Abstract
Background and Purpose:
We examined the association of chronic liver disease with cognition and brain imaging markers of cognitive impairment using data from two large randomized controlled trials that included participants based on diabetes and hypertension, two common systemic risk factors for cognitive impairment and dementia.
Methods:
We performed post hoc analyses using data from the Action to Control Cardiovascular Risk in Diabetes (ACCORD) and Systolic Blood Pressure Intervention Trial (SPRINT) studies, which included participants with diabetes and hypertension, respectively. Data were from the NHLBI Biologic Specimen and Data Repository Information Coordinating Center. In ACCORD, our measure of chronic liver disease was the Dallas Steatosis Index (DSI). In SPRINT, we used self-reported chronic liver disease. We used linear regression to evaluate the association between the measure of chronic liver disease and both baseline and longitudinal cognitive test performance and brain magnetic resonance imaging volume measurements.
Results:
Among 2,969 diabetic participants in ACCORD, the mean age of participants was 62 years, 47% were women. The median DSI was 1.0 (IQR, 0.2–1.8); a DSI of 1.0 corresponds to approximately a >70% probability of having NAFLD. Among 2,890 hypertensive participants in SPRINT, the mean age was 68 years, and 37% were women, and 60 (2.1%) had chronic liver disease. There were no consistent associations between liver disease and cognitive performance or brain volumes at baseline or longitudinally after adjustment.
Conclusion:
Markers of chronic liver disease were not associated with cognitive impairment or related brain imaging markers among individuals with diabetes and hypertension.
Keywords: liver disease, nonalcoholic fatty liver disease, brain health, cognition
Introduction
Common neurodegenerative etiologies of dementia such as Alzheimer’s disease, Lewy Body, and cerebrovascular disease account for only about 40% of the variance in age-related cognitive impairment, suggesting other mechanisms at play in cognitive decline.1 Chronic liver disease is increasingly prevalent and may have an under-recognized impact on brain health.
The epidemiology of chronic liver disease has dramatically changed over the past several decades, and approximately 25% of people may have a chronic liver disease.2, 3 There is emerging interest in the impact of chronic liver disease on brain health.4–6 The most common chronic liver diseases are nonalcoholic fatty liver disease (NAFLD), alcoholic liver disease, and viral hepatitis,3 and these conditions frequently overlap.7 Apart from what is known about the impact of end-stage liver disease on cognition in terms of hepatic encephalopathy,8, 9 there is growing evidence that less severe forms of chronic liver disease may impact brain health as well. Though inconsistent, prior studies suggest that NAFLD and other forms of chronic liver disease are associated cognitive measures and brain imaging markers related to cognitive impairment.10–15 The impact of chronic liver disease on brain health in people vulnerable to cognitive impairment and dementia, such as people with diabetes14, 16, 17 or hypertension,14, 17 is unclear.
We examined the association of measures of chronic liver disease with cognition and brain imaging markers of cognitive impairment using two large randomized controlled trial datasets, Action to Control Cardiovascular Risk in Diabetes (ACCORD) and Systolic Blood Pressure Intervention Trial (SPRINT) studies, which included participants with diabetes and hypertension, respectively.
Materials and Methods
Design
We performed retrospective analyses using data from two randomized controlled trials. First, we evaluated participants in ACCORD, which evaluated whether intensive compared to standard management of hyperglycemia, blood pressure, and lipid levels reduced cardiovascular event rates and mortality.18 Specifically, we used data from participants in the ACCORD Memory in Diabetes (ACCORD-MIND) sub-study, which evaluated cognition and brain imaging in 2,977 patients from within the overall ACCORD cohort.19 Second, we evaluated participants in the SPRINT trial, which randomly assigned 9,361 patients with high blood pressure and increased cardiovascular risk to an intensive (SBP<120 mm Hg) or standard (SBP<140 mm Hg) blood pressure control group and monitored the primary outcomes of myocardial infarction, stroke, heart failure, or death from cardiovascular causes.20 Specifically, we used data from two SPRINT sub-studies. The first was a sub-study of cognition in a subset of SPRINT participants; these participants were administered a cognitive screening battery and extended battery consisting of tests evaluating memory, recall, and processing speed.21 Of the 9,361 SPRINT participants overall, 2,921 (31%) were included in this cognitive function sub-study. The second was the SPRINT-MIND study which investigated the effects of intensive blood pressure control on brain imaging outcomes among 670 SPRINT participants.22 The Weill Cornell Medicine Institutional Review Board granted an exemption for the ACCORD-MIND analyses. The University of Michigan Institutional Review Board granted an exemption for the SPRINT analyses.
