Explore the Role of Frailty as a Contributor to the Association Between AT(N) Profiles and Cognition in Alzheimer’s Disease

Abstract

Background:

The relationship between Alzheimer’s disease (AD)-related pathology and cognition was not exactly consistent.

Objective:

To explore whether the association between AD pathology and cognition can be moderated by frailty.

Methods:

We included 1711 participants from the Alzheimer’s Disease Neuroimaging Initiative database. Levels of cerebrospinal fluid amyloid-β, p-tau, and t-tau were identified for AD-related pathology based on the amyloid-β/tau/neurodegeneration (AT[N]) framework. Frailty was measured using a modified Frailty Index-11 (mFI-11). Regression and interaction models were utilized to assess the relationship among frailty, AT(N) profiles, and cognition. Moderation models analyzed the correlation between AT(N) profiles and cognition across three frailty levels. All analyses were corrected for age, sex, education, and APOEɛ4 status.

Results:

In this study, frailty (odds ratio [OR] = 1.71, p < 0.001) and AT(N) profiles (OR = 2.00, p < 0.001) were independently associated with cognitive status. The model fit was improved when frailty was added to the model examining the relationship between AT(N) profiles and cognition (p < 0.001). There was a significant interaction between frailty and AT(N) profiles in relation to cognitive status (OR = 1.12, p_interaction = 0.028). Comparable results were obtained when Mini-Mental State Examination scores were utilized as the measure of cognitive performance. The association between AT(N) profiles and cognition was stronger with the levels of frailty.

Conclusions:

Frailty may diminish patients’ resilience to AD pathology and accelerate cognitive decline resulting from abnormal AD-related pathology. In summary, frailty contributes to elucidating the relationship between AD-related pathology and cognitive impairment.

Keywords

Alzheimer’s disease amyloid/tau/neurodegeneration (AT[N]) framework cognition frailty

INTRODUCTION

There are complex connections between Alzheimer’s disease (AD) pathology and cognition.1,2, 1,2 Previous studies have demonstrated that AD-related pathology is significantly associated with cognition, but the contribution of AD-related pathology to cognitive decline is less than 50%. 3 This indicates that participants with low AD-related pathology burdens can have cognitive impairment, while many patients with normal cognition have a high burden of AD-related pathology. 4 According to this discrepancy, it is hypothesized that the correlation between AD-related pathology and cognition may be influenced by some latent factors. Patients with cognitive impairment tend to be older and often have multiple comorbidities, which can impact cognitive changes. 5 The accumulation of health deficits with age can be represented by frailty.6,7, 6,7 Individuals with frailty are more fragile and unstable, and thus may be vulnerable to abnormal pathology. Consequently, frailty may help elucidate the relationship between AD-related pathology and cognition.

Frailty is defined as a decrease in physiological reserve and the resistance of individuals to stressors, which increases the susceptibility and incidence of adverse outcomes such as falls, hospitalization, and mortality.7 –9 A large-scale study has reported that frailty and cognitive decline are linked by brain structural changes. 10 Furthermore, there are considerable overlaps between the risk factors and clinical features of cognitive decline and frailty, including age, activities of daily living, metabolism, chronic disease, etc. 11

According to the study by Wallace et al., cognition is impacted by the interaction between frailty and neuropathology. 12 However, autopsy pathology is not readily available, and its clinical application is limited. Monitoring core pathological changes in living individuals is now achievable with the development of validated and widely used AD-related biomarkers, which are less invasive, more feasible, and can serve as proxies for neuropathological changes in AD. 13 The National Institute on Aging and the Alzheimer’s Association (NIA-AA) proposed the amyloid-β/tau/neurodegeneration (AT[N]) framework in 2018, which combines multiple biomarkers and views AD as a continuum.13,14, 13,14 According to the AT(N) framework, it is possible to quantify the cumulative degree of AD-relatedpathology.

Based on these premises, we hypothesized that frailty might be a contributor that moderates the association between AD-related pathology and cognition. Cognitive problems may become more severe in individuals with frailty due to reduced tolerance to external stimuli. We aimed to explore this moderating effect using more accessible AD-related biomarkers and focused on the entire progression of cognitive changes.

METHODS

Data source

The participants were included and their clinical data extracted from the Alzheimer’s Disease Neuroimaging Initiative (ADNI) database (http://adni.loni.usc.edu). The ADNI project, a public-private collaboration initiated in 2003, aims to predict the progression of mild cognitive impairment (MCI) and early AD utilizing serial magnetic resonance imaging (MRI), positron emission tomography (PET), clinical and neuropsychological assessments, and various AD-related biomarkers. The project has received written approval from all participating institutions and their respective institutional review boards.

Participants

A total of 1,711 participants were included in our research, of whom 1,079 had complete information for AT(N) profiles. Participants were excluded if they lacked complete clinical characteristics and information required to assess frailty. Those with the profile of suspected non-Alzheimer’s pathologies (SNAP) were also precluded. Following predefined criteria in the ADNI, participants were classified into cognitively normal (CN), subjective memory complaints (SMC), MCI, and AD. 15 They underwent neuroimaging examinations, cognitive assessments, and cerebrospinal fluid (CSF) samples were collected by lumbar puncture.

