Late-Life Metabolic Dysfunction-Associated Steatotic Liver Disease and its Association With Physical Disability and Dementia

Abstract

Background

The burden of metabolic dysfunction-associated steatotic liver disease (MASLD) is growing rapidly, including among older adults. The number of older adults is also rising with concomitantly increasing rates of age-related physical and cognitive dysfunction. However, data on whether MASLD affects physical and cognitive function in older adults is limited. As such, we aimed to identify whether prevalent MASLD influences the risk of incident physical disability or dementia in initially healthy older adults.

Methods

A post-hoc analysis of participants from the ASPREE-XT cohort study, which recruited community-dwelling older adults without a history of cardiovascular disease, dementia, or independence-limiting functional impairment. The Fatty Liver Index (to identify MASLD) was calculated in those with complete data. Cox proportional-hazards models were used to investigate the outcomes of dementia and persistent physical disability in participants with MASLD vs those without.

Results

Of the 9 097 individuals included (mean age 75.1 ± 4.2 years; 45.0% men), 341 (3.7%) developed persistent physical disability and 370 (4.1%) developed dementia over a median follow-up of 6.4 years (IQR 5.3–7.5 years). When adjusting for known contributors including age, gender, education, comorbidity, and functional measures, MASLD was associated with an increased risk of persistent physical disability (HR 1.41 [95% CI: 1.07–1.87]) and reduced risk of incident dementia (HR 0.63 [95% CI: 0.48–0.83]).

Conclusions

Prevalent MASLD is associated with reduced rates of incident dementia but increased risk of persistent physical disability in initially relatively healthy older adults. Understanding the mechanisms underlying these divergent results to allow appropriate risk stratification and counseling is important.

Metabolic dysfunction-associated steatotic liver disease (MASLD) (1) is characterized by ≥5% of hepatocytes containing fat with a comorbid cardiometabolic disease in the absence of excess alcohol use and other concomitant liver diseases. MASLD (formerly known as nonalcoholic fatty liver disease [NAFLD]) has been previously associated with dementia (2) (in older adults) and impaired cognitive function (3) (in middle-aged adults). There are some reports that MASLD/NAFLD is associated with reduced physical function (4), though there is less data on whether this leads to persistent loss of activities of daily living (ADL) functioning or a requirement for residential aged-care facility placement. These associations between MASLD/NAFLD and reduced physical and cognitive functioning are of particular importance given the significant and growing prevalence of steatotic liver disease in the United States (5,6), Australia (7), and globally (8).

As the burden of MASLD grows, so too does the number of older adults globally (9). With increasing rates of dementia (10) and long-term community support and residential care requirements (11,12), maintaining physical and cognitive function into older age is an important public health goal. However, the link between NAFLD/MASLD and these important clinical outcomes is less well studied in older adults than in middle-aged populations, and recent work has shown some discrepancies in outcomes. Although 1 study has shown a moderately increased risk of dementia in middle-aged to older NAFLD patients compared with healthy controls (2), another based on biopsy-proven NAFLD in middle-aged adults showed no significant impact of baseline NAFLD on incident dementia aside from in a subgroup with advanced liver fibrosis (13). This has been corroborated by other work from both the Rotterdam study (14) and UK Biobank data (15); indeed, the Rotterdam Study provided some evidence of an early protective effect of NAFLD presence on incident dementia (14). Similarly, a recent systematic review suggested that the presence of NAFLD was associated with an increased risk of cognitive impairment but a reduced risk of vascular dementia (16). However, the studies evaluating cognitive impairment were both on middle or younger-aged adults rather than older subjects, and causality is unable to be adequately inferred. In addition, while there are data supporting a link between steatotic liver disease and reduced physical activity (17) and sarcopenia (18), there is a paucity of longitudinal work evaluating the potential impact of steatotic liver disease on the development of persistent independence-limiting physical disability.

Determining the rates of these age-related outcomes on individuals with MASLD is important both from a public health perspective via appropriate resource allocation and planning, as well as patient level, in relation to identifying whether this rapidly growing group may benefit from targeted intervention to prolong independence, physical function, and residence in the general community. As such, our study aimed to evaluate whether MASLD confers an increased risk of dementia and/or sustained physical disability in relatively healthy community-dwelling adults.

Method

Study Population

We performed a post-hoc analysis of Australian participants included in the ASPirin in Reducing Events in the Elderly (ASPREE) randomized trial and the ASPREE-extension (ASPREE-XT) cohort study; the study designs and findings from the main trial have been previously published in detail (19–22). In brief, between 2010 and 2014, ASPREE recruited 16 703 participants from Australia via their usual primary care providers. The participants were aged 70 years or older with key exclusions including: self-reported or physician-diagnosed dementia, a Modified Mini-Mental State Examination (3MS) (23) score less than 78, established or previous cardiovascular or cerebrovascular disease, an inability to independently perform any 1 or more of 6 basic ADL (20,24) (defined as “a lot of difficulty” or “unable to perform” an ADL), and/or a serious illness likely to cause death within the following 5 years. Participants were followed up with annual in-person visits and medical record reviews as well as additional telephone contact. All participants provided written informed consent. The ASPREE trial and ASPREE-XT study were approved by local ethics committees and are registered on ClinicalTrials.gov (NCT01038583) and the International Standard Randomized Controlled Trial Number Registry (ISRCTN83772183).

