Coffee consumption is not associated with prevalent subclinical cardiovascular disease (CVD) or the risk of CVD events, in nonalcoholic fatty liver disease: results from the Multi-Ethnic Study of Atherosclerosis

Abstract

Background

Atherosclerosis and its clinical sequelae represent the leading cause of mortality among patients with nonalcoholic fatty liver disease (NAFLD). While epidemiologic data support the hepatoprotective benefits of coffee in NAFLD, whether coffee improves NAFLD-associated CVD risk is unknown.

Methods

We examined 3,710 ethnically-diverse participants from the Multi-Ethnic Study of Atherosclerosis (MESA) cohort, without history of known liver disease, and with available coffee data from a validated 120-item food frequency questionnaire. All participants underwent baseline non-contrast cardiac CT from which NAFLD was defined by liver:spleen ratio (L:S<1.0), and subclinical CVD was defined by coronary artery calcium (CAC)>0. Major CVD events were defined by the first occurrence of myocardial infarction, cardiac arrest, angina, stroke, or CVD death. We used log-binomial regression to calculate the adjusted prevalence ratio (PR) for CAC>0 by coffee intake and NAFLD status, and events were compared between groups using frequency of events within adjusted Cox proportional hazard regression models.

Results

Seventeen percent (N=637) of participants met criteria for NAFLD. NAFLD participants were more likely to have elevated BMI (mean 31.1 ± 5.5 kg/m² vs. 28.0 ± 5.2 kg/m², p<0.0001), and diabetes (22% vs. 11%, p<0.0001), but did not differ in daily coffee consumption (p=0.97). Among NAFLD participants, coffee consumption was not associated with prevalent, baseline CAC>0 (PR=1.02 [0.98-1.07]). Over 12.8 years of follow-up, 93 NAFLD and 415 non-NAFLD participants experienced a CV event. However, coffee intake was not associated with incident CVD events, in either NAFLD (HR=1.05 [0.91-1.21]) or non-NAFLD participants (HR=1.03 [0.97-1.11]).

Conclusion

In a large, population-based cohort, coffee consumption was not associated with the prevalence of subclinical CVD, nor did coffee impact the future risk of major CVD events, regardless of underlying NAFLD status.

Introduction

Among patients with nonalcoholic fatty liver disease (NAFLD), atherosclerosis and its clinical sequelae represent the leading cause of mortality ^{1, 2}, and NAFLD has also been linked to subclinical CVD, including coronary calcification, carotid intima-media thickness, and arterial hypertension ³. Recent data suggest that coffee exerts unique metabolic and hepatoprotective benefits in NAFLD, improving insulin sensitivity and reducing fibrosis progression ^4-6. Although the relationship between coffee and CVD risk is well studied within the general population, with modest coffee consumption associated with a neutral to mildly decreased CVD risk ^7-9, little is known about the CVD benefits of coffee in NAFLD. We explored whether coffee consumption is related to prevalent coronary artery calcification (CAC) and the risk of incident CVD events among participants with NAFLD in the Multi-Ethnic Study of Atherosclerosis (MESA).

Methods

MESA is a population-based, prospective study of 6,814 adults, who were free of known CVD at enrollment. After providing written informed consent, all participants completed examination 1, including a validated, 120-item food frequency questionnaire that assesses diet over the preceding year, including average serving size and frequency of coffee, alcohol intake and other dietary variables (grams/day), and each were adjusted for daily energy expenditure ¹⁰. Coffee consumption was analyzed in two separate models: (1) coffee intake (yes/no) and (2) daily servings of coffee (range: 0-9 servings/day). Due to non-random missing data from one center, imputed dietary data were used for a small subset of individuals to avoid biasing parameter estimates.

The methodology for measuring coronary artery calcification and other covariates within MESA, and for defining comorbidities including hypertension, dyslipidemia and diabetes, have previously been described¹¹. Inflammatory biomarkers were assessed in a subset of MESA participants using methodology that has also been reported, and biomarkers included gamma-glutamyltransferase (GGT), Creactive protein (CRP), and interleukin-6 (IL-6)¹². Finally, each participant underwent non-contrast cardiac computed tomography (CT) for measurement of coronary artery calcium (CAC) and liver fat, with steatosis defined by the validated liver-to-spleen (L:S) attenuation ratio<1.0, as previously described by our group and others¹³.