Frailty, cognitive assessment, and mental problems

In our study, the modified Frailty Index-11 (mFI-11) was utilized to evaluate frailty severity. Supplementary Table 1 provides additional information on this scale. The mFI-11, a previously validated retrospective tool, is a modified version of the Canadian Study of Health and Aging Frailty Index (CSHA-FI). 16 Developed by Velanovich et al., the mFI-11 integrates elements from the CSHA-FI with the National Surgical Quality Improvement (NSQIP) database of the American College of Surgeons. 17 It evaluates multi-system disease and individual life competence, which are among the eleven factors it comprises. These factors are not only integral but also readily available in clinical data, facilitating its use in various medical contexts. Significantly, an elevated mFI-11 score has been associated with increased hospital days, mortality, and complications. 18 The mFI-11 has become widely utilized for risk stratification of diseases, prognostic prediction, and surgical-related assessment, establishing itself as a key tool for frailty assessment across different medical specialties.19,20, 19,20 In the ADNI database, Soon et al. have implemented and validated the viability of mFI-11 for frailtyassessment. 21

The deficit accumulation approach is a common method for measuring frailty. 6 The mFI-11 comprises 11 health variables, each scored on a binary scale. If the health variable is not in a risk state, the deficit is coded as 0; otherwise, it is coded as 1. Based on total mFI-11 scores, participants were classified as non-frail (mFI-11 = 0), mildly frail (mFI-11 = 1 or 2), or severely frail (mFI-11≥3). All information fulfilling mFI-11 criteria was extracted from medical histories in the ADNI database.

The Mini-Mental State Examination (MMSE) and Alzheimer’s Disease Assessment Scale-Cognitive Subscale (ADAS-Cog) 11 were utilized to evaluate general cognition. Anxiety and depression were assessed through the utilization of the Neuropsychiatric Inventory Questionnaire (NPI).

Neuroimaging

¹⁸F-fluorodeoxyglucose (FDG) PET was utilized to assess brain hypometabolism, derived from the average of five metaROIs (bilateral angular gyrus, bilateral posterior cingulate gyrus, and left middle/inferior temporal gyrus). PET data were provided by UC Berkeley and Lawrence Berkeley National Laboratory. 22

Hippocampal volumetry was performed using 1.5T or 3.0T MRI scanners in collaboration with the ADNI imaging core at UCSF. 1.5T MRI images were processed utilizing the FreeSurfer version 4.3 image processing framework, while 3.0T MRI images were processed with FreeSurfer version 5.1. 23 Details on the ADNI FreeSurfer-based pipelines are available online (http://adni.loni.usc.edu/).

CSF biomarker measurements

CSF samples were collected by lumbar puncture as described in the ADNI procedures manual (http://www.adni-info.org/). Following collection, the levels of CSF amyloid-β (Aβ₄₂), phosphorylated (181P) tau (p-tau), and total tau (t-tau) were quantitatively assessed. This analysis was conducted on a fully automated cobas e 601 instrument using Roche Elecsys^® β-amyloid(1–42) CSF, Elecsys^® phosphotau (181P) CSF, and Elecsys^® total-tau CSF immunoassays, respectively. 24 All measurements were processed at the University of Pennsylvania Biomarker Laboratory following the acceptance criteria of the Roche protocol.

AT(N) profiles

Based on a cutoff value of 976.6 pg/ml for CSF Aβ₄₂, we determined abnormal (A+) and normal (A–) levels of amyloid. 25 To assess whether tau pathology was abnormal (T+) or normal (T–), the cutoff value for CSF p-tau was 23 pg/ml. 26 Moreover, for CSF t-tau (N), which is considered a more general biomarker validated in the ADNI population, the cutoff value was established at 300 pg/ml. 27 For sensitivity analysis, the cutoff of 1.21 was used for FDG PET (average of angular, temporal, and posterior cingulate), and 6,723 mm³ was used for hippocampal volume corrected for intracranialvolume. 26

According to the AT(N) framework proposed by NIA-AA in 2018, the AD-related biomarkers and neuroimaging data were classified into eight biomarker profiles: A–T–N–, A+T–N–, A+T–N+, A+T+N–, A+T+N+, A–T+N–, A–T–N+, and A–T+N+. Typically, A–T+N–, A–T–N+, and A–T+N+ were considered to indicate suspected non-Alzheimer’s pathologies (SNAP). 13 Participants with SNAP do not exhibit deposition of specific Aβ pathology but show alterations in FDG PET and structural MRI, along with changes in t-tau levels. Conversely, these pathological features may also manifest in various non-AD neurological disorders. This overlap in manifestations contributes to the frequent observation of similar FDG PET, structural MRI, and t-tau changes in individuals with non-AD conditions. 28 Additionally, it is important to note that the SNAP group is not classified or considered as probable AD dementia according to both the NIA-AA and the 2014 IWG classification systems. 14 We focused only on the relationship between AD-related pathological changes and cognitive status, so individuals with SNAP were excluded from the analysis.

