A number of chemicals including persistent organic pollutants such as dioxins, furans, chlorinated pesticides, polychlorinated biphenyls (PCB) flame-retardant polybrominated diphenyl ethers, organochlorine bisphenyls, perfluorinated acids (PFOA, PFOS), non-persistent pollutants such as phthalates, bisphenol A or tributyltin, or heavy metals such as cadmium, arsenic or mercury, have been shown to induce biological effects that alter glucose homeostasis after acute or chronic exposure in mice and rats. Adult mice exposed to TCDD (Seveso Dioxin) exhibited reduced glucose-stimulated insulin release, an effect absent in AhR (Aryl hydrocarbon receptor)-null mice [13]. Concerning pesticides, atrazine promotes insulin resistance [14] and malathion exposure results in increases in both glucose and insulin levels in rats [15]. PCB126 and PCB77 were shown to impair glucose tolerance with induction of insulin resistance when coupled with a low-fat diet [16]. PCB mixture aroclor 1254 induced insulin resistance in both lean and diet-induced obese states [17]. Aroclor1260 administered to mice fed a high-fat diet altered carbohydrate metabolism at multiple levels, highlighting the additive role of dietary metabolic stressors in the effects of EDCs [18].

While acute exposure of male mice to BPA led to a rapid increase in plasma insulin and in a corresponding decrease in plasma glucose, chronic exposure to BPA was found to induce hyperinsulinemia and also to reduce insulin sensitivity, suggesting that BPA may be a diabetogenic factor per se. Interestingly, the impairment in insulin action occurred despite an increase in β-cell insulin content and especially appeared when mice were exposed to BPA orally or by subcutaneous injection [19]. One explanation is that BPA operates through several pathways and mechanisms that independently increase insulin synthesis, while simultaneously inducing peripheral insulin resistance in adipose tissue, liver and especially in skeletal muscle due to impairments of insulin signaling [20]. Furthermore, insulin resistance observed after chronic exposure to BPA may consist in an adaptive cruise to higher insulin levels in order to limit hypoglycemia [21]. Chronic exposure to BPA led also to changes in whole-body energy homeostasis, by a direct effect of BPA on the central nervous system responsible for lower energy intake and for lower energy expenditure [20]. These results are similar to those observed in experimental models exposed with TCDD, diethylhexyl phthalate (DEHP) or tributyltin, an organotin used as biocide in anti-fouling paint in order to prevent the settling of organisms on the hulls of large ships. Cadmium exposure has been shown to promote glucose intolerance with a specific reduction in adipose expression of GLUT4 [22] and in vivo exposure to arsenic promotes glucose intolerance with concordant insulin resistance [23].

Results from experiments using animal models have demonstrated that nutritional changes during pregnancy can directly affect the programming of metabolically active tissues and can induce type-2 diabetes, in their offspring [24]. In addition to nutritional changes, exposure to EDCs in animal models also alters glucose homeostasis in mothers and their offspring when they become adults. For example, prenatal exposure of mice to DES resulted in a significant decrease in body weight during treatment, which was followed by a “catch-up” period around puberty, and then finally resulted in a significant increase in body weight associated with an increase in the percent of body fat after two months of age. Increased body weight was maintained throughout adulthood and DES-treated mice exhibited higher levels of adiponectin, leptin and IL-6, as well as altered glucose levels [9]. Perinatal DDT exposure in rats impairs thermogenesis and the metabolism of carbohydrates and lipids, which may increase susceptibility to the metabolic syndrome in adult female offspring [25]. Exposure to low doses of PFOS during gestation and early postnatal development resulted in glucose intolerance and insulin resistance [26]. In a rat model, females exposed to phthalates (DEHP) throughout gestation and perinatal development, exhibited in offspring, hyperglycemia in the presence of reduced insulin levels and reduction in β cell mass [27]. Results with perinatal exposure to BPA are controversial. While initial studies reported that oral exposure to BPA during pregnancy and lactation increased the body weight of adult male and female offspring [28], more recent reports showed very mild phenotypes. These discrepancies between the studies may be explained by all the differences in experimental design, dosage or both [19]. More recently, effects of BPA exposure during mice pregnancy were shown to mimic the effects of a high-fat diet altering glucose and lipid metabolism [29]. In any case, BPA exposure during a short time during pregnancy may affect the metabolic programming (weight and glucose metabolism) of offspring later in life, but also the metabolic state of the mother in the long-term.