Participant Assessment

At baseline and during follow-up, in-person interviews and assessments by study investigators collated self-reported information on lifestyle, social, and medical history; cognitive function assessments; anthropometry and markers of physical function were collected; and laboratory parameters were requested and stored. Specific questionnaires included the 3MS (23) (repeated every 2 years), the Lifestyle questionnaire with 6 ADL questions embedded (repeated every 6 months) (20,24), and the Center for Epidemiologic Studies Depression Scale (CES-D-10) (25,26). Baseline anthropometric and physical markers were recorded including weight, height, body mass index (BMI), abdominal circumference, grip strength, and gait speed. Gait speed was measured as the time to walk 3 meters at a habitual pace indoors on a flat surface, performed twice and averaged (27). Grip strength was measured using spring-type or hydraulic hand-held dynamometers (Jamar, Chicago, IL; Lafayette Instruments, Lafayette, IN) while seated. The arm was at the side of the body with a neutral wrist and there was no support for the hand or arm during the test. The average of 3 readings (separated by 20–30 seconds rest) using the dominant hand was included.

Laboratory Data

Initial baseline fasting laboratory data collected included glucose, triglycerides, and a lipid profile, repeated annually at local pathology laboratories. In addition, as part of the ASPREE Healthy Ageing Biobank, Australian participants were invited to provide serum and plasma for storage at −80° C; plasma was thawed and used for biochemical analysis to determine the alanine aminotransferase (ALT), aspartate aminotransferase (AST), gamma-glutamyl-transferase (GGT) and additional tests. The analysis of the Healthy Ageing Biobank plasma was performed centrally using an Abbott Alinity ci analyzer (Abbott Diagnostics, Macquarie Park, Australia).

Identifying MASLD

The identification of the MASLD and no-MASLD groups in ASPREE-XT has been previously described (28). In brief, in participants where complete data was available and for whom plasma for biochemistry analysis was collected within 90 days of their baseline visit (9 847 of the 16 703), the Fatty Liver Index (FLI) was calculated using BMI, abdominal circumference, triglycerides, and GGT as described (29); a score of ≥60 was considered to represent hepatic steatosis. The FLI was originally used and validated for identifying NAFLD—however, subsequent work has shown a similar accuracy in metabolic-dysfunction-associated fatty liver disease (30). Given this, as well as recent work of us and others showing near total concordance between NAFLD and the newer MASLD definition (31,32), the FLI was used as originally published (29). To identify a subgroup with MASLD, participants with a FLI ≥ 60 were excluded if they drank excess alcohol (men more than 210 g/week, women more than 140 g/week) or were on medications typically considered to be steatogenic (methotrexate, amiodarone, glucocorticoids, and/or tamoxifen) so as to meet the traditional NAFLD definition (33). Subsequently, those with a FLI ≥ 60 meeting the NAFLD definition were only classified as MASLD if they had 1 or more requisite cardiometabolic comorbidities (1). Individuals with a FLI of <30 were considered a no-MASLD comparator irrespective of cardiometabolic comorbidity, medication use, or alcohol consumption (see Figure 1).

Figure 1.

Patient Flow Diagram. *Cardiometabolic features include those previously described (1): BMI ≥ 25 kg/m² (or BMI ≥ 23 kg/m² in Asians); elevated abdominal circumference (>94 cm in men, >80 cm in women); fasting serum glucose ≥ 100 mg/dL; self-described T2DM; drug treatment for T2DM; blood pressure ≥ 130/85 mmHg; treatment with an antihypertensive; plasma triglycerides ≥ 150 mg/dL; low HDL cholesterol (≤40 mg/dL in men, ≤50 mg/dL in women); treatment with a fibrate or HMG-CoA reductase inhibitor. NAFLD = nonalcoholic fatty liver disease.

Baseline Characteristics and Cardiometabolic Comorbidities

For the purposes of this study, obesity was defined as BMI ≥ 25 kg/m² for Asians and ≥30 kg/m² for non-Asians; and a large waist circumference (34) was defined as ≥90 cm for Asian men, ≥102 cm for non-Asian men, ≥80 cm for Asian women, and ≥88 cm for non-Asian women. Type 2 diabetes mellitus (T2DM) was defined as 1 or more of: self- or physician-reported T2DM, fasting glucose ≥7 mmol/L, and/or the prescription of hypoglycaemic medications. Chronic kidney disease was defined as 1 or both of eGFR < 60 ml/kg/1.73 m² or an elevated urinary albumin-creatinine ratio (≥35 mg/g for women or ≥25 mg/g for men). Smoking status was defined as currently smoking vs never or formerly smoking.