Statistical analysis

The characteristics of the research sample were analyzed using descriptive methods. Based on baseline diagnoses in ADNI, participants were categorized into three groups (non-frail, mildly frail, and severely frail) according to mFI-11 scores. Mean (standard deviation, SD) or number (percentage, %) were used to describe the characteristics of participants. One-way ANOVA and Chi-square tests were utilized to identify differences between groups for continuous and categorical variables at baseline, respectively.

To identify discrepancies in frailty across different cognitive statuses and AT(N) profiles, one-way ANOVA was utilized. Conventional models, interaction models, and moderation models were constructed for the primary analyses. In conventional models, we evaluated the relationships between AT(N) profiles, mFI-11 scores, and cognitive status using ordinal logistic regression. We then examined model fit between conventional and interaction models via ANOVA to evaluate whether adding mFI-11 scores improved model fit. In interaction models, mFI-11 scores were included as an interaction term to examine the interaction between mFI-11 and AT(N) profiles in relation to cognitive status using ordinal logistic regression on a multiplicative scale. In moderation models, comparisons of non-frail or mildly frail individuals to those meeting severe frailty criteria were conducted to assess whether the strength of the association between AT(N) profiles and cognitive status varied.

The identical analyses were then replicated using MMSE scores as the cognitive measure. One-way ANOVA was utilized to examine the relationships between MMSE scores, mFI-11 scores, and AT(N) profiles. In conventional models, linear regression was employed to assess the relationships among AT(N) profiles, mFI-11 scores, and MMSE scores. After evaluating model fit between conventional and interaction models, linear regression on a multiplicative scale was utilized to examine the association between the interaction term and MMSE scores. Furthermore, the strength of the association between AT(N) profiles and MMSE scores across different levels of frailty was analyzed.

For sensitivity analyses, (i) FDG PET and hippocampal volume were utilized in place of CSF t-tau as biomarkers of neurodegeneration in the AT(N) framework, respectively; (ii) ADAS-Cog 11 was used to measure cognitive performance. All analyses controlled for age, sex, education, and APOEɛ4 status. Statistical significance was determined at a two-sided p-value of 0.05, and standardized z-scores were calculated for the dependent variable in each model. R version 4.2.2 was utilized for all statistical analyses.

RESULTS

Baseline characteristics

Baseline characteristics are displayed in Table 1. The mean (SD) age of the 1711 individuals enrolled was 73.54 (7.17) years, with 957 (55.9%) females. Of the participants, 563 were non-frail, 993 mildly frail, and 155 severely frail. In summary, 67.10% met frailty criteria as defined by the mFI-11. Severely frail individuals tend to be older, are more likely to be female, possess less education, and are more likely to have mental health problems compared to non-frail and mildly frail participants, highlighting the multifaceted nature of frailty. Severely frail individuals also exhibited lower CSF Aβ₄₂, higher CSF t-tau and p-tau, reduced glucose metabolism on FDG PET imaging, smaller hippocampal volumes, and worse MMSE scores. All items and their distribution in the mFI-11 are presented in Supplementary Table 1.

Table 1

Baseline characteristics of the study population in ADNI

	Overall	Non-frail	Mildly frail	Severely frail	p
	(N = 1,711)	(N = 563)	(N = 993)	(N = 155)
Age, y	73.54 (7.17)	72.41 (7.23)	73.89 (7.07)	75.40 (7.10)	<0.001^a
Sex					0.008^b
Male	754 (44.1)	275 (48.8)	422 (42.5)	57 (36.8)
Female	957 (55.9)	288 (51.2)	571 (57.5)	98 (63.2)
Education, y	15.92 (2.82)	16.24 (2.71)	15.86 (2.83)	15.08 (2.95)	<0.001^a
APOEɛ4 genotype					0.022^b
0 ɛ4 allele	892 (52.1)	321 (57.0)	490 (49.3)	81 (52.3)
1 ɛ4 allele	634 (37.1)	188 (33.4)	383 (38.6)	63 (40.6)
2 ɛ4 allele	185 (10.8)	54 (9.6)	120 (12.1)	11 (7.1)
CSF Aβ₄₂, pg/mL	937.14 (499.11)	1021.33 (518.43)	902.31 (494.52)	839.82 (399.74)	<0.001^a
CSF t-tau, pg/mL	274.01 (129.73)	253.90 (119.97)	281.36 (131.61)	304.29 (144.00)	<0.001^a
CSF p-tau, pg/mL	26.58 (14.54)	24.43 (13.50)	27.39 (14.79)	29.68 (15.81)	0.001^a
FDG PET	1.19 (0.16)	1.25 (0.15)	1.18 (0.16)	1.11 (0.15)	<0.001^a
Hippocampus, mm²	6,754.77 (1173.96)	7,035.53 (1127.69)	6,631.86 (1165.56)	6,424.93 (1205.34)	<0.001^a
MMSE scores	27.06 (2.74)	28.02 (2.04)	26.73 (2.83)	25.68 (3.24)	<0.001^a
ADAS-Cog 11	11.07 (6.67)	8.57 (4.72)	11.97 (7.06)	14.35 (7.32)	<0.001^a
Depression	0.52 (1.22)	0.37 (1.02)	0.53 (1.21)	0.94 (1.67)	<0.001^a
Anxiety	0.43 (1.30)	0.31 (1.20)	0.44 (1.26)	0.68 (1.70)	0.039^a

Data are mean (SD) or number (%). Aβ₄₂: amyloid-β; ADAS-Cog 11, Alzheimer’s Disease Assessment Scale-Cognitive Subscale 11; APOEɛ4, apolipoprotein E ɛ4; CSF, cerebrospinal fluid; FDG, ¹⁸F-fluorodeoxyglucose; MMSE, Mini-Mental State Examination; p-tau: phosphorylated tau; t-tau, total tau. P values based on: ^a ANOVA test; ^b Chi-Square test.

