosis in the circle of Willis, arteriolosclerosis, cerebral amyloid angiopathy (CAA), and white matter rarefaction (WMR) graded as none, mild, moderate, or severe using standardized protocols across sites by trained neuropathologists. Details of these assessments are provided in Table S1 in Supporting Information. We also assessed the presence of microinfarcts (in cortical and subcortical regions), infarcts and lacunes (a harmonized composite of large cerebral artery infarcts, small artery ischemic and hemorrhagic lesions, and infarcts visible on gross examinations), hemorrhages, and microbleeds. To evaluate the overall cerebrovascular pathological burden, we summed the number of the presence of these pathologies as a composite measure of cerebrovascular pathology. A composite score ≥ 4 (out of 7) was defined as high cerebrovascular burden.
Among these neuropathological assessments included in the current study, the Thal phase for Aβ plaques and white matter lesions were assessed since 2014 and the data before 2014 were imputed based on other available neuropathological measurements. We defined cognitive resilience as having a high AD pathological burden (≥ 3 of 4 AD pathologies) or a high cerebrovascular pathological burden (≥ 4 of 7 cerebrovascular pathologies) without dementia before death. 16 , 17
2.5. Covariate assessment
Age at death and age at each visit were calculated using the date of birth, visit date, and recorded date of death (year and month). The years before death at each visit was used throughout longitudinal analyses.
Data on demographic characteristics (sex, race, and education), apolipoprotein E (APOE) genotype, smoking habits, history of alcohol abuse, body mass index (BMI, kg/m2) status (normal weight: BMI < 25, overweight: BMI 25–29.9, and obese: BMI ≥ 30), living independency, hearing and vision function, depression symptoms (assessed by geriatric depression scale), medical history (e.g., stroke, hypertension, diabetes), and use of anti‐hypertensive medication (taken within the previous 2 weeks) were collected at initial visit following the NACC‐UDS data collection protocol. 9 APOE genotype was categorized as ε3/ε3; ε2/ε2 and ε2/ε3 (ε2 carriers); and ε4/ε4, ε3/ε4, and ε2/ε4 (ε4 carriers). We included ε2/ε4 in the ε4 carrier group based on prior evidence indicating that the ε2/ε4 genotype is associated with AD pathologies similar to ε3/ε4. 18 Sensitivity analyses excluding ε2/ε4 individuals yielded consistent results. Missing data in covariates (0.4% in race, 1.0% in education, 0.7% in living independency, 2.1% in smoking status, 0.7% in history of alcohol abuse, 0.5% in history of hypertension, 0.4% in history of diabetes, 1.4% in history of hypercholesterolemia, 0.2% in history of cardiovascular disease (CVD), 0.7% in history of stroke, 1.3% in history of transient ischemic attack (TIA), 4.3% in hearing status, 4.6% in vision status, 16% in APOE genotype, and 17% in BMI) were imputed using random forest methods before modeling. 19 For covariates with > 10% missing data (APOE genotype and BMI), we further did a sensitivity analysis by excluding participants with missing values.
2.6. Statistical analyses
2.6.1. Cognitive trajectory analysis
Participants were categorized into five groups based on age at death: 50 to 70, 70 to 80, 80 to 90, 90 to 100 (nonagenarians), and ≥ 100 (centenarians) years old. Considering that cognitive decline may accelerate before death, we modeled cognitive trajectories with years before death as the time scale using a generalized linear mixed regression model. This approach allows for modeling non‐linear trajectories in cognitive function by incorporating the years before death as an independent variable using natural cubic splines. A product term of years before death and the age‐at‐death group was included in the model to allow the varying trajectories of cognitive decline across different age groups. We adjusted for time between initial visit and death, sex, race, and years of education, and included both a random intercept and slope for each participant. The Bayesian information criterion (BIC) was used to determine the optimal degrees of freedom for the splines, with six degrees selected for the final model. Given the non‐negative, bounded, and right‐skewed distribution of CDR‐SB score, we standardized the scores for model fitting, with predicted values back‐transformed for easier interpretation. Based on this model, we estimated the predicted cognitive function for each participant within a decade before death. We further assessed terminal cognitive decline as the rate of decline within the 5 years before death, estimated using a linear regression model for each participant.
In secondary analyses, we additionally examined cognitive trajectories across different lifespans by sex, education level, and APOE genotype. We also conducted the following sensitivity analyses: (1) we stratified the analyses by cognitive status at initial visit (dementia or MCI, normal cognition) to explore the potential influence of baseline cognitive impairment; (2) we further adjusted the calendar years of birth (before 1920, 1920–1929, 1930–1939, 1940–1949, 1950 or later) to account for the potential birth cohort effect.
