Interaction between alcohol consumption and PNPLA3 variant in the prevalence of hepatic steatosis in the U.S. Population

Abstract

Background & aims:

To our knowledge, the interaction between alcohol consumption and PNPLA3 genotype on hepatic steatosis has not been explored in a representative sample. To examine the interaction between alcohol consumption and PNPLA3 genotype on hepatic steatosis in the U.S. adult population.

Methods:

Cross-sectional study of 4,674 adult participants of the Third National Health and Nutrition Examination Survey, Phase 2 (1991-1994) with data on PNPLA3 genotype, self-reported alcohol consumption, ultrasound-defined hepatic steatosis and socio-demographic characteristics.

Results:

In 1991-1994 in the U.S. population, the weighted allele frequency of the G (risk) allele of the rs738409 at PNPLA3 was 25.4%. We confirmed both a J shaped association between alcohol consumption and hepatic steatosis among those with the CC genotype of PNPLA3, and a higher prevalence of hepatic steatosis among those with PNPLA3 gene G variant. We found evidence of an interaction of PNPLA3 G allele presence on the association between moderate alcohol consumption and hepatic steatosis on both the multiplicative (relative prevalence ratio [RPR]=1.95, 95% confidence interval [CI] 1.04-3.65) and additive scales (relative excess risk due to interaction=0.49, 95% CI 0.13-0.85). Compared to never drinkers, moderate alcohol drinking was associated with a 48% decreased risk of hepatic steatosis only among those without PNPLA3 G allele (PR=0.52, 95% CI 0.26-1.05), with no association among those with at least one copy of the PNPLA3 G allele (PR=1.02, 95% CI 0.68-1.54).

Conclusions:

Our results suggest that a highly common and strong genetic susceptibility to liver disease is modifiable by the level of alcohol consumption. Keeping alcohol consumption low may offset genetic predisposition to liver disease.

Introduction

The burden of liver diseases in the U.S. and other countries is substantial.^{1, 2} In the U.S., liver-related mortality has been found to be a contributor to the recent decline in life expectancy. ³ Key risk factors for chronic liver disease include viral hepatitis, obesity and insulin resistance, and alcohol consumption.^4–7 In particular, alcohol-related liver disease accounts for around 50% of liver-related mortality in the U.S.⁸ and >70% in the U.K.⁹

Hepatic steatosis is a key early histopathological feature of both non-alcoholic and alcoholic fatty liver diseases (NAFLD and ALD, respectively).^{6, 10} Approximately 25-30% of the US adult population have hepatic steatosis.¹¹ Currently, there are no FDA-approved treatments available for these conditions and therefore prevention is key. Both, obesity and alcohol consumption are highly prevalent modifiable risk factors in the Western hemisphere. However, even in the context of heavy alcohol consumption or obesity, some individuals do not develop liver disease. These findings are not surprising given the multifactorial nature of the disease, and support the potential importance of interactions between genetic and environmental factors in explaining the variation of the occurrence of liver disease.^{12, 13}

In 2008, a very strong association between a common genetic variation in the PNPLA3 gene, the G-allele at rs738409, and hepatic steatosis was described.¹⁴ The effect allele frequency (EAF) in the world population is 0.26.¹⁵ We and others have documented that in the U.S. adult population the prevalence of the PNPLA3 risk variant varies considerably by race/ethnicity, with the highest prevalence among Hispanics (EAF 0.54), intermediate among non-Hispanic Whites (EAF 0.25) and lowest among non-Hispanic Blacks (EAF 0.14), mirroring the observed race/ethnic disparities in NAFLD.¹⁶ The strong association between PNPLA3 gene mutation and NAFLD has been replicated in multiple settings, with a very large meta-analysis documenting that individuals homozygous (GG) individuals have 3.3 times higher odds of NAFLD compared to the reference group (CC).¹⁷ Further studies demonstrated associations between PNPLA3 genetic variation and ALD.^18–21

