Association of liver related biomarkers with incident cardiovascular disease and all-cause mortality in the Hispanic community health study/study of Latinos (HCHS/SOL), a population-based cohort study | BMC Gastroenterology

Study sample

HCHS/SOL is a multi-center study of 16,415 US Hispanic/Latino adults which include Central American, Cuban, Dominican, Mexican, Puerto Rican, and South American backgrounds [11]. Participants were recruited from four field centers (Bronx, NY, Chicago, IL, Miami, FL, and San Diego, CA) between 2008 and 2011 using a two-stage area household sampling design. Persons who self-identified as being Hispanic/Latino, ages 18–74, had no plans to move out of the study region, and were able to attend a clinic examination were eligible to participate. 42% of screened, eligible persons enrolled under informed consent. HCHS/SOL participation included up to three in person examinations at approximately 6-year intervals and annual telephone contacts to update major health related variables.

We identified individuals with prevalent CVD at baseline using electrocardiograms (ECGs) from baseline examinations. Major Q wave abnormalities, which are indicative of old myocardial infarction, and minor Q or QS waves with ST or T abnormalities, suggestive of potential old myocardial infarction, were categorized as having prevalent CVD. Additionally, we incorporated self-reported medical histories into our classification of prevalent CVD. Individuals reporting past myocardial infarction, coronary artery revascularization procedures (including balloon angioplasty, stent placement, or bypass surgery), stroke, or heart failure were also categorized as having prevalent CVD. Consequently, we excluded a total of 1,199 participants who met these criteria for prevalent CVD at baseline. A total of 15,216 participants were included in the analysis (Fig. 1).

Outcome

Our outcome was a composite of incident CVD and all-cause mortality. Event surveillance for HCHS/SOL relies on data gathered from annual phone follow-up interviews. When an annual interview attempt indicates that a participant has experienced outcomes such as death, hospital admission, or emergency department visits, measures are taken to obtain relevant medical records or death certificates. HCHS/SOL systematically records all hospitalizations and emergency department visits, capturing information on discharge diagnoses and procedure codes (using ICD-9 and ICD-10 codes) associated with cardiovascular events. For participants with potential cardiovascular events, medical records are abstracted, de-identified, and prepared for review by the HCHS/SOL event classification committee. Independent review of events is conducted by two physicians, with any discrepancies resolved through adjudication by a third reviewer. All-cause mortality was also ascertained through annual follow-up interviews conducted with family members or other proxy respondents as well as reviews of vital statistics lists, obituaries and matching with the National Death Index (NDI). Participants were followed until December 31 st, 2017, the date of death, the date of their first CVD event, or the date of the last completed annual follow-up interview, whichever occurred first.

Covariates and key exposure variables

Recruitment center was used as a covariate to adjust for potential differences across the four centers. Participants provided information on age, sex, Hispanic/Latino background, education, annual household income, behaviors, and medical history. Alcohol intake was categorized as high-level use if more than 7 drinks per week were consumed by females or more than 14 drinks per week by males. Dietary intake was assessed using the National Cancer Institute method which involved using two 24-hour food recalls to account for within subject variability [12]. The initial 24-hour food recall was administered in person at the baseline examination, and the second 24-hour was administered at least 5 days later but ideally within 45 days. Additional attempts were made after the 45 days to complete the second 24-hour recall assuring completeness of questionnaires. The Alternative Healthy Eating Index (AHEI-2010) was calculated based on 11 dietary components with evidence of association with chronic diseases, including: vegetables, whole fruit, whole grains, sugar-sweetened beverages and fruit juice, nuts and legumes, red/processed meat, trans fat, long-chain fatty acids, polyunsaturated fatty acids, sodium, and alcohol [13].

Physical activity was evaluated using the global physical activity questionnaire (GPAQ) developed by the World Health Organization. Total physical activity and activity intensity were used to calculate metabolic equivalent of task per day (MET-minutes/day) [14] which was categorized into four activity levels ranging from inactive to high activity [15]. Body weight was measured using a Tanita scale (model TBF-300 A; Tanita Corporation, Arlington Heights, IL) to the nearest 0.1 kg, and standing height was measured to the nearest centimeter. Body mass index (BMI) was calculated as weight (kg) divided by height squared (m [2]). Resting brachial blood pressure was computed as the average of three measurements obtained on the right arm in a seated position using an automated blood pressure device (OMRON HEM-907 XL, OMRON Corporation, Kyoto, Japan). Waist circumference was measured to the nearest centimeter. Participants were instructed to bring all medications taken in the last 4 weeks, which were scanned into a digital system to perform medical therapeutic classification.