Population
The ACCORD-MIND analysis included adult participants from the ACCORD cohort, aged 55–80 years and recruited from clinics in North America. All participants had a diagnosis of type 2 diabetes, high HbA1c concentrations (>7.5%), and were at increased risk of cardiovascular events.19 A subset of ACCORD-MIND participants was recruited for the MRI portion of the study, with exclusions largely related to MRI safety and tolerability.19
The SPRINT analysis included all participants with cognitive and/or brain imaging data from the SPRINT trial.23 The SPRINT trial population consisted of participants aged >= 50 years who had a systolic blood pressure (SBP) between 130–180 mm Hg at their screening visit. Those with a prior diagnosis of dementia, medications used for dementia therapy, diabetes mellitus, history or prior stroke, and residence in a nursing home were excluded from SPRINT. A subset of SPRINT participants were recruited for the cognitive function and brain MRI sub-study, which we also included.22
Measurements
In our analysis of the ACCORD-MIND data, the chronic liver disease measure was the Dallas Steatosis Index (DSI). The Dallas Steatosis Index is a tool for identifying patients with NAFLD; it has superior discrimination in screening for NAFLD compared to other risk scores, and it was derived in a population-based, multi-ethnic sample.24 The DSI risk thresholds for low-risk, intermediate-risk and high-risk of NAFLD are established at <20%, 20–50% and ≥50%, validated to have approximately 90% sensitivity and 90% specificity for predicting likelihood of NAFLD.24 In our analyses, we used the DSI as a continuous variable in our primary approach, and as a categorical variable in our secondary approach. When using the DSI as a categorical variable, we categorized DSI in two ways: as high (> 1.4; >80% probability of NAFLD) versus low (<−1.4; <20% probability of NAFLD), and quintile 1 versus quintile 5. The outcomes of interest were four cognitive tests (Mini-mental status Examination, Digital Symbol Substitution Test (DSST), Stroop Color-Word Test, Rey Auditory Verbal Learning Test), and two brain MRI measures (total brain volume and abnormal white matter volume).19, 25–27 This cognitive battery was designed by the ACCORD-MIND investigators to assess various domains including global mental status, memory, psychomotor speed, and executive function.19 Higher test scores indicate better cognition and brain health, with the exception of the Stroop test and white matter volume changes on MRI. Total brain volume can reflect brain atrophy, a measure that may reflect neurodegeneration, while white matter hyperintensity volume may reflect the presence of cerebrovascular disease linked to vascular cognitive impairment and dementia.28 DSST assessments were given at baseline, 20-month, and 40-month follow-up points. Brain MRI was conducted at baseline and at 40 months. We analyzed both baseline measures and change in measures between baseline and the 40-month follow-up visit. Covariates included age, sex, race/ethnicity, education, body mass index, hypertension, hyperlipidemia, baseline hemoglobin A1c, smoking, alcohol overuse (>14 drinks per week for men, >7 drinks per week for women), and serum creatinine.
In our analysis of the SPRINT data, the chronic liver disease measure was any self-reported chronic liver disease. Specifically, patients were asked: “Have you ever been told by a physician that you have: chronic hepatitis or cirrhosis of the liver?” The outcomes of interest were cognitive function tests and brain MRI data, obtained at baseline and at a 4-year follow-up. Specifically, six cognitive tests (Trail Making Tests A and B, Animal Naming Test, Boston Naming Test, Digit Span Forwards and Backwards Tests), and abnormal white matter volume from brain MRI were assessed. These cognitive tests were components of an extended cognitive battery to measure attention, memory, language, and executive function.21, 23 We chose these tests from the extended SPRINT panel to align with the tests chosen for the ACCORD-MIND analyses. For these tests and imaging parameters, higher values indicate better cognitive health, except for the Trails tests and abnormal white matter brain volume. Cognitive testing and brain MRI were conducted at baseline and at 4 years. Both baseline measures and change in measures between baseline and the 4-year follow-up visit were analyzed. Covariates included age, sex, race/ethnicity, education, body mass index, hypertension, hyperlipidemia, diabetes, subclinical cardiovascular disease, smoking, alcohol overuse (>14 drinks per week for men, >7 drinks per week for women), and serum creatinine. Abnormal brain white matter volume model was additionally adjusted for total intracranial volume.