Complex relationship between AT(N) profiles and cognition

The AT(N) profiles comprised data from 1,079 participants, including 358 in the A–T–N–profile, 235 in A+T–N–, 141 in A+T+N–, and 345 in A+T+N+ (Table 2). Based on the AT(N) profiles, 239 (22.15%) participants without any positive AD-related biomarkers still exhibited varying degrees of cognitive impairment, with 11 even diagnosed with dementia. In comparison, 98 (9.08%) participants with more than one abnormal biomarker were ultimately classified as cognitively normal. Only 206 (19.09%) participants with AD-related pathological changes were diagnosed with dementia (Table 2). In summary, the correlation between cognitive decline and accumulation of AD-related pathology was not entirely consistent.

Table 2

Modified frailty index-11 scores by AT(N) profiles and cognitive status

	CN		SMC		MCI		AD		Total
	Number of participants	Mean (SD)	Number of participants	Mean (SD)	Number of participants	Mean (SD)	Number of participants	Mean (SD)	Number of participants	Mean (SD)
A–T–N–	119	0.66 (0.72)	50	0.92 (1.03)	178	0.87 (0.89)	11	1.73 (1.01)	358	0.83 (1.00)
A+T–N–	61	0.82 (0.76)	18	0.83 (0.92)	124	0.98 (1.02)	32	1.56 (0.95)	235	1.00 (0.99)
A+T+N–	17	0.88 (0.60)	4	1.25 (0.96)	77	0.95 (0.96)	43	1.67 (0.99)	141	1.17 (0.99)
A+T+N+	20	0.80 (0.89)	11	0.64 (0.67)	183	0.97 (1.00)	131	1.58 (1.01)	345	1.18 (1.00)
Total	217	0.74 (0.98)	83	0.88 (1.08)	562	0.93 (1.00)	217	1.60 (0.99)	1079	1.03 (1.00)

A+, amyloid normal using CSF Aβ; A–, amyloid abnormal using CSF Aβ; AD, Alzheimer’s disease; CN, cognitively normal; MCI, mild cognitive impairment; N+, neurodegeneration or neuronal injury normal using t-tau; N–, neurodegeneration or neuronal injury abnormal using t-tau; SD, standard deviation; SMC, subjective memory complaints; T+, tau normal using CSF p-tau; T-, tau abnormal using CSF p-tau.

The relationship between AT(N) profiles, frailty, and cognitive status

One-way ANOVA analysis revealed distinct frailty levels across individuals with different cognitive statuses (Fig. 1A-D). Moreover, AD participants exhibited the highest mFI-11 scores within each AT(N) profile. Nevertheless, the association between mFI-11 scores and AT(N) profiles in each cognitive group was not statistically significant(Fig. 1E-H).

Fig. 1

The association of mFI-11 scores with cognitive status and AT(N) profiles. AD, Alzheimer’s disease; CN, cognitively normal; MCI, mild cognitive impairment; mFI-11, modified frailty index-11; SMC, subjective memory complaints. A-D) The association of mFI-11 scores with clinical cognitive statuses; E-H) The association of mFI-11 scores with AT(N) profiles. All analyses were assessed by the one-way ANOVA test with the covariates regressed out including age, sex, education, and APOEɛ4 status, and calculated after Bonferroni correction.

In conventional models, mFI-11 scores (OR = 1.71, 95% CI = 1.51 to 1.94; p < 0.001; Table 3) and AT(N) profiles (OR = 2.00, 95% CI = 1.78 to 2.25; p < 0.001; Table 3) were significantly associated with cognitive status. The fit of interaction models was substantially better than conventional models (p < 0.001). In interaction models, the interaction between mFI-11 scores and AT(N) profiles was significant for cognitive status (OR = 1.12, 95% CI = 1.01 to 1.24; p_interaction = 0.028; Table 3). This interaction indicated that participants with frailty and AD-related pathology were more likely to exhibit cognitive impairment.

Table 3

Associations and interactions of AT(N) profiles and frailty with the cognition

	Cognitive diagnosis		MMSE scores
	OR (95% CI)	p	Estimate (95% CI)	p
AT(N)	2.00 (1.78, 2.25)	<0.001	–0.79 (–0.92, –0.66)	<0.001
mFI-11 scores	1.71 (1.51, 1.94)	<0.001	–0.62 (–0.77, –0.47)	<0.001
AT(N)*mFI-11	1.12 (1.01, 1.24)	0.028	–0.18 (–0.29, –0.06)	0.002

Neurodegeneration or neuronal injury of the AT(N) were classified using CSF t-tau. All analyses were adjusted for age, sex, education and APOEɛ4 status. mFI-11, modified frailty index-11; MMSE, Mini-Mental State Examination; OR, odds ratio.