Whether risk factors, including alcohol consumption and obesity, modify the effects of the PNPLA3 genotype in the population is incompletely understood. Existing evidence supports an interaction between PNPLA3 genotype and adiposity on hepatic steatosis, by which body mass index (BMI) amplifies the effect of the G allele.^22–24 The GG phenotype confers little risk of hepatic steatosis among lean individuals, and but has a strong effect in individuals with obesity.²² However, to our knowledge, there has been no study of the interaction between alcohol consumption and PNPLA3 genotype on the occurrence of hepatic steatosis, in a large representative sample, with one prior small study suggesting an interaction.²³ The objective of our study is to examine the role of alcohol consumption modulating this genetic vulnerability of PNPLA3 genotype on hepatic steatosis in a large representative sample of the US.

Methods

Study population

We used data from participants of the NHANES III, Phase 2, a cross-sectional survey of the U.S. civilian population conducted in 1988-1994 by the National Center of Health Statistics (NCHS). NHANES III was divided into two phases: phase 1 conducted from 1988-1990 and phase 2 from 1991-1994. Genetic data are available for Phase 2 participants only. We limited our analyses to participants who were 20-74 years of age (eligible for ultrasound to determine hepatic steatosis), with PNPLA3 genotype information, hepatic steatosis assessment and information regarding alcohol consumption. The total analytical sample was 4,674 (unweighted). The NHANES III was approved by the institutional review board of the NCHS.

Exposures:

Self-reported past and current alcohol consumption was categorized into 4 mutually exclusive groups: never, former, current low-moderate alcohol consumption (≤1/≤2 drinks per day in women/men), and current heavy alcohol consumption (>1/>2 drinks per day in women/men). These cut points of alcohol consumption are consistent with the literature on fatty liver disease to distinguish “alcoholic” and “non-alcoholic” nature.⁶ Genotyping of rs738409 (PNPLA3) was performed using the iPLEX Sequenom platform.

Outcome:

The protocol used to ascertain the presence of hepatic steatosis using gallbladder ultrasounds in NHANES III can be found elsewhere.²⁵ For the current analyses, to maximize specificity and accuracy of the outcome, we defined hepatic steatosis as the presence of severe hepatic steatosis by ultrasound.

Additional details regarding methods are described in the supplement

Statistical Methods

We characterized the study population by PNPLA3 genotype (CC, CG, GG) using descriptive statistics. We used robust Poisson regression to estimate adjusted prevalence estimates of hepatic steatosis by alcohol consumption categories and PNPLA3 genotype and adjusted prevalence ratios (PR) and 95% confidence intervals. We adjusted for the following potential confounders: race-ethnicity, sex, age, education and income; a priori, we did not adjust for BMI given that PNPLA3 genotype is not associated with BMI.¹⁴ For simplicity and given the scope of the current manuscript results for former drinkers are only shown in the supplemental material.

We examined the interaction between alcohol consumption categories and PNPLA3 genotype in both the multiplicative scale (using the relative prevalence ratio, or RPR) and the additive scale (using the relative excess risk due to interaction, or RERI). Additional details about these measures can be found in the supplemental material.

For the main analyses, we present the results using never drinkers and those with no copy of the allele as the reference group. In sensitivity analyses we present the results using moderate drinkers with no copy of allele, the group with the lowest risk, as the reference group.

Results

In 1991-1994 in the U.S., 57.3% of the U.S. population had the CC genotype, 34.8% CG, and 7.9% had GG genotype, for a weighted allele frequency of the G allele of the rs738409 at PNPLA3 in the overall U.S. population of 25.4%. There were notable race/ethnic differences in the distribution of the PNPLA3 genotypes, with much higher prevalence of the GG genotype among Mexican Americans and lowest among non-Hispanic Blacks. Importantly, there were no differences in BMI, diabetes prevalence and mean number of alcoholic drinks/week by PNPLA3 genotype. (Table 1)

Table 1.

Characteristics of the US Adult Population (20-74 years old) by PNPLA3 Genotype. The Third National and Nutrition Examination Survey. Phase 2 (1991-1994).

