To identify dietary characteristics associated with variation in plasma per- and polyfluoroalkyl substances (PFAS) concentrations among pre-diabetic adults in the US.
We evaluated cross-sectional associations between diet and plasma PFAS concentrations among pre-diabetic adults in the US enrolled in the Diabetic Prevention Program 1996-1999. Participants reported their diet over the previous year using a 117-item food frequency questionnaire. Plasma concentrations of 6 PFASs were measured using SPE-LC/MS/MS. We calculated daily intake for major food groups and characterized dietary patterns using principal component analysis (PCA). We estimated differences in plasma PFAS concentrations relative to diet using multivariate linear regression models adjusting for age, sex, race, marital status, education, income, smoking, waist circumference and total caloric intake.
Participants ( = 941, 65% female , 58% Caucasian, 68% married, 75% with higher education, 95% nonsmoker) had similar plasma PFAS concentrations compared to the general US population measured in 1999-2000. Mean (SD) daily energy intake was 2091 (975) kcal. Using single food group approach, the strongest predictor for plasma PFAS concentrations was fried fish; each SD increase (0.1 serving/day) was associated with 13.6% (95% CI: 7.7, 19.9) and 6.9% (95% CI: 2.2, 11.8) higher median plasma concentration of perfluorononanoic acid (PFNA) and straight-chain perfluorooctanesulfonic acid (n-PFOS), respectively. Some food groups had inverse associations, with the strongest predictor being cruciferous vegetable [11.5% (95% CI: -16.1, -6.6) decrease in median concentration of plasma 2-(N-methyl-perfluorooctane sulfonamido) acetic acid (MeFOSAA) per SD (0.3 serving/day) increase in intake]. PCA showed a to be associated with lower plasma PFAS concentrations, and diet patterns of and to be associated with higher PFAS concentrations.
Dietary patterns may influence PFAS exposures and/or excretion and absorption pathways. Studies of PFASs with diet-sensitive disease outcomes should account for confounding by diet.
This work was supported by the US National Institutes of Health grant R01ES024765.