Moderation analyses further revealed that the strength of the association between AT(N) profiles and cognitive status varied across different levels of frailty, becoming more robust with increased frailty (Fig. 2). More specifically, participants with greater frailty severity were more likely to develop cognitive impairment as AD-related pathology accumulated.

Fig. 2

The values of the association between AT(N) profiles and cognition at different levels of frailty. The values of the association between AT(N) profiles and cognition (cognitive status and MMSE scores) at three levels of frailty. The association of AT(N) profiles with cognitive status and MMSE scores were investigated using ordinal logistic regression and linear regression, respectively. All analyses were adjusted for age, sex, education, and APOEɛ4 status.

Validation by cognitive performance

This analysis was repeated using MMSE scores as the cognitive measure. The results demonstrated that MMSE scores were lower in those with higher mFI-11 scores (p < 0.001; Fig. 3A) and with greater abnormal AD-related pathology (p < 0.001; Fig. 3B).

Fig. 3

MMSE scores are associated with mFI-11 scores and AT(N) profiles. mFI-11, modified frailty index-11 (range 0–11); MMSE, Mini-Mental State Examination. p values were assessed by the one-way ANOVA test and calculated after Bonferroni correction.

Similar results were exhibited in conventional models, showing that mFI-11 scores (β= –0.62, 95% CI = –0.77 to –0.47; p < 0.001; Table 3) and AT(N) profiles (β= –0.79, 95% CI = –0.92 to –0.66; p < 0.001; Table 3) correlated with cognitive performance. Cognitive performance also improved model fit compared to conventional models (p < 0.001). In interaction models, the interaction term was associated with a decline in MMSE scores (β= –0.18, 95% CI = –0.29 to –0.06; p_interaction = 0.002; Table 3). Meanwhile, the association between AT(N) profiles and MMSE scores was more pronounced in patients with severe frailty (Fig. 2).

Sensitivity analyses

Sensitivity analyses were conducted to explore whether AT(N) profiles with different AD-related biomarkers had varying effects on the results. FDG PET data were utilized as a biomarker of neurodegeneration in the AT(N) framework, and the interaction between mFI-11 scores and AT(N) profiles was significantly associated with cognitive status (OR = 1.14, 95% CI = 1.05 to 1.24, p_interaction = 0.002; Supplementary Table 2) and MMSE scores (β= –0.17, 95% CI = –0.27 to –0.07, p_interaction < 0.001; Supplementary Table 2). When hippocampal volume was utilized to assess “N” in the AT(N) framework, comparable results were obtained between the interaction term and cognition (Supplementary Table 2). Regardless of whether “N” in the AT(N) framework was measured by FDG PET or hippocampal volume, the association between AT(N) profiles and cognitive status became stronger with increased frailty severity (Supplementary Figures 1 and 2). Aligning with our initial findings, the interaction between frailty and AT(N) profiles was significantly associated with ADAS-Cog 11 scores (β= 0.44, 95% CI = 0.16 to 0.72, p_interaction = 0.002; Supplementary Table 2).

DISCUSSION

In this study, the following conclusions were reached. First, at similar levels of AD-related pathology, frailty severity varied across different cognitive statuses, with the highest mFI-11 scores occurring in AD participants. This indicates that frailty may interfere with cognitive expression. Second, frailty as an interaction term improved model fit for the association between AD pathology and cognition. The potential mechanism is that frailty may account for multiple pathways leading to cognitive decline with diverse underlying etiologies.29,30, 29,30 Finally, frailty moderates the relationship between AT(N) profiles and cognition, playing a crucial role in the association of AD-related pathology with the cognitive expression.

Based on these findings, frailty may account for some of the variance in the associations between AD-related pathology and clinical cognitive features. These results align with the interpretation proposed by Wallace et al. 12 Frailty, as a functional state of a reduced physiological reserve associated with aging and time development, limits the tolerance of the organism to the appearance of pathological abnormalities and variations. While the accumulation of AD-related pathology increases the risk of cognitive impairment, frailty might reduce the tolerance of individuals to AD-related pathology. Hence, participants with severe frailty are more likely to suffer cognitive impairment, although their AD-related pathology changes slightly. All of these indicate that frailty can intensify cognitive decline among individuals with abnormal AD-related pathology and lower the threshold for AD-related pathology that results in cognitive impairment.

Frailty has repeatedly been associated with cognition.31 –33 Most research indicates that individuals exhibit poorer cognitive performance with increased frailty severity. Meanwhile, longitudinal analyses also reveal faster cognitive decline linked to greater baseline frailty and rising frailty over time.34,35, 34,35 In recent, a large and multidimensional research show that cognitive functions and grey matter volumes derived from almost half of all brain areas exhibited significant negative correlations with the severity of frailty. 10 In a recent large multidimensional study, cognitive functions and grey matter volumes in nearly half of all brain areas showed significant negative correlations with frailty severity. 36 However, varied viewpoints remain regarding how frailty impacts cognitive expression. 37 These findings suggest potential anomalies in aging-related biological processes or a shared pathological origin.33,35, 33,35 Possible shared mechanisms include subcellular pathophysiology (e.g., oxidative stress and protein misfolding) and impaired repair (e.g., failures in chaperone proteins and autophagy). 38 However, frailty may also influence other risk factors or pathophysiological mechanisms underlying cognitive impairment. 32 Caution is warranted when interpreting results, especially in frail individuals with complex comorbidities. Therefore, we focused on risk stratification by frailty, which aids identification of cognitive decline, prognostic management of cognitive impairment, and prevention of cognitive decline in cognitively normal populations.

