CHILDHOOD OBESITY AND ENVIRONMENTAL CHEMCALS

Abstract

Childhood and adolescent rates of obesity and overweight are continuing to increase in much of the world. Risk factors such as diet composition, excess caloric intake, decreased exercise, genetics, and the built environment are active areas of etiologic research. The obesogen hypothesis, which postulates that pre- and peri- natal chemical exposure can contribute to risk of childhood and adolescent obesity, remains relatively under-examined. This review surveys numerous classes of chemicals for which this hypothesis has been explored. We focus on human data where they exist and also discuss the findings of rodent and cell culture studies. Organochlorine chemicals as well as several classes of chemicals that are PPAR agonists are identified as possible risk factors for obesity. Recommendations for future epidemiologic and experimental research on the chemical origins of obesity are also given.

The number of children who are obese and overweight continues to rise in most countries across the world.¹ In the United States, the prevalence of obesity and overweight is growing rapidly among children and adolescents.^{2, 3} For instance West Virginia is among the states with highest adult obesity prevalence and in a West Virginian adolescent medicine clinic, 37% of adolescents had an age- and gender- adjusted body mass index (BMI; body weight in kg divided by height in meters squared) greater than the National Health and Nutrition Examination (NHANES) III 95^th percentile.³ Rising obesity prevalence is a concern for many reasons. The risk of life threatening diseases, such as diabetes, cardiovascular disease and cancer, is increased in obese persons.^4-6 Further, obesity has overtaken cigarette smoking as the most costly and detrimental preventive cause of terminal diseases in the United States, with latest estimates suggesting that obesity accounts for 17% of all US medical costs each year.⁷

Although the increasing prevalence of obesity is usually attributed to changes in diet, physical activity, and underlying genetic susceptibility, the possibility that environmental chemicals could influence obesity is relatively underexplored. Early life exposure to environmental chemicals is beginning to be examined as a contributing cause of the obesity epidemic due to the potentially critical role of pre- and peri- natal metabolic programming in later risk of obesity. Thus, we find it may be useful to think about obesity not just in terms of genetics and lifestyle, but also in terms of how early life exposure to these “obesogenic” chemicals might be setting the stage for weight gain later in life. In this review, we first discuss metabolic programming and the unique physiology of obese children and adolescents. We next examine the evidence of associations between chemicals and obesity with an emphasis on human research results where they exist. There are numerous recent reviews focused on experimental evidence of the mechanisms of obesity caused by chemicals and the interested reader is referred to these publications.^{8, 9} This review concludes with a discussion of the strengths and weaknesses of obesogen research as well as recommendations on future directions of obesogen epidemiological and experimental research. While obesity is closely associated with metabolic syndrome and diabetes, these topics are outside the scope of this review.

UNIQUE SUSCEPTIBILITY OF OBESE CHILDREN AND ADOLESCENTS

Pre- and Peri-natal Metabolic Programming

Metabolic programming during pre- and peri- natal development has become an active area of obesity etiology research that is rife with seeming contradictions. Caloric restriction during pregnancy at the time of the Dutch famine during World War II is associated with a greater occurrence of obesity in adult offspring.¹⁰ In contrast to a nutritionally-limited environment, higher maternal pre-pregnancy BMI and higher gestational weight gain are associated with increased birth weight and fat mass at birth, and increased BMI in young and adult offspring.^{11, 12} Similarly, dietary fat- induced paternal obesity also is also associated with a disruption in insulin secretion and glucose tolerance in offspring.¹³ Maternal diabetes is also associated with increased birth weight as well as childhood overweight and obesity.^{14, 15} Paradoxically, children and adults born small for gestational age also have an increased risk of obesity, which may be mediated by rapid compensatory postnatal growth.^{16, 17} Even studies in which adiposity outcomes are only measured in children capture adult risk because prepubertal BMI correlates to BMI in young adults, and BMI in young adults predicts BMI in mature adults.¹⁸

Related to the field of metabolic programming are the concepts of the developmental origins of human adult disease (DOHAD) and windows of susceptibility to toxicants.^{19, 20} Research in animals and humans shows that the developmental processes that occur at embryonic, fetal and infantile stages are especially vulnerable to disruption from relatively low doses of certain chemicals (Figure 1). When organs and tissues are developing, they are particularly at risk to toxic insult.²⁰ This was first observed decades ago in the case of lead and other metals which could harm neurological development as a result of in utero and childhood exposures. This concept also applies to agents that alter metabolic homeostasis during development, which can lead to obesity, diabetes, and metabolic syndrome.^{19, 21-24} In particular, exposure to toxicants during the organogenesis of tissues involved in metabolic homeostasis, e.g. adipose, liver, skeletal muscle, pancreas and brain, may play an important pathophysiological role in the development of childhood obesity (Figure 1). While much of organogenesis occurs prenatally, adipose, skeletal muscle, pancreas, and brain continue to develop postnatally.¹⁹ It remains possible that fetal adaptations to toxic metabolic insults restrict the scope of adaptive responses to a toxic postnatal environment. If this were the case, one could envision DOHAD similar to the multi-stage carcinogenesis hypothesis, where risk of obesity results from multiple toxic insults that temporally span the various stages in which metabolic tissues are developing.