Fasting morning blood samples and 2-hour glucose tolerance test blood samples were centrifuged and frozen within 45 min of collection. Blood levels of glucose and insulin (fasting and 2-hr), hemoglobin A1c, triglycerides, high-density lipoprotein cholesterol (HDL-c), low density lipoprotein cholesterol (LDL-c), high sensitivity C-reactive protein (hs-CRP), cystatin C, creatinine and complete blood count were quantified in the central laboratory. Homeostatic model assessment of insulin resistance (HOMA-IR) was calculated as fasting glucose (mg/dl) x Fasting insulin (mU/L)/405 [16]. Serum ALT, AST, and GGT were measured at baseline on a Roche Modular P Chemistry Analyzer (Roche Diagnostics) using an alpha-ketoglutaratic enzymatic method [9]. Serum specimens were tested for antibodies against hepatitis B and C antigens as previously described [17].

Laboratory approach to classifying suspected liver disease

Serum ALT, AST and GGT were assessed untransformed, log transformed and as binary variables (elevated vs. not elevated) using cut points which have been used in previous studies [18,19,20]. A composite variable for elevated ALT or AST (“elevated ALT/AST”) was defined as those having AST > 37 IU/mL or ALT > 40 IU/mL for males, and AST or ALT > 31 IU/mL for females; this well-established surrogate marker for liver disease [21] was previously shown to correlate with degree of Amerindian ancestry in Hispanic and Latino populations [22]. Elevated GGT included values above 36 U/L.

The FIB-4 index was calculated to estimate the severity of liver fibrosis as [20]:

FIB-4 was assessed a continuous variable, log transformed and categorized as a binary variable with values lower than or equal to 2.67 categorized as no to intermediate fibrosis likely and values greater 2.67 categorized as high fibrosis likely [20, 22].

Metabolic dysfunction-associated steatotic liver disease (MASLD) was defined by the presence of steatosis using the fatty liver index (FLI) and at least one metabolic risk abnormality [10]. Values of the FLI index range from 0 to 100 with higher values indicating more fatty liver and scores over 60 were categorized as having steatosis [23]. The FLI was calculated using the following formula:

For defining MASLD, metabolic risk abnormalities were defined any of the following: type 2 diabetes, body mass index greater than 25 kg/m², waist circumference ≥ 94 cm for males and ≥ 80 cm for females, blood pressure ≥ 130/85 mmHg or hypertension drug treatment, plasma HDL-c < 40 mg/dl for males and < 50 mg/dl for females, plasma triglycerides ≥ 150 mg/dl or hyperlipidemic drug treatment, prediabetes (fasting glucose in range 100–125 mg/dL or 2hr glucose in range 140–199 mg/dL or A1C in range 5.7-6.5%), HOMA-IR ≥ 2.5, or plasma hs-CRP level > 2 mg/L.

Statistical analysis

Unadjusted and serially adjusted Cox proportional hazards models accounting for complex survey features in HCHS/SOL assessed the relationship between elevated AST/ALT, FIB-4, and MASLD with CVD or all-cause mortality. All adjusted survey-weighted Cox proportional hazards models adjusted for age, sex and recruitment center. Model 2 also adjusted for Hispanic/Latino background, AHEI-2010, physical activity, smoking and alcohol use. Model 3 additionally adjusted for BMI, waist to hip ratio, hypertension and diabetes. In additional models to detect statistical interaction, we assessed for departure from a multiplicative combined effect of elevated ALT/AST and MASLD on the risk of CVD or all-cause mortality. Association tests involving MASLD were performed using models 1 and 2 only, since model 3 covariates make up the definition of MASLD. Finally, we assessed the relationships between our exposures and CVD or all-cause mortality separately as well as potential differences by sex. In subsequent analyses, we used stratified analyses by Hispanic/Latino background to assess potential differences in the association of liver disease with CVD and all-cause mortality risk. All analyses were conducted in SAS software version 9.4 (SAS institute, Cary, NC) and statistical significance was set at p < 0.05 for main effects and p < 0.10 for interactions. Variables were log-transformed when not normally distributed and medians with interquartile ranges were presented. Variables included in our analytical models were missing for less than 5% of participants and were excluded from analyses which included the missing variable.