Current cognitive literature centers on risk factors and pathology but raises several unresolved issues. First, there are numerous cognitive risk factors. 39 Second, the contribution of AD pathology to cognitive expression is insufficient. Third, the relationship between risk factors and AD pathology requires clarification. Finally, cognitive variation is evident both between and within individuals. 40 Our findings contribute to better understanding cognitive impairment as a multifactorial pathophysiological process, opening a new avenue for cognitive research.

Our study elucidated the relationship between frailty and cognition in two key ways. First, the interaction between frailty and AD pathology impacts cognitive expression. Second, frailty severity influences the strength of association between AD pathology and cognition, with the association intensifying as severity increases. Our findings align with the current understanding that cognitive decline results from combined effects of multiple risks. 41 We confirmed that frailty contributes to the process of cognitive impairment caused by AD pathology. To avoid bias from a single biomarker, we selected multiple commonly used biomarkers in living individuals and classified pathological burden utilizing the AT(N) framework.

The mechanism underlying the relationships between frailty, AD pathology, and cognition remains unclear. Some studies suggest frailty may be a non-cognitive manifestation of AD pathology. 11 Accumulation of AD pathology in brain regions supporting cognition may impair neural systems involved in planning and managing even basic responses, thereby affecting aspects of frailty. Conversely, frailty may contribute to AD pathology development.42 –44 However, it likely involves additional biological mechanisms that synergistically influence cognitive expression. 32 Collectively, this indicates frailty and AD pathology are substantially interrelated during AD progression. We help explain this phenomenon and validate the conclusion by Wallace et al. that cognitive impairment results from interaction between frailty and AD pathology. Our study suggested frailty and AD pathology are independently associated with cognition. We then demonstrated frailty severity aids in explaining the relationship between AD pathology and cognitive expression. Frailty increases susceptibility to cognitive impairment from greater AD pathology. Thus, lower frailty may reduce incidence of cognitive impairment.

Several limitations warrant consideration. First, only cross-sectional data were analyzed lacking longitudinal analysis. Although we showed frailty moderates the association between AD pathology and cognition, this limits inference of a causal relationship. Second, all participants were recruited from the highly selective ADNI cohort and may not fully represent the general elderly population with heterogeneous cognitive and physical conditions. 45 Frailty assessment was constrained by ADNI inclusion/exclusion criteria. Individuals unable to undergo AD biomarker testing due to medical instability or diagnosed neurological conditions are ineligible for ADNI. 46 This may exclude those with greater frailty severity, underestimating frailty levels overall. 46 Third, AT(N) framework cutoffs are not uniformly defined, with various values used in different studies. Employing variable standard cutoffs to classify AT(N) profiles may bias results. To minimize this, all cutoffs were verified in the ADNI cohort.

Our findings indicate frailty contributes to elucidating the association between AD pathology and cognition. Some individuals with low AD pathology burden may be at risk for cognitive impairment if frail. This factor enhances understanding of the biological processes underlying cognitive impairment resulting from abnormal AD pathology. Meanwhile, these findings support conceptualizing cognitive impairment as a complex condition rather than a single disease characterized by specific pathological changes.

AUTHOR CONTRIBUTIONS

Bao-Lin Han (Data curation; Formal analysis; Visualization; Writing – original draft; Writing – review & editing); Ling-Zhi Ma (Conceptualization); Shuang-Ling Han (Writing – review & editing); Yin-Chu Mi (Data curation); Jia-Yao Liu (Data curation); Ze-Hu Sheng (Data curation); Hui-Fu Wang (Conceptualization; Writing – review & editing); Lan Tan, MD (Conceptualization; Writing – review & editing).

Footnotes

ACKNOWLEDGMENTS

The authors thank all the colleagues who have made contributions to build the ADNI database. The authors also thank the subjects and their families for their cooperation in this study.

Data collection and sharing for this project was funded by the Alzheimer’s Disease Neuroimaging Initiative (ADNI) (National Institutes of Health Grant U01 AG024904) and DOD ADNI (Department of Defense award number W81XWH-12-2-0012). ADNI is funded by the National Institute on Aging, the National Institute of Biomedical Imaging and Bioengineering, and through generous contributions from the following: AbbVie, Alzheimer’s Association; Alzheimer’s Drug Discovery Foundation; Araclon Biotech; BioClinica, Inc.; Biogen; Bristol-Myers Squibb Company; CereSpir, Inc.; Cogstate; Eisai Inc.; Elan Pharmaceuticals, Inc.; Eli Lilly and Company; EuroImmun; F. Hoffmann-La Roche Ltd. and its affiliated company Genentech, Inc.; Fujirebio; GE Healthcare; IXICO Ltd.; Janssen Alzheimer Immunotherapy Research & Development, LLC.; Johnson & Johnson Pharmaceutical Research & Development LLC.; Lumosity; Lundbeck; Merck & Co., Inc.; Meso Scale Diagnostics, LLC.; NeuroRx Research; Neurotrack Technologies; Novartis Pharmaceuticals Corporation; Pfizer Inc.; Piramal Imaging; Servier; Takeda Pharmaceutical Company; and Transition Therapeutics.

