Associations Between Alcohol and Hepatic Steatosis

In This Article

Abstract and Introduction

Abstract

Despite being widely recognized as a common cause of fatty liver, the exact impact of alcohol consumption on hepatic steatosis in the general population is elusive. The recent National Health and Nutrition Examination Survey (NHANES) allowed us to examine this relationship among US adults. Herein, we extracted data on detailed alcohol consumption and controlled attenuation parameter (CAP) by FibroScan from 4509 participants in NHANES 2017–2018. Compared to metabolic risk factors such as diabetes and obesity, the association between alcohol consumption and CAP was less significant. In multivariable analysis, only those drinking 5–7 times per week showed significant increases in CAP scores. Although both frequency and quantity of drinking were positively associated with CAP score, only frequency remained significant after adjustment for quantity and binge drinking. These epidemiological observations suggested that the impact of alcohol on hepatic steatosis was much smaller than metabolic factors and dependent upon the frequency of drinking.

Introduction

Metabolic-associated fatty liver disease (MAFLD) and alcohol-associated liver disease (ALD) are two common causes of fatty liver.^[1] While the mechanism underlying hepatic steatosis in MAFLD is well described, how alcohol consumption causes fatty liver is not well understood.^[2,3] The level of alcohol consumption causing steatosis in humans also remains elusive.

Guidelines regarding appropriate alcohol consumption were sometimes inconsistent. The National Institute of Alcohol Abuse and Alcoholism (NIAAA) defined heavy drinking as more than three drinks for women and four drinks for men on a given day.^[4] Meanwhile, the latest Dietary Guidelines for Americans by the US Department of Health and Human Services set the upper limit of safe drinking at one drink for women and two drinks for men.^[5] Yet, epidemiological data supporting the adverse effect of alcohol on hepatic steatosis in the general public were lacking. A prospective study in Japan showed that light to moderate drinking protected against fatty liver, while even excessive alcohol consumption (≥280 g/wk) did not exacerbate fatty liver.^[6] Recently, Unalp-Arida and colleagues analysed transient elastography data from the 2017–2018 National Health and Nutrition Examination Survey (NHANES).^[7] Their analysis showed that heavy alcohol consumption was not associated with fatty liver measured by controlled attenuation parameter (CAP). Meanwhile, moderate consumption was associated with a lower CAP in multivariate analysis. Herein, we performed a targeted analysis to examine the relationship between alcohol consumption and hepatic steatosis using the same NHANES 2017–2018 data and reached a more nuanced conclusion.

Table 1. Differential impact of frequency and quantity of alcohol consumption on CAP score
	Model 1^e			Model 2^f		Model 3^g
	Coeff. (95% CI)^d	P	Coeff. (95% CI)	P	P _interaction	Coeff. (95% CI)	P
Frequency^a	2.2 (1.2–3.1)	<.001	1.9 (0.8–3.0)	.002	.05	1.7 (0.6–2.9)	.006
Quantity^b	1.2 (0.2–2.2)	.01	0.8 (−0.3 to 1.9)	.1		0.6 (−0.6 to 1.8)	.2
Binge^c	9.9 (4.5–15.3)	.001				3.8 (−3.1 to 10.7)	.2

^aFrequency of alcohol consumption is measured by days of drinking in 1 week (0–7).
^bQuantity is measured by the average number of standard drinks on drinking days (0–15).
^cBinge drinking is defined by 5+ drinks for men and 4+ drinks for women at least once a week (0–1).
^dBeta coefficients are calculated by linear regression for CAP score (dB/m) while adjusting for sampling weights for the US population.
^eModel 1 examines the individual association of frequency, quantity, binge, and CAP, while adjusting for age, gender, race, smoking history, BMI, and diabetes. Beta coefficients are calculated by linear regression for CAP score (dB/m). A total of 3,790 participants were included in the analysis representing 183.2 million US adults.
^fModel 2 includes drinking frequency and quantity per drinking day while adjusting for all covariates in Model 1. P _interaction calculates the interaction between frequency and quantity per drinking day in Model 2.
^gModel 3 includes drinking frequency, alcohol quantity per drinking day, and binge drinking as well as covariates in Model 1 and 2.