Women with higher urine levels of MnBP, MiBP, MBzP, MCPP, and ΣDEHP were more likely to have reported diabetes than were women with the lowest levels, even after accounting for sociodemographic, behavioral, and dietary factors. The addition of BMI attenuated some associations, whereas others became stronger. Nonmonotonic associations were seen for MnBP and ΣDEHP, with the greatest increased odds being seen among women in Q3. Although evidence of a threshold effect appeared to be present for MBzP and MCPP, increasing levels of MiBP were associated with an increasing odds of diabetes. The strongest associations were seen among women with high levels of MiBP and MBzP, who had almost twice the odds of prevalent diabetes as women with the lowest level of exposure. Among women without self-reported diabetes, MiBP was positively associated with FBG, and MiBP and ΣDEHP were positively associated with HOMA-IR. No associations were present for urinary phthalate metabolites and hemoglobin A1c levels.
Our findings for phthalate metabolites and prevalent diabetes are similar to Svensson et al. (2011)
who reported that higher levels of the ΣDEHP metabolites MEHHP and MEOHP were associated with diabetes in a study of 221 Mexican women. In the present study, ΣDEHP was associated with insulin resistance among women who did not report a diagnosis of diabetes. Because insulin resistance often precedes T2DM, this finding suggests that DEHP might affect T2DM risk via effects on insulin resistance. In addition, we found other phthalate metabolites to be associated with an increased odds of diabetes, as well as FBG levels and insulin resistance. These findings suggest that other phthalates might affect T2DM risk either through glucose dysregulation or insulin resistance.
A previous study positing metabolic effects of high phthalate metabolite concentrations found associations between phthalate metabolites and a measure of insulin resistance in male NHANES participants (Stahlhut et al. 2007
). Animal and cellular models provide support for causal effects of phthalate exposures on diabetes risk. For example, rats given DEHP had altered insulin and glycogen levels, as well as increased blood glucose (Gayathri et al. 2004
). In vitro
models using liver cells have also found DEHP to reduce insulin receptor concentrations and glucose oxidation, suggesting that phthalates may lead to insulin resistance (Rengarajan et al. 2007
). These previous findings could provide useful information about the present study’s findings of an association between higher levels of ΣDEHP and ln-HOMA-IR.
Other phthalates associated with diabetes did not appear to be associated with insulin resistance or FBG among women without diagnosed diabetes. In fact, MnBP was not significantly associated with FBG, HOMA-IR, or A1c. MBzP and MCPP were inversely associated with FBG and A1c, respectively, after controlling for BMI and waist circumference. These conflicting results could be due to residual confounding or to an artifact of the cross-sectional study design.
The present exploratory research study had a number of limitations. As a cross-sectional study, we cannot rule out the possibility of reverse causation. Phthalates are known to be in certain types of medications and medical devices, including medical products containing polyvinyl chloride used in intravenous bags and medical tubing. As such, it is possible that some of these associations are due to greater exposure to phthalates through increased use of certain medical devices and medications among women with diabetes (Hauser et al. 2004
; Kelley et al. 2012
). Future studies should longitudinally evaluate the association between phthalate levels and markers of insulin resistance and beta-cell functioning among nondiabetic women to better understand how phthalates could alter normal glucose metabolism and diabetes risk. Additionally, phthalate levels were measured using spot urines and do not account for temporal changes in exposure levels. Despite the rapid excretion of phthalates, which typically occurs within 24–48 hr (Hauser and Calafat 2005
; Swan 2008
) and vary within a person over time (Fromme et al. 2007
), a number of studies show phthalate levels measured at one point to be modestly predictive of levels measured over the course of weeks or months (Hoppin et al. 2002
; Peck et al 2010
; Svensson et al. 2011
). Although within-woman variability in phthalate levels is undoubtedly present, this type of measurement error would likely lead to nondifferential misclassification and a bias toward the null. Future studies should examine multiple urine samples taken over time to better estimate long-term phthalate exposure.
We were also unable to assess combinations of phthalates or adjust for exposure to different types of phthalates in this analysis. Exposure to a particular phthalate is unlikely to occur in isolation. Different phthalates may bind and activate different sets of genes, leading to potentially opposing effects on the normal functioning endocrine system. Correlation between phthalates and complexity of their interaction make it difficult to assess potential combinations of phthalates and their impact on human health outcomes, including diabetes. Future studies may need to assess this association, as high levels of various combinations of phthalates may increase the risk of diabetes.
Finally, type of diabetes was assessed by self-report and did not distinguish between type 1 and T2DM. Although we cannot differentiate diabetes type, we believe most (90–95%) of the persons had T2DM (Harris 1995
). Also, self-reported diabetes may vary by education, age, and other factors. If more respondents of lower socioeconomic status inaccurately report not having diabetes, the resulting bias would be toward the null. Furthermore, a study using NHANES 2003–2006 reported that approximately 30% of diabetes cases were undiagnosed (Danaei et al. 2009
), which could result in an underestimate of the true association between phthalate exposure and diabetes.
Despite the limitations of this study, there are many strengths. First, we explored this research question in a large study population of women who participated in NHANES over an 8-year period evaluating five individual phthalate metabolites and ΣDEHP metabolites. Second, our study population consisted of a representative sample of U.S. women. Third, we were able to control for several potential confounders, including sociodemographic, dietary, and behavioral factors. Fourth, to expand upon our findings of phthalates and prevalent diabetes, we conducted a secondary analysis among women without diagnosed diabetes. This analysis allowed us to explore the association between phthalates and markers of diabetes risk (FBG, HOMA-IR, and A1c). To our knowledge, this is the first study to examine the association between phthalates and diabetes in a large sample of women living in the United States.
In conclusion, urinary levels of MnBP, MiBP, MBzP, MCPP, and ΣDEHP were associated with diabetes among women. For women without diagnosed diabetes, some phthalate metabolites were positively associated with FBG and HOMA-IR. These findings suggest the need to further explore the association between phthalates, insulin resistance, and diabetes. If future studies determine causal links between phthalates and diabetes, then reducing phthalate exposure could decrease the risk of diabetes among women.