Statistical Analyses
We used standard descriptive statistics to summarize participant characteristics. In our analysis of the ACCORD-MIND data, we evaluated the association between DSI and 1) baseline cognition and brain volumes and 2) change in cognition and brain volumes over 40 months of follow-up. We treated DSI as a continuous measure in the primary analysis. In secondary analyses, we categorized DSI: 1) high (> 1.4; >80% probability of NAFLD) versus low (<−1.4; <20% probability of NAFLD) and 2) quintile 5 versus quintile 1. Using multiple linear regression, we evaluated the association of the DSI with cognitive and brain imaging markers while adjusting for demographics and cardiometabolic risk factors. Regression models were incrementally adjusted: Model 1 (unadjusted), Model 2 (adjusted for age, sex, race/ethnicity), and Model 3 (additionally adjusted for body mass index, hypertension, hyperlipidemia, baseline hemoglobin A1c, smoking, alcohol overuse (>14 drinks per week for men, >7 drinks per week for women), and serum creatinine). Waist circumference was not adjusted for because of high collinearity (variation inflation factor>4) with body mass index; there was no other significant collinearity. In a post-hoc sensitivity analysis, we added baseline LDL to the most adjusted models. In the ACCORD-MIND analyses, the beta regression coefficients represent the relationship between each 1.0 unit increase in DSI and individual cognitive test or brain imaging parameters. We tested for interaction by ACCORD-MIND treatment arm (intensive versus standard glycemic control) in longitudinal analyses. In the analysis of SPRINT data, we evaluated the association between chronic liver disease and 1) baseline cognition and abnormal brain white matter volume and 2) change in cognition and abnormal brain white matter volume after 4 years. In all analyses, chronic liver disease was modeled as a categorical variable – absent or present chronic liver disease. Using multiple linear regression, we evaluated the association of the chronic liver disease with cognitive and brain imaging markers while adjusting for age, sex, race/ethnicity, education, body mass index, hypertension, hyperlipidemia, diabetes, subclinical cardiovascular disease, smoking, alcohol overuse, and serum creatinine. Models were incrementally adjusted in a similar manner as done for the ACCORD-MIND analyses. In the SPRINT analyses, the beta regression coefficients represent the difference in outcome parameters in those with versus without chronic liver disease. Brain volume models were adjusted for total brain volume by adjusting for total intracranial volume.
The threshold of statistical significance was set at α = 0.05. Multiple comparison penalties were not applied for these post-hoc secondary analyses. Statistical analyses for ACCORD-MIND were performed using SAS (version 9.4) by C.Z. with NP, and statistical analyses for SPRINT were performed using RStudio (Version 1.3) by M.M. with E.T.
Results
ACCORD-MIND: Sample characteristics
In our analysis of the ACCORD-MIND data, 2,969 participants had complete data for this analysis. Of these participants, 614 had brain MRI data. The mean age of participants was 62 years, and 47% were women. The median hemoglobin A1c was 8.1% (IQR, 7.5–8.8). The median DSI was 1.0 (IQR, 0.2–1.8); a DSI of 1.0 corresponds to approximately a >70% probability of having NAFLD. Using validated cut-offs, 1,144 participants (38.4%) had a high DSI. Participants with a high quintile DSI had higher BMI and a greater prevalence of metabolic comorbidities than people with a low quintile DSI (Table 1).
Table 1.
Baseline Characteristics of Primary Analysis Study Sample (ACCORD-MIND)
| Characteristic* (n=2,969) | All subjects | Quintile 5 | Quintile 1 |
|---|---|---|---|
| Age (median, IQR) | 62 (58–67) | 61 (58–64) | 63 (59–69) |
| Female (%) | 47 | 56 | 42 |
| Black (%) | 16 | 2 | 48 |
| White (%) | 70 | 82 | 42 |
| Education (%) | |||
| Less than high school graduate | 13 | 13 | 17 |
| High school graduate or GED | 26 | 24 | 29 |
| Some college or technical school | 35 | 39 | 27 |
| College graduate or more | 27 | 25 | 26 |
| BMI, kg/m2 (median, IQR) | 33 (29–37) | 38 (35–41) | 27 (25–31) |
| Hypertension (%) | 93 | 97 | 88 |
| Hyperlipidemia (%) | 85 | 85 | 81 |
| Baseline LDL (median, IQR) | 99 (80–123) | 101 (79–125) | 99 (81–120) |
| Baseline A1c, % (median, IQR) | 8.1 (7.5–8.8) | 8.2 (7.6–8.8) | 8.1 (7.5–8.9) |
| Serum creatinine, mg/dL (median, IQR) | 0.9 (0.7–1.0) | 0.8 (0.7–1.0) | 0.9 (0.8–1.0) |
| Alcohol overuse (%) | 1 | 1 | 0 |
| ALT†, IU/L (median, IQR) | 25 (19–33) | 30 (24–42) | 18 (15–23) |
| Dallas Steatosis Index (median, IQR) | 1.05 (0.23–1.81) | 2.4 (2.1–2.7) | −0.8 (−1.2--0.4) |
| Baseline MMSE‡ (median, IQR) | 28 (26–29) | 28 (27–29) | 27 (25–29) |
| Baseline DSST§ (median, IQR) | 53 (42–63) | 56 (46–66) | 48 (36–57) |
| Baseline Stroop (median, IQR) | 28 (21–38) | 27 (20–35) | 30 (22–42) |
| Baseline RAVLT∥ (median, IQR) | 8 (6–10) | 8.1 (6.3–9.9) | 6.9 (5.1–8.7) |
| Baseline Total brain volume | 923 (858–998) | 916 (867–991) | 916 (844995) |
| Baseline WM Volume, cm3 (median, IQR) | 0.9 (0.3–1.9) | 0.6 (0.2–1.2) | 1.2 (0.4–2.8) |
Data presented as % unless otherwise stated. Quintile 5 refers to participants with a DSI in 5th quintile, Quintile 1 refers to participants in 1st quintile.