Several studies reported that organic pollutants could enhance insulin resistance in vitro by a direct action on adipocytes. For example, cultured and differentiated mouse adipocytes 3 T3-L1 cells exposed to a mixture of persistent organic pollutants had an impaired response to insulin (especially with organochlorine pesticides) and exhibited a down regulation of insulin-induced gene-1 (Insig-1) and Lpin1, two key regulators of lipid metabolism [30]. Exposure of mouse and human cultured cells to phthalates, especially mono-(2-ethylhexyl)-phthalate (MEHP) and monobenzyl-phthalate (MBzP), led to activation of PPARα and PPARγ, and then to fatty acid oxidation and to a strong adipocyte differentiation [31]. Despite controversial data, some dioxins (especially 2,3,7,7-tetrachlorodibenzo-p-dioxin, TCDD) were demonstrated to impair glucose uptake by the adipose tissue and the pancreas (due to a reduction in GLUT4 and glucokinase expression), as well as insulin secretion [32]. Bisphenol A (BPA), or 2,2′-bis-4-hydroxyphenyl-propane, is one of the highest volume chemicals produced worldwide. It was first designed in the late 1890s as a synthetic estrogenic compound and was investigated for potential commercial use in the 1930s; although its estrogenic activity had been confirmed, BPA has been substituted for DES, which exhibited far more potent estrogenic effects [33]. Due to its heat resistance and its elasticity, a use for BPA was identified in the plastics industry during the 1950s. It is currently used in the manufacturing of polycarbonate plastic and epoxy resins, notably contained in bottles, food storage containers, and dental sealants [34], explaining why most people have detectable levels of BPA in their urine [35] and in their serum [36]. Although much of the concern regarding BPA toxicity focused on reproductive health and development, more recently DES and BPA were shown to induce in freshly isolated islets of Langerhans a rapid glucose-induced Ca²⁺ signal, which is a key messenger responsible for insulin secretion [37]. Exposure to a single low dose of BPA decreased blood glucose levels and simultaneously increased serum insulin levels, but led to post-prandial hyperinsulinemia leading to insulin resistance, especially in peripheral tissues such as adipocytes, liver and skeletal muscle [19]. The augmentation in β-cell insulin content and release after BPA exposure appears to be a direct result of its estrogenic properties, as these effects were not observed in ERα-knockout animals [19]. However, these rapid actions of BPA occurred via a non-classical estrogen-activated pathway, as they were not counteracted by the antiestrogen fulvestrant [38], involving probably GPER [39] and/or ERβ [40] and/or estrogen-related receptor γ (ERRγ) [41]. Human adipocytes have been demonstrated to be also an important target of BPA, which could reduce adiponectin production and secretion [42] and enhance adipocyte differentiation and lipid accumulation [43], leading to an insulin-resistant state that may result in type-2 diabetes when combined with a genetic predisposition, an excess caloric consumption and/or a sedentary lifestyle.

However, effects of EDCs are not limited to interactions with ERs and PPAR. Indeed, EDCs are classically able to induce changes in gene expression that may persist throughout life, despite no change in DNA sequence, also known as epigenetic modifications, including DNA methylation, histone modifications and expression of non-coding RNAs (microRNAs, picoRNAs…). Multiple lines of evidences from animal models have established that epigenetic modifications due to in utero exposure to EDCs, especially DNA methylation, can induce alterations in gene expression that may persist throughout life [44,45]. The epigenetic effect of BPA was clearly demonstrated as maternal exposure to BPA shifted the coat color distribution of viable yellow mouse offspring toward yellow by decreasing CpG methylation of the Agouti gene [44]. Thus, in utero and perinatal exposure to BPA may induce DNA hypermethylation and/or hypomethylation of several gene promoters, as well as histone modifications and expression of miRNAs, which may contribute to alter glucose metabolism and β-cell failure, as observed in type-2 diabetes. These non-genomic mechanisms may also be involved in the transgenerational transmission of this programmed defect [46,47], as previously described for reproductive tract abnormalities and subfertility induced by in utero exposure to DES [48]. Epigenetics represents likely the molecular link [49] between environmental factors and T2D as illustrated by fetal BPA exposure in rat, which results in glucokinase gene promoter hypermethylation, contributing to insulin resistance in adulthood [50] and in F2 generation [51].

Initial data linking diabetes to environmental pollutants have come from epidemiological studies performed after acute and accidental releases of EDCs. The explosion of a chemical plant in Seveso, Italy, was responsible for a massive and acute release of dioxins in the summer of 1976; follow-up studies showed, several years after this environmental disaster, an increased risk of diabetes in exposed people, especially in women [52]. The study of military personnel exposed to dioxins contained in herbicide orange used by the US military program, Operation Ranch Hand, during the Vietnam War, revealed the same association between serum TCDD concentration and the prevalence of type-2 diabetes (relative risk [RR] = 1.5; [95% CI: 1.2–2.0]) [53]. Consumption of adulterated rice bran cooking oil in a group of Taiwanese, also known as the in Yu-cheng incident, was related to a higher prevalence in type-2 diabetes, especially in women (odd ratio [OR] = 2.1; [95% CI: 1.1–4.5]), due to a contamination with polychlorinated biphenyl ethers (PCBs), dioxins and furans [54]. Same recent observations were performed with indirect consumption of persistent organic pollutants (POPs) including organochlorine pollutants, heavy metals and metabolites of the insecticide DDT, despite its ban in 1972 because of its toxicity on fertility [55].

