Non-Alcoholic Fatty Liver Disease, Liver Fibrosis Score and Cognitive Function in Middle-Aged Adults: The Framingham Study

Abstract

Background:

Non-alcoholic fatty liver disease (NAFLD) is common and has been recently related to brain health. We aimed to assess the relationships of NAFLD and its severity, using the NAFLD fibrosis score (NFS), with cognitive performance.

Methods:

Framingham study Offspring and 3^rd generation participants were included if they attended exams 9 (2002–2008) and 2 (2008–2011), respectively, were free of dementia and stroke, and did not have excessive alcohol intake. Between 2008 and 2011, participants underwent Multi-detector computed tomography scans of the abdomen to determine NAFLD diagnosis and the NFS was used to categorize the severity of fibrosis. Cross-sectional relationships of NAFLD and the NFS with cognitive testing of memory, abstract reasoning, visual perception, attention, and executive function were assessed, while adjusting for multiple cardiometabolic variables including visceral adipose tissue diabetes and insulin resistance.

Results:

Of the 1,287 participants (mean age=61±12, 48% men), 378 (29%) had NAFLD. The presence of NAFLD was not associated with cognitive function. However, among those with NAFLD (mean age=61±12; 58% men), high compared to low risk of advanced fibrosis was associated with poorer performance on similarities (β=−2.22±0.83; p=0.009) and trail-making B minus A (β=−0.11±0.05; p=0.028), independently of potential confounders.

Conclusions:

Participants with high risk of advanced fibrosis may have poorer cognitive function compared to those with low risk, particularly in executive function and abstract reasoning. Future findings are necessary to evaluate the value of the NFS as a biomarker that predicts cognitive impairment and dementia and to explore the role of hepatic fibrosis in brain health.

Introduction

Accumulating evidence suggests that liver diseases, even in pre-cirrhotic stages, may be linked with brain aging.^2,3 More specifically, there is a growing interest in the implications of non-alcoholic fatty liver disease (NAFLD) on brain health, as these conditions share common risk factors (e.g. cardiovascular disease, diabetes and obesity) and possibly also common pathophysiological mechanisms, including inflammation and endothelial dysfunction³. Recently, NAFLD has been linked with poor cognitive function⁴ and with decreased brain activity.⁵ In addition, a study from our group has demonstrated an association between NAFLD and low total cerebral brain volume independently of multiple cardio-metabolic factors, including diabetes, cardiovascular diseases and visceral fat.⁶ These findings are important since NAFLD prevalence is high, with estimations of around 30% in the general population and 80% among the morbidly obese.^7,8 Furthermore, NAFLD is a promising target for intervention aiming to prevent cognitive impairment because it can be improved through life-style changes including exercise⁹, healthy nutrition and weight reduction.¹⁰

Yet, it is important to note that NAFLD comprises a wide range of liver pathologies, ranging from simple steatosis, to non-alcoholic steatohepatitis (NASH) with varying stages of fibrosis, and cirrhosis¹¹. Hence, it still remains to clarify whether the presence of NAFLD per se is related to cognitive aging, or alternatively, if this association is driven by the degree of liver fibrosis.

Liver fibrosis stage is an important prognostic factor in patients with NAFLD. It is associated with end-stage liver disease and with liver-related mortality even in its initial stages, and regardless of steatosis or inflammatory status¹². The gold standard to detect liver fibrosis is liver biopsy, with characterization of liver histology. However, this is an invasive approach, with acknowledged limitations including high cost, risk of procedure-related morbidity and mortality and the possibility of sampling error¹³. To overcome these limitations, several non-invasive approaches, including serum-based biomarkers, have been developed¹⁴. One of them, the NAFLD Fibrosis Score (NFS), is based on a formula which incorporates six readily available variables: age, BMI, hyperglycemia, platelet count, albumin and AST/ALT ratio.¹⁵ In a meta-analysis of 13 studies consisting of 3,064 participants, the NFS had an area under the curve (AUC), sensitivity and specificity for predicting advanced fibrosis of 0.85 (0.80–0.93), 0.90 (0.82–0.99) and 0.97 (0.94–0.99), respectively.¹⁶ Using previously published cutoffs of the NFS to categorize probability of fibrosis severity, a large population-based cohort study showed that individuals with NAFLD who had NFS values indicative of advanced fibrosis had an increased risk of all-cause and cardiovascular mortality compared to those without indication of significant fibrosis¹⁷. Additionally, in the Framingham study, NAFLD with high vs. low or intermediate probability of advanced fibrosis according to the NFS was related to worse vascular function as evident by lower diastolic blood pressure, wider pulse pressure, and increased odds of hypertension¹⁸.