Figure 1.

Maternal chemical exposures are associated with childhood obesity. Maternal exposure to chemicals may target offspring through gametes, placenta, or milk. Potential target tissues of obesogens in offspring can arise through all three germ cell layers of the blastocyst, which continue to differentiate post-natally.

While much of organogenesis occurs prenatally, adipose, skeletal muscle, pancreas, and brain continue to develop postnatally.

Unique Physiology of Obesity

There are many reasons why children are not merely small adults in terms of the ways that their environment affects them. Similarly, the physiology of obese persons is not the same as the physiology of lean persons. Thus it follows that obese children have unique physiology. In obese children, glucose and lipid metabolism tend to be dysfunctional, as evidenced by the highly prevalent comorbidites of insulin resistance, hyperlipidemia, and metabolic syndrome.²⁵ Obesity is a chronic inflammatory state, which likely explains part of the increased incidence of asthma in obese children.^{25, 26} The endocrine system also functions differently in obese persons. For instance, adipocytes produce hormones such as estrogen and leptin which are produced in excess in obese people. It is not surprising that obesity is associated with decreased reproductive health and there is no indication that adolescent reproductive health is an exception. The fact that polycystic ovarian syndrome is more common in obese females may in part be due to the altered endocrine state of obesity. Childhood obesity is also associated with entering puberty earlier.²⁷

The physiology of obese persons is not the same as the physiology of lean persons.

The pharmacokinetic (PK) and pharmacodynamic (PD) properties of environmental chemicals are different in obese children compared to lean children. PKPD studies of humans tend to be overly simplified in part due to the ethical concerns of deliberate human exposures to toxicants. Thus PKPD studies in obese and/or developing mammals have mostly been explored in rodents.

A given quantity of a lipophilic exposure will be diluted in an obese individual because their total adipose mass is greater than their lean peers. Body and adipose tissue weight gain over time furthers the dilution effect to lower serum levels independently of elimination of the chemical.²⁸ Although all people can gain mass, the temporal dilution of chemicals by mass gain is particularly salient in children because of their rapid growth. Further, the metabolism of chemicals such as 2,3,7,8-tetrachlododibenzo-p-dioxin (TCDD) and Dichlorodiphenyltrichloroethane (DDE) is delayed and half-lives extended in obese mammals relative to lean mammals.^29-31 Perhaps counter intuitively, as TCDD exposure levels decrease the half-life actually increases.²⁹ Thus the chemical concentration in blood may be lower in obese persons due to dilution but the cumulative exposure may be higher because of the extended half-life (particularly if exposure is relatively low in the case of dioxin-like compounds).

Many candidate obesogens are lipophilic chemicals and are therefore deposited in fat tissues. In the case of TCDD, its distribution into adipose tissue is greater at low exposure levels.²⁹ Obese persons tend to have higher lipid levels in circulation and in the case of lipophilic chemicals whose blood concentrations are dependent on blood lipid levels, whole blood chemical quantities may appear higher in obese persons than lean persons even if total body burdens are equivalent.^{32, 33} In other words, the dilution effect of higher body mass may be masked by higher blood lipids when sampling serum or whole blood to assess lipophilic chemical exposure. In some cases, storage of lipophilic chemicals in fatty tissues appears to sequester the chemicals from their toxicity target tissue, and thus obesity may be protective.³⁴ However, if the target tissue of toxicity has a high fat content, the lipophilic chemical will be stored where it is most able to cause toxicity. Thus, the notion of protective sequestration of lipophilic chemicals may be irrelevant in the case of obesogens since metabolic homeostasis is partially regulated by adipose tissue.

In some cases, storage of lipophilic chemicals in fatty tissues appears to sequester the chemicals from their toxicity target tissue, and thus obesity may be protective. However, if the target tissue of toxicity has a high fat content, the lipophilic chemical will be stored where it is most able to cause toxicity.

CANDIDATE OBESOGENS

Persistent Organic Pollutants

Dioxins and Dioxin-like Compounds

Dioxins are persistent organic pollutants (POPs) that are primarily byproducts of industrial activities. The most potent dioxin is 2,3,7,8-Tetrachlorodibenzo-p-dioxin (TCDD). TCDD is a high affinity ligand of the aryl hydrocarbon receptor (AhR), and as such its toxicity is mostly attributed to AhR binding as well as its persistence. Other AhR ligands include polychlorinated dibenzofurans, and some “dioxin-like” polychlorinated biphenyls (DL-PCBs), which were used extensively in industrial applications. These AhR ligands are frequently analyzed according to toxic equivalents (TEQs), which multiplies the mass of dioxins, DL-PCBs and furans by a potency factor that ranks their toxicity relative to TCDD.³⁵ All PCBs that do not bind AhR are called non-dioxin like (NDL).