Alanine aminotransferase
Mini-mental status examination
Digital Symbol Substitution Test
Rey Auditory Verbal Learning Test
ACCORD-MIND: Neurocognitive and imaging measures
There were no consistent associations of DSI with baseline and longitudinal cognitive testing and brain volume measures among individuals with diabetes in ACCORD MIND once adjusting for demographics (Table 2). Higher DSI was also not consistently associated with worse cognition or brain volume measures at baseline, or changes in these measures over 40 months, once adjusting for demographics, comorbidities, and other potential confounders (Table 2). There were no associations for the Mini-mental status Examination, Stroop Color-Word Test, and Rey Auditory Verbal Learning Test; however, higher DSI was associated with better performance on the baseline DSST (β=0.89; P=0.01). Overall results were unchanged when treating DSI as a categorical variable, specifically comparing high (> 1.4; >80% probability of NAFLD) versus low (<−1.4; <20% probability of NAFLD) DSI, and when comparing quintile 5 versus quintile 1 DSI (Supplemental Table 1 and 2). There were no significant interactions between DSI and ACCORD-MIND treatment arm in analyses of change in cognition or brain imaging (P>0.05 for all interaction terms). Results were unchanged when adding baseline LDL to models in a post-hoc sensitivity analysis (Table 2). Supplemental Table 3 shows parameter estimates for all covariates.
Table 2.
Association* of Dallas Steatosis Index† with Baseline and Longitudinal Cognitive and Brain Volume Measures among Individuals with Diabetes in ACCORD-MIND
| Baseline | Change | ||||
|---|---|---|---|---|---|
| Outcome | Model‡ | β | P | β | P |
| Mini-Mental Status Exam | 1 | 0.26 | <0.01 | 0.05 | 0.13 |
| 2 | 0.05 | 0.25 | 0.03 | 0.44 | |
| 3 | −0.01 | 0.83 | 0.12 | 0.04 | |
| 4 | −0.01 | 0.85 | −0.04 | 0.90 | |
| Digit-Symbol Substitution Test | 1 | 2.63 | <0.01 | −0.05 | 0.71 |
| 2 | 0.39 | 0.09 | −0.06 | 0.67 | |
| 3 | 0.90 | <0.01 | 0.08 | 0.72 | |
| 4 | 0.94 | <0.01 | 0.56 | 0.71 | |
| Stroop Color-Word Test | 1 | −1.45 | <0.01 | −0.11 | 0.62 |
| 2 | 0.28 | 0.30 | −0.01 | 0.97 | |
| 3 | −0.26 | 0.47 | −0.01 | 0.97 | |
| 4 | −0.30 | 0.41 | −0.43 | 0.86 | |
| Rey Auditory Verbal Learning | 1 | 0.32 | <0.01 | 0.06 | 0.06 |
| 2 | 0.06 | 0.18 | −0.01 | 0.78 | |
| 3 | 0.01 | 0.81 | <0.01 | 0.98 | |
| 4 | 0.01 | 0.87 | 0.02 | 0.96 | |
| Total brain volume | 1 | 4.08 | <0.01 | 1.67 | <0.01 |
| 2 | 1.34 | 0.23 | 1.10 | 0.13 | |
| 3 | 0.66 | 0.65 | −0.31 | 0.74 | |
| 4 | 0.59 | 0.69 | −4.27 | 0.53 | |
| Abnormal WM volume§ | 1 | −0.15 | <0.01 | −0.01 | 0.83 |
| 2 | −0.10 | 0.02 | 0.04 | 0.43 | |
| 3 | −0.04 | 0.49 | 0.02 | 0.76 | |
| 4 | −0.08 | 0.84 | −0.41 | 0.47 | |
Higher test scores and brain volumes indicate better cognition and brain health, with the exception of the Stoop Color-Word test and abnormal white matter volume.
DSI measure as a continuous variable in primary approach
Model 1 was unadjusted. Model 2 was adjusted for age, sex, race/ethnicity, and education. Model 3 was additionally adjusted for body mass index, hypertension, hyperlipidemia, baseline hemoglobin A1c, smoking, alcohol overuse (>14 drinks per week for men, >7 drinks per week for women), and serum creatinine. Model 4 was additionally adjusted for baseline LDL. All brain volume models were also adjusted for total intracranial volume.
Baseline abnormal white matter volume was log transformed.
SPRINT: Sample characteristics
With regards to the SPRINT trial, a total of 9,361 participants were randomized into the trial. Within our analysis of the cognitive sub-study, 2,890 participants completed baseline cognitive tests, 2,268 completed the 4-year follow-up cognitive testing, 670 completed the baseline MRI, and 448 completed the 4-year follow-up MRI. The mean age of the 2,890 participants with complete baseline cognitive testing was 68 years, and 37% were women. Mean systolic blood pressure was 138.9 mmHg, and mean diastolic blood pressure was 77.3 mm Hg. Of the 2,890 participants in the baseline cohort, 60 (2.1%) had chronic liver disease, who had less education and more smoking than those without liver disease (Table 3).
Table 3.