In the current study, we aimed to assess the cross-sectional relationship of NAFLD and the NFS with cognitive performance among middle-aged community-dwelling participants of the Framingham Study.

Methods

Study sample

The Framingham Heart Study (FHS) is a population-based, multi-generational study which was initiated in 1948¹⁹. Figure 1 represents a flow chart for the study sample. The sample for the current analyses is based on a subgroup of participants from the Offspring- and 3^rd generation cohorts who attended exams 9 (2011–2014) and 2 (2008–2011), respectively, and underwent multi-detector computed tomography (CT) for evaluation of ectopic fat, including liver fat, between September 2008 and December 2011 (N=2,655). We excluded 416 individuals with significant alcohol consumption (> 7 drinks per week for women and > 14 drink per week for men) as recorded at the exams, and 4 individuals missing information on alcohol use. Of the remaining 2,235 subjects, 520 had no available measure of cognitive performance and 110 subjects were excluded due to dementia, stroke, or other neurological conditions, leaving a sample of 1,605 participants. We additionally excluded participants with missing components of the NFS yielding a final sample of 1,287 individuals. Subset of participants with NAFLD (N=378) was studied separately and also as part of the full sample. Data were obtained under a protocol approved by the Institutional review board of the Boston University Medical Center, and written informed consent was obtained from all participants.

Figure 1.

Flow chart of the study sample

NAFLD=Non-alcoholic Fatty Liver Disease

Assessment of fatty liver disease and risk for advanced fibrosis

Fatty liver was assessed using multi-detector CT with 8-slice MDCT technology (LightSpeed Ultra, General Electric, Milwaukee, WI). A calibration phantom (Image Analysis, Lexington, KY, USA) with a water equivalent compound (CT-Water, Light Speed Ultra, General Electric, Milwaukee, WI, USA) and calcium hydroxyapatite at 0, 75, and 150 mg/cm3 was placed under each participant. Three areas from the liver and one from an external phantom were measured, and the average of the liver measures were then used to create liver/phantom ratios. NAFLD was defined as having a liver/phantom ratio ≤0.33, consistent with prior Framingham Heart Study publications²⁰.

Severity of liver fibrosis in individuals with NAFLD was estimated using clinical and serum markers recorded at exams 9 and 2 for Offspring- and 3^rd generation cohorts, respectively, through the following NFS formula: NFS = −1.675 +0.037 × age [years] +0.094 × BMI [kg/m2] + 1.13 × impaired fasting glucose (IFG) or diabetes [yes = 1, no = 0] + 0.99 × AST/ALT ratio– 0.013 × platelet [x 109/L]– 0.66 × albumin g/dL) [10]. Participants were characterized into three categories based on the following, previously published cut-offs: NFS > 0.676 high probability advanced fibrosis (F3-F4), −1.455 < NFS ≤ 0.676 indeterminate probability of advanced fibrosis, and NFS < −1.455 as low probability of advanced fibrosis¹⁵.

Cognitive assessment

A standardized neuropsychological test was administered after exams 8 and 2 of the Offspring- and 3^rd generation cohorts, respectively, using standard administration protocols and trained examiners. We selected a subset of tests from the cognitive battery that were representative of several cognitive domains: The Wechsler Memory Scale (Logical Memory and Visual Reproduction) assessed verbal and visual memory. Time to complete Trail-making B minus time to complete Trail-making A was used to measure executive function. The difference in scores between Trail Making B and A is considered a pure measure of the executive abilities, as it subtracts out the simple sequencing and psychomotor demands common to both Trail Making Test A and B. The Similarities test (SIM) measured abstract reasoning skills. Lastly, the Hooper visual organization test (HVOT) was a measure of visual perception. Data derived from these tests can provide meaningful information for identification of early, subtle signs of cognitive impairment.^21,22 Details of these tests have been published previously.²³

