BackgroundPersistent organic pollutants (POPs) have been associated with type 2 diabetes (T2D), but causality is uncertain. ObjectiveWithin longitudinal population-based data from northern Sweden, we assessed how POPs associated with T2D prospectively and cross-sectionally, and further investigated factors related to individual changes in POP concentrations. MethodsFor 129 case-controls pairs matched by age, sex and date of sampling, plasma concentrations of hexachlorobenzene (HCB), dichlorodiphenyl-dichloroethylene (p,p'-DDE), dioxin-like (DL) polychlorinated biphenyl congeners (PCB-118 and PCB-156), and non-dioxin like (NDL-PCB: PCB-74, -99, -138 -153, −170, −180, −183 and PCB-187) were analyzed twice (baseline and follow-up, 9–20 years apart). The cases received their T2D diagnose between baseline and follow-up. Prospective (using baseline data) and cross-sectional (using follow-up data) odds ratios (ORs) for T2D on lipid standardized POPs (HCB, p,p'-DDE, ∑DL-PCBs, ∑NDL-PCBs) were estimated using conditional logistic regression, adjusting for body mass index (BMI) and plasma lipids. The influence of BMI, weight-change, and plasma lipids on longitudinal changes in POP concentrations were evaluated among non-diabetic individuals (n = 306). ResultsPOPs were associated with T2D in both the prospective and cross-sectional assessments. Of a standard deviation increase in POPs, prospective ORs ranged 1.42 (95% CI: 0.99, 2.06) for ∑NDL-PCBs to 1.55 (95% CI: 1.01, 2.38) for HCB (p < 0.05 only for HCB), and cross-sectional ORs ranged 1.62 (95% CI: 1.13; 2.32) for p,p'-DDE to 2.06 (95% CI: 1.29, 3.28) for ∑DL-PCBs (p < 0.05 for all POPs).In analyses of non-diabetic individuals, higher baseline BMI, decreased weight and decreased plasma lipid concentrations were associated with a slower decrease of POPs. Cases had, besides a higher BMI, reduced cholesterol and weight gain at follow-up compared to controls, which can explain the higher ORs in the cross-sectional assessments. DiscussionThe association between POPs and T2D was confirmed, but an indication that individuals body fat history might influence POP-T2D associations weakens the epidemiological support for a causal association. It also warrants studies based on other exposure metrics than biomonitoring. In addition, we note that a cross-sectional design overestimates the ORs if T2D cases have successfully intervened on weight and/or blood lipids, as changes in these factors cause changes in POPs.