2.6.2. Cognitive resilience analysis
To compare cognitive resilience to neuropathology across different lifespans, we performed the following analyses stratified by the age‐at‐death groups defined above: (1) we first described the neuropathological burdens and cognitive resilience; and (2) we then evaluated the associations between neuropathological burdens and dementia using logistic regression models, adjusting for age at death, sex, race, years of education, and APOE genotype.
We examined the associations between baseline characteristics and cognitive resilience using multivariate logistic regression models with adjustment for age at initial visit, sex, race, years of education, APOE genotype, and cognitive status at initial visit. Due to the low proportion of cognitive resilience in the age‐at‐death groups of 50 to 70 years (2%) and 70 to 80 years (5%), this analysis was restricted to participants who died at age ≥ 80 years, comparing those who died between ages 80 and 90 to those aged ≥ 90. The centenarian group (aged ≥ 100 years, n = 190) was combined with the 90 to 100 age‐at‐death group in the main analyses for statistical power considerations. We further performed a secondary analysis examining the centenarians separately. We tested the difference in associations across age‐at‐death groups by including an interaction term between the exposure of interest and the age group in the model.
All data analyses were performed using R software (version 4.3.1; R Core Team). We considered a two‐sided p value < 0.05 as significant.
3. RESULTS
Table 1 presents the characteristics of 13,999 participants by lifespan, including 276 centenarians with a mean age at death of 102.3 years (standard deviation: 2.1) –and a median follow‐up duration of 8.5 years (25th‐75th percentiles: 5.0–13.0). Generally, individuals with a longer lifespan were more likely to be women, APOE ε2 carriers, and less likely to be APOE ε4 carriers. Among 276 centenarians, 76.8% were women, and only 15.9% were APOE ε4 carriers. Centenarians tended to have a healthier lifestyle (never smoked and no alcohol abuse), lower prevalence of dementia and metabolic disease (diabetes and hypercholesterolemia), and higher prevalence of non‐fatal cerebrovascular disease (stroke and transient ischemic attack) at the initial visit. The characteristics of the 8,146 participants with autopsy‐based neuropathology data were similar to those of the overall population (Table S2 in Supporting Information).
TABLE 1.
Characteristics of 13,999 NACC participants by age groups at death.
| Age at death, years old | |||||
|---|---|---|---|---|---|
| Characteristics † | 50–70 (N = 2147) | 70–80 (N = 3518) | 80–90 (N = 5161) | 90–100 (N = 2897) | ≥100 (N = 276) |
| Age at initial visit, mean (SD), years | 60.4 (4.6) | 70.7 (4.3) | 79.3 (4.5) | 86.1 (4.8) | 93.5 (5.3) |
| Age at death, mean (SD), years | 64.2 (4.2) | 75.5 (2.8) | 85.0 (2.8) | 93.7 (2.6) | 102.3 (2.1) |
| Years of follow‐up, median (IQR) | 3.5 (1.7, 5.5) | 4.3 (2.2, 6.8) | 5.0 (2.7, 8.1) | 7.2 (4.1, 10.9) | 8.5 (5.0, 13.0) |
| Interval years between last visit and death, median (IQR) | 1.1 (0.5, 2.5) | 1.1 (0.6, 2.5) | 1.1 (0.6, 2.5) | 1.2 (0.6, 2.8) | 1.1 (0.7, 2.6) |
| Female | 915 (42.6%) | 1514 (43.0%) | 2450 (47.5%) | 1697 (58.6%) | 212 (76.8%) |
| White race | 1941/2127 (91.3%) | 3086/3499 (88.2%) | 4481/5149 (87.0%) | 2513/2891 (86.9%) | 243/276 (88.0%) |
| APOE genotypes | |||||
| ε3/ε3 | 849/1767 (48.0%) | 1174/2915 (40.3%) | 1975/4305 (45.9%) | 1419/2530 (56.1%) | 164/245 (66.9%) |
