Maternal smoking and nicotine.
The strongest conclusion from the workshop was that nicotine likely acts as a developmental obesogen in humans. This conclusion was based on the very consistent pattern of overweight/obesity observed in epidemiology studies of children of mothers who smoked during pregnancy () and was supported by findings from laboratory animals exposed to nicotine during prenatal development. Crude and adjusted odds ratios (ORs) were similar within the individual epidemiological studies, suggesting that the social and behavioral characteristics that were included in models did not account for the observed differences in the prevalence of overweight (Oken et al. 2008
). Two recent meta-analyses concluded there was some evidence for publication bias, but not enough to negate the overall conclusion of increased risk (Ino 2010
; Oken et al. 2008
). The body weight and adiposity-related changes reported in the animal studies recapitulated to a large extent those seen in children of mothers who smoke (Levin 2005
; Newman et al. 1999
; Oliveira et al. 2009
; Santos-Silva et al. 2010
; Somm et al. 2008
; Williams and Kanagasabai 1984
). The breakout group recognized that other components in cigarette smoke may also be contributing to the association between maternal smoking and childhood overweight/obesity; however, the studies of nicotine in experimental animals provided compelling evidence that nicotine alone was the causal agent.
Figure 1 Association between maternal smoking during pregnancy and overweight/obesity in offspring. Studies are grouped by study design and then sorted by health outcome (overweight or obesity); studies are presented alphabetically by author within the health (more ...) Arsenic.
The breakout group participants that evaluated this literature concluded that the existing human data were limited to sufficient in support of an association between arsenic and diabetes in populations with high exposure levels, namely, regions in Taiwan and Bangladesh with historical problems with arsenic contamination of drinking water (). Although most members of the group considered the evidence sufficient for an association, additional research is needed to determine whether the relationship is causal. Workshop participants concluded that current evidence was insufficient for an association with diabetes and arsenic in lower--exposure areas (< 150 ppb in drinking water), such as the United States and Mexico, although recent studies with better measures of exposure and outcome provided increased evidence for an association (Coronado-Gonzalez et al. 2007
; Del Razo et al. 2011
; Ettinger 2009
Figure 2 Association between arsenic and diabetes in areas of relatively high exposures (> 150 ppm drinking water). Studies are sorted by quality of the diagnostic from worse to better. Abbreviations: adj PR, adjusted prevalence ratio; As, arsenic; avg., (more ...)
The literature on arsenic and diabetes in experimental animals was judged inconclusive. The body of existing studies is highly diverse, with considerable variation in the duration of treatment (1 day to 2 years), routes of adminis-tration, and dose levels used in the studies. Most of the studies treated animals with sodium arsenite [As(III); arsenic tri-oxide], but other arsenicals have also been studied (Aguilar et al. 1997
; Arnold et al. 2003
; Hill et al. 2009
; Paul et al. 2008
). The studies also vary in experimental design and model systems used to assess end points relevant to diabetes as a health effect. Most of the studies were not designed to examine the diabeto-genic effects of chronic arsenic exposure. Although the literature as a whole was judged inconclusive, findings from recent studies that were designed to focus more specifically on glucose homeostasis appear consistent with those human studies that link arsenic exposure to diabetes. Supportive findings include impaired glucose tolerance in studies of mice or rats treated with As(III) for several months at drinking water concentrations from 5 to 50 ppm (Cobo and Castineira 1997
; Paul et al. 2007
; Wang et al. 2009
). In addition, measures of insulin regulation [e.g., homeo-static model assessment (HOMA) insulin resistance)] were affected in Wistar rats treated with 3.4 mg/kg body weight/day As(III) by oral gavage for 90 days (Izquierdo-Vega et al. 2006
) and in pregnant female LM/Bc/Fnn mice treated with 9.6 mg/kg As(V) by intra-peritoneal injection on gestational days 7.5 and 8.5 (Hill et al. 2009
Most in vitro
or mechanistic studies were not designed specifically to study the diabetogenic or adipogenic effects of arsenic. Nevertheless, these studies suggest several pathways by which arsenic could influence pancreatic β-cell function and insulin sensitivity, including oxidative stress and effects on glucose uptake and transport, gluconeogenesis, adipocyte differentiation, and Ca2+
signaling (reviewed by Diaz-Villasenor et al. 2007
; Druwe and Vaillancourt 2010
; Tseng 2004
). Studies suggest that arsenic may exert adverse effects on β-cell function in vitro
through several mechanisms, depending on the concentration tested (Fu et al. 2010
Epidemiological studies of POPs and -diabetes.