Defining Outcomes

Individuals with a suspected diagnosis of dementia (“trigger”) were referred for further standardized cognitive and functional assessments (35). Dementia triggers included a 3MS score <78, a drop in 3MS score of >10.15 from their predicted score, self-reported memory concerns, a clinician diagnosis of dementia, and/or the prescription of cholinesterase inhibitors. Standardized cognitive and functional assessments were then administered at least 6 weeks post-trigger event to minimize the risk of delirium confounding the results (35). Other documentation relevant to the dementia assessment was also sought (including laboratory tests, neuroimaging, and clinical notes).

All cases were then adjudicated by an international expert panel of neurologists, neuropsychologists, and geriatricians, who were blinded to treatment allocation. Dementia was adjudicated based on criteria specified in the Diagnostic and Statistical Manual for Mental Disorders Version IV (DSM-IV) (36). This required evidence of memory impairment, and evidence of at least 1 of the following: aphasia, apraxia, agnosia, and/or disturbances in executive functioning. The cognitive impairments needed to have caused significant impairment in social or occupational functioning, and to have represented a significant decline from a previous level of functioning. Time-to-event was defined as the time to the dementia trigger that resulted in a confirmed dementia diagnosis by the adjudication committee (35).

Disability was defined as persistent loss of capacity to perform at least 1 of the 6 basic ADLs (20), which included walking across a room, transferring from bed or chair, toileting, bathing, dressing and eating, or assessment confirming the need or eligibility for admission to a nursing care facility for a physical disability. The ADL response options were coded as: 1 (no difficulty), 2 (a little difficulty), 3 (some difficulty), 4 (a lot of difficulty), or 5 (unable to perform). As previously described (37), this data was collected every 6 months, and persistent loss of function was determined as loss of capacity to perform the same ADL for at least 6 months. Time to event was defined as the time to the first report of the ADL loss or assessment that the participant required residential care admission (37).

Statistics

Baseline data were compared using a 1-way ANOVA or Student’s t test (for continuous variables), or a Chi-squared test (for categorical variables). The co-primary outcomes for this study were whether MASLD increases the hazard for the development of dementia and/or disability. The secondary outcome was whether there are differences in the rates of admission to longer-term residential care in older adults with MASLD. In the primary analysis, unadjusted Cox proportional hazard regression models were utilized to evaluate these outcomes. Secondary analysis adjusted for age and sex. The final model for predicting disability utilized clinically relevant variables previously shown to be predictors in the ASPREE population (38), including age, sex, baseline 3MS score, gait speed, grip strength, smoking status, renal function, a CES-D-10 score of ≥8, and baseline T2DM. The final model for predicting dementia utilized age, sex, education, baseline 3MS score, smoking status, alcohol use, baseline T2DM, a baseline CES-D-10 score of ≥8, grip strength, and gait speed, as previously utilized in the ASPREE population (27). BMI and abdominal circumference were not included as these 2 variables were incorporated in the calculation of the FLI used to define MASLD. Sensitivity analyses were performed using a FLI cut-off of <60 to define no-MASLD (rather than <30 for the primary analysis). A p value of <.05 was considered statistically significant. Statistical analyses were performed using Stata software v17.0 (StataCorp LLC, College Station, TX).

Results

Study Populations and Follow-Up

Of the 16 703 Australian ASPREE participants, 9 847 participants had a calculable FLI using biochemistry within 90 days post-recruitment. Of these, 498 with a FLI ≥ 60 were excluded due to excess alcohol consumption and 284 were excluded due to steatogenic medication use, leaving 9 097 for the final analysis (Figure 1). All the FLI ≥ 60 group had at least 1 of the requisite cardiometabolic criteria present as required to meet the definition of MASLD [1]—therefore, all those diagnosed as MASLD also met the diagnostic criteria for NAFLD. The mean age of all 9 097 participants was 75.1 ± 4.3 years, 55% were women, and 98.6% were Caucasian. The general characteristics of the participants stratified by FLI can be seen in Table 1 (comparing no-MASLD [FLI < 30], an indeterminate FLI [30–60], and MASLD [FLI ≥ 60]). In the final analysis group, 33.0% had a diagnosis of MASLD. The 9 097 participants were followed for a median of 6.4 (IQR: 5.3–7.5) years.

Table 1.