| ε2/ε2 and ε2/ε3 | 135/1767 (7.6%) | 199/2915 (6.8%) | 305/4305 (7.1%) | 301/2530 (11.9%) | 42/245 (17.1%) |
| ε4/ε4, ε2/ε4, ε3/ε4 | 783/1767 (44.3%) | 1542/2915 (52.9%) | 2025/4305 (47.0%) | 810/2530 (32.0%) | 39/245 (15.9%) |
| Education years, mean (SD) | 15.2 (3.0) | 15.3 (3.2) | 15.1 (3.5) | 14.8 (3.6) | 14.7 (3.6) |
| Able to live independently | 472 (22.3%) | 1267 (36.2%) | 2376 (46.3%) | 1750 (60.7%) | 181 (65.6%) |
| Functionally normal hearing | 1883 (91.9%) | 2979 (87.9%) | 4191 (84.8%) | 2267 (82.2%) | 206 (79.8%) |
| Functionally normal vision | 1818 (89.1%) | 3046 (90.7%) | 4495 (91.0%) | 2464 (89.5%) | 217 (84.1%) |
| Suggested depression a | 463/1624 (28.5%) | 613/2926 (21.0%) | 670/4533 (14.8%) | 308/2622 (11.8%) | 28/243 (11.5%) |
| Body mass index, kg/m2, mean (SD) | 27.1 (5.7) | 27.1 (5.2) | 26.4 (4.9) | 25.7 (4.2) | 24.7 (3.4) |
| Normal weight (<25) | 655/1725 (38.0%) | 1116/2922 (38.2%) | 1869/4379 (42.7%) | 1125/2365 (47.6%) | 114/214 (53.3%) |
| Overweight (25‐29.9) | 630/1725 (36.5%) | 1109/2922 (38.0%) | 1644/4379 (37.5%) | 901/2365 (38.1%) | 87/214 (40.7%) |
| Obesity (≥30) | 440/1725 (25.5%) | 697/2922 (23.9%) | 866/4379 (19.8%) | 339/2365 (14.3%) | 13/214 (6.1%) |
| Never smoked | 1255/1992 (63.0%) | 1742/3246 (53.7%) | 2558/4797 (53.3%) | 1627/2753 (59.1%) | 188/263 (71.5%) |
| History of alcohol abuse | 206/2119 (9.7%) | 267/3500 (7.6%) | 298/5129 (5.8%) | 80/2875 (2.8%) | 6 (2.2%) |
| History of hypertension | 1109/1836 (60.4%) | 2271/3054 (74.4%) | 3694/4590 (80.5%) | 2098/2515 (83.4%) | 176/223 (78.9%) |
| History of diabetes | 200/2133 (9.4%) | 526/3508 (15.0%) | 757/5141 (14.7%) | 326/2884 (11.3%) | 10 (3.6%) |
| History of hypercholesterolemia | 864/2106 (41.0%) | 1854/3481 (53.3%) | 2753/5100 (54.0%) | 1333/2846 (46.8%) | 70/273 (25.6%) |
| History of cardiovascular disease | 190/2139 (8.9%) | 702/3511 (20.0%) | 1563/5153 (30.3%) | 895/2891 (31.0%) | 72/276 (26.1%) |
| History of stroke | 48/2133 (2.3%) | 164/3491 (4.7%) | 422/5123 (8.2%) | 231/2881 (8.0%) | 25/275 (9.1%) |
| History of TIA | 31/2131 (1.5%) | 135/3476 (3.9%) | 368/5074 (7.3%) | 277/2856 (9.7%) | 35/275 (12.7%) |
| Cognitive status at initial visit | |||||
| Normal cognition | 124 (5.8%) | 464 (13.2%) | 1201 (23.3%) | 1195 (41.2%) | 165 (59.8%) |
| MCI | 255 (11.9%) | 678 (19.3%) | 1204 (23.3%) | 698 (24.1%) | 51 (18.5%) |
| Dementia | 1768 (82.3%) | 2376 (67.5%) | 2756 (53.4%) | 1004 (34.7%) | 60 (21.7%) |
| Cognitive status before death | |||||
| Normal cognition | 111 (5.2%) | 380 (10.8%) | 757 (14.7%) | 648 (22.4%) | 73 (26.4%) |
| MCI | 154 (7.2%) | 337 (9.6%) | 738 (14.3%) | 533 (18.4%) | 63 (22.8%) |
| Dementia | 1882 (87.7%) | 2801 (79.6%) | 3666 (71.0%) | 1716 (59.2%) | 140 (50.7%) |
| AD as primary or contributing etiology of cognitive impairment before death | 1034 (48.2%) | 1991 (56.6%) | 3538 (68.6%) | 1906 (65.8%) | 176 (63.8%) |
| VD as primary or contributing etiology of cognitive impairment before death | 59 (2.7%) | 200 (5.7%) | 616 (11.9%) | 420 (14.5%) | 38 (13.8%) |
Note: Data are presented as mean (SD) or median (IQR) for continuous variables, and as N (%) for categorical variables if no missing values, or as N/Total (%) indicating the proportion after excluding missing values.
Abbreviations: AD, Alzheimer's disease; APOE, apolipoprotein E; GDS, Geriatric Depression Scale; IQR, interquartile range; MCI, mild cognitive impairment; SD, standard deviation; TIA, transient ischemic attack; VD, vascular dementia.
Characteristics were assessed at initial visit if not specified.
Suggested depression was defined as GDS score ≥ 5.