Covariate assessment

Information on covariates was derived from exams 8 and 2 of the Offspring- and 3^rd generation cohorts, respectively. Educational achievement was defined as a three-class variable (high-school degree or less, some college and at least a college degree). Fasting blood glucose (FBG), insulin, lipid markers and serum alanine aminotransferase (ALT) and aspartate aminotransferase (AST) were measured in fasting morning blood samples. Diabetes was defined as FBG ≥126 mg/dL or use of an antidiabetic therapy, and impaired fasting glucose was defined as FBG ≥126 mg/dL without antidiabetic medications use. History of cardiovascular disease (myocardial infarction, angina, coronary insufficiency, stroke, transient ischemic attack, intermittent claudication, or congestive heart failure) was assessed by a review panel. The physical activity index (PAI) was calculated as a composite score based on information collected from a structured questionnaire¹. Body mass index (BMI) was defined by weight (in kilograms) divided by the square of height (in meters). Women were considered menopausal if their periods had stopped for at least 1 year. Alcohol consumption was defined as number of drinks per week through a series of physician administered questions. Participants were considered current smokers if they had smoked at least 1 cigarette per day in the year preceding the Framingham Heart Study examination. Volume of visceral adipose tissue was assessed using a 8-slice supine multidetector CT as previously described²⁴. Homeostasis model assessment of insulin resistance (HOMA-IR) was defined as fasting insulin (μIU/ml) × fasting glucose (mmol/l), and serum C-reactive protein (CRP) was measured using high-sensitivity assay. Total plasma homocysteine was measured using high-performance liquid chromatography with fluorometric detection²⁵. Data were obtained under a protocol approved by the institutional review board of the Boston University Medical Center, and written informed consent was obtained from all participants.

Statistical analysis

Descriptive statistics were calculated stratified by prevalent NAFLD (in the total sample) and fibrosis severity (in the subsample of individuals with NAFLD), with variables presented as mean±SD, median [intra-quartile range] or N (percent), as appropriate.

Linear regression models were constructed to estimate beta coefficients and standard errors for the association between NAFLD and cognitive performance in the total sample and between the 3 categories of risk for advanced fibrosis in individuals with NAFLD. Due to numerous potential confounders, we used 3 adjustment models: the basic model included age, sex, time between exam and cognitive assessment, education and study cohort. Model 2 included further adjustment for BMI, alcohol consumption, systolic blood pressure, smoking, history of cardiovascular disease, physical activity, CRP, the ratio of total- to High-density lipoprotein (HDL) cholesterol, visceral adipose tissue and HOMA-IR. Lastly, model 3 additionally controlled for prevalent diabetes. Since age, BMI, and IFG are components of the NFS score, we were concerned about possible collinearity when adjusting for age, BMI, and/or diabetes status. Therefore, collinearity was assessed in all models using the variance inflation factor (VIF) with a cut-off of 10.

Results

The prevalence of NAFLD in the current sample was 29%. The characteristics of the study sample according to the presence of NAFLD are presented in table 1. Compared to individuals without NAFLD, those with NAFLD were more likely to be men and to have prevalent diabetes, impaired fasting glucose, insulin resistance and hypertension. In addition, those with NAFLD had higher BMI, VAT and systolic blood pressure as well as elevated levels of CRP, ALT, AST and triglycerides. Participants with compared to those without NAFLD were also less engaged in physical activity and had decreased total- and HDL-cholesterol. Table 2 presents the characteristics of individuals with NAFLD, across categories of low, intermediate and high risk for advanced fibrosis as determined using the NFS. As expected, increased risk for advanced fibrosis among the subsample of participants with NAFLD was significantly associated with components of the NFS formula, i.e. older age, increased BMI, higher prevalence of diabetes and impaired fasting glucose, and reduced platelets count and levels of albumin (p<0.001 for all). Additionally, time between abdomen CT examination and cognitive assessment and frequency of alcohol consumption differed between the fibrosis groups, with non-linear associations. Risk for advanced fibrosis was also linked with decreased physical activity, increased insulin resistance, increased VAT, decreased ALT (but not AST), decreased total cholesterol and increased prevalence of hypertension and cardiovascular disease.