	Overall	PNPLA3 genotype CC (57.3%)	PNPLA3 genotype CG (34.8%)	PNPLA3 genotype GG (7.9%)
Unweighted N	4674	2402	1700	572
Mean Age (95%CI)	42.49 (41.39;43.59)	42.85 (41.48;44.21)	42.44 (40.89;43.98)	40.14 (38.48;41.80)
Gender
% Female	51.5%	51.5%	51.5%	51.1%
Race-Ethnicity
Non-Hispanic White	81.7%	81.9%	83.1%	74.6%
Non-Hispanic Black	12.1%	15.7%	8.6%	2.4%
Mexican American	6.2%	2.5%	8.4%	23.0%
Education
% <12 years	18.8%	18.9%	18.2%	21.0%
% 12 years	35.3%	33.1%	38.9%	36.3%
% >12 years	45.8%	48.0%	42.9%	42.7%
Metabolic parameters
Mean BMI (95%CI)	26.95 (26.61;27.30)	26.83 (26.47;27.19)	27.27 (26.75;27.79)	26.48 (25.92;27.03)
BMI categories (%)
Normal	40.4%	41.9%	37.7%	41.3%
Overweight	32.9%	31.30%	33.6%	41.2%
Obese	25.0%	24.80%	27.2%	16.9%
Mean Glucose mg/dL (95%CI)	97.28 (95.51;99.05)	96.94 (94.82;99.07)	98.06 (96.13;100.00)	96.28 (94.51;98.06)
Diabetes (%)	6.6%	6.4%	7.1%	6.3%
Median Triglycerides mg/dL [P25;P75]	108 [76;165]	106 [76;160]	108 [78;168]	108 [73;170]
Liver related measures
Severe Hepatic Steatosis (%)	8.5%	7.5%	9.1%	12.8%
Median ALT IU/L [P25;P75]	15 [11;21]	14 [10;20]	15 [11;22]	17 [11;26]
Median AST IU/L [P25;P75]	19 [16;24]	19 [16;23]	19 [16;24]	20 [17;25]
Median GGT IU/L [P25;P75]	19 [13;29]	19 [13;28]	19 [13;29]	18 [13;33]
Alcohol Consumption
Never	10.5%	9.7%	12.1%	9.2%
Former	33.4%	32.6%	34.5%	34.8%
Low-Moderate [≤7 (W) and ≤14 (M) drinks/week]	47.2%	47.8%	46.2%	47.2%
Heavy [≥7 (W) and ≥14 (M) drinks/week]	8.9%	9.9%	7.3%	8.8%
Median drinks per week [P25;P75]	3.0 [1.0;8.0]	3.0 [1.0;8.0]	3.0 [0.9;6.0]	3.0 [0.9;8.0]

Main effects of alcohol and PNPLA3 genotype on severe hepatic steatosis

Compared to never drinkers, low-moderate drinkers had lower prevalence of severe hepatic steatosis (6.7% vs. 8.8%), whereas those with heavy drinking had higher prevalence (11.4%). (Figure 1) The corresponding prevalence ratios and 95% confidence intervals for the whole population and by race/ethnicity are shown in the Supplement Table 1. Compared to those with the CC genotype, those with CG and GG genotype had higher prevalence of hepatic steatosis: 6.6%, 8.0% and 11.2%, respectively. (Figure 2) The corresponding prevalence ratios and 95% confidence intervals for the whole population and by race/ethnicity are shown in the Supplement Table 2.

Figure 1.

Adjusted prevalence (95% confidence intervals) of severe hepatic steatosis by alcohol consumption categories, overall and stratified Race/Ethnicity. The Third National and Nutrition Examination Survey. Phase 2 (1991-1994).

Figure 2.

Adjusted prevalence (95% confidence intervals) of severe hepatic steatosis across PNPLA3 genotypes, overall and by Race/Ethnicity. The Third National and Nutrition Examination Survey. Phase 2 (1991-1994).