Most of the prospective studies of developmental exposures to dioxins and PCBs had no association with child, adolescent, or adult obesity (Table 1). For instance, there was no association of either TEQ (calculated from exposure assessment of dioxins and PCBs) or PCB mass in milk on the body weights of toddlers (Table 1).³⁶ Summed PCBs in maternal and cord plasma samples were associated with a transient decreased change in the standard deviation score (SDS, a descriptive statistic used to describe variability^{37, 38}) of body weight from birth to 3 months old in formula fed infants (Table 1).³⁶ Further, prenatal PCB exposure was not associated with BMI or body mass of adult daughters.³⁹ While these two studies were fairly small, a larger prospective study of in utero exposure to PCBs had a trend of increased association to higher body weights in adolescent girls, yet this was not statistically significant nor was this seen in boys (Table 1).⁴⁰

Table 1.

Human organochlorine exposures and obesity. Organized by chemical class with ascending ages per chemical.

Chemical	Age at exposure (n)	Design	Duration (and gender) of Follow- up	Exposure	Outcome
Dioxin-like compounds
TEQ of PCDDs and DL-PCBs 36	2 wk old (105)	Prospective	Term birth- 42 mo old (male and female)	28.0- 155.0 ng TEQ/kg milk fat	NS BW
TEQ of PCDDs43	15- 73 yr old (1,374)	Cross- sectional	0 (male and female)	4.6- 11.2 pg TEQ/g whole blood lipid	NS BMI ≥ 25
TEQ of PCDFs43	15- 73 yr old (1,374)	Cross- sectional	0 (male and female)	2.9- 6.8 pg TEQ/g whole blood lipid	NS BMI ≥ 25
TEQ of PCB11841	Birth (138)	Prospective	Term birth- 3 yr old (male and female)	6.0- 78.7 pg TEQ/g cord plasma lipid	NS change in BMI SDSa
TEQ DL-PCBs43	15- 73 (1,374)	Cross- sectional	0 (male and female)	4.4- 13.0 pg TEQ/g whole blood lipid	Increased trend of BMI ≥ 25 (OR = 2.6 between Q4 and Q1)
PCB11842	14- 15 yr old (887)	Cross- sectional	0 (male)	2.8- 13.6 ng/g serum lipid	Increased BMI (βb = 0.56 kg/m2 per doubled exposure)
PCB11842	14- 15 yr old (792)	Cross- sectional	0 (female)	2.4- 11.6 ng/g serum lipid	Increased BMI (βb = 0.74 kg/m2 per doubled exposure)
NDL-PCBs
PCBsc60	1st trimester (518)	Prospective	14 mo old (male and female)	18.2- 67.0 ng/g maternal serum lipidd	NS rapid growth, NS BMI z-scorea
PCBsc36	Birth (207)	Prospective	Term birth- 42 mo old (male and female)	0.1- 2.1 μg/L cord plasma	Decreased change in BW SDS birth- 3 mo old (βb = −0.4 change in BW SDS per μg/L)
PCBse41	Birth (138)	Prospective	1- 3 yr old (male and female)	9- 442 ng/g cord plasma lipid	Increased change in BMI SDS 1- 3 yr old (βb = 0.003 kg/m2 SDS per ng/g lipid)
PCBsf40	Prenatal (594)	Prospective	14 yr old (male and female)	0.5- 5.5 ppm transplacentag	NS BW
PCBsf39	Prenatal (169)	Prospective	20- 50 yr old (female)	Quintiles: 0.1, 1.9, 3.5, 7.1 μg/L maternal serumg	NS BMI, NS BW
PCBsf40	Postnatal (594)	Prospective	14 yr old (male and female)	0.2- 23.1 total mg consumed from milkg	NS BW
PCBsh42	14- 15 (887)	Cross- sectional	0 (male)	42.7- 141.3 ng/g serum lipid	Decreased BMI (βb = −2.4 kg/m2 per doubled exposure)i
PCBsh42	14- 15 (792)	Cross- sectional	0 (female)	30.3- 98.5 ng/g serum lipid	Decreased BMI (βb = −2.0 kg/m2 per doubled exposure)
Organochlorine pesticides
DDTs59	Prenatal (304)	Prospective	10.8- 20.0 yr old (male)	1.8- 33.1 μg/g maternal serum lipid	NS BMI, NS tricep skinfold thickness, NS central adiposity
DDE60	1st trimester (518)	Prospective	14 mo old (male and female)	Quartiles: 71.7, 116.9, 186.2 ng/g maternal serum lipid	Increased rapid growth 6 mo old (RR = 2.4 between Q2-4 and Q1) increased BMI z-score 14 mo old (RR = 1.2 per log ng/g lipid)
DDE39	Prenatal (169)	Prospective	20- 50 yr old (female)	Quintiles: 1.5, 2.9, 6.1, 9.4 μg/L maternal serumg	Increased BMI (βb = 2.88 kg/m2 per μg/L between Q2-5 and Q1), increased BW (βb = 9.22 kg per μg/L between Q2-5 and Q1)
DDE40	Prenatal (315)	Prospective	14 yr old (male)	0.3- 23.8 ppm transplacentag	Increased BW 14 yr old (> 4 ppm group mean = 60.6 kg, ≤ 1 ppm group mean = 53.7 kg)
DDE40	Prenatal (277)	Prospective	14 yr old (female)	0.3- 23.8 ppm transplacentag	NS BW
DDE41	Birth (138)	Prospective	Term birth- 3 yr old (male and female)	24- 1,816 ng/g cord plasma lipid	Increased BMI SDSa 3 yr old (450 ng/g group mean = 0.1 kg/m2 SDSa, 63.7 ng/g group mean = −0.7 kg/m2 SDSa)j
DDE40	Postnatal (594)	Prospective	14 yr old (male and female)	0.2- 96.3 total mg consumed from milkg	NS BW
DDE42	14- 15 yr old (887)	Cross- sectional	0 (male)	46.8- 403.9 ng/g serum lipid	NS BMI
DDE42	14- 15 yr old (792)	Cross- sectional	0 (female)	39.3- 247.1 ng/g serum lipid	NS BMI
HCB60	1st trimester (518)	Prospective	14 mo old (male and female)	Quartiles: 22.8, 41.0, 66.3 ng/g maternal serum lipid	NS rapid growth, NS BMI z-scorea
HCB61	Birth (482)	Prospective	Term birth- 6.5 yr old (male and female)	0.5- 1.0 ng HCB/mL cord serum interquartile range	Increased BW 6.5 yr old (βb = 1.9 kg between Q4 and Q1), increased BMI 6.5 yr old (βb = 1.0 kg/m2 between Q4 and Q1), increased overweight risk (RR = 1.7 per log ng/mL), increased obese risk (RR = 2.0 per log ng/mL)
HCB42	14- 15 yr old (887)	Cross- sectional	0 (male)	15.2- 34.5 ng/g serum lipid	Decreased BMI (βb = −0.7 kg/m2 per doubled exposure)k
HCB42	14- 15 yr old (792)	Cross- sectional	0 (female)	12.3- 26.6 ng/g serum lipid	Decreased BMI (βb −0.6 kg/m2 per doubled exposure)l
βHCH60	1st trimester (518)	Prospective	14 mo old (male and female)	Quartiles: 21.70, 32.23, 47.28 ng/g maternal serum lipid	NS rapid growth 6 mo old, NS BMI z-scorea 14 mo old

