Combined Effects of PFAS, Social, and Behavioral Factors on Liver Health

Abstract

Background: Environmental exposures, such as per- and polyfluoroalkyl substances (PFAS), in conjunction with social and behavioral factors, can significantly impact liver health. This research investigates the combined effects of PFAS (perfluorooctanoic acid (PFOA) and perfluorooctane sulfonate (PFOS), alcohol consumption, smoking, income, and education on liver function among the U.S. population, utilizing data from the 2017-2018 National Health and Nutrition Examination Survey (NHANES).

Methods: PFAS concentrations in blood samples were analyzed using online solid-phase extraction combined with liquid chromatography-tandem mass spectrometry (LC-MS/MS), a highly sensitive and specific method for detecting levels of PFAS. Liver function was evaluated using biomarkers such as alanine aminotransferase (ALT), aspartate aminotransferase (AST), alkaline phosphatase (ALP), gamma-glutamyltransferase (GGT), total bilirubin, and the fatty liver index (FLI). Descriptive statistics and multivariable linear regression analyses were employed to assess the associations between exposures and liver outcomes. Bayesian Kernel Machine Regression (BKMR) was utilized to explore the nonlinear and interactive effects of these exposures. To determine the relative influence of each factor on liver health, Posterior Inclusion Probabilities (PIPs) were calculated.

Results: Linear regression analyses indicated that income and education were inversely associated with several liver injury biomarkers, while alcohol use and smoking demonstrated stronger and more consistent associations. Bayesian Kernel Machine Regression (BKMR) further highlighted alcohol and smoking as the most influential predictors, particularly for GGT and total bilirubin, with posterior inclusion probabilities (PIPs) close to 1.0. In contrast, PFAS showed weaker associations. Regression coefficients were small and largely non-significant, and PIPs were comparatively lower across most liver outcomes. Notably, education had a higher PIP for ALT and GGT than PFAS, suggesting a more protective role in liver health. People with higher education levels tend to live healthier lifestyles, have better access to healthcare, and are generally more aware of health risks. These factors can all help reduce the risk of liver problems. Overall mixture effects demonstrated nonlinear trends, including U-shaped relationships for ALT and GGT, and inverse associations for AST, FLI, and ALP.

Conclusion: These findings underscore the importance of considering both environmental and social-behavioral determinants in liver health. While PFAS exposures remain a long-term concern, modifiable lifestyle and structural factors, particularly alcohol, smoking, income, and education, exert more immediate and pronounced effects on hepatic biomarkers in the general population.

Keywords: PFAS; behavior; environmental health; exposome; liver function; socioeconomic status.

Figure 1

Spearman correlation matrix of key predictor variables and liver biomarkers of interest. Correlation coefficients reflect monotonic associations between variables without assuming linearity or normality.

Figure 1

Spearman correlation matrix of key predictor variables and liver biomarkers of interest. Correlation coefficients reflect monotonic associations between variables without assuming linearity or normality.

Figure 1

Spearman correlation matrix of key predictor variables and liver biomarkers of interest. Correlation coefficients reflect monotonic associations between variables without assuming linearity or normality.

Figure 2

Univariate exposure–response and 95-percent credible interval (gray area) for each predictor variable when all other predictor variables are fixed at the 50th percentile. Adjusted for age, race, and sex.

Figure 2

Univariate exposure–response and 95-percent credible interval (gray area) for each predictor variable when all other predictor variables are fixed at the 50th percentile. Adjusted for age, race, and sex.

Figure 3

Bivariate exposure–response for each predictor variable with the first predictor variable (x-axis) increasing from left to right and the second predictor variable fixed at the 0.25, 0.50, and 0.75 quantiles while all other exposures are fixed at the 50th percentile. Adjusted for age, race, and sex.

Figure 3

Bivariate exposure–response for each predictor variable with the first predictor variable (x-axis) increasing from left to right and the second predictor variable fixed at the 0.25, 0.50, and 0.75 quantiles while all other exposures are fixed at the 50th percentile. Adjusted for age, race, and sex.

Figure 4

Single-exposure effect of the individual exposure on liver markers; examining the change in response associated with a change in a single exposure from its 25th to 75th quantile while all other exposures are fixed at a specific quantile (25th, 50th, and 75th). In image blue is the 0.75 quantile, green is the 0.5 quantile, and red is the 0.25 quantile. Adjusted for age, race, and sex. The X-axis represents the estimated change in liver dysfunction, with estimates further to the right indicating higher dysfunction.

Figure 5

Overall exposure effect estimates for each liver biomarker. Each point represents the estimated change in the biomarker level as all exposures (PFAS, alcohol, smoking, income, and education) increase from the 25th to the 75th percentile. Adjusted for age, race, and sex. Vertical lines denote the 95% credible intervals around each estimate. Higher estimates on the Y-axis denote worse liver health.

Figure 5

Overall exposure effect estimates for each liver biomarker. Each point represents the estimated change in the biomarker level as all exposures (PFAS, alcohol, smoking, income, and education) increase from the 25th to the 75th percentile. Adjusted for age, race, and sex. Vertical lines denote the 95% credible intervals around each estimate. Higher estimates on the Y-axis denote worse liver health.