Interaction between alcohol with PNPLA3 Genotype on severe hepatic steatosis

At the population level and after adjusting for socio-demographic characteristics, the lowest prevalence of severe hepatic steatosis was among individual with no copies of the PNPLA3 G variant (CC) and with low-moderate alcohol consumption (5.3%, 95% 2.0%-8.6%), with the highest prevalence observed among those with heavy drinking and homozygous GG (17.3%, 95% CI 0-36.4%). Furthermore, when we compare the shape of the association between alcohol consumption given the PNPLA genotype, we observe how the “J shaped” association between alcohol levels and hepatic steatosis disappears in the presence of the G variant. Among those heterozygous (CG), low-moderate drinkers no longer have a lower prevalence of hepatic steatosis compared to never drinkers (8.0% vs. 7.9%, for moderate and never drinkers, respectively), and among those with 2 copies (GG), low-moderate drinkers and never drinkers have equally high prevalence of severe hepatic steatosis (12.1% vs 6%, for low-moderate and never drinkers, respectively). For heavy drinkers, the prevalence of hepatic steatosis increases monotonically with increases in copies of the G variant, and at any given PNPLA3 genotype, heavy drinkers have the highest prevalence. (Figure 3)

Figure 3.

Adjusted prevalence (95% confidence intervals) of severe hepatic steatosis by alcohol consumption and PNPLA3 genotypes. The Third National and Nutrition Examination Survey. Phase 2 (1991-1994).

Additive and multiplicative interaction metrics

We found evidence of an interaction of PNPLA3 G allele presence on the association between low-moderate alcohol consumption and severe hepatic steatosis on both the multiplicative (RPR=1.95, 95% CI: 1.04-3.65, p=0.048) and additive scales (RERI=0.49, 95% CI 0.13-0.85). Compared to never drinkers, low- moderate alcohol drinking was associated with a 48% decreased risk of hepatic steatosis only among those without PNPLA3 G allele (PR=0.52, 95% CI 0.26-1.05), with no association among those with at least one copy of the PNPLA3 G allele (PR=1.02, 95% CI 0.68-1.54).

For heavy alcohol drinking, compared to never drinkers, heavy alcohol drinking was associated with no increased risk of hepatic steatosis among those without any PNPLA3 G allele (PR=0.98, 95% CI 0.38-2.53), but among those with at least one copy of the PNPLA3 G, there was a non-significant but increased risk of hepatic steatosis (PR= 1.29, 95% CI 0.66-2.53). The results of the interaction metrics were generally consistent with the ones observed for low-moderate drinking, but did not reach statistical significance (RPR=1.71, 95% CI: 0.65-4.51, p=0.29; RERI=0.54, 95% CI −0.38-1.46). (Table 2) In sensitivity analyses, changing the reference to low-moderate drinkers with no PNPLA3 G allele, the results were consistent (Supplement Table 3).

Table 2.

Adjusted^a prevalence ratios (aPRs) and 95% confidence intervals for the interaction between alcohol consumption and PNPLA3 genotype and severe hepatic steatosis. The Third National and Nutrition Examination Survey. Phase 2 (1991-1994).

PNPLA3 G allele	Alcohol consumption
	Never	Low-Moderateb	Heavyb
Stratified PR	aPR (95%CI)	aPR (95%CI)	aPR (95%CI)
0	1.0 (reference)	0.52 (0.26-1.05)	0.98 (0.38-2.53)
+1 copy	1.0 (reference)	1.02 (0.68-1.54)	1.67 (0.93-3.00)
Multiplicative interaction metric
RPRc	1.0 (reference)	1.95 (1.04-3.65)	1.71 (0.65-4.52)
Single referent PR	aPR (95%CI)	aPR (95%CI)	aPR (95%CI)
0	1.0 (reference)	0.52 (0.26-1.05)	0.98 (0.38-2.53)
+1 copy	0.77 (0.45-1.32)	0.79 (0.43-1.44)	1.29 (0.66-2.53)
Additive interaction metrics
RERI d	0 (reference)	0.49 (0.13-0.85)	0.54 (−0.38-1.46)

^aAdjusted for race/ethnicity, sex, age, age squared, education, and income.