Characteristics of Study Participants

Characteristic	All participants	FLI < 30	FLI 30–60	FLI ≥ 60	p Value
				(MASLD)
Number of participants	9 097	2 968	3 131	2 998
Sex (men)	4 093 (45.0%)	921 (31.0%)	1 624 (51.9%)	1 548 (51.6%)	<.001¶
Age (y) (mean ± SD)	75.06 ± 4.25	75.28 ± 4.42	75.23 ± 4.36	74.66 ± 3.91	<.001#
Age category					<.001¶
70–72 y (n, %)	3 776 (41.5%)	1 184 (39.9%)	1 277 (40.8%)	1 315 (43.9%)
73–75 y (n, %)	2 349 (25.8%)	764 (25.7%)	782 (25.0%)	803 (26.8%)
76–78 y (n, %)	1 363 (15.0%)	436 (14.7%)	484 (15.5%)	443 (14.8%)
79–81 y (n, %)	856 (9.4%)	300 (10.1%)	305 (9.7%)	251 (8.4%)
82–84 y (n, %)	449 (4.9%)	156 (5.3%)	171 (5.5%)	122 (4.1%)
≥85 y (n, %)	304 (3.3%)	128 (4.3)	112 (3.6%)	64 (2.1%)
Alcohol use					.027¶
Current drinker (n, %)	7 148 (78.6%)	2 359 (79.5%)	2 489 (79.5%)	2 300 (76.7%)
Former drinker (n, %)	437 (4.8%)	129 (4.3%)	140 (4.5%)	168 (5.6%)
Never drinker (n, %)	1 512 (16.6%)	480 (16.2%)	502 (16.0%)	530 (17.7%)
Ethnic background					.003¶
White/Caucasian*	8 961 (98.6%)	2 907 (98.0%)	3 090 (98.8%)	2 964 (99.0%)
Non-White	130 (1.4%)	60 (2.0%)	39 (1.2%)	31 (1.0%)
Formal education					<.001¶
≤11 y (n, %)	4 425 (48.6%)	1 311 (44.2%)	1 523 (48.6%)	1 591 (53.1%)
12–15 y (n, %)	2 420 (26.6%)	839 (28.3%)	812 (25.9%)	769 (25.7%)
≥16 y (n, %)	2 252 (24.8%)	818 (27.6%)	796 (25.4%)	638 (21.3%)
CES-D-10 Score ≥ 8 (n, %)	825 (9.1%)	272 (9.2%)	261 (8.3%)	292 (9.2%)	.157¶
Baseline 3MS score (mean ± SD)	93.65 ± 4.40	94.11 ± 4.26	93.51 ± 4.43	93.33 ± 4.48	.017#
Weight (kg) (mean ± SD)	76.12 ± 14.03	64.07 ± 8.69	75.75 ± 8.68	88.44 ± 12.16	<.001#
BMI (kg/m2) (mean ± SD)	27.75 ± 4.43	23.90 ± 2.37	27.33 ± 2.23	31.99 ± 4.01	<.001#
BMI category†					<.001¶
Underweight	43 (0.5%)	43 (1.4%)	0 (0.0%)	0 (0.0%)
Healthy weight	2 422 (26.6%)	1 971 (66.4%)	419 (13.4%)	32 (1.1%)
Overweight	4 224 (46.4%)	929 (31.3%)	2 328 (74.4%)	967 (32.3%)
Obese	2 408 (26.5%)	25 (0.8%)	384 (12.3%)	1 999 (66.7%)
Waist circumference (cm) (mean ± SD)	96.07 ± 12.29	84.27 ± 7.84	96.22 ± 6.38	107.59 ± 9.18	<.001#
					<.001¶
Large waist circumference (34) (n, %)‡	4 971 (54.6%)	499 (16.8%)	1 743 (55.7%)	2 729 (91.0%)
Grip strength (kg) (mean ± SD)	27.55 ± 9.69	25.41 ± 8.76	28.59 ± 9.82	28.57 ± 10.08	<.001#
Gait speed (m/s) (mean ± SD)	1.04 ± 0.21	1.07 ± 0.21	1.06 ± 0.21	1.01 ± 0.21	.127#
Diabetes mellitus (n, %)§	839 (9.2%)	109 (3.7%)	227 (7.3%)	503 (16.8%)	<.001¶
Insulin requiring (n, % of diabetes)	67 (8.0%)	11 (10.1%)	19 (8.4%)	37 (7.4%)
Smoking history					<.001¶
Never-smoker (n, %)	5 215 (57.3%)	1 863 (62.8%)	1 735 (55.4%)	1 617 (53.9%)
Former smoker (n, %)	3 604 (39.6%)	1 000 (33.7%)	1 293 (41.3%)	1 311 (43.7%)
Current smoker (n, %)	278 (3.1%)	105 (3.5%)	103 (3.3%)	70 (2.3%)
Pack-year history (mean ± SD)	21.45 ± 23.70	17.62 ± 20.86	21.51 ± 23.05	24.45 ± 25.95	<.001#
Laboratory values∥
GGT (U/L) (mean ± SD)	27.54 ± 29.92	18.67 ± 10.94	24.93 ± 15.94	39.04 ± 46.01	<.001#
ALT (U/L) (mean ± SD)	20.12 ± 10.94	17.15 ± 6.77	19.42 ± 7.91	23.77 ± 15.15	<.001#
AST (U/L) (mean ± SD)	21.76 ± 7.73	21.31 ± 5.72	21.20 ± 6.04	22.79 ± 10.46	<.001#
Total cholesterol (mmol/L) (mean ± SD)	5.25 ± 0.97	5.34 ± 0.94	5.25 ± 0.97	5.14 ± 0.99	<.001#
HDL cholesterol (mean ± SD)	1.58 ± 0.45	1.82 ± 0.46	1.55 ± 0.41	1.36 ± 0.37	<.001#
Renal function
eGFR (mL/min/m2) (mean ± SD)	75.18 ± 13.52	77.19 ± 12.61	74.84 ± 13.29	73.53 ± 14.35	<.001#
Chronic kidney disease (n, %)	1 622 (17.8%)	413 (13.9%)	538 (17.2%)	671 (22.4%)	<.001¶

