Trends in serum levels of emerging and legacy per- and polyfluoroalkyl substances from 2008 to 2018: a longitudinal study in China

China is an important producer and consumer of per- and polyfluoroalkyl substances (PFAS), but there is limited understanding of longitudinal trends of PFAS in Chinese people. This study investigated decadal trends in PFAS and potential sources of exposure in the Dongfeng-Tongji cohort of Chinese people and explored potential misclassification bias. We repeatedly measured serum PFAS (containing 11 perfluorocarboxylic acids (PFCAs), 9 perfluorosulfonic acids (PFSAs), 6 PFAS precursors, and 4 PFAS alternatives) in 648 samples (collected in 2008, 2013, and 2018) from 216 Chinese residents in Hubei Province. We used linear mixed effect model to examine trends in PFAS concentrations over survey time. We also assessed the potential exposure sources of PFAS using principal component analysis-multiple linear regression (PCA-MLR) model. Eleven PFAS were detected in at least 80% of the population, PFOS, PFOA, and 6:2CL-PFESA being the predominant PFAS. We found a slight decrease in linear-PFOS during the 10-year period; at the same time, the levels of PFOA doubled and the detection rate of short-chain PFCAs (PFBA, PFHpA) rose. 6:2CL-PFESA (China specific PFAS alternative) remained high for a decade. Similar correlation trends between PFAS were observed at three timepoints. Comparison of our findings with those from other studies suggests the study population was exposed to three potential sources of contamination, which may represent ingestion of contaminated diets, long-range transport of atmospheric PFAS, and exposures acquired from dust, drinking water, and daily commodities. We found that serum PFAS measurement at on timepoint was similar to levels over a 5-year period but not a 10-year period. Our study provides information on serum PFAS levels in Chinese retired workers in the Dongfeng-Tongji cohort, and highlights that the risk of exposure to novel PFCAs and PFAS alternatives deserves continued attention. For longitudinal studies with long-term follow-up (e.g. more than 5 years), relying only on a single baseline serum PFAS level may generate misclassification bias resulting in expected bias towards the null and affect the estimation of health effects. https://doi.org/10.1289/EHP15340.