Abbreviations: BMI, body mass index; BW, body weight; DDT, dichlorodiphenyltrichloroethane; dichlorodiphenyldichloroethylene; DL, dioxin-like; HCB, hexachlorobenzene; HCH, hexachlorocyclohexane; mo, month; n, sample size; NDL, non-dioxin-like; NS, not significant; OR, odds ratio; PCB, polychlorinated biphenyl; PCDD, polychlorinated dibenzodioxin; PCDF, polychlorinated dibenzofuran; ppm, parts per million; r = correlation coefficient; RR, relative risk; SD, standard deviation; SDS, standard deviation score; TEQ, toxic equivalents; WHR, waist to hip ratio; yr, year

^aDescriptive statistic used to describe variability³⁸

^bChange in the outcome (e.g. BMI, BW) per one-unit change in the exposure

^cPCB118 + PCB138 + PCB153 + PCB180; note PCB118 is a DL-PCB

^dRange of means across subgroups

^ePCB118 + PCB138 + PCB153 + PCB170 +PCB180; note PCB118 is a DL-PCB

^fPCB congeners not reported

^gExtrapolated

^hPCB138 + PCB153 + PCB180

ⁱLarger change in β in children with low exposure (below PCB median) compared to high exposure (above PCB median); above- and below- median PCB groups are both associated with decreased β

^jLarger change in BMI SDS in children of smoking mothers compared to non smoking mothers

^kLarger change in β in children with low exposure (below HCB median) compared to high exposure (above HCB median); above HCB median group associated with increased β and below median HCB group associated with decreased β

^lLarger change in β in children with low exposure (below median) compared to high exposure (above median); above- and below- median HCB groups both associated with decreased β