^bLow-Moderate alcohol consumption: ≤1/≤2 drinks/day for women/men. Heavy alcohol consumption: >1/2 drink per day in women/men.

^cRPR: Relative prevalence ratio. The null hypothesis is that RPR=1. A RPR>1/RPR<1 indicate positive/negative interaction on the multiplicative scale.

^dRERI: Relative excess risk due to interaction. Calculated as: PR₊₊-PR₊₋-PR₋₊+1. The null hypothesis is that RERI=0. RERI>0/RERI<0 indicate positive/negative interaction on the additive scale.

Discussion

In a representative sample of the adults in the U.S. civilian population, we found a significant interaction between the PNPLA3 gene G variant and alcohol consumption on hepatic steatosis in which we observed a dose-response association between alcohol consumption and hepatic steatosis among those with the GG genotype, and a “J shaped” association suggesting potential beneficial effect of moderate drinking among those without G variant copies (CC genotype). These results imply that abstinence from alcohol may offset the genetic risk of fatty liver disease. Furthermore, the dose-response association of alcohol consumption and hepatic steatosis among individuals with PNPLA3 GG genotype, helps support recent guidelines emphasizing that there is no safe drinking level for patients with NAFLD.

From a clinical point of view, these findings are important given that moderate alcohol consumption is widespread worldwide, yet there is substantial and increasing controversy regarding its health effects.²⁶ Current guidelines for patients with nonalcoholic fatty liver disease concede a potential benefit of moderate drinking and give equivocal recommendations.^{6, 27} Our results, although observational in nature, suggest that any potential beneficial effect of moderate alcohol consumption on fatty liver disease may be limited to those with the CC genotype (just over half the population) and therefore emphasize the role of keeping alcohol consumption low as an effective and modifiable way to protect against liver damage. Indeed, our findings help to further support the recommendations included in the recently published “Diagnosis and Treatment of Alcohol-Related Liver Disease: 2019 Practice Guidance from the American Association for the Study of Liver Diseases” that concludes “Patients with ALD or other liver diseases, in particular NAFLD, NASH, viral hepatitis, and hemochromatosis, should be counseled that there is no safe level of drinking, and that they should abstain”.²⁸

In addition, the findings are of significant medical and public health relevance as the PNPLA3 risk variant is highly common and strongly associated with liver-related outcomes; carriers of at least one copy of the risk alleles have a two-fold risk of liver disease, and carriers of at least two copies having a greater than threefold risk.^{17, 29, 30} In that context, findings related to gene and environment interactions are important. Our findings extend prior studies observing interaction between PNPLA3 variant and adiposity²², PNPLA3 variant and adiposity and alcohol²³, PNPLA3 variant and chronic hepatis c and alcohol³¹. Together, these results imply that even though the effect of PNPLA3 genetic variant on liver disease is strong, this effect may be modulated by environmental factors. These findings emphasize the potential benefits of interventions to lower alcohol consumption and for weight loss at the population level to offset genetic risk for liver disease. Future research is needed to determine to what extent interventions to lower alcohol consumption and for weight loss based on PNPLA3 genotype screening are more efficacious than those without genetic testing.

Moreover, given that the PNPLA3 risk variant has marked race/ethnic differences (49% of individuals with Hispanic ancestry, and 23% among individuals with European ancestry)¹⁴ our findings have substantial relevance for reducing race/ethnic disparities in liver disease among Hispanics, who have experienced large and sustained disparities in liver-related morbidity and mortality.³²