Table 1:

Characteristics of the study sample by NAFLD status (N=1,287)

	No NAFLD (N=909)	NAFLD (N=378)	p-value
Age, years	60.60±12.28	61.10±12.16	0.5046
Men, n (%)	398 (43.78)	221 (58.47)	<0.0001
Time between CT and NP (years)	1.57±1.00	1.55±1.02	0.8185
Alcohol, drinks/week	1.25 [0,4.5]	1 [0,5]	0.3175
BMI, kg/m2	26.86 [24.04,29.94]	31.14 [27.95,34.75]	<0.0001
Education			0.4638
< High-school degree	12 (1.32)	3 (0.79)
High-school degree	182 (20.02)	80 (21.16)
Some college	271 (29.81)	125 (33.07)
≥ College graduate	444 (48.84)	170 (44.97)
Physical Activity Index	34.75 [31.85,38.60]	33.70 [30.60,36.90]	0.0007
menopause, n(%)†	405 (95.29)	135 (97.83)	0.1917
Smoking, n (%)	37 (4.07)	16 (4.23)	0.8938
Diabetes, n (%)	62 (6.86)	77 (20.53)	<0.0001
IFG, n(%)‡	242 (26.62)	208 (55.03)	<0.0001
HOMA-IR	2.25 [1.55,3.18]	4.37 [2.76,6.05]	<0.0001
Statins, n (%)	296 (35.84)	161 (45.61)	0.0016
Stage 1 hypertension, n (%)	347 (38.22)	224 (59.26)	<0.0001
Systolic blood pressure	121.3±15.2	124.8±13.0	<0.0001
Prevalent CVD, n (%)	65 (7.15)	38 (10.05)	0.0805
CRP, mg/L	1.09 [0.59,2.41]	2.27 [1.13,4.65]	<0.0001
VAT, cm3	1846.95 [1074.05,2742.75]	3358.80 [2405.10,4404.40]	<0.0001
ALT, U/L	20 [15,25]	24 [19,34]	<0.0001
AST, U/L	21 [18,24]	22 [18,26]	0.0002
Platelets, [x 109/L]	233 [200,273]	234.5 [206,275]	0.3094
Albumin, g/dL	4.45±0.25	4.45±0.26	0.8384
Total cholesterol, mg/dL	189.73±35.50	183.10±38.70	0.0042
HDL-cholesterol, mg/dL	60 [49,74]	49 [41,63]	<0.0001
LDL-cholesterol, mg/dL	104 [84,125]	99 [79,120]	0.0110
Triglycerides, mg/dL	93 [70,127]	123 [91,174]	<0.0001

CT=computed tomography; MRI=Magnetic Resonance Imaging; BMI=Body Mass Index; IFG=Impaired Fasting Glucose; HOMA-IR=Homeostatic Model Assessment- Insulin Resistance; CVD=Cardiovascular Disease; CRP= C-Reactive Protein; VAT=Visceral Adipose Tissue; ALT= Alanine aminotransferase; AST= Aspartate aminotransferase; HDL=High-density lipoprotein; LDL=Low-density lipoprotein

^†Denominator is number of women

^‡Use of anti-diabetic medications was also included

Table 2:

Characteristics of participants with NAFLD by risk of fibrosis according to NFS (N=378)