3.1. Compression of cognitive decline with longevity
We observed a clear compression of cognitive decline with longer lifespans. Figure 1A shows the trajectories of cognitive function, assessed by CDR‐SB, during the decade before death across age‐at‐death groups. On average, participants with longer lifespans maintained better cognitive function throughout the last decade of life, had slower terminal cognitive decline, and spent a shorter period living with dementia compared to those who died at a younger age (Figure 1). There was a significant interaction between age groups and time of cognitive assessment before death (p < 0.001). Specifically, the estimated median rate of terminal increase in CDR‐SB score among centenarians was 0.9 per year, significantly slower than the rates observed in younger age groups (Figure 1B). The estimated years living with dementia before death became shorter with longer lifespans in a dose–response manner. The estimated median period with dementia before death was 1.1 years (1.1% of the lifespan) in centenarians, followed by 2.4 years (2.6% of lifespan) in nonagenarians (Figure 1C). Similar patterns of compression in cognitive decline were observed when cognitive function was assessed using the MMSE (Figure S2 in Supporting Information), supporting the robustness of our findings. In sensitivity analyses stratified by cognitive impairment (a diagnosis of dementia or MCI) at initial visit, the compression of cognitive decline was also observed among individuals with baseline cognitive impairment (Figure S3 in Supporting Information). The cognitive trajectories remained largely unchanged with further adjustment for birth cohort (Figure S4 in Supporting Information). In stratified analyses by sex, we observed significantly more pronounced compression in men compared to women when lifespan exceeds 90 years (p for interaction < 0.01, Figure S5 in Supporting Information). Compression of cognitive decline was also more pronounced among individuals with APOE ε2 alleles and with higher education across different lifespans (Figures S6 and S7 in Supporting Information).
FIGURE 1.
Cognitive trajectories and compression of cognitive decline before death, stratified by age at death. CDR‐SB, Clinical Dementia Rating Sum of Boxes.
Although there was a compression of cognitive impairment at group level, distinct cognitive trajectories were observed across cognitive status before death among centenarians. As shown in Figure 2A, compression in cognitive decline was more pronounced among those with less cognitive impairment. Centenarians with MCI or mild dementia exhibited significantly slower cognitive decline during the last decade of life compared to those with moderate to severe dementia. Meanwhile, 73 centenarians (26%) remained normal cognition throughout their last decade and did not show an obvious terminal decline before death (Figure 2A). Figure 2B further showed the associations between baseline factors and CDR‐SB score assessed before death. The results suggested that older age at initial visit, APOE ε4 carrier status, impaired hearing, and obesity were associated with worse cognitive function in centenarians. In contrast, APOE ε2 carrier status, higher education attainment, and living independently were associated with better cognition before death (Figure 2B).
FIGURE 2.
Variations in cognitive decline among centenarians. APOE, apolipoprotein E; CDR‐SB, Clinical Dementia Rating Sum of Boxes; CI, confidence interval; CVD, cardiovascular disease; GDS, Geriatric Depression Scale; TIA, transient ischemic attack.
3.2. Cognitive resilience to neuropathology with longevity
Cognitive resilience to high neuropathological burden substantially increased with longer lifespans, from 2% among participants who died at 50 to 70 years old to 35% among centenarians (Figure 3A). Participants who lived to a more advanced age had a lower burden of AD pathology but a higher burden of cerebrovascular pathology (Figures S8 and S9 in Supporting Information). Notably, centenarians had a lower proportion of individuals with high burden of both AD and cerebrovascular pathology compared to nonagenarians (26% vs. 31%, Figure S9C). When analyzed separately, the proportion of cognitive resilience to both high AD pathology and high cerebrovascular pathology increased with longer lifespans (Figure S10A,S10C, in Supporting Information). A similar trend was observed when high AD burden was defined as intermediate or high ADNC (Figure S10B). Despite the high neuropathological burden, centenarians exhibited significantly higher resilience to dementia risk, showing attenuated association of both cerebrovascular and AD pathology with the odds of developing dementia, compared to groups with shorter lifespans (Figure 3B).
FIGURE 3.
Cognitive resilience to neuropathology. AD, Alzheimer's disease.
Several factors, including sex, race, and history of cardiovascular disease (CVD), exhibited lifespan‐dependent associations with cognitive resilience (Figure 4). Specifically, among individuals who died between ages 80 and 90, women had greater cognitive resilience compared to men (odds ratio [OR] = 1.20, 95% confidence interval [CI]: 0.93, 1.55), as did non‐White participants (White vs. non‐White, OR = 0.62, 95% CI: 0.38, 1.02) and those without a history of CVD (with vs. without, OR = 0.81, 95% CI: 0.60, 1.08). However, these factors showed opposite associations with cognitive resilience in those who lived beyond age 90 (all p for interaction < 0.01, Figure 4). We also observed that living independently was associated with higher cognitive resilience in both age groups (Figure 4). In secondary analyses, we additionally observed that APOE ε2 was associated with higher cognitive resilience only among centenarians (OR = 2.70, 95% CI: 0.99, 7.39, Figure S11 in Supporting Information).