Only one prospective study of developmental PCB exposure found a positive association with adiposity: PCBs in umbilical cord levels of Belgian children were positively associated with BMI SDS between the ages of one and three years old, which was when the study ended.⁴¹ Although the TEQ was not associated with BMI SDS in this study, the only dioxin-like compound that contributed to the TEQ was PCB118, which has a relatively low contribution to total TEQ.⁴¹

The majority of the epidemiology evidence in favor of a positive association between PCBs and obesity comes from cross-sectional studies of adults, although a few studies that included children also exist (Table 1).^{42, 43} Cross sectional studies of adult PCB exposures have consistently shown a positive association with measures of obesity.^43-46 Further, there are 3 epidemiology studies of adults that found a positive association between dioxins and obesity and only one study of adults that did not.^{43-45, 47}

Experimental research on dioxins and dioxin-like compounds as obesogens is quite sparse and may be biased by high doses. High doses of a PCB mixture (30 mg/kg/day on GD 10-18) caused a transient decrease in offspring body mass ⁴⁸ Elsewhere prenatal and lactational TCDD (single exposure to 1 μg/kg on GD 12) had no effect on adiposity in several mouse strains.^{49, 50} Perhaps the dose of TCDD and PCBs in these studies was too high to detect obesogenic effects as TCDD induces adipocyte differentiation at low doses and suppresses it at high doses.^51-53 Adipocyte differentiation is considered to play a role in the etiology of obesity primarily during childhood, as adipocyte numbers are currently thought to be in a steady state in adults, regardless of whether they are lean or obese.^{54, 55} As was seen with TCDD, low dose of dioxin- like PCB77 also stimulated adipocyte differentiation.⁵³ Further, recent evidence suggests that peroxisome proliferator activated receptor (PPAR) γ expression may be elevated by lower doses of TCDD and DL-PCBs.^{53, 56} Consistent with these findings, exposure to a PCB mixture (6 mg/kg/day on GD 6- PND 21) was associated with a transient increase in body weights of offspring on PND 16-20, in a 34 day study.⁵⁷ While adult exposure to PCB126 (a DL-PCB) had no impact on body mass in mice in one study, ⁵⁸ another study found that adult mice exposed to PCBs have AhR-mediated increased body mass as well as adipocyte hypertrophy.⁵³

Organochlorine pesticides

Several persistent organochlorine pesticides have been implicated in obesity. While they have not been in commercial use in the United States for over 20 years, they are used abroad. Dichlorodiphenyltrichloroethane (DDT) is rapidly metabolized to dichlorodiphenyldichloroethylene (p,p′-DDE), which has a half-life of about ten years in humans.³¹ Thus, the presence of DDT indicates a recent or current exposure and p,p′-DDE tends to imply a long-term body burden. Hexachlorobenzene (HCB) and hexachlorocyclohexane (HCH, often called lindane) also have long half-lives.

Of the five prospective studies of maternal exposure to DDE, four found a positive association with measures of obesity in offspring (Table 1). A prospective study of 169 women in the Michigan fish-eater cohort (1973-1991) revealed that prenatal DDE exposure significantly increased the BMI and body mass of adult daughters in a dose dependent manner.³⁹ Further, as Belgian umbilical cord DDE increased, BMI SDS increased in two- and three- year old children; the effect of increasing DDE levels on the BMI SDS was greater in children born to women who ever smoked compared children of non-smoking mothers.⁴¹ Transplacental exposure to DDE was associated with increased body weight of adolescent boys, however this was not seen in girls.⁴⁰ This was not seen in a population with higher DDT + DDE exposure levels, where summed DDT and DDE were not significantly associated with the BMI, tricep skinfold thickness, or central adiposity of adolescent boys.⁵⁹

A prospective study of 169 women in the Michigan fish-eater cohort (1973-1991) revealed that prenatal DDE exposure significantly increased the BMI and body mass of adult daughters in a dose dependent manner.

It was recently reported that the risk of rapid infant growth among infants born of women above the lower quartile of DDE exposure was 2.4 times the risk of rapid growth in those born of women in the lowest quartile of DDE exposure (Table 1).⁶⁰ As was seen in the studies by Gladen et al., this effect may have been limited to children born of women with moderate DDE exposures because when prenatal DDE exposures exceeded 750 ng/g, no rapidly growing infants were observed.^{40, 59, 60} When the children of this study were 14 months old, the risk of elevated BMI z-scores (another measure of variability, elevated defined here as ≥ 1.44) increased 1.40 for each unit increase in log ng DDE /g lipid.⁶⁰

There are far fewer studies of prospective developmental exposures to HCB and HCH (Table 1). For instance, the association between cord blood HCB levels and childhood obesity was consistent with a positive dose effect of HCB on body weight and BMI when children were 6.5 years old (Table 1).⁶¹ Cord blood HCB levels in excess of 0.46 ng HCB/ml were associated with a 70% increase in the risk of being overweight and doubled the risk of being obese at age 6.5 years.⁶¹