Baseline Characteristics of Primary Analysis Study Sample (SPRINT)
| Characteristic* | No Liver Disease | Liver Disease | P |
|---|---|---|---|
| n | 2,828 | 60 | |
| Age, age (median, IQR) | 68 (62–75) | 63 (58–73) | 0.02 |
| Female | 37% | 30% | 0.35 |
| Black | 30% | 50% | 0.002 |
| White | 67% | 47% | |
| Education | |||
| Less than high school graduate | 8% | 13% | |
| High school graduate or GED | 16% | 22% | |
| Some college or technical school | 29% | 33% | |
| College graduate or more | 47% | 32% | |
| BMI, kg/m2 (median, IQR) | 29 (26–33) | 29 (27–33) | 0.55 |
| Systolic blood pressure, mmHg (median, IQR) | 137 (128–148) | 136 (128–147) | 0.96 |
| Diastolic blood pressure, mmHg (median, IQR) | 77 (69–85) | 80 (70–91) | 0.10 |
| Hyperlipidemia | 46% | 45% | 0.99 |
| Diabetes | 1% | 2% | 1.00 |
| Serum creatinine mg/dL (median, IQR) | 1.0 (0.9–1.2) | 1.0 (0.9–1.3) | 0.49 |
| Alcohol overuse | 4% | 5% | 0.89 |
| Cigarette Smoker | 13% | 33% | <0.001 |
| Baseline Trail A (seconds) (median, IQR) | 35 (27–44) | 32 (27–45) | 0.90 |
| Baseline Trail B (seconds) (median, IQR) | 25 (9–45) | 29 (4–43) | 0.70 |
| Baseline Animals Naming Test (median, IQR) | 18 (14–21) | 17 (15–23) | 0.72 |
| Baseline Digit Span Forwards (median, IQR) | 9 (8–11) | 9 (8–11) | 0.51 |
| Baseline Digit Span Backwards (median, IQR) | 7 (6–9) | 7 (6–8) | 0.14 |
| Baseline Boston Naming Test (median, IQR) | 13 (10–14) | 13 (10–14) | 0.92 |
| Baseline WM Volume, cm3 (median, IQR) | 1.87 (0.80–4.36) | 1.99 (1.21–5.07) | 0.69 |
Data presented as % unless otherwise stated. Liver history data not available for 2 participants.
SPRINT: Neurocognitive and imaging measures
In unadjusted model results, there were no significant associations of self-reported liver disease with baseline cognitive performance and brain volume measures among participants in SPRINT. After adjustment for demographics and confounders, there remained no association between liver disease and performance on any baseline cognitive test (Table 4). There were no statistically significant associations between chronic liver disease and change in cognitive function over the 4-year period in all models, except for opposite patterns for the Boston Naming Test (β=−0.62, P=0.03) and the Digit Span Backwards (β=0.76, P=0.01). There was also no statistically significant interaction between chronic liver disease and abnormal white matter brain volume at both baseline and over time in all models.
Table 4.
Association* of Liver Disease with Baseline and Longitudinal Cognitive Performance and Brain Volume Measures in SPRINT
| Baseline | Change | ||||
|---|---|---|---|---|---|
| Outcome | Model† | β | P value | β | P value |
| Trail A (seconds) | 1 | 0.29 | 0.86 | 2.20 | 0.58 |
| 2 | 0.06 | 0.97 | 1.70 | 0.65 | |
| 3 | −0.18 | 0.91 | 1.68 | 0.66 | |
| Trail B (seconds) | 1 | −0.55 | 0.83 | 21.69 | 0.08 |
| 2 | −0.06 | 0.98 | 15.88 | 0.15 | |
| 3 | −0.05 | 0.98 | 14.35 | 0.19 | |
| Animal Naming Test | 1 | 0.41 | 0.54 | −0.12 | 0.86 |
| 2 | 0.95 | 0.12 | −0.21 | 0.76 | |
| 3 | 1.03 | 0.09 | −0.15 | 0.83 | |
| Boston Naming Test | 1 | 0.15 | 0.72 | −0.61 | 0.03 |
| 2 | 0.75 | 0.04 | −0.63 | 0.02 | |
| 3 | 0.66 | 0.06 | −0.62 | 0.03 | |
| Digit Span Forwards | 1 | −0.13 | 0.68 | −0.18 | 0.58 |
| 2 | 0.02 | 0.94 | −0.18 | 0.59 | |
| 3 | −0.03 | 0.92 | −0.19 | 0.57 | |
| Digit Span Backwards | 1 | −0.51 | 0.09 | 0.78 | 0.01 |
| 2 | −0.19 | 0.49 | 0.77 | 0.01 | |
| 3 | −0.18 | 0.53 | 0.76 | 0.01 | |
| Abnormal WM volume | 1 | −0.95 | 0.51 | −0.97 | 0.62 |
| 2 | −0.04 | 0.98 | −0.14 | 0.94 | |
| 3 | −0.26 | 0.85 | −1.17 | 0.50 | |
Higher score and brain volumes indicate better cognition and brain health, with the exception of Trails tests and abnormal white matter volume.