FIGURE 4.
Associations between baseline factors and cognitive resilience, stratified by lifespan. APOE, apolipoprotein E; CI, confidence interval; CVD, cardiovascular disease; GDS, Geriatric Depression Scale; OR, odds ratio; TIA, transient ischemic attack.
4. DISCUSSION
In this longitudinal study of participants enrolled from the US national ADRCs, we found that older adults who had a longer lifespan, particularly centenarians, maintained better cognitive function and experienced slower cognitive decline toward the end of life, suggesting a compression of cognitive impairment with longer lifespans. Consistently, the associations between neuropathological burden and dementia risk were attenuated with longer lifespans and the proportion of individuals with cognitive resilience to neuropathology increased substantially with longer lifespans. Women, non‐white people, and individuals without CVD had greater cognitive resilience among those with lifespans < 90, but among centenarians, only APOE ε2 was associated with greater cognitive resilience. These findings may support the hypothesis of the compression of morbidity with longer lifespans and that cognitive resilience may play an important role in extreme longevity.
Our study observed that participants with longer lifespans exhibited both higher levels of preserved cognitive function and slower terminal decline, indicating the presence of compressed cognitive impairment with longevity. This aligns with the New England Centenarian Study (NECS), which reported delayed onset of cognitive decline and dementia cross‐sectionally in the oldest‐old, 5 and a previous study that observed a compression of cognitive morbidity in the US older population from 1993 to 2004. 20 Our study extends this evidence by using a large longitudinal cohort with a longer follow‐up, demonstrating that terminal cognitive decline is a common phenomenon across different lifespans, but that its rate and duration vary. While our findings support the overall compression of cognitive decline with increasing lifespan, especially among centenarians, substantial inter‐individual variation in cognitive trajectories should be noted within this exceptionally long‐lived group. Some centenarians maintained intact cognitive function until shortly before death, while those with greater cognitive impairment experienced more rapid decline despite similar lifespans. These findings underscore the importance of continued cognitive assessment and personalized interventions in the oldest‐old population to promote cognitive health. 21
We also observed that individuals with longer lifespans generally had a lower AD pathological burden but a higher cerebrovascular burden. This is in line with previous reports from the 100+ Study showing that cognitively normal centenarians exhibited only moderate AD pathology, 22 , 23 and with another report from the 90+ Study showing that cerebrovascular lesions were common in the oldest‐old population. 24 The relatively lower burden of AD pathology in the oldest‐old may partially reflect selective survival, as individuals with severe AD pathology are less likely to reach extreme old age. 25 Meanwhile, the higher burden of cerebrovascular pathologies, especially atherosclerosis, white matter lesions, and microinfarcts, might reflect the cumulative effects of vascular aging over a longer lifespan. Our data on the distinct cognitive trajectories at the very end of life across different lifespans, together with the underlying neuropathological patterns, provide new insights into cognitive decline before death, especially among individuals at high risk of dementia.
The compression of cognitive impairment with longer lifespans is further evidenced by our finding of increased cognitive resilience, as defined based on post‐mortem neuropathology. We found that both AD and cerebrovascular pathology showed a weakened association with clinical diagnosis of dementia among centenarians compared to younger lifespan groups. Data from the 100+ Study also showed that some centenarians with the highest neuropathological burden maintained high cognitive performance. 23 Our finding is in line with the report in 456 individuals aged 69 to 103 years from the Cognitive Function and Ageing Study (CFAS), showing age‐dependent attenuation of the association between AD pathology and dementia risk. 26 These data together suggest that cognitive resilience may be more common in individuals with extreme longevity.
Moreover, we observed lifespan‐dependent associations of both genetic and modifiable risk factors with cognitive resilience. Female sex and absence of CVD, factors typically associated with better cognition and lower dementia risk, as well as non‐White race, were linked to higher cognitive resilience among individuals with shorter lifespans. However, these associations attenuated or reversed among those living beyond age 90. Notably, APOE ε2, an established longevity allele, 27 was associated with greater cognitive resilience only in centenarians. Additionally, we observed that centenarians with higher education exhibited better cognition and more pronounced compression of cognitive decline before death. Individuals who survive to extreme age represent a highly selective population with reduced susceptibility to major chronic diseases. 4 , 28 It is also plausible that individuals with cognitive resilience share potential protective traits, such as favorable genetic profiles, healthier lifestyles, or greater physiological reserve, which contribute to their ability to avoid or survive cardiovascular and metabolic diseases and reach extreme old age. These mechanisms may help explain the diminished association with traditional risk factors in the oldest‐old. Our results, combined with previous data on cognitively normal centenarians, 29 suggest that cognitive resilience may be a potential target for promoting cognitive health in extreme longevity.