The only cross-sectional study of children and DDTs found a negative association between DDE and BMI in adolescents of both sexes (Table 1).⁴² However, in the majority (5/6) of cross-sectional studies of adults, DDT and/or DDE exposures are associated with increased obesity.^{42, 44, 45, 62-64} Similarly, serum HCB was associated with decreased BMI in adolescent boys and girls cross-sectionally.⁴² The opposite trend was seen in adults from this same region of Belgium, where serum HCB levels in adult men and women were associated with increased BMI after adjustment for other environmental exposures.⁴²

There is surprisingly little evidence in the rodent literature to support the seeming relationship between perinatal organochlorine pesticides and offspring obesity. In utero exposure to HCB did not consistently affect the body weights of rats across their 100 days of life.⁶⁵ Similarly, adult female rats exposed to HCB did not experience a change in body mass after one month.^{66, 67} Mice prenatally exposed to 100 mg DDT/kg maternal body weight/day had higher body weights in the week after birth when the study ended.⁶⁸ Male rats exposed to DDT during puberty had no change in body weights from puberty to 12 weeks of age when the study ended.⁶⁹ Likewise female rat pups exposed to DDT throughout their gestation and nursing had no change in body weights through six weeks old.⁷⁰ None of these studies examined fat mass. They also did not observe animals through middle-age, when obesity is more likely to be evidenced.

There is surprisingly little evidence in the rodent literature to support the seeming relationship between perinatal organochlorine pesticides and offspring obesity. However, none of the relevant studies observed animals through middle-age, when obesity is more likely to be evidenced.

Despite largely null findings of developmental DDT effects on postnatal growth, one perinatal DDT study stands out in light of recent obesity research. Rats prenatally exposed to 50 mg DDT/kg bw/day for 3 days had transient fetal growth restriction, yet birth weights were similar to control rats.⁷¹ When this same exposure paradigm was used across various postnatal windows, lactational transfer of DDTs also caused no change in offspring body mass.⁷¹ While no changes in offspring liver weights were observed, late gestational or lactational exposure to DDT resulted in excessive and disorganized endoplasmic reticulum as well as excess lipid droplets and a higher RNA/DNA ratio in hepatocytes as early as the day of birth.⁷¹ What is striking about this study is that nearly 30 years later, it was discovered that endoplasmic reticulum stress, which can result from excess protein translation, is tightly coupled to obesity.^{72, 73}

Adult primate research suggests that the association between prenatal DDE and offspring obesity in human may be due to effects of DDE on lipid metabolism. Rhesus monkeys exposed to DDT had decreased cholesterol and phospholipids in the brain, increased hepatic lipogenesis, increased hepatic triglycerides, as well as increased cholesterol and triglycerides in both serum and adipose of primates.^{74, 75}

Polyfluoroalkyls

Perfluoroalkyls (PFOA) are surfactants that act through PPARα and PPAR γ, and perhaps other nuclear receptors.^{76, 77} PPARs are critical in the regulation of fat metabolism and storage, adipocyte differentiation and insulin sensitivity.⁷⁸ As has been observed in other cross-sectional studies of POPs reviewed here, perfluoroalkyls in adolescents are associated with lower waist circumference or have no association with either waist circumference or BMI, while perfluoroalkyls in adults are associated with increased BMI and waist circumference (Table 2).^79-81 While the direction of causality cannot be inferred from cross-sectional studies, two retrospective adult studies using retrospective BMI data with PFOA and PFOS in serum suggest that positive associations between adiposity and perfluroalkyls may only reflect unique exposure of obese people to perfluoroalkyls and/or unique PBPK of perfluoroalkyls in obese people.^{82, 83}

Table 2.

Child and adolescent exposures to peroxisome proliferator activated receptor agonists and obesity. Evidence is presented by chemical class with ascending ages per chemical.