L:S measurements were available for 4,389 participants. Six hundred seventy-nine individuals were excluded: 405 had a history of heavy alcohol use (>7 drinks/week in women or >14 drinks/week in men), 59 reported alcohol consumption but did not quantify intake, 22 self-reported Hepatitis B virus, 16 self-reported Hepatitis C virus, 8 reported cirrhosis, 2 were amiodarone users and 167 had missing dietary information, leaving a final population of 3,710 individuals.

Subclinical CVD was defined by CAC>0 at examination 1. Clinical events were ascertained by follow-up telephone interviews every 9-12 months, with all events adjudicated by an independent MESA committee¹⁴. Incident major CVD events included the first occurrence of myocardial infarction, resuscitated cardiac arrest, angina, stroke or CVD death. Prevalence ratios were calculated for the risk of baseline CAC>0 by servings/day of coffee, and compared in participants with and without radiographic NAFLD, after adjusting for cardiometabolic risk factors. Frequency of events were compared using proportional hazard regression models to estimate the relationship between coffee consumption and risk of incident major CV events, in participants with and without radiographic NAFLD.

Results

The baseline demographic and clinical characteristics of the NAFLD (n=637) and non-NAFLD (n=3,073) participants are shown in Table S1. Compared to individuals without NAFLD, those with NAFLD were younger (mean age 61±10 vs. 63±11 years, p<0.0001), but had higher mean body mass index (BMI) (31.1±5.5 vs. 28.0±5.2 kg/m², p<0.0001), and were more likely to have diabetes (22% vs 11%, p<0.0001) and the metabolic syndrome (65% vs. 33%, p<0.0001; Table S1). Mean daily coffee consumption did not differ between NAFLD and non-NAFLD participants (mean 1.1±1.5 servings/day in both groups; p=0.97).

At baseline, 335 NAFLD participants (52.6%) had CAC>0, compared to 1522 (49.5%) in the non-NAFLD group (p=0.164). In fully-adjusted, multivariable log-binomial regression analysis, coffee servings/day were not associated with prevalent CAC>0 (PR=1.00 [95% CI 0.98-1.02]). This association was unchanged when the cohort was limited just to those with NAFLD (N=637; PR=1.02 [0.98-1.07]; Table 1A), and also when the cohort was limited to those with NAFLD and CRP ≥5 mg/L (N=196) (RP=1.10 [0.98-1.23]). The relationship between coffee servings/day and prevalent CAC>0 similarly did not differ between NAFLD (PR=1.02 [95% CI 0.98-1.07]) and non-NAFLD participants (PR=0.99 [95% CI 0.97-1.01]). Coffee consumption also was not associated with differences in log-transformed values of the circulating inflammatory biomarkers, GGT, CRP or IL-6 (all p>0.05; Table 1B).

Table 1a, 1b. Cross-sectional association between coffee consumption and the risk of prevalent coronary artery calcification (CAC) (Panel 1A), and mean log-transformed serum inflammatory biomarker levels (Panel 1B), among MESA participants with radiographic NAFLD (N=637).

1A. Multivariable model	Adjusted1 prevalence ratio [95% CI] for CAC > 0
	No coffee	Any coffee	Servings per day
• Model 1†	Ref.	1.24 [1.03, 1.50]	1.05 [1.01, 1.10]
• Model 2‡	Ref.	1.15 [0.97, 1.37]	1.04 [0.996, 1.08]
• Model 3∫	Ref.	1.09 [0.92, 1.30]	1.03 [0.98, 1.07]
• Model 4¥	Ref.	1.08 [0.91-1.29]	1.02 [0.98-1.07]
• Model 5€	Ref.	1.08 [0.91-1.29]	1.02 [0.98-1.07]
1B. Inflammatory marker level2	Adjusted* β coefficient [95% CI]
	No coffee	Any coffee	Servings per day
• Gamma glutamyltransferase, log IU/L, (N=634) (N=633)	Ref.	-0.0638 [-0.0739, 0.0201]	-0.0127 [-0.0558, 0.0304]
• C-reactive protein, log mg/L, (N=634)	Ref.	0.0739[-0.108, 0.256]	-0.0059 [-0.0629, 0.0511]
• Interleukin-6, log pg/mL, (N=627)	Ref.	0.0305 [-0.0789, 0.140]	-0.0138 [-0.0480, 0.0204]

¹Multivariable adjusted prevalence ratios are shown.