This study has several strengths. First, it is based on a large, well‐characterized longitudinal cohort with standardized annual assessments of cognitive function, allowing for robust modeling of cognitive trajectories up to the end of life, over up to a decade, across a wide lifespan spectrum, including centenarians. Second, the large number of individuals with neuropathology data enabled us to assess neuropathological burden and characterize cognitive resilience across different lifespans. Several limitations should also be noted. First, the NACC cohort is composed of volunteer participants, most of whom are highly educated White participants, and includes individuals at higher risk of dementia than the general population. This may limit the generalizability of our findings, although the proportions of women (77%), APOE ε4 carriers (16%), and dementia prevalence (51%) among centenarians were comparable to those observed in community‐based cohorts. 30 , 31 , 32 Second, the majority of younger participants (aged 50–80 years) in our study were diagnosed with MCI or dementia prior to death, suggesting the cohort may overrepresent older adults at a higher risk of cognitive decline. Third, the inherently cross‐sectional nature of post‐mortem neuropathological data limits the ability to establish causal or temporal relationships between pathological burden and cognitive decline. Reverse causation is also possible in our analysis of baseline predictors of cognitive resilience, as preserved cognitive function may itself reflect lifelong biological or behavioral advantages. Additionally, differential survival bias may have influenced associations; for instance, men reaching extreme old age likely represent a highly selective subgroup with more favorable health profiles than their female peers. Lastly, autopsy‐based studies are susceptible to participation bias.
In this longitudinal study, we demonstrated a substantial compression of cognitive decline with advancing lifespan, particularly among centenarians. This pattern was characterized by better‐preserved cognition, slower terminal decline, and a shorter period living with dementia toward the end of life. In addition, individuals with longer lifespans had greater cognitive resilience to both neuropathology and clinical risk factors. These findings support the hypothesis of the compression of cognitive morbidity and highlight the importance of cognitive resilience in achieving healthy aging. Intervention strategies that improve cognitive resilience might help extend the lifespan with healthy cognitive function.
CONFLICT OF INTEREST STATEMENT
The authors have no conflicts of interest to disclose. Author disclosures are available in the Supporting Information.
CONSENT STATEMENT
Written informed consent was obtained from all participants and co‐participants at each Alzheimer's Disease Research Center (ADRC) and approved by the ADRC institutional review boards.
Supporting information
Supporting Information
Supporting Information
Supporting Information
ACKNOWLEDGMENTS
The NACC database is funded by NIA/NIH Grant U24 AG072122. NACC data are contributed by the NIA‐funded ADRCs: P30 AG062429 (PI James Brewer, MD, PhD), P30 AG066468 (PI Oscar Lopez, MD), P30 AG062421 (PI Bradley Hyman, MD, PhD), P30 AG066509 (PI Thomas Grabowski, MD), P30 AG066514 (PI Mary Sano, PhD), P30 AG066530 (PI Helena Chui, MD), P30 AG066507 (PI Marilyn Albert, PhD), P30 AG066444 (PI John Morris, MD), P30 AG066518 (PI Jef‐ frey Kaye, MD), P30 AG066512 (PI Thomas Wisniewski, MD), P30 AG066462 (PI Scott Small, MD), P30 AG072979 (PI David Wolk, MD), P30 AG072972 (PI Charles DeCarli, MD), P30 AG072976 (PI Andrew Saykin, PsyD), P30 AG072975 (PI David Bennett, MD), P30 AG072978 (PI Neil Kowall, MD), P30 AG072977 (PI Robert Vassar, PhD), P30 AG066519 (PI Frank LaFerla, PhD), P30 AG062677 (PI Ronald Petersen, MD, PhD), P30 AG079280 (PI Eric Reiman, MD), P30 AG062422 (PI Gil Rabinovici, MD), P30 AG066511 (PI Allan Levey, MD, PhD), P30 AG072946 (PI Linda Van Eldik, PhD), P30 AG062715 (PI Sanjay Asthana, MD, FRCP), P30 AG072973 (PI Russell Swerdlow, MD), P30 AG066506 (PI Todd Golde, MD, PhD), P30 AG066508 (PI Stephen Strittmatter, MD, PhD), P30 AG066515 (PI Victor Henderson, MD, MS), P30 AG072947 (PI Suzanne Craft, PhD), P30 AG072931 (PI Henry Paulson, MD, PhD), P30 AG066546 (PI Sudha Seshadri, MD), P20 AG068024 (PI Erik Roberson, MD, PhD), P20 AG068053 (PI Justin Miller, PhD), P20 AG068077 (PI Gary Rosenberg, MD), P20 AG068082 (PI Angela Jefferson, PhD), P30 AG072958 (PI Heather Whitson, MD), P30 AG072959 (PI James Leverenz, MD). The Alzheimer's Disease Genetic Consortium (ADGC) is funded by NIA/NIH Grant U01 AG032984 (PI Gerard D. Schellenberg). This work was supported by grant R00AG071742 to Y. Ma from the National Institute on Aging.