Notes:

^*Where ethnicity was not recorded/reported, White/Caucasian was assumed.

^†Underweight = BMI < 18.5 kg/m², Healthy weight = BMI 18.5–22.9 kg/m² (Asians) or 18.5–24.9 kg/m² (non-Asian), Overweight = BMI 23–24.9 kg/m² (Asians) or 25–29.9 kg/m² (non-Asian), Obese = BMI ≥ 25 kg/m² (Asians), or ≥ 30 kg/m² (non-Asian).

^‡Large waist circumference = If Asian, ≥88 cm (men) and ≥80 cm (women); if non-Asian, ≥102 cm (men), and ≥90 cm (women).

^§Defined as 1 or more of: (a) self-reported diabetes mellitus, (b) prescription of at least 1 glucose lowering therapy at baseline, (c) a fasting blood sugar of ≥7.0 mmol/L.

^∥GGT is gamma-glutamyltransferase, ALT is alanine aminotransferase, AST is aspartate aminotransferase.

^¶Chi-square p Value.

^#1-way ANOVA p Value.

Dementia

During a median follow-up of 6.3 years (IQR: 5.3–7.4 years), 370 (4.1%) of the 9 097 participants developed dementia (Table 2), of which 56 also developed persistent disability. Those who developed dementia were, at baseline, more likely to be men (51.6% vs 44.7%, p = .01), older (77.7 ± 5.0 vs 74.9 ± 4.2 years, p < .01), have lower baseline 3MS scores (89.7 ± 5.4 vs 93.8 ± 4.3, p < .01), and have a CES-D-10 score consistent with depression (13.8% vs 8.9%, p < .01). Similarly, there were higher rates of T2DM (12.4% vs 9.1%, p = .03) and chronic kidney disease (CKD; 23.8% vs 17.6%, p < .01) in the group that developed dementia. There was no relationship between educational attainment and the development of incident dementia (p = .89). There was a reduced rate of dementia in the group with baseline MASLD (27.6% vs 33.2%, p < .01); consistent with this, the BMI was also lower in the group who developed dementia (26.8 ± 4.3 kg/m² vs 27.8 ± 4.4 kg/m², p < .01).

Table 2.