A major limitation of these studies using occupational cohorts is that in many cases, the exposure occurs many years before the analysis is undertaken; thus levels of POPs are usually estimated by questionnaires or by back calculation from serum or urine concentrations measured many years later, while the half-life of each pollutant may easily vary according to the conditions of sample preservation. Furthermore, most of these occupational studies focused on exposure to one EDC at a time, while the worker cohorts are usually exposed to a whole mixture of toxicants, often in large concentrations; therefore, the results could be confounded by exposure to other chemicals, so that they may not reflect the exact risk incurred by the general population.

Epidemiological studies performed on non-occupational populations are very suggestive of an independent association between POPs exposure and diabetes. Nevertheless, they have been focused on individuals who live in heavily polluted areas (Scandinavia, Great Lakes), raising the matter of the control group that has been chosen [56]. Indeed, the diet could act as a confounder in establishing the causality between exposure and diabetes and between exposure and obesity, as most of those people (the cases as well as the controls) are daily exposed to POPs, via food intake and water drink. For this reason, the spectacular concordance between the evolution of diabetes prevalence in the United States and the national production of synthetic organic compounds should be carefully considered [7].

The data showing a link between EDCs exposure and diabetes in general population studies mainly comes from the US National Health and Nutrition Examination Survey (NHANES), which is an American food consumption database program that began in 1960 and was designed to assess the health and nutritional status of adults and children in this country. Since 1999, this program includes testing of blood and urine for an extensive number of chemicals and monitors yearly 5000 to 10,000 representative persons, with a cross-sectional analysis performed each two years. Those analyses showed that most participants had detectable blood and/or urine levels of several chemicals (in particular BPA) and that diagnosed or self-reported diabetes was strongly associated with exposure to PCBs, dioxins, p,p′-DDE, phthalates and also BPA (with an OR reaching 2.74; [95% CI: 1.44–5.23]) after adjustment for age, gender, body mass index and ethnicity [57,58]. However, it is interesting to notice that a recent systematic review showed that the positive correlation observed between EDCs exposure (in particular BPA or phthalates exposure) and diabetes was found only in the 2003/2004 NHANES survey or in pooled data that included the 2003/2004 survey [59,60]. Other cross-sectional studies among the world reported suggestive hints of strong association between diabetes and prediabetes and EDCs exposure [61,62].

Recently, Sun et al. compared the prevalence of diabetes in US women from the Nurses’ Health Study (NHS an NHSII) according to their levels of BPA and phthalates assessed from blood and urine samples collected 5 to 10 years before [63]. After adjustment for BMI, they showed in this prospective, longitudinal case/control study, a positive correlation between urinary BPA and butyl-phthalate concentrations and incident type-2 diabetes in the premenopausal participants (NHSII), but not in the older NHS counterparts [63]. However, some several limitations with these epidemiological studies should be highlighted, as detailed by Magliano et al. [56]. As in any observational study, the association between EDCs exposure and diabetes does not establish causality. Furthermore, many of the above studies (excepted for the NHS and the NHSII) were designed to correlate diabetes prevalence with current EDCs levels, while their current levels may strongly differ from concentrations during disease development. Moreover, EDCs are usually highly correlated with each other, and it is not always possible to precisely determine the independent effect of each chemical compound when people are daily exposed to a whole mixture. Finally, the association between EDCs exposure and diabetes may be due to confounding by fat mass, as most of EDCs are stored in adipose tissue, and/or energy dense diet, especially sweet beverages from tinned cans, which contain a large amount of EDCs. Further higher levels of POPs were related to diabetes only in overweight patients, suggesting that EDCs exposure may have a synergistic effect with increased adipose tissue on the risk of type-2 diabetes [64]. An interesting case/control Canadian study compared recently menopausal obese women (BMI > 30 kg/m²). For similar age, BMI, and fat mass index metabolically healthy obese (MHO) women (30 to 40% of obese patients) showed less visceral adipose tissue, higher insulin sensitivity levels and a more favorable cardiometabolic profile than in metabolically abnormal (MAO) obese women. MHO was associated with lower plasma levels of several dioxin-like and non-dioxin-like persistent organic pollutants, suggesting that POPs should be considered as real risk factors for visceral obesity, insulin resistance, and DT2 [65].