^†Model 1: unadjusted

^‡Model 2: age, sex and race/ethnicity

^∫Model 3: Model 2 + body mass index (BMI, kg/m²), total cholesterol, use of any lipid-lowering medication, high-density lipoprotein (HDL) cholesterol, diabetes, hypertension, alcohol use (servings/day) and smoking status

^¥Model 4: Model 3 + energy-adjusted intake of regular soda (servings/day), diet soda (servings per day), red meat (servings per day) and fructose (grams)

^€Model 5: Model 4 + Chronic Kidney Disease (CKD), defined by glomerular filtration rate (GFR) < 60

²All serum inflammatory markers were log-transformed

^*Multivariable linear regression model, adjusted for age, race, gender, body mass index (BMI), waist circumference, diabetes, hypertension, CKD, use of lipid-lowering medications, smoking status, alcohol use (servings/day), and energy-adjusted intake of diet soda and regular soda, fructose and red meat.

Over a median 12.8 years, 93 NAFLD and 415 non-NAFLD MESA participants experienced CVD events (Table 2). In fully-adjusted Cox proportional hazard regression models accounting for the presence of NAFLD, no association was found between coffee consumption and risk of incident CV events (HR=1.03 [95% CI 0.97-1.10]). In stratified analysis, a null association was found between servings/day of coffee and incident CV events, in both NAFLD (HR=1.05, [95% CI 0.91-1.21]) and non-NAFLD participants (HR=1.03 [95% CI 0.97-1.11) (Table 2). No significant interactions were observed between coffee and age category, sex, BMI, race, smoking, alcohol intake or diabetes.

Table 2. Unadjusted and adjusted hazard ratios (95% confidence intervals) between coffee consumption and incident cardiovascular events over 12.8 years of median follow-up, according to NAFLD status.

Daily coffee intake	No NAFLD (N=3,710)			NAFLD (N=637)
	No coffee	Any coffee	Servings / day	No coffee	Any coffee	Servings / day
No. events / no. total	105/870	310/2203	-	16/164	77/473	-
IR per 1000 p-y.	10.7	12.8	-	8.7	15.2	-
• Model 1†	1.00	1.19 [0.96, 1.49]	1.04 [0.98, 1.10]	1.00	1.74 [1.02, 2.98]	1.09 [0.96, 1.24]
• Model 2‡	1.00	1.08 [0.86, 1.36]	1.04 [0.97, 1.11]	1.00	1.51 [0.85, 2.68]	1.04 [0.91, 1.20]
• Model 3∫	1.00	1.09 [0.87, 1.37]	1.03 [0.97, 1.10]	1.00	1.53 [0.85, 2.74]	1.05 [0.92, 1.21]
• Model 4¥	1.00	1.09 [0.86-1.37]	1.03 [0.97-1.11]	1.00	1.61 [0.89-2.89]	1.05 [0.91-1.21]

Abbreviations: NAFLD, nonalcoholic fatty liver disease; IR, incidence rate; p-y, person-years; No., number

^†Model 1: unadjusted

^‡Model 2: adjusted for age, sex and race/ethnicity

^∫Model 3: Model 2 + body mass index (BMI, kg/m²), total cholesterol, use of any lipid-lowering medication, high-density lipoprotein (HDL) cholesterol, diabetes, hypertension, alcohol use (servings/day) and smoking status

^¥Model 4: Model 3 + energy-adjusted estimated daily intake of diet soda and regular soda, fructose and red meat.

Discussion

This represents the first published study from a large, population-based, multi-ethnic cohort to describe the impact of coffee consumption on subclinical and overt CVD risk, in participants with NAFLD, compared to the general population. We found that among participants with NAFLD, coffee consumption was not associated with prevalent coronary calcification or with risk of incident CVD events. This null association persisted in dose-response analyses that accounted for daily servings of coffee, arguing against an inflection point at which a certain daily intake of coffee might give rise to protective effects.