Model 1 was unadjusted. Model 2 was adjusted for age, sex, race/ethnicity, and education. Model 3 was additionally adjusted for body mass index, hypertension, hyperlipidemia, diabetes, sub-clinical cardiovascular disease, smoking, and alcohol overuse (>14 drinks per week for men, >7 drinks per week for women). Brain volume models were adjusted for total intracranial volume.
Discussion
Among participants with diabetes and hypertension in ACCORD-MIND and SPRINT, respectively, chronic liver disease was not associated with cognitive impairment or worse brain volumes measures. In our analysis of ACCORD-MIND, a validated NAFLD score was not associated with a range of brain health measures. In our analysis of SPRINT, self-reported liver disease similarly had no consistent associations with brain health parameters, with no compelling evidence of an impact on longitudinal outcomes brain imaging outcomes.
Taken together, our analyses of ACCORD-MIND and SPRINT did not identify detrimental associations of chronic liver disease with cognitive function or brain volumes within diabetic and hypertensive study populations. Our findings have several possible explanations. First, it is possible that liver disease itself does not contribute independently to brain health in the context of its etiologic drivers, in this case metabolic syndrome (diabetes and/or hypertension). Second, it is possible that our measures of chronic liver disease – the DSI and self-reported chronic liver disease – are not the correct measures. The DSI identified 1,144 people in the ACCORD MIND cohort as having possible NAFLD based on high DSI. The high prevalence of NAFLD in the ACCORD MIND cohort is related to the fact that all participants in ACCORD had diabetes, and is in line with the global prevalence of NAFLD.29 The smaller number of people identified by self-report in SPRINT (2%) is likely a reflection of the specific but insensitive nature of self-report for chronic liver disease, and the narrower range of conditions queried by the self-report question. These measures are thus complementary features of the two analyses that help us interrogate the association of liver markers with brain health, acknowledging that neither DSI or self-report are ideal measures, but rather important steps in refining the approach to studying the brain-liver axis. The study of the brain-liver axis in cognition and brain health is a nascent field.4 Though it remains possible that chronic liver disease is an independent risk factor for cognitive impairment, our findings suggest that future studies should utilize direct, sensitive markers of liver disease such as transient elastography rather than markers of steatosis alone or self-report.30 Future studies should also utilize a comprehensive neuropsychological test battery designed to be sensitive to the earliest changes of cognitive decline for neurodegenerative processes to minimize the risk of falsely negative results. Similarly, more specific brain imaging markers, rather than total brain volume and abnormal white matter volume, may be needed to identify the impact of liver disease on brain health.
NAFLD’s association with cognitive function as demonstrated in prior studies may be an epiphenomenon, related to factors such as obesity, insulin resistance, hypertension, and other comorbidities.17 Indeed, there is conflicting data on the impact of chronic liver disease and specifically NAFLD on cognition. In a large cross-sectional study of 4,472 participants below the age of 59,10 NAFLD defined by ‘moderate-to-severe ‘steatosis by ultrasound was associated with reduced cognitive learning, poor memory, attention, and concentration independent of metabolic risk factors. However, there was no association between NAFLD and psychomotor speed or visuospatial function after adjustments for metabolic covariates.10 It may be statistically invalid, however, to ‘adjust for confounders ‘when the steatosis which defines NAFLD is actually distal to the covariates in the causal pathway. Indeed, in a separate analysis of the same data, we found that among subjects without liver disease (including NAFLD), the same effects on cognitive testing were observed from education, diabetes, body mass index, and smoking.31 Alternatively, this variation in association between different cognitive tests may suggest that NAFLD might affect cognitive function through region-specific processes as opposed to diffuse cortical dysfunction. Yet, in the Framingham Heart Study, computed tomography evidence of NAFLD was not associated with cognitive function.13 Instead, only participants with higher fibrosis scores (indirect evidence of risk for fibrotic liver disease) had worse performance in tests of executive function and reasoning, suggesting that more advanced liver disease impacts cognition.12, 13 In another recent analysis of the Coronary Artery Risk Development in Young Adults (CARDIA) study by Gerber et al., the association between NAFLD and cognitive function was attenuated after adjustment for cardiovascular risk factors, and NAFLD was not predictive of cognitive function during follow-up assessment.17 Overall, the substantial heterogeneity in study design, liver disease definitions, and outcomes precludes confident conclusions regarding the presence or absence of an impact of chronic liver disease on brain health.