POPs comprise a broad class of organohalides (i.e., organochlorines, organofluorines, and organobromines). The POP literature related to diabetes and other metabolic disorders is complex, consisting of approximately 75 epidemiological studies that report hundreds of findings relating to diabetes, altered glucose homeostasis, insulin resistance, or metabolic syndrome. Often results for multiple POPs are reported in the same study. Because of time constraints at the workshop, breakout group participants focused on diabetes outcomes, although findings related to glucose homeostasis, insulin resistance, and metabolic syndrome will be summarized in supplemental materials that accompany the POPs breakout group report. The breakout group developed a quality rating for each study based primarily on the methods used to classify or measure exposure, and the diagnostic used to ascertain diabetes status. Studies received a lower rating if the diagnoses of diabetes came from death certificates, if diabetes was self--reported, if exposure was self-reported, or if exposure was not clearly measured. The breakout group then used the Meta Data Viewer program to assess patterns of association between various POPs chemicals or chemical classes and diabetes (Boyles et al. 2011
The group concluded that there is evidence for a positive association of diabetes with certain organochlorine POPs. Initial data mining indicated the strongest associations of diabetes with trans
-nonachlor, DDT (dichloro-diphenyltrichloroethane)/DDE (dichloro-diphenyldichloroethylene)/DDD (dichloro-chlorophenylethane), and dioxins/dioxin-like chemicals, including polychlorinated biphenyl (PCBs; ). In no case was the body of data considered sufficient to establish causality. The very strong exposure correlations among some POPs [correlation coefficients of 0.50–0.90 (Lee et al. 2006
)] make it difficult to identify individual POPs as potential causal agents.
Figure 3 Association between PCBs and diabetes. Studies are grouped by study design and then sorted alphabetically by first author within each study design category. Abbreviations: CARDIA, Coronary Artery Risk Development in Young Adults; CS, cross-sectional; (more ...) Peroxisome proliferator–activated receptor (PPAR) activators (organotins and phthalates).
Organotins and phthalates were considered together in a breakout group session because these compounds both interact with PPARs. The PPARs are intimately involved in the regu-lation of adipocyte differentiation, production of adipokines, insulin responsiveness, and other biological processes related to glucose and lipid regulation (Janesick and Blumberg 2011
; Kahn and McGraw 2010
; Li et al. 2011
; Wang 2010
). In addition, there is the potential for coexposures to these two chemical classes because both are commonly used as plasticizers in PVC (polyvinylchloride) plastics. The extent and magnitude of exposure are assumed to be higher for phthalates than for organotins, but exposure to organotins is not well characterized (Kannan et al. 2010
The phthalates are less potent activators of PPARγ than are organotins, with agonist activity occurring at concentrations 1,000 times higher (~ 10–100 µM vs. ~ 10–100
nM), and phthalates have not been identified as agonists for RXRα. In contrast to the organotins, the phthalates are more potent agonists for PPARα than for PPARγ. The organotins are not considered activators of PPARα (Blumberg B, personal communication, 28 November 2010). In rodent models, PPARα appears to mediate high-dose di(2-ethylhexyl) phthalate (DEHP)-induced body weight loss, but its role in regulating adipogenesis is less clear (Wang 2010
Phthalates. Three cross-sectional human studies of exposure to phthalates were discussed by the breakout group (Boas et al. 2010
; Hatch et al. 2008
; Stahlhut et al. 2007
). These studies reported some positive associations but did not provide sufficient evidence to conclude there is an association with diabetes or obesity. Therefore, findings suggesting the possibility of sex differences in associations and different effects of individual phthalates were considered preliminary. In these studies, the urinary phthalate metabolite monoethyl phthalate was the phthalate metabolite most often associated with higher body mass index (BMI) (Hatch et al. 2008
), waist circumference (Stahlhut et al. 2007
), or HOMA (Stahlhut et al. 2007
). Mono-2-ethylhexyl phthalate was associated with decreased BMI in females > 12 years of age (Hatch et al. 2008
Understanding differences in PPARα activity between humans and rodents is important with respect to understanding potential effects of phthalates on body weight and metabolic end points. Phthalate monoester metabolite concentrations required to activate human PPARα are two to three times higher than the concentrations required to activate mouse PPARα, and the maximum-fold induction is less for human PPARα than for mouse PPARα (Bility et al. 2004
; Hurst and Waxman 2003
; Maloney and Waxman 1999
). Animals treated with relatively high doses of phthalates such as DEHP typically display decreased body weight and fat mass (Itsuki-Yoneda et al. 2007
; Sakurai et al. 1978
). These effects were not observed in PPARα-knockout mice (Feige et al. 2010
), which suggests they are largely mediated via the PPARα agonist activities of DEHP metabolites (Feige et al. 2010
; Martinelli et al. 2010
). However, when the normal mouse PPARα gene was replaced with the human PPARα gene, mice treated with DEHP gained weight and had increased epididymal white adipose mass compared with wild-type animals (Feige et al. 2010
). PPARγ activity is similar in rodents and humans, but stronger PPARα activity in mice compared with humans may mask effects mediated through PPARγ.