The Canadian Institutes of Health Research is providing funds to support ADNI clinical sites in Canada. Private sector contributions are facilitated by the Foundation for the National Institutes of Health (). The grantee organization is the Northern California Institute for Research and Education, and the study is coordinated by the Alzheimer’s Therapeutic Research Institute at the University of Southern California. ADNI data are disseminated by the Laboratory for Neuro Imaging at the University of Southern California.

FUNDING

This study was supported by grants from the Science and Technology Innovation 2030 Major Projects (2022ZD0211600), National Natural Science Foundation of China (81971032), and the Taishan Scholars Program of Shandong Province (tsqn201812157).

CONFLICT OF INTEREST

Lan Tan is Editorial Board Members of this journal but was not involved in the peer-review process nor had access to any information regarding its peer-review.

All other authors have no conflict of interest to report.

DATA AVAILABILITY

The data used and analyzed in this study are available from the corresponding authors on reasonable request.

The supplementary material is available in the electronic version of this article: .

References

Wallace

LMK

, Theou

, Darvesh

, et al.Neuropathologic burden and the degree of frailty in relation to global cognition and dementia. Neurology2020; 95: e3269–e3279.

Espay

, Vizcarra

, Marsili

, et al.Revisiting protein aggregation as pathogenic in sporadic Parkinson and Alzheimer diseases. Neurology2019; 92: 329–337.

Sierksma

, Escott-Price

, De Strooper

. Translating genetic risk of Alzheimer’s disease into mechanistic insight and drug targets. Science2020; 370: 61–66.

Illán-Gala

, Pegueroles

, Montal

, et al.Challenges associated with biomarker-based classification systems for Alzheimer’s disease. Alzheimers Dement (Amst)2018; 10: 346–357.

Scheltens

, Strooper

, Kivipelto

, et al.Alzheimer’s disease. Lancet2021; 397: 1577–1590.

Rockwood

and Howlett

. Age-related deficit accumulation and the diseases of ageing. Mech Ageing Dev2019; 180: 107–116.

Clegg

, Young

, Iliffe

, et al.Frailty in elderly people. Lancet2013; 381: 752–762.

Eeles

EMP

, White

, O’Mahony

, et al.The impact of frailty and delirium on mortality in older inpatients. Age Ageing2012; 41: 412–416.

Fried

, Tangen

, Walston

, et al.Frailty in older adults: evidence for a phenotype. J Gerontol A Biol Sci Med Sci2001; 56: M146–M157.

10.

Jiang

, Noble

, Sui

, et al.Associations of physical frailty with health outcomes and brain structure in 483 033 middle-aged and older adults: a population-based study from the UK Biobank. Lancet Digit Health2023;S2589750023000432.

11.

Buchman

, Schneider

, Leurgans

, et al.Physical frailty in older persons is associated with Alzheimer disease pathology. Neurology2008; 71: 499–504.

12.

Wallace

LMK

, Theou

, Godin

, et al.Investigation of frailty as a moderator of the relationship between neuropathology and dementia in Alzheimer’s disease: a cross-sectional analysis of data from the Rush Memory and Aging Project. Lancet Neurol2019; 18: 177–184.

13.

Jack

, Bennett

, Blennow

, et al.NIA-AA Research Framework: Toward a biological definition of Alzheimer’s disease. Alzheimers Dement2018; 14: 535–562.

14.

Jack

, Bennett

, Blennow

, et al.A/T/N: An unbiased descriptive classification scheme for Alzheimer disease biomarkers. Neurology2016; 87: 539–547.

15.

Aisen

, Petersen

, Donohue

, et al.Alzheimer’s Disease Neuroimaging Initiative 2 Clinical Core: Progress and plans. Alzheimers Dement2015; 11: 734–739.

16.

Fang

, Hu

, Arya

, et al.Preoperative frailty is predictive of complications after major lower extremity amputation. J Vasc Surg2017; 65: 804–811.

17.

Velanovich

, Antoine

, Swartz

, et al.Accumulating deficits model of frailty and postoperative mortality and morbidity: its application to a national database. J Surg Res2013; 183: 104–110.

18.

Pazniokas

, Gandhi

, Theriault

, et al.The immense heterogeneity of frailty in neurosurgery: a systematic literature review. Neurosurg Rev2021; 44: 189–201.

19.

Pichatechaiyoot

, Thannil

, Boonyapipat

, et al.Preoperative modified frailty index to predict surgical complications in endometrial cancer patients. Obstet Gynecol Sci2022; 65: 513–521.

20.

Holzgrefe

, Wilson

, Staley

, et al.Modified frailty index is an effective risk-stratification tool for patients undergoing total shoulder arthroplasty. J Shoulder Elbow Surg2019; 28: 1232–1240.