The strengths of our analyses include the use of randomized clinical trial data sets with large study populations. By focusing on patients with diabetes and hypertension, we narrowed our population of interest to those at risk for cognitive decline and brain volume changes. Our analyses should be interpreted considering some limitations. First, generalizability to other populations should be considered with caution, as the SPRINT study excluded patients with baseline diabetes, stroke, dementia, and severe heart failure. Similarly, ACCORD-MIND excluded patients with serious illness, hypoglycemic events, or patients with a creatinine >1.5mg/dL. Both are randomized trials with study populations that may not be representative of the general population. Second, our use of the DSI and self-reported chronic liver disease may have led to misclassification error. Non-differential misclassification (for example with the DSI) can bias analyses towards the null. Conversely, differential misclassification (for example, subjects who self-report chronic liver disease may have higher cognitive performance as indicated by their comorbidity awareness or other factors) can bias results towards spurious associations. Similarly, the cognitive and brain imaging measures used in ACCORD-MIND and SPRINT may be insensitive to early changes or changes that are specifically related to chronic liver disease. Future large-scale epidemiological studies of liver disease and cognitive impairment should endeavor to capture more detailed and accurate liver disease assessments, in addition to more comprehensive neuropsychological and imaging batteries, to facilitate further research in this area.
Conclusion
In conclusion, chronic liver disease was not associated with cognitive impairment or worse brain volumes among individuals with diabetes or hypertension in our analyses of participant data from ACCORD-MIND and SPRINT, respectively. Further work with more precise measures of chronic liver disease is needed to confirm our work.
Supplementary Material
Highlights:
Chronic liver disease is increasingly prevalent.
Chronic liver disease may impact brain health.
We analyzed data from two randomized trials.
There was no relationship between liver disease and brain health among people with hypertension and diabetes.
Acknowledgements
This manuscript was prepared using ACCORD Research Materials obtained from the NHLBI Biologic Specimen and Data Repository Information Coordinating Center and does not necessarily reflect the opinions or views of the ACCORD or the NHLBI. This manuscript was prepared using SPRINT Research Materials obtained from the NHLBI Biologic Specimen and Data Repository Information Coordinating Center. The views expressed in this paper are those of the authors and do not represent the official position of the National Institutes of Health (NIH), NHLBI, the Department of Veterans Affairs, the U.S. Government, or the SPRINT Research Group.
Sources of Funding
Dr. Parikh reports funding support from the NIH/NIA (AG073524) and Leon Levy Foundation.
Conflicts-of-Interest/Disclosures
Dr. Parikh: personal fees for medicolegal consulting and unrelated research support from the Florence Gould Foundation and the NY State Empire Clinical Research Investigator Program.
Footnotes
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
References
- 1.Boyle PA, Wilson RS, Yu L, Barr AM, Honer WG, Schneider JA, et al. Much of late life cognitive decline is not due to common neurodegenerative pathologies. Ann Neurol. 2013;74:478–489 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Asrani SK, Devarbhavi H, Eaton J, Kamath PS. Burden of liver diseases in the world. J Hepatol. 2019;70:151–171 [DOI] [PubMed] [Google Scholar]
- 3.Moon AM, Singal AG, Tapper EB. Contemporary epidemiology of chronic liver disease and cirrhosis. Clin Gastroenterol Hepatol. 2019;18(12):2650–2666 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Bordet R, Deplanque D. Brain-liver axis: A new pathway for cognitive disorders related to hepatic fibrosis. Eur J Neurol. 2020;27:2111–2112 [DOI] [PubMed] [Google Scholar]
- 5.Bassendine MF, Taylor-Robinson SD, Fertleman M, Khan M, Neely D. Is alzheimer’s disease a liver disease of the brain? J Alzheimers Dis. 2020;75:1–14 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Horowitz AM, Fan X, Bieri G, Smith LK, Sanchez-Diaz CI, Schroer AB, et al. Blood factors transfer beneficial effects of exercise on neurogenesis and cognition to the aged brain. Science. 2020;369:167–173 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Yeh MM, Brunt EM. Pathological features of fatty liver disease. Gastroenterology. 2014;147:754–764 [DOI] [PubMed] [Google Scholar]
- 8.Bajaj JS, Lauridsen M, Tapper EB, Duarte-Rojo A, Rahimi RS, Tandon P, et al. Important unresolved questions in the management of hepatic encephalopathy: An ishen consensus. Am J Gastroenterol. 2020;115:989–1002 [DOI] [PubMed] [Google Scholar]
- 9.Tapper EB, Jiang ZG, Patwardhan VR. Refining the ammonia hypothesis: A physiology-driven approach to the treatment of hepatic encephalopathy. Mayo Clin Proc. 2015;90:646–658 [DOI] [PubMed] [Google Scholar]
