OBJECTIVE: To examine whether concentrations of specific per- and polyfluoroalkyl substances (PFAS) and a mixture of PFAS relate to incident knee osteoarthritis (KOA) and knee pain progression.

METHODS: Among a case-cohort sample from the Osteoarthritis Initiative (n=1,878), we examined associations of serum concentrations of 8 PFAS with odds of developing over the subsequent 48 months (1) symptomatic KOA and (2) a clinically meaningful change in WOMAC knee pain score. We used weighted logistic regression to assess single PFAS (continuous and quartiles) and quantile g-computation to assess the PFAS-mixture in relation to our outcomes, accounting for sociodemographics.

RESULTS: Fifty-eight percent of participants were female with a mean age of 62 years and BMI of 28.6 kg/m. Participants with higher perfluorooctanoic acid (continuous) had greater odds of incident symptomatic KOA [OR(95%CI) per interquartile-range: 1.12(1.04-1.21)]. Participants in the highest quartiles of perfluorononanoic acid (PFNA) had greater odds of incident symptomatic KOA [OR(95%CI): third quartile 1.63(1.29-2.07); fourth quartile 1.55(1.19-2.01)]. perfluorooctanoic acid may have a nonlinear relationship with the greatest chance of incident symptomatic KOA with serum concentrations between 5.19-6.88 ng/mL [OR (95%CI): 1.55(1.21-1.97)]. Participants with greater perfluorooctanoic acid and PFNA (continuous) had greater odds of knee pain progression [OR (95%CI) per interquartile-range: perfluorooctanoic acid 1.10(1.03-1.18) and PFNA 1.06(1.01-1.11)]. Mixture models showed perfluorooctanoic acid and PFNA were among the PFAS with the strongest positive weights for both outcomes.

CONCLUSION: In a large cohort at risk for KOA, higher perfluorooctanoic acid and PFNA serum concentrations may be associated with a greater likelihood of symptomatic KOA incidence and knee pain progression.