Zhang W, Cai W, Zhang Y, et al. Compression of cognitive decline and cognitive resilience in extreme longevity. Alzheimer's Dement. 2025;21:e70683. 10.1002/alz.70683
REFERENCES
- 1. Fries JF. Aging, natural death, and the compression of morbidity. N Engl J Med. 1980;303(3):130–135. doi: 10.1056/NEJM198007173030304 [DOI] [PubMed] [Google Scholar]
- 2. McGuire LC, Ford ES, Ajani UA. Cognitive functioning as a predictor of functional disability in later life. Am J Geriatr Psychiatry. 2006;14(1):36–42. doi: 10.1097/01.JGP.0000192502.10692.d6 [DOI] [PubMed] [Google Scholar]
- 3. Collaborators GBDDF . Estimation of the global prevalence of dementia in 2019 and forecasted prevalence in 2050: an analysis for the Global Burden of Disease Study. Lancet Public Health. 2019;7(2):e105–e125. doi: 10.1016/S2468-2667(21)00249-8 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4. Evert J, Lawler E, Bogan H, Perls T. Morbidity profiles of centenarians: survivors, delayers, and escapers. J Gerontol A Biol Sci Med Sci. 2003;58(3):232–237. doi: 10.1093/gerona/58.3.m232 [DOI] [PubMed] [Google Scholar]
- 5. Andersen SL, Sebastiani P, Dworkis DA, Feldman L, Perls TT. Health span approximates life span among many supercentenarians: compression of morbidity at the approximate limit of life span. J Gerontol A Biol Sci Med Sci. 2012;67(4):395–405. doi: 10.1093/gerona/glr223 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6. Perls T. Dementia‐free centenarians. Exp Gerontol. 2004;39(11‐12):1587–1593. doi: 10.1016/j.exger.2004.08.015 [DOI] [PubMed] [Google Scholar]
- 7. Perls T. Centenarians who avoid dementia. Trends Neurosci. 2004;27(10):633–636. doi: 10.1016/j.tins.2004.07.012 [DOI] [PubMed] [Google Scholar]
- 8. Andersen SL. Centenarians as models of resistance and resilience to Alzheimer's disease and related dementias. Adv Geriatr Med Res. 2020;2(3):e200018. doi: 10.20900/agmr20200018 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9. Morris JC, Weintraub S, Chui HC, et al. The Uniform Data Set (UDS): clinical and cognitive variables and descriptive data from Alzheimer Disease Centers. Alzheimer Dis Assoc Disord. 2006;20(4):210–216. doi: 10.1097/01.wad.0000213865.09806.92 [DOI] [PubMed] [Google Scholar]
- 10. Beekly DL, Ramos EM, Lee WW, et al. The National Alzheimer's Coordinating Center (NACC) database: the Uniform Data Set. Alzheimer Dis Assoc Disord. 2007;21(3):249–258. doi: 10.1097/WAD.0b013e318142774e [DOI] [PubMed] [Google Scholar]
- 11. Lynch CA, Walsh C, Blanco A, et al. The clinical dementia rating sum of box score in mild dementia. Dement Geriatr Cogn Disord. 2006;21(1):40–43. doi: 10.1159/000089218 [DOI] [PubMed] [Google Scholar]
- 12. O'Bryant SE, Waring SC, Cullum CM, et al. Staging dementia using clinical dementia rating scale sum of boxes scores: a Texas Alzheimer's research consortium study. Arch Neurol. 2008;65(8):1091–1095. doi: 10.1001/archneur.65.8.1091 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13. Besser LM, Kukull WA, Teylan MA, et al. The revised National Alzheimer's Coordinating Center's Neuropathology form‐available data and new analyses. J Neuropathol Exp Neurol. 2018;77(8):717–726. doi: 10.1093/jnen/nly049 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14. Montine TJ, Phelps CH, Beach TG, et al. National Institute on Aging‐Alzheimer's Association guidelines for the neuropathologic assessment of Alzheimer's disease: a practical approach. Acta Neuropathol. 2012;123(1):1–11. doi: 10.1007/s00401-011-0910-3 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15. Hyman BT, Phelps CH, Beach TG, et al. National Institute on Aging‐Alzheimer's Association guidelines for the neuropathologic assessment of Alzheimer's disease. Alzheimers Dement. 2012;8(1):1–13. doi: 10.1016/j.jalz.2011.10.007 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16. Arenaza‐Urquijo EM, Vemuri P. Resistance vs resilience to Alzheimer disease: clarifying terminology for preclinical studies. Neurology. 