- 10.Seo SW, Gottesman RF, Clark JM, Hernaez R, Chang Y, Kim C, et al. Nonalcoholic fatty liver disease is associated with cognitive function in adults. Neurology. 2016;86:1136–1142 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Solfrizzi V, Scafato E, Custodero C, Loparco F, Ciavarella A, Panza F, et al. Liver fibrosis score, physical frailty, and the risk of dementia in older adults: The italian longitudinal study on aging. Alzheimers Dement (N Y). 2020;6:e12065. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Parikh NS, Kumar S, Rosenblatt R, Zhao C, Cohen DE, Iadecola C, et al. Association between liver fibrosis and cognition in a nationally representative sample of older adults. Eur J Neurol. 2020;27:1895–1903 [DOI] [PubMed] [Google Scholar]
- 13.Weinstein G, Davis-Plourde K, Himali JJ, Zelber-Sagi S, Beiser AS, Seshadri S. Non-alcoholic fatty liver disease, liver fibrosis score and cognitive function in middle-aged adults: The Framingham study. Liver Int. 2019;39:1713–1721 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.VanWagner LB, Terry JG, Chow LS, Alman AC, Kang H, Ingram KH, et al. Nonalcoholic fatty liver disease and measures of early brain health in middle-aged adults: The CARDIA study. Obesity (Silver Spring). 2017;25:642–651 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Weinstein G, Zelber-Sagi S, Preis SR, Beiser AS, DeCarli C, Speliotes EK, et al. Association of nonalcoholic fatty liver disease with lower brain volume in healthy middle-aged adults in the Framingham study. JAMA Neurol. 2018;75:97–104 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Weinstein AA, de Avila L, Paik J, Golabi P, Escheik C, Gerber L, et al. Cognitive performance in individuals with non-alcoholic fatty liver disease and/or type 2 diabetes mellitus. Psychosomatics. 2018;59:567–574 [DOI] [PubMed] [Google Scholar]
- 17.Gerber Y, VanWagner LB, Yaffe K, Terry JG, Rana JS, Reis JP, et al. Non-alcoholic fatty liver disease and cognitive function in middle-aged adults: The CARDIA study. BMC Gastroenterol. 2021;21:96. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Action to Control Cardiovascular Risk in Diabetes Study G, Gerstein HC, Miller ME, Byington RP, Goff DC Jr., Bigger JT, et al. Effects of intensive glucose lowering in type 2 diabetes. N Engl J Med. 2008;358:2545–2559 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Launer LJ, Miller ME, Williamson JD, Lazar RM, Gerstein HC, Murray AM, et al. Effects of intensive glucose lowering on brain structure and function in people with type 2 diabetes (ACCORD MIND): A randomised open-label substudy. Lancet Neurol. 2011;10:969–977 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.SPRINT Research Group Wright JT Jr., Williamson JD Whelton PK, Snyder JK Sink KM, et al. A randomized trial of intensive versus standard blood-pressure control. N Engl J Med. 2015;373:2103–2116 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Rapp SR, Gaussoin SA, Sachs BC, Chelune G, Supiano MA, Lerner AJ, et al. Effects of intensive versus standard blood pressure control on domain-specific cognitive function: A substudy of the SPRINT randomised controlled trial. Lancet Neurol. 2020;19:899–907 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Nasrallah IM, Gaussoin SA, Pomponio R, Dolui S, Erus G, Wright CB, et al. Association of intensive vs standard blood pressure control with magnetic resonance imaging biomarkers of alzheimer disease: Secondary analysis of the SPRINT MIND randomized trial. JAMA Neurol. 2021;78:568–577 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.SPRINT MIND Investigators for the SPRINT Research Group, Williamson JD, Pajewski NM, Auchus AP, Bryan RN, Chelune G, et al. Effect of intensive vs standard blood pressure control on probable dementia: A randomized clinical trial. JAMA. 2019;321:553–561 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.McHenry S, Park Y, Browning JD, Sayuk G, Davidson NO. Dallas steatosis index identifies patients with nonalcoholic fatty liver disease. Clin Gastroenterol Hepatol. 2020;18:2073–2080 e2077 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.Folstein MF, Folstein SE, McHugh PR. “Mini-mental state”. A practical method for grading the cognitive state of patients for the clinician. J Psychiatr Res. 1975;12:189–198 [DOI] [PubMed] [Google Scholar]
- 26.Houx PJ, Jolles J, Vreeling FW. Stroop interference: Aging effects assessed with the stroop color-word test. Exp Aging Res. 1993;19:209–224 [DOI] [PubMed] [Google Scholar]
- 27.Jaeger J Digit symbol substitution test: The case for sensitivity over specificity in neuropsychological testing. J Clin Psychopharmacol. 2018;38:513–519 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28.Parikh NS, Gottesman RF. Midlife cardiovascular risk factors, subclinical atherosclerosis, and cerebral hypometabolism. J Am Coll Cardiol. 2021;77:899–901 [DOI] [PubMed] [Google Scholar]
- 29.Younossi ZM, Koenig AB, Abdelatif D, Fazel Y, Henry L, Wymer M. Global epidemiology of nonalcoholic fatty liver disease-meta-analytic assessment of prevalence, incidence, and outcomes. Hepatology. 2016;64:73–84 [DOI] [PubMed] [Google Scholar]
- 30.Tapper EB, Lok AS. Use of liver imaging and biopsy in clinical practice. N Engl J Med. 2017;377:756–768 [DOI] [PubMed] [Google Scholar]
- 31.Perez-Matos MC, Jiang ZG, Tapper EB. Factors that affect results of psychometric tests to identify patients with minimal hepatic encephalopathy. Clin Gastroenterol Hepatol. 2018;16:1836–1838 [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.