Incident Dementia and Disability Stratified by Baseline Data

Characteristic	No dementia	Dementia	p Value	No disability	Disability	p Value
Number of participants	8 727	370		8 756	341
MASLD (n, %)	2 896 (33.2%)	102 (27.6%)	.005¶	2 836 (32.4%)	162 (47.5%)	<.001¶
Sex (men)	3 902 (44.7%)	191 (51.6%)	.009¶	3 958 (45.2%)	135 (39.6%)	.041¶
Age (y) (mean ± SD)	74.94 ± 4.17	77.69 ± 5.01	<.001#	75.06 ± 4.16	77.76 ± 5.33	<.001#
Age category			<.001¶			<.001¶
70–72 y (n, %)	3 693 (42.3%)	83 (22.4%)		3 698 (42.2%)	78 (22.9%)
73–75 y (n, %)	2 274 (26.1%)	75 (20.3%)		2 279 (26.0%)	70 (20.5%)
76–78 y (n, %)	1 295 (14.8%)	68 (18.4%)		1 297 (14.8%)	66 (19.4%)
79–81 y (n, %)	791 (9.1%)	65 (17.6%)		808 (9.2%)	48 (14.1%)
82–84 y (n, %)	406 (4.7%)	43 (11.6%)		414 (4.7%)	35 (10.3%)
≥85 y (n, %)	268 (3.1%)	36 (9.7%)		260 (3.0%)	44 (12.9%)
Alcohol use			.021¶			<.001¶
Never drinker (n, %)	1 437 (16.5%)	75 (20.3%)		1 427 (16.3%)	15 (24.9%)
Former drinker (n, %)	412 (4.7%)	25 (6.8%)		422 (4.8%)	15 (4.4%)
Current drinker (n, %)	6 878 (78.8%)	270 (73.0%)		6 907 (78.9%)	241 (70.7%)
Ethnic background			.111**			.459**
White/Caucasian*	8 606 (98.6%)	361 (97.6%)		8 630 (98.6%)	337 (98.8%)
Non-White	121 (1.4%)	9 (2.4%)		126 (1.4%)	4 (1.2%)
Formal education			.886¶			.012¶
≤11 y (n, %)	4 241 (48.6%)	184 (49.7%)		4 233 (48.3%)	192 (56.3%)
12–15 y (n, %)	2 322 (26.6%)	98 (26.5%)		2 338 (26.7%)	82 (24.0%)
≥16 y (n, %)	2 164 (24.8%)	88 (23.8%)		2 185 (25.0%)	67 (19.6%)
CES-D-10 score ≥ 8 (n, %)	774 (8.9%)	51 (13.8%)	.001¶	769 (8.8%)	56 (16.4%)	<.001¶
Baseline 3MS score (mean ± SD)	93.81 ± 4.28	89.67 ± 5.42	<.001#	93.69 ± 4.39	92.52 ± 4.67	<.001#
Weight (kg) (mean ± SD)	76.25 ± 14.03	73.12 ± 13.72	<.001#	75.99 ± 13.83	79.40 ± 18.05	<.001#
BMI (kg/m2) (mean ± SD)	27.78 ± 4.43	26.83 ± 4.31	<.001#	27.67 ± 4.31	29.67 ± 6.52	<.001#
BMI category†			.001¶			<.001¶
Underweight	40 (0.5%)	3 (0.8%)		40 (0.5%)	3 (0.9%)
Healthy weight	2 290 (26.2%)	132 (35.7%)		2 347 (26.8%)	75 (22.0%)
Overweight	4 072 (46.7%)	152 (41.1%)		4 100 (46.8%)	124 (36.4%)
Obese	2 325 (26.6%)	83 (22.4%)		2 269 (25.9%)	139 (40.8%)
Waist circumference (cm) (mean ± SD)	96.12 ± 12.28	94.45 ± 12.40	.073#	95.88 ± 12.10	100.89 ± 15.65	<.001#
Large waist circumference (34) (n, %)‡	4 788 (54.9%)	183 (49.5%)	.041¶	4 734 (54.1%)	237 (69.5%)	<.001¶
Grip strength (kg) (mean ± SD)	27.61 ± 9.69	26.03 ± 9.70	.002#	27.70 ± 9.69	23.56 ± 8.71	<.001#
Gait speed (m/s) (mean ± SD)	1.05 ± 0.21	0.98 ± 0.24	<.001#	1.05 ± 0.21	0.87 ± 0.22	<.001#
Diabetes mellitus (n, %)§	793 (9.1%)	46 (12.4%)	.029¶	782 (8.9%)	57 (16.7%)	<.001¶
Insulin requiring (n, % of diabetes)	64 (0.7%)	3 (0.8%)	.864¶	63 (0.7%)	4 (1.2%)	.337¶
Smoking history			.498¶			.074¶
Never-smoker (n, %)	4 992 (57.2%)	223 (60.3%)		5 031 (57.5%)	184 (54.0%)
Former smoker (n, %)	3 467 (39.7%)	137 (37.0%)		3 464 (39.6%)	140 (14.1%)
Current smoker (n, %)	268 (3.1%)	10 (2.7%)		261 (3.0%)	17 (5.0%)
Pack-year history (mean ± SD)	21.54 ± 23.86	19.21 ± 19.11	.248#	21.30 ± 23.44	24.95 ± 29.08	.062#
Laboratory values∥
GGT (U/L) (mean ± SD)	27.41 ± 27.75	30.61 ± 62.06	.043#	27.46 ± 30.06	29.56 ± 26.01	.204#
ALT (U/L) (mean ± SD)	20.13 ± 10.57	19.67 ± 17.66	.428#	20.16 ± 10.93	18.94 ± 11.14	.042#
AST (U/L) (mean ± SD)	21.74 ± 7.68	22.17 ± 8.96	.294#	21.78 ± 7.70	21.24 ± 8.57	.206#
Total cholesterol (mmol/L) (mean ± SD)	5.25 ± 0.97	5.22 ± 1.01	.583#	5.25 ± 0.97	5.21 ± 0.99	.436#
HDL cholesterol (mean ± SD)	1.58 ± 0.45	1.59 ± 0.47	.566#	1.58 ± 0.45	1.54 ± 0.45	.142#
Renal function
eGFR (mL/min/m2) (mean ± SD)	75.26 ± 13.48	73.29 ± 14.34	.008#	75.31 ± 13.44	71.64 ± 14.99	<.001#
Chronic kidney disease (n, %)	1 534 (17.6%)	88 (23.8%)	.002¶	1 518 (17.3%)	104 (30.5%)	<.001¶

Notes:

^*Where ethnicity was not recorded/reported, White/Caucasian was assumed.

^†Underweight = BMI < 18.5 kg/m², Healthy weight = BMI 18.5–22.9 kg/m² (Asians) or 18.5–24.9 kg/m² (non-Asian), Overweight = BMI 23–24.9 kg/m² (Asians) or 25-29.9 kg/m² (non-Asian), Obese = BMI ≥ 25 kg/m² (Asians) or ≥ 30 kg/m² (non-Asian).