Chemical	Age at exposure (n)	Design	Duration (and gender) of Follow- up	Exposure	Outcome
Perfluoroalkyls
PFOA79	12- 19 yr old (585)	Cross- sectional	0 (male and female)	0.1-37.3 μg/L seruma	NS BMI, NS WC
PFOA80	12-19 yr old (474)	Cross- sectional	0 (male and female)	1.5 ± 0.1 log ng/mL serum	Decreased WC (OR = 0.6 per log ng/mL)
PFOS79	12-19 yr old (322)	Cross- sectional	0 (male)	1.4-392.0 μg/L seruma	Decreased BMI (βb = −2.8 kg/m2 between Q4 and Q1), decreased WC (βb = −9.0 cm between Q4 and Q1)
PFOS79	12-19 (263)	Cross- sectional	0 (female)	1.4-392.0 μg/L seruma	NS BMI, decreased WC (βb = −4.8 cm between Q4 and Q1)
PFOS80	12-19 (474)	Cross- sectional	0 (male and female)	3.11 ± 0.05 log ng/mL serum	Decreased WC (OR = 0.4 per log ng/mL)
PFNA79	12-19 (585)	Cross- sectional	0 (male and female)	0.1-10.3 μg/L seruma	NS BMI, NS WC
PFNA80	12-19 (474)	Cross- sectional	0 (male and female)	−0.3 ± 0.1 log ng/mL serum	NS WC
PFHxS79	12-19 (585)	Cross- sectional	0 (male and female)	0.2-27.1 μg/L seruma	NS BMI, NS WC
PFHxS80	12-19 (474)	Cross- sectional	0 (male and female)	0.9 ± 0.1 log ng/ml	Decreased WC (OR = 0.6 per log ng/mL)
Short lived, but ubiquitous pollutants
MEP100	6- 9 (90)	Cross- sectional	0 (female)	5.3- 2,580.0 μg/L urine	NS BMI
MEP99	6- 11 (656)	Cross- sectional	0 (male and female)	0.6- 9,043.6 μg/L urine	NS BMI and WC
MEP99	12- 19 (662)	Cross- sectional	0 (male)	0.6- 12,359.0 μg/L urine	NS BMI, NS WC
MEP99	12- 19 (682)	Cross- sectional	0 (female)	5.9- 39,631.7 μg/L urine	Increased BMI (βb = 1.7 kg/m2 between Q4 and Q1), increased WC (βb = 4.1 cm between Q4 and Q1)
MECPP100	6- 9 (90)	Cross- sectional	0 (female)	5.9- 2,260.0 μg/L urine	NS BMI
MEHHP100	6- 9 (90)	Cross- sectional	0 (female)	1.4- 1,699.0 μg/L urine	NS BMI
MEHHP99	6- 19 (1,030)	Cross- sectional	0 (male and female)	0.7- 2,118.3 μg/L urine	NS BMI, NS WC
MEHP100	6- 9 (90)	Cross- sectional	0 (female)	0.6- 110.0 μg/L urine	NS BMI
MEHP99	6- 11 (656)	Cross- sectional	0 (male and female)	0.6- 9,043.6 μg/L urine	NS BMI, NS WC
MEHP99	12-19 (662)	Cross- sectional	0 (male)	0.7- 273.4 μg/L urine	NS BMI, NS WC
MEHP99	12- 19 (682)	Cross- sectional	0 (female)	0.7- 549.2 μg/L urine	Decreased BMI (βb = −1.5 kg/m2 between Q4 and Q1), NS WC
MBzP100	6- 9 (90)	Cross- sectional	0 (female)	0.1- 191.0 μg/L urine	NS BMI
MBzP99	6- 19 (2,000)	Cross- sectional	0 (female)	0.2- 1,685.0 μg/L urine	NS BMI, NS WC
MiBP100	6- 9 (90)	Cross- sectional	0 (female)	0.2- 144.0 μg/L urine	NS BMI
MCPP100	6- 9 (90)	Cross- sectional	0 (female)	0.4- 76.9 μg/L μg/L urine	NS BMI
MBP100	6- 9 (90)	Cross- sectional	0 (female)	0.3- 363.0 μg/L urine	NS BMI
MBP99	6- 19 (2,000)	Cross- sectional	0 (male and female)	0.6-2,595.3 μg/L urine	NS BMI, NS WC
MEOHP100	6- 9 (90)	Cross- sectional	0 (female)	1.3- 1,070.0 μg/L urine	NS BMI
MEOHP99	6- 19 (2,000)	Cross- sectional	0 (male and female)	0.8-1,380.1 μg/L urine	NS BMI, NS WC
BPA100	6- 9 (90)	Cross- sectional	0 (female)	<0.2- 26,700.0 μg/L urine	Decreased BMI (≥ 85th %tile group mean = 41.8 μg/L, < 85th %tile group mean = 26.9 μg/L)
Thiazolidinediones
Pioglitazone163	14 ± 1.9 (35)	Randomized control trial	6 months	15 mg orally daily for 3 wks, 30 mg daily thereafter if tolerated	Increased change in BMI- z-scorec (pioglitazone group mean = 0.3 kg/m2 z- scorec, placebo group mean = 0.0 kg/m2 z- scorec)
Rosiglitazone164	13.6 ± 1.6 SD (36)	Randomized control cross-over trial	24 wks each, with 4 wk wash-out	4 mg orally twice daily	NS BMI-SDS, NS WC, NS skin folds

Abbreviations: BMI, body mass index; BPA, bisphenol A; MBP, mono-n-butyl phthalate; MBzP, mono-benzyl phthalate; MCPP, mono(2-ethylhexyl)phthalate; MEHHP, mono-2-ethyl-5-hydroxyhexyl phthalate; MEOHP, mono-2-ethyl-5-oxohexyl phthalate; MEP, monoethyl phthalate; MEHP, mono-2-ethylhexyl phthalate; MiBP, mono-isolbutyl phthalate; MMP, monomethyl phthalate; NS, not-significant; PFOA, perfluorooctanoic acid; PFOS, perfluorooctane sulfate; PFNA, perfluorononanoic acid; PFHxS, perfluorohexane sulfonic acid; OR, odds ratio; SD, standard deviation; SDS, standard deviation score; WC, waist circumference.