Variable	Risk for advanced fibrosis
	Low (<1.455) (n=134)	Intermediate (−1.455 ≤NFS≤ 0.676) (n=205)	High (>0.676) (n=39)	p-value
Age, years	54.53±10.92	63.14±10.75	72.97±10.43	<0.0001
Men, n (%)	81 (60.45)	122 (59.51)	18 (46.15)	0.2536
Time between CT and NP (years)	1.49±0.95	1.67±1.02	1.15±1.17	0.0096
Alcohol, drinks/week	1 [0,6]	1.25 [0,5]	0 [0,2]	0.0227
BMI, kg/m2	29.49 [26.32,32.34]	31.83 [28.65,34.75]	36.14 [30.86,40.33]	<0.0001
Education				0.1280
< High-school degree	1 (0.75)	1 (0.49)	1 (2.56)
High-school degree	26 (19.40)	40 (19.61)	14 (35.90)
Some college	43 (32.09)	72 (35.29)	14 (35.90)
≥ College graduate	64 (47.76)	91 (44.61)	10 (25.64)
Physical Activity Index	34.10 [31.30,37.90]	33.70 [30.60,36.40]	31.65 [29.30,35.10]	0.0209
menopause, n(%)†	41 (100)	74 (97.37)	20 (95.24)	0.4388
Smoking, n (%)	6 (4.48)	8 (3.90)	2 (5.13)	0.9268
Diabetes, n (%)	9 (6.72)	48 (23.65)	20 (52.63)	<0.0001
IFG, n(%)‡	31 (23.13)	139 (67.80)	38 (97.44)	<0.0001
HOMA-IR	3.37 [2.30,5.18]	4.67 [3.19,6.44]	5.82 [4.28,7.57]	<0.0001
Statins, n (%)	45 (38.14)	92 (46.94)	24 (61.54)	0.0336
Stage 1 hypertension, n (%)	61 (45.52)	129 (62.93)	34 (87.18)	<0.0001
Prevalent CVD, n (%)	5 (3.73)	20 (9.76)	13 (33.33)	<0.0001
CRP, mg/L	2.14 [1.13,4.09]	2.08 [1.03,4.79]	2.88 [2.10,5.79]	0.0592
VAT, cm3	3010.4 [2210.9,3815.4]	3533.7 [2551.6,4556.3]	3910.5 [3298.1,5102.3]	<0.0001
ALT, U/L	29 [20,43]	24 [19,32]	18 [14,23]	<0.0001
AST, U/L	22 [19,28]	21 [18,26]	21 [18,26]	0.4501
Platelets, [x 109/L]	271.50 [233,317]	222 [200,251]	190 [179,213]	<0.0001
Albumin, g/dL	4.52±0.26	4.43±0.25	4.27±0.22	<0.0001
Total cholesterol, mg/dL	193.17±47.35	179.40±32.22	167.87±27.73	0.0002
HDL-cholesterol, mg/dL	49 [41,63]	48 [41,61]	57 [43,66]	0.5142
LDL-cholesterol, mg/dL	110 [90,131]	95 [76,115]	82 [69,97]	<0.0001
Triglycerides, mg/dL	115 [88,166]	129 [93,185]	133 [90,179]	0.3810

NFS=NAFLD Fibrosis Score; CT=computed tomography; MRI=Magnetic Resonance Imaging; BMI=Body Mass Index; IFG=Impaired Fasting Glucose; HOMA-IR=Homeostatic Model Assessment- Insulin Resistance; CVD=Cardiovascular Disease; CRP= C-Reactive Protein; VAT=Visceral Adipose Tissue; ALT= Alanine aminotransferase; AST= Aspartate aminotransferase; HDL=High-density lipoprotein; LDL=Low-density lipoprotein

^†Denominator is number of women

^‡Including diabetes treatment

In our sample, no associations between NAFLD and cognitive performance were identified (table 3). Yet, among participants with NAFLD, increased risk for advanced fibrosis was associated with poorer performance on the Trail-making test (TrB-TrA) (β=−0.11±0.05; p=0.028) and on the similarities test (β=−2.22±0.83; p=0.009) after adjustment for multiple potential confounders. No significant association was found between liver fibrosis severity and logical memory or visual reproduction delayed recall or the Hooper Visual Organization test (table 4). No evidence of collinearity was observed in our models.

Table 3:

Association of NAFLD with cognitive measures (N=1,287)

	Model 1		Model 2		Model 3
	β±SE	p-value	β±SE	p-value	β±SE	p-value
LMd	−0.01±0.22	0.954	−0.10±0.26	0.705	−0.13±0.26	0.610
VRd	−0.11±0.16	0.518	−0.15±0.19	0.425	−0.15±0.19	0.434
TrB-TrA†	−0.01±0.01	0.619	−0.01±0.02	0.472	−0.01±0.02	0.418
SIM	−0.09±0.19	0.644	−0.03±0.23	0.887	−0.07±0.23	0.746
HVOT†	0.05±0.03	0.138	0.03±0.04	0.477	0.02±0.04	0.528

^†Variable are log-transformed

Model 1: adjusted for age, sex, education, time between abdominal CT and cognitive tests, cohort

Model 2: adjusted for model 1 covariates plus BMI, alcohol, SBP, smoking, history of CVD, physical activity, CRP, the ratio of total cholesterol to HDL cholesterol, visceral adipose tissue and insulin resistance

Model 3: adjusted for model 2 covariates plus diabetes status

LMd=Logical Memory Delayed Recall; VRd= Visual Reproduction Delayed Recall; TrB-TrA=Trail making B minus Trail making A; SIM=Similarities test; HVOT= Hooper Visual Organization test; CT= computed tomography; BMI=Body Mass Index; SBP=Systolic Blood Pressure; CVD=Cardiovascular Disease, CRP=C-Reactive Protein; HDL=High-density lipoprotein