^‡Large waist circumference = If Asian, ≥88 cm (men) and ≥80 cm (women); if non-Asian, ≥102 cm (men), and ≥90 cm (women).

^§Defined as 1 or more of: (a) self-reported diabetes mellitus, (b) prescription of at least 1 glucose lowering therapy at baseline, (c) a fasting blood sugar of ≥7.0 mmol/L.

^∥GGT is gamma-glutamyltransferase, ALT is alanine aminotransferase, AST is aspartate aminotransferase.

^¶Chi-Square p Value.

^#Student’s t Test p Value.

^**Fisher’s exact test p Value.

Utilizing Cox proportional hazards modeling, MASLD was significantly associated with a reduced hazard of incident dementia (HR 0.73, 95% CI: 0.57 to 0.94; Table 3). This remained significant when adjusted for age and sex (HR: 0.76, 95% CI: 0.59 to 0.99), as well as in a fully adjusted model (HR 0.63, 95% CI: 0.48 to 0.83; Table 3). When including only those who did not develop persistent disability, MASLD remained protective (HR 0.64, 95% CI: 0.47 to 0.87) against incident dementia. In a sensitivity analysis stratifying MASLD (FLI ≥ 60) versus no-MASLD (FLI < 60) the results are similar, with MASLD remaining protective against the development of dementia in the fully adjusted model (HR 0.73, 95% CI: 0.57 to 0.93) (Supplementary Table 1).

Table 3.

Association Between Prevalent MASLD (FLI ≥ 60) versus no-MASLD (FLI < 30) and Incident Dementia and Physical Disability

	Dementia		Physical disability
MASLD (FLI ≥ 60) vs no-MASLD (FLI < 30)	Hazard ratio (95% CI)	p Value	Hazard ratio (95% CI)	p Value
Unadjusted/crude model	0.73 (0.57–0.94)	.016	1.87 (1.45–2.42)	<.001
Age and gender adjusted	0.76 (0.59–0.99)	.040	2.21 (1.70–2.88)	<.001
ASPREE adjusted model*,†	0.63* (0.48–0.83)	.001	1.41† (1.07–1.86)	.016

Notes:

^*ASPREE adjusted model for dementia included: age, sex, education, baseline 3MS score, current smoking status, alcohol use (current/former/never), baseline T2DM, a baseline CES-D-10 score of ≥8, grip strength, and gait speed.

^†ASPREE adjusted model for disability included: age, sex, baseline 3MS score, gait speed, grip strength, smoking status, eGFR, a CES-D-10 score of ≥8, and baseline T2DM.

Disability

Three hundred and forty-one of the 9 097 participants developed a persistent disability and/or an assessment that the participant was eligible for nursing home placement due to a physical disability (3.7%) over a median of 6.3 years (IQR: 5.3–7.4 years) follow-up (Table 2). Of these 341, 56 also developed dementia (16.4%). The group with a persistent physical disability was more likely to be older (77.8 ± 5.3 vs 75.1 ± 4.2 years, p < .01), have lower educational attainment (p = .01), higher rates of a CES-D-10 score ≥ 8 (16.4% vs 8.8%, p < .01), lower gait speed (0.9 ± 0.2 vs 1.1 ± 0.2 m/s, p < .01), and reduced grip strength (23.6 ± 8.7 vs 27.7 ± 9.7 kg, p < .01) at baseline. They had higher rates of MASLD (47.5% vs 32.4%, p < .01), T2DM (16.7% vs 8.9%, p < .01), and CKD (30.5% vs 17.3%, p < .01). Consistent with these increased rates of dysmetabolic features, the group had higher rates of obesity and increased abdominal adiposity.

When utilizing Cox proportional hazard modeling, MASLD was significantly associated with an increased hazard of disability (HR 1.87, 95% CI: 1.45 to 2.42; Table 3). This persisted when adjusting for sex and age (HR 2.21, 95% CI: 1.70 to 2.88), and also when adjusting for all known contributors to physical disability in the ASPREE cohort (HR 1.41, 95% CI: 1.07 to 1.86). The association between MASLD and persistent physical disability is even stronger when excluding those who developed dementia during follow-up (HR 1.68, 95% CI: 1.23 to 2.29). These results remain consistent when utilizing a more liberal comparator of FLI < 60 as the no-MASLD cohort, with MASLD associated with an increased hazard of persistent physical disability (HR 1.59, 95% CI: 1.27 to 2.00; Supplementary Table 1).

Discussion

Despite the known clinical and public health challenges associated with an increasing number of older adults globally (9) as well as an increasing burden of MASLD (6–8), there has been a relative paucity of studies on the interplay between MASLD and important age-related outcomes in older adults. To address this important clinical question, we have investigated the impact of MASLD on incident dementia and persistent physical disability in a cohort of relatively well-community-dwelling older adults. The main findings are that in this large prospective cohort of community-dwelling older adults participating in the ASPREE randomized clinical trial and ASPREE-XT cohort study, the presence of MASLD is associated with an increased risk of persistent physical disability but a reduced risk of development of dementia.