From a biological perspective, the results of this study extend our current understanding of the mechanisms by which PNPLA3 mutation lead to hepatic steatosis, and ethanol leads to triglycerides accumulation. It has been suggested that the presence of the PNPLA3 is not sufficient to cause hepatic steatosis,³³ and that, over-expression of PNPLA3 is also required.³⁴ The lipid droplets coated by mutant PNPLA3 protein resists ubiquitylation and prevents degradation of triglycerides.^{35, 36} Based on animal studies, there is evidence that ethanol increases triglyceride synthesis independently of insulin action, for example thru SREBP-1 expression³⁷, which in turn, can increase expression of PNPLA3.³⁸ In-vivo studies in mice and in vitro experiments have shown that diets rich in either carbohydrates or alcohol induce the expression of PNPLA3.^{38, 39} The extent to which these factors induce the expression of PNPLA3 in humans is an area that deserves further attention. Moreover, recent evidence from in-vivo studies demonstrated the role of downregulation of the PNPLA3 mutant allele in decreasing hepatic steatosis and inflammation,⁴⁰ implying that factors that target this gene variant may be considered as part of the therapeutic armamentarium for treating fatty liver.

While this study had a number of strengths, such as its large size and representativeness of the U.S. adult population, there are also limitations to keep in mind when interpreting the results. These data were collected over 20 years ago. However, this is the only nationally representative study in the US with data on PNPLA3 genotype, alcohol consumption and hepatic steatosis. The prevalence estimates of both hepatic steatosis and the PNPLA3 G allele in the population are likely underestimated given the steep increase in NAFLD and increases in the Latino population.⁴¹ However, given that the mechanistic effects of alcohol and/or the PNPLA3 gene should not be dependent on their distribution in the population, the main inferences of this study should not be affected. This study had an observational cross-sectional design, limiting our ability to draw inferences about causality regarding the effect of alcohol consumption on liver steatosis. However, our study aim was to examine the role of alcohol consumption modulating the genetic vulnerability of PNPLA3 genotype on hepatic steatosis and the main exposure was a genetic variant, not susceptible to confounding. Moreover, since we focused on a single self-reported measure of alcohol consumption, and heavy alcohol consumption is associated with higher morbidity and mortality, individuals with susceptibility to alcohol damage may have been selected out of our sample. This measure may also be subject to recall bias and misclassification. In both cases, the extent to which these biases affect our estimate of the interaction is likely to be towards the null. We used ultrasound to determine hepatic steatosis. While this is not the gold standard, it has shown to be a reliable and accurate tool for the detection of moderate-severe fatty liver and is widely used in large epidemiological studies.⁴² To maximize validity, we focused in the more extreme phenotype (severe steatosis) as the outcome. Future studies using liver biopsy, ultrasound elastography, magnetic resonance measures of hepatic steatosis and clinical outcomes, and with alcohol consumption biomarkers are needed to allow for a more accurate estimation of the effect of the interaction, as well as the characterization of the interaction with liver fibrosis, a more advanced manifestation of fatty liver disease. Last, other genetic determinants of fatty liver such as TM6SF2 and MBOAT7 were not available in NHANES and could not be examined.

Conclusion

Our results suggest that strategies to reduce alcohol consumption may be an effective measure to offset genetic risk of liver disease. Furthermore, our results of a dose-response association of alcohol consumption and hepatic steatosis among individuals with PNPLA3 GG genotype, help support recent guidelines emphasizing that there is no safe drinking level for patients with NAFLD.

What you need to know:

Background:

PNPLA3 variant (I148M) is of significant relevance given the strong association with fatty liver disease and because of the high prevalence in the population and specially among some racial/ethnic minorities. The empirical evidence of the role of alcohol consumption modulating this genetic vulnerability is extremely limited.

Findings:

In this cross-sectional study of a U.S. nationally representative sample we found evidence of an interaction of PNPLA3 G variant on the association between alcohol consumption on hepatic steatosis in which we observe a dose-response association between alcohol consumption and hepatic steatosis among those with the GG genotype, and the “J shaped” association suggesting potential beneficial effect of moderate drinking only observed among those without G variant copies (CC genotype).

Implications for patient care:

Our results suggest that strategies to reduce alcohol consumption may be an effective measure to offset genetic risk of liver disease. Furthermore, our results of a dose-response association of alcohol consumption and hepatic steatosis among individuals with PNPLA3 GG genotype, help support recent guidelines emphasizing that there is no safe drinking level for patients with NAFLD.