Table 4:

Association of fibrosis severity with cognitive measures (N=378)

		Model 1		Model 2		Model 3
		β±SE	p-value	β±SE	p-value	β±SE	p-value
LMd	Int vs. Low	0.24±0.40	0.553	0.13±0.46	0.772	0.06±0.47	0.899
	High vs. low	0.28±0.69	0.680	−0.14±0.88	0.871	−0.37±0.90	0.679
VRd	Int vs. Low	0.47±0.31	0.130	0.25±0.35	0.474	0.28±0.36	0.429
	High vs. low	−0.04±0.53	0.934	−0.65±0.66	0.326	−0.54±0.67	0.424
TrB-TrA†	Int vs. Low	−0.004±0.024	0.868	−0.02±0.03	0.372	−0.02±0.03	0.375
	High vs. low	−0.05±0.04	0.239	−0.11±0.05	0.034	−0.11±0.05	0.028
SIM	Int vs. Low	0.25±0.38	0.51	0.04±0.44	0.923	0.009±0.440	0.984
	High vs. low	−1.67±0.65	0.011	−2.05±0.83	0.013	−2.22±0.83	0.009
HVOT†	Int vs. Low	0.06±0.06	0.367	0.03±0.07	0.666	0.02±0.07	0.736
	High vs. low	0.03±0.11	0.780	−0.003±0.14	0.981	−0.01±0.14	0.971

Values in bold indicate results with p-value≤0.05

^†Variable are log-transformed

Model 1: adjusted for age, sex, education, time between abdominal CT and cognitive tests, cohort

Model 2: adjusted for model 1 covariates plus BMI, alcohol, SBP, smoking, history of CVD, physical activity, CRP, the ratio of total cholesterol to HDL cholesterol, visceral adipose tissue, HOMA-IR

Model 3: adjusted for model 2 covariates plus diabetes status

LMd=Logical Memory Delayed Recall; VRd= Visual Reproduction Delayed Recall; TrB-TrA=Trail making B minus Trail making A; SIM=Similarities test; HVOT= Hooper Visual Organization test; CT=computed tomography; BMI=Body Mass Index; SBP=Systolic Blood Pressure; CVD=Cardiovascular Disease, CRP=C-Reactive Protein; HDL=high-density lipoprotein

Protein; HDL=high-density lipoprotein

Discussion

In this sample of community-dwelling individuals, NAFLD per se was not associated with cognitive performance. However, evidence of advanced fibrosis according to the NAFLD fibrosis score was associated with poorer performance on tests assessing executive function and abstract reasoning.

Studies addressing the relationship between NAFLD and cognitive function are scarce. In contrast to our findings of no association between NAFLD and cognition, a previous study found that individuals with NAFLD had poorer performance on the Serial Digit Learning Test, a measure of learning and memory, but not on other cognitive tests (i.e. the Simple Reaction Time Test and the Symbol-Digit Substitution Test, which measure visual-motor speed and visual attention, respectively)⁴. Among other explanations, this discrepancy may be due to the assessment of different cognitive domains as well as by the older age of participants in the current vs. the previous study (mean age 61±12 vs. 37±0.3, respectively).

The observation that severity of liver fibrosis, but not NAFLD per se, may be linked with poorer cognitive function is in line with a growing body of literature stressing the important role of fibrosis degree in various health conditions, including vascular function,^18,26 morphological and functional cardiac alterations,²⁷ carotid atherosclerosis,^28,29 chronic kidney disease³⁰ and stroke.³¹ Even stronger support to our findings arrives from evidence suggesting that fibrosis severity vs. NAFLD per se is a stronger risk factor for various health outcomes. Indeed, in a cohort of young and middle-aged Korean adults, a strong association was found between the NFS and risk of diabetes, while a weaker association with the presence of NAFLD per se was also demonstrated.³² Similarly, fibrosis stage, but not,^12,17 or to a lesser extent³³ NAFLD per se was suggested as a strong risk factor for mortality, mainly from cardiovascular diseases. In addition, fibrosis severity, but not NAFLD, were linked with the presence of white matter lesions.³⁴ In the context of these prior findings, our study suggests that cognitive impairment may be an additional consequence of NAFLD with liver fibrosis even in the pre-cirrhotic stages of the disease, and not merely a consequence of covert hepatic encephalopathy.^35,36 While we cannot rule out the possibility that the poor cognitive performance observed in individuals with high probability of advanced fibrosis in our sample was due to minimal hepatic encephalopathy,³⁷ this explanation is less likely in a sample of community-dwelling individuals who arrive for routine examinations.