We have shown that MASLD increases the hazard of persistent physical disability in relatively well-community-dwelling older adults by 41%, even when adjusting for multiple known contributors to physical disability and decline in this population (38). This novel result is, to our knowledge, the first time that MASLD has been directly linked to clinically relevant worsened independent functioning, an outcome of significant concern to older adults (39). Previous work has previously described an association between steatotic liver disease and myosteatosis (40), sarcopenia (18), and frailty (28,41); it is thus unsurprising that one of the consequences of MASLD would include an increased risk of losing physical functional independence.

However, there is a concomitant decrease in the hazard of dementia in the MASLD group of 37%. These results concord with other recent data showing either minimal (13,15) influence or even protection (14) from incident dementia in individuals with steatotic liver disease. While data from the Framingham offspring study has shown an association between NAFLD and lower brain volume, the mean age of these individuals at the time of NAFLD diagnosis was 63.8 years, more than 10 years younger than our cohort (42), and the calculated additional years of brain aging attributed to NAFLD (estimated by brain volume measurement) was attenuated with increasing years. It’s possible that middle-aged adults with NAFLD have an increased risk of cognitive dysfunction and incident dementia (3), but in older adults free from dementia by the age of 70 years, this association is lost. Plausibly, those with more severe consequences of MASLD in middle age may be less likely to meet enrolment criteria in this study, potentially preselecting a relatively more benign MASLD phenotype in these older adults. Furthermore, losing weight is associated with an improvement in hepatic steatosis and is generally recommended for MASLD treatment (43). However, given that dementia itself is associated with pre-diagnosis loss of weight (44) as well as with lower BMIs during relatively shorter-term follow-up (45), it is possible that liver steatosis has partly or completely resolved during the preclinical phase of dementia such that the potential association with MASLD is lost.

This study has numerous strengths, including its large sample size, rigorous data collection at baseline and during follow-up, and international expert panel adjudication for determining the endpoints. However, there are potential weaknesses to consider. First, the use of the FLI to determine steatosis is less accurate than imaging or biopsy-based data; however, the FLI has been shown to be accurate in a very similar population to ours (46) and so the stratification of groups into MASLD vs no-MASLD is likely quite valid. Additionally, the use of the DSM-IV criteria for dementia (36) is relatively conservative, and it is also possible that participants were less likely to follow-up with study visits if they had developed cognitive decline/dementia during the study period (particularly during the coronavirus disease-2019 [COVID-19] pandemic), potentially underestimating the rates of dementia development in this group. The original ASPREE recruitment period was from 2010 to 2014 and therefore pre-dated any influence of COVID-19 on alcohol intake (and thus the diagnosis of baseline MASLD) or physical activity. Although follow-up for this study did extend into December, 2020 (9 months after the first confirmed COVID-19 case in Australia), the vast majority of follow-up time preceded the pandemic, and due to Australia’s COVID-19 policies, the rates of COVID-19 through 2020 were low. Although it is not possible to directly account for any potential impact of the pandemic on the diagnosis of dementia nor of the impact of COVID-19 infection on accelerated physical disability, there is no reason to suppose that this would bias results towards or away from the MASLD group specifically, and so the overarching direction of results should still hold true.

In summary, we have shown that in older community-dwelling adults, MASLD is associated with an increased risk of physical disability development and loss of independence with ADLs and is inversely associated with incident dementia. These data have important public health and clinical implications regarding emphasis on retaining physical functioning and fitness in older adults, suggesting that proactively screening for sarcopenia/frailty in older adults with MASLD and intervening early may afford the opportunity to improve independence, rates of community living, and overall physical health. This important issue should be further evaluated in more detailed long-term prospective cohort studies.

Funding

The ASPREE clinical trial was supported by the National Institute on Aging and the National Cancer Institute at the National Institutes of Health (U01AG029824, U19AG062682); the NHMRC (334047, 1127060); Monash University; and the Victorian Cancer Agency. The ASPREE Biobank was supported by research grants from the Australian Government’s CSIRO (Commonwealth Scientific and Industrial Research Organization; Preventative Health Flagship 2009) and the National Cancer Institute/National Institutes of Health (5U01AG029824-02). No funding sources were involved in the design or conduct of the study; collection, management, or analysis of the data; interpretation of the results; preparation, review, or approval of the manuscript; or decision to submit it for publication.

Conflict of Interest

None.

Author Contributions

All authors have read and approved the submission of this manuscript. D.C.-C., W.W.K., and S.K.R. contributed to the study concept and design. All authors contributed to the acquisition, analysis, and interpretation of data. D.C.-C. drafted the manuscript. All authors contributed to critical revisions of the manuscript for important intellectual content. D.C.-C. was responsible for statistical analysis. S.K.R. provided study supervision.