Overall, this breakout group concluded that the existing data, primarily based on animal and in vitro
studies, are suggestive of an effect of BPA on glucose homeostasis, insulin release, cellular signaling in pancreatic β cells, and adipogenesis (Alonso-Magdalena et al. 2010
; Miyawaki et al. 2007
; Ryan et al. 2010
; Somm et al. 2009
). The existing human data on BPA and diabetes (Lang et al. 2008
; Melzer et al. 2010
) available at the time of the workshop were considered too limited to draw meaningful conclusions. Similarly, data were insufficient to evaluate BPA as a potential risk factor for childhood obesity: Only one pilot study was available at the time of the workshop (Wolff et al. 2008
It was not possible to reach clear conclusions about BPA and obesity from the existing animal data. Although several studies report body weight gain after developmental exposure, the overall pattern across studies is inconsistent. However, breakout group participants emphasized that body weight is not considered a good measure of obesity in rodents and noted that only a few studies have assessed obesity using the preferred metrics such as fat mass, fat pad weight, and cell adipose tissue cellularity. There is inconsistency in the in vivo
findings that may relate to differences in experimental designsuch as differences in diet, route of administration, and species/strain. Understanding the basis for these inconsistencies was considered a research priority. The group also noted that the mechanisms of BPA action are not fully understood but extend beyond its activity as an estrogen receptor agonist. A number of in vitro
findings suggest interactions with other receptor systems involved in metabolic regulation (Wetherill et al. 2007
), including anti-androgen effects at low concentrations and high binding affinity for estrogen-related receptor-γ (Takayanagi et al. 2006
The pesticide breakout group concluded the epidemiological, animal, and mechanistic data support the biological plausibility that exposure to multiple classes of pesticides may affect risk factors for diabetes and obesity, although many significant data gaps remain. Some active ingredients of pesticides, and of insecticides in particular, affect neurotransmitter and/or ion channel systems that are also involved in regulating pancreatic function, including acetylcholine (e.g., organophosphate, carbamate, neonicotinoids), sodium channels (e.g., pyrethroids), γ-aminobutyric acid (e.g., organochlorine), catecholamine (e.g., amidine/-formamidine), and mitochondrial function (e.g., rotenone). This raises the possibility that these compounds might affect glucose homeostasis, at least at dose levels where they are effective as pesticides (Franklin and Wollheim 2004
; Satin and Kinard 1998
). Much less research has focused on whether pesticides have activities that might affect adiposity or other components for metabolic syndrome.
There have been numerous reports of intoxication with organophosphate insecticides on blood glucose in laboratory animals, generally finding hyperglycemia at high dose levels (see reviews by Karami-Mohajeri and Abdollahi 2010
; Rahimi and Abdollahi 2007
). Recently, the focus of investigations has shifted toward studies designed to understand the consequences of developmental exposure to lower doses of organophosphates, and the long-term effects of these exposures on metabolic dysfunction, diabetes, and obesity later in life (Adigun et al. 2010a
; Icenogle et al. 2004
; Lassiter et al. 2008
; Levin et al. 2002
; Roegge et al. 2008
; Slotkin et al. 2005
; reviewed by Slotkin 2010
). The general findings are that early-life exposure to otherwise subtoxic levels of organophosphates results in pre-diabetes, abnormalities of lipid metabolism, and promotion of obesity in response to increased dietary fat.
The EPA Toxicity Reference Database (ToxRefDB; U.S. EPA 2011b), was also used as a resource for the pesticide breakout group. The current version of the ToxRefDB contains detailed study and effect information on 474 chemicals, primarily the data-rich pesticide active ingredients. Many of these studies were conducted for regulatory purposes and are not available in the peer-reviewed literature. ToxRefDB was queried for chemicals that caused increased body weight (or body weight gain), increased blood glucose, and pancreatic effects, including changes in mass, adenomas, atrophy, congestion, hyperplasia, hypertrophy, inflammation, fatty change, degeneration, and cellular infiltration. Approximately 100 chemicals were causes of at least one of these effects (see NTP 2011b, appendix B). Six of the studies identified increased body weight as a result of treatment with several organophosphates, including two separate studies for fenthion, one conducted in rats and the other in mice (). Several sulfonylurea herbicides and imidazole fungicides were also identified by the ToxRefDB search. These pesticides belong to the same general chemical class as agents used to manage type 2 diabetes or that are being investigated as potential therapeutic agents.
Selected results from ToxRefDB search for chemicals that caused increased body weight, increased blood glucose, or pancreatic effects.