Besides cell death, which by itself triggers not only fibrinogenic but also inflammatory signaling cascade, inflammation is the main underlying process promoting the development of liver fibrosis.³⁸ Such local overexpression of inflammatory mediators may lead to systemic endothelial dysfunction and atherosclerosis,³⁹ as supported by multiple clinical observations linking fibrosis severity to vascular pathology^18,29,34, as well as to stroke³¹, cardiovascular disease⁴⁰ and its risk factors.⁴¹ In line with these observations, the link we found between fibrosis severity and measures of executive function but not with other domains tested further stresses the possible vascular involvement, as executive function reflects the integrity of the frontal lobe,⁴² which is more prone to subclinical vascular injury.⁴³ Alternative explanations for the association between liver fibrosis and poor cognitive function include oxidative stress, insulin resistance and the secretion of adipokines, with the latter affecting not only lipid metabolism but also inflammatory and fibrotic processes in NAFLD.⁴⁴ These possible underlying mechanisms have been implicated both in liver fibrosis⁴⁵ and cognitive impairment.^46–48

A major strength of our study is the assessment of liver fibrosis in a large community-based cohort with NAFLD based on ultrasound. Additionally, our study included a wealth of data on potential covariates as well as validated and comprehensive cognitive tests. There were also several limitations: First, this is an observational study with a cross-sectional design which does not permit the determination of a temporal sequence between fibrosis score and cognitive measures, hence, cause-effect inferences cannot be made. Second, the low sensitivity of the NFS is low (between 50–67%)⁴⁹, thus a proportion of participants identified as having advanced fibrosis were misclassified. We expect this to result in attenuation of the association. Moreover, we used NFS cutoffs which were validated in hospital-based or case-control studies but their validity in the general population is unknown. Third, the number of participants with the highest risk for advanced fibrosis was small, thus statistical power was limited. Lastly, the study population was restricted to individuals of European ancestry, from one geographic area and of a relatively high socioeconomic status, the external validity for other ethnicities and varying populations is uncertain.

In conclusion, our findings add to the large bulk of evidence supporting the importance of liver fibrosis severity as a marker of various health outcomes, by showing that among NAFLD patients detected by ultrasonography, the presence of significant liver damage may be also linked with brain health. The NFS can serve as a useful biomarker because it is noninvasive and incorporates routinely determined and easily available clinical and biochemical variables¹⁵. Future prospective studies are necessary to evaluate the value of this system scoring in predicting incident cognitive impairment and dementia, and whether it can improve previously validated risk prediction models. In addition, it remains to explore the possible role of liver fibrosis in brain aging and pathologies and to test whether treatments that directly target inflammation and fibrosis in NAFLD can ameliorate the risk for cognitive decline.

Key points.

• Non-alcoholic fatty liver disease and the degree of liver fibrosis are both associated with multiple adverse health outcomes, yet their association with cognitive function in old adults is not clear.

• We demonstrate that cognitive function of participants in the community-based Framingham Study who have non-alcoholic fatty liver disease do not differ significantly from those who do not have this condition.

• However, our study demonstrates that participants with non-alcoholic fatty liver disease who are at high risk for advanced liver fibrosis show poorer executive function compared to those with lower risk for fibrosis.

• Our findings suggest that liver fibrosis, even prior to its clinical manifestation, may be related to impairment in cognitive function.

List of abbreviations:

NAFLD

Non-alcoholic fatty liver disease

NFS

NAFLD fibrosis score

FHS

Framingham Heart Study

IFG

impaired fasting glucose

HVOT

Hooper visual organization test

ALT

alanine aminotransferase

AST

aspartate aminotransferase

on information collected from a structured questionnaire¹

BMI

Body mass index

HOMA-IR

Homeostasis model assessment of insulin resistance

CRP

C-reactive protein

HDL

High-density lipoprotein