21.

Soon

SXY

, Kumar

, Tan

AJL

, et al.The impact of multimorbidity burden, frailty risk scoring, and 3-directional morphological indices vs. testing for CSF responsiveness in normal pressure hydrocephalus. Front Neurosci2021; 15: 751145.

22.

Landau

and Jagust

. UC Berkeley FDG MetaROI methods. https://adni.bitbucket.io/reference/docs/UCBERKELEYFDG/ADNI_UC_Berkeley_FDG_Methods_20220323.pdf.

23.

Hsu

Y-Y

, Schuff

, Du

A-T

, et al.Comparison of automated and manual MRI volumetry of hippocampus in normal aging and dementia. J Magn Reson Imaging2002; 16: 305–310.

24.

Hansson

, Seibyl

, Stomrud

, et al.CSF biomarkers of Alzheimer’s disease concord with amyloid-β PETand predict clinical progression: A study of fully automated immunoassays in BioFINDER and ADNI cohorts. Alzheimers Dement2018; 14: 1470–1481.

25.

Tang

, Wang

Z-B

, Ma

L-Z

, et al.Dynamic changes of CSF clusterin levels across the Alzheimer’s disease continuum. BMC Neurol2022; 22: 508.

26.

, Li

, Suckling

, et al.Frequency and longitudinal clinical outcomes of Alzheimer’s AT(N) biomarker profiles: A longitudinal study. Alzheimers Dement2019; 15: 1208–1217.

27.

Bucci

, Chiotis

and Nordberg

. Alzheimer’s disease profiled by fluid and imaging markers: tau PET best predicts cognitive decline. Mol Psychiatry2021; 26: 5888–5898.

28.

Wirth

, Villeneuve

, Haase

, et al.Associations between Alzheimer disease biomarkers, neurodegeneration, and cognition in cognitively normal older people. JAMA Neurol2013; 70: 1512–1519.

29.

Canevelli

, Cesari

, van Kan

. Frailty and cognitive decline: how do they relate?Curr Opin Clin Nutr Metab Care2015; 18: 43–50.

30.

Robertson

, Savva

and Kenny

. Frailty and cognitive impairment— A review of the evidence and causal mechanisms. Ageing Res Rev2013; 12: 840–851.

31.

Waite

, Maitland

, Thomas

, et al.Sarcopenia and frailty in individuals with dementia: A systematic review. Arch Gerontol Geriatr2021; 92: 104268.

32.

Bisset

and Howlett

. The biology of frailty in humans and animals: Understanding frailty and promoting translation. Aging Med (Milton)2019; 2: 27–34.

33.

Fabrício

D de M

, Chagas

MHN

and Diniz

. Frailty and cognitive decline. Transl Res2020; 221: 58–64.

34.

Solfrizzi

, Scafato

, Lozupone

, et al.Biopsychosocial frailty and the risk of incident dementia: The Italian longitudinal study on aging. Alzheimers Dement2019; 15: 1019–1028.

35.

Buchman Aron

, Yu

, Wilson

, et al.Brain pathology contributes to simultaneous change in physical frailty and cognition in old age. J Gerontol A Biol Sci Med Sci2014; 69: 1536–1544.

36.

Gómez-Gómez

and Zapico

. Frailty, cognitive decline, neurodegenerative diseases and nutrition interventions. Int J Mol Sci2019; 20: 2842.

37.

Amboni

, Barone

and Hausdorff

. Cognitive contributions to gait and falls: evidence and implications. Mov Disord2013; 28: 1520–1533.

38.

Searle

and Rockwood

. Frailty and the risk of cognitive impairment. Alzheimers Res Ther2015; 7: 54.

39.

Hrast

, Khachatryan

, Šuput

, et al.Effects of risk factors on longitudinal changes in brain structure and function in the progression of AD. Alzheimers Dement2023; 19: 2666–2676.

40.

Fierini

. Mixed dementia: Neglected clinical entity or nosographic artifice?J Neurol Sci2020; 410: 116662.

41.

Barnes

and Yaffe

. The projected impact of risk factor reduction on Alzheimer’s disease prevalence. Lancet Neurol2011; 10: 819–828.

42.

Wallace

, Theou

, Rockwood

, et al.Relationship between frailty and Alzheimer’s disease biomarkers: A scoping review. Alzheimers Dement (Amst)2018; 10: 394–401.

43.

Jacob

, O’Donnell

, Samra

, et al.Grip strength, gait speed and plasma markers of neurodegeneration in asymptomatic middle-aged and older adults. J Frailty Aging2022; 11: 291–298.

44.

Maltais

, de Souto Barreto

, Moon

, et al.Prospective association of white matter hyperintensity volume and frailty in older adults. Exp Gerontol2019; 118: 51–54.

45.

Weiner

, Aisen

, Jack

, et al.The Alzheimer’s Disease Neuroimaging Initiative: Progress report and future plans. Alzheimers Dement2010; 6: 202–11.e7.

46.

Boyle

, Yu

, Wilson

, et al.Person-specific contribution of neuropathologies to cognitive loss in old age. Ann Neurol2018; 83: 74–83.