Published data in the general population regarding the benefits of coffee on CVD outcomes are conflicting, with most studies describing neutral to perhaps beneficial effects ^{9,7, 8}. Available reports from population-based cohorts with dose-response relationships are heterogeneous, related likely to variations in statistical analyses, inadequately-measured dietary or lifestyle factors, such as smoking, which could confound coffee intake. In addition, most studies to date have relied upon primarily Caucasian populations, limiting potential generalizability¹⁵. Among the few available multi-ethnic analyses, a recent report from MESA demonstrated a neutral overall association between coffee, CAC prevalence, progression and incident CVD events, to date none have included NAFLD in their analyses ¹⁰.

In contrast to the general population, little is known about the relationship between coffee and CVD in patients with NAFLD. However, there is a growing appreciation of the close ties between NAFLD and both subclinical^{13, 16} and overt CVD risk^{1, 2}. Given that coffee has been shown to attenuate NAFLD fibrosis progression, minimize hepatocarcinogenesis and exert anti-inflammatory effects in patients with chronic liver disease, it has been hypothesized that coffee could exert uniquely beneficial effects in NAFLD, beyond that observed in the general population. However, in the present study, we found a neutral association between coffee intake and both prevalent coronary calcification and risk of incident CV events, among participants with NAFLD, that did not differ significantly from the non-NAFLD MESA population.

The proposed benefits of coffee in NAFLD are thought to relate to caffeine-mediated inhibition of hepatic stellate cell proliferation¹⁷, and to anti-fibrotic and anti-inflammatory hepatoprotective effects of polyphenols¹⁸. In epidemiologic studies, coffee consumption has been linked to numerous beneficial metabolic effects, including reduced systemic inflammation and improved insulin sensitivity^4-6, and among patients with chronic liver disease, coffee has been associated with a reduced risk of hepatic fibrogenesis and hepatocellular carcinoma (HCC)^{5, 19}. However, the precise pathophysiologic mechanisms that underpin this relationship remain incompletely characterized. In a recent prospective study by Zelber-Sagi and colleagues, coffee was associated only with reduced serum markers of fibrosis over time, but did not impact steatosis ²⁰. Further studies are needed to determine whether coffee affects steatogenesis, or rather if the hepatoprotective benefits of coffee relate primarily to prevention of fibrosis and hepatocarcinogenesis ²⁰.

This study benefited from a large, population-based cohort with well-characterized dietary data and adjudicated cardiovascular outcomes. Nonetheless, several limitations must be highlighted. First, its observational nature renders this study subject to residual confounding. Second, while CT is well-validated for steatosis, it cannot reliably identify NASH or fibrosis, and without hepatic histology or liver chemistry data, we could not explore coffeerelated effects by NAFLD severity. Third, our study lacked echocardiography and future analyses will be needed to explore these relationships as they relate to cardiac structure and function. Finally, although power was a concern, we achieved 80% power to detect an association of HR=0.82 between servings/day of coffee and incident CVD in NAFLD participants. This effect size is favorable to the neutral overall effect previously reported from the general MESA population¹⁰, and roughly equivalent to estimates from a recent meta-analysis (HR=0.85)⁹, suggesting that our study was adequately powered to detect previously-reported associations.

In conclusion, within this large, well-characterized cohort, we provide the first report that coffee consumption is not associated with prevalent subclinical CVD or with future risk of CVD events, in individuals with NAFLD. Future long-term studies with well-phenotyped NAFLD populations, focused upon the dose and timing of coffee consumption with respect to NAFLD severity, will allow for more complete characterization of the relationship between coffee intake and NAFLD-associated CVD risk.

Highlights.

• This study is the first to explore the CV effects of coffee in adults with NAFLD.

• In those with NAFLD, no association was found between coffee and subclinical CVD.

• In those with NAFLD, coffee was not associated with reduced risk of major CV events.

• These results suggest coffee may not have unique cardioprotective benefits in NAFLD.