2018;90(15):695–703. doi: 10.1212/WNL.0000000000005303 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17. Pappalettera C, Carrarini C, Miraglia F, Vecchio F, Rossini PM. Cognitive resilience/reserve: myth or reality? A review of definitions and measurement methods. Alzheimers Dement. 2024;20(5):3567–3586. doi: 10.1002/alz.13744 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18. Goldberg TE, Huey ED, Devanand DP. Association of APOE e2 genotype with Alzheimer's and non‐Alzheimer's neurodegenerative pathologies. Nat Commun. 2020;11(1):4727. doi: 10.1038/s41467-020-18198-x [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19. Stekhoven DJ, Buhlmann P. MissForest–non‐parametric missing value imputation for mixed‐type data. Bioinformatics. 2012;28(1):112–118. doi: 10.1093/bioinformatics/btr597 [DOI] [PubMed] [Google Scholar]
- 20. Langa KM, Larson EB, Karlawish JH, et al. Trends in the prevalence and mortality of cognitive impairment in the United States: is there evidence of a compression of cognitive morbidity?. Alzheimers Dement. 2008;4(2):134–144. doi: 10.1016/j.jalz.2008.01.001 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21. Culig L, Sahbaz BD, Bohr VA. Effects of lifespan‐extending interventions on cognitive healthspan. Expert Rev Mol Med. 2022;25:e2. doi: 10.1017/erm.2022.36 [DOI] [PubMed] [Google Scholar]
- 22. Ganz AB, Beker N, Hulsman M, et al. Neuropathology and cognitive performance in self‐reported cognitively healthy centenarians. Acta Neuropathol Commun. 2018;6(1):64. doi: 10.1186/s40478-018-0558-5 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23. Zhang M, Ganz AB, Rohde S, et al. The correlation between neuropathology levels and cognitive performance in centenarians. Alzheimers Dement. 2023;19(11):5036–5047. doi: 10.1002/alz.13087 [DOI] [PubMed] [Google Scholar]
- 24. Robinson JL, Corrada MM, Kovacs GG, et al. Non‐Alzheimer's contributions to dementia and cognitive resilience in The 90+ Study. Acta Neuropathol. 2018;136(3):377–388. doi: 10.1007/s00401-018-1872-5 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25. James BD, Leurgans SE, Hebert LE, Scherr PA, Yaffe K, Bennett DA. Contribution of Alzheimer disease to mortality in the United States. Neurology. 2014;82(12):1045–1050. doi: 10.1212/WNL.0000000000000240 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26. Savva GM, Wharton SB, Ince PG, et al. Age, neuropathology, and dementia. N Engl J Med. 2009;360(22):2302–2309. doi: 10.1056/NEJMoa0806142 [DOI] [PubMed] [Google Scholar]
- 27. Shinohara M, Kanekiyo T, Tachibana M, et al. APOE2 is associated with longevity independent of Alzheimer's disease. Elife. 2020;9:e62199. doi: 10.7554/eLife.62199 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28. Borras C, Ingles M, Mas‐Bargues C, et al. Centenarians: an excellent example of resilience for successful ageing. Mech Ageing Dev. 2020;186:111199. doi: 10.1016/j.mad.2019.111199 [DOI] [PubMed] [Google Scholar]
- 29. Beker N, Ganz A, Hulsman M, et al. Association of Cognitive Function Trajectories in centenarians with postmortem neuropathology, physical health, and other risk factors for cognitive decline. JAMA Netw Open. 2021;4(1):e2031654. doi: 10.1001/jamanetworkopen.2020.31654 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30. Beker N, Sikkes SAM, Hulsman M, et al. Longitudinal maintenance of Cognitive Health in centenarians in the 100‐plus Study. JAMA Netw Open. 2020;3(2):e200094. doi: 10.1001/jamanetworkopen.2020.0094 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31. Neuville RS, Biswas R, Ho CC, et al. Study of neuropathological changes and dementia in 100 centenarians in The 90+ Study. Alzheimers Dement. 2023;19(8):3417–3425. doi: 10.1002/alz.12981 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 32. Leung Y, Barzilai N, Batko‐Szwaczka A, et al. Cognition, function, and prevalent dementia in centenarians and near‐centenarians: an individual participant data (IPD) meta‐analysis of 18 studies. Alzheimers Dement. 2023;19(6):2265–2275. doi: 10.1002/alz.12828 [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Supporting Information
Supporting Information
Supporting Information