^aRange includes adults

^bChange in the outcome (e.g. BMI, BW) per one-unit change in the exposure

^cDescriptive statistic used to describe variability³⁸

Despite the lack of prospective studies in humans of developmental exposure to perfluoroalkyls and offspring body mass and fat, mice exposed to PFOA echo the theme seen with POPs reviewed above, where low developmental doses cause obesity and high doses do not. Mice exposed to low levels of PFOA in utero had significantly increased body mass by 10 weeks old which persisted through mid-life.²⁴

Mice exposed to low levels of perfluoroalkyls in utero had significantly increased body mass by 10 weeks old which persisted through mid-life.

When these mice were 18 months old there was an inverse and direct dose response relationship between in utero PFOA doses and abdominal white- and brown- adipose tissue masses, respectively.²⁴ Mice exposed to high doses of PFOA during gestation had decreased body mass.^{24, 84, 85} As to be expected from our conclusions regarding adult human exposure to perfluoroalkyls, mice exposed to PFOA as adults experienced no change in body- or fat- mass across PFOA doses and ages.²⁴

Polybrominated Diphenyl Ethers

While there are many kinds of brominated flame retardants which are been used to consumer products for the past 30+ years to provide fire safety, one of the major classes, the polybrominated diphenyl ethers (PBDEs)₋were used in many products in the United States and either have been or are being voluntarily phased out, while they are banned in the European Union. Levels of PBDEs in children are 2-6 times higher than those found in adults across the world.^86-91 There are no known studies of developmental exposures to PBDE and obesity in humans. As to be expected with cross-sectional observations of POPs, women with high BMI have higher levels of PBDEs.^{92, 93}

Experimental studies have linked developmental exposure to the commercial penta-PBDE mixture, or to congeners mainly present in that mixture, to changes in body weight, yet the direction of change is inconsistent and observation periods of most of the studies are quite short.^{48, 57, 94-96} In the longest study of developmental PBDE exposure to look at body weights, male mice exposed to BDE47 (2,2′,4,4′-tetraBDE) 10 days after birth had increased body weights from PND 47 until the end of the study, at 4 months of age.⁹⁴ The effects of BDE47 were not as clean cut in another study.⁹⁶ When rats were exposed to 200 μg BDE47/kg every five days from GD 15- PND 20, their offspring weighed more from birth to the end of the study at PND 47 (when p-value = 0.06).⁹⁶ Because these BDE47-exposed offspring were also consistently longer, their BMI was lower than controls on PND 15. In contrast, if the same protocol was used for 2 μg BDE47/kg doses, offspring body weights and lengths did not differ from controls but their BMI were higher than controls at PND 5 and lower than controls at PND 15 and 25.⁹⁶ Kestrels, a bird of prey, exposed to a mixture of penta-PBDEs in ovo and during development weighed more, gain weight more quickly and ate more as nestlings.⁹⁵ However, during the first 3-4 weeks of life, body weights of mouse offspring were unaffected by maternal exposure to PBDE99 (2,2′,4,4′,5-pentaBDE) (0.6, 6, or 30 mg/kg/day GD 6- PND 21; or 1 or 10 mg/kg/day GD 10- 18).⁴⁸, 57 These PBDE induced growth changes might be linked to changes in lipid metabolism. For instance, developmental exposure to BDE47 increased cholesterol levels in shrimp, and exposure to a commercial Penta-PBDE mixture in rats increased lipolysis in their isolated adipocytes.^{97, 98}

Short-lived, but Ubiquitous, Pollutants

Phthalates

Phthalates are used in plastics and fragrances and are known activators of PPARα and PPARγ.⁷⁸ In adolescent girls of the 1999-2002 NHANES, BMI and waist circumference increased with increasing urinary monoethyl phthalate (MEP) levels but BMI decreased in association with monoethylhexyl phthalate (MEHP; Table 2).⁹⁹ Other urinary phthalates detected in the 1999-2002 NHANES were not associated with these measures of adiposity in children or adolescents (Table 2).⁹⁹ In a smaller cross sectional study of adolescent girls, urinary phthalates were not associated with BMI above the national 85^th percentile (Table 2).¹⁰⁰ However in contrast to the trend seen with cross-sectional studies of POPs, cross sectional studies of phthalate metabolites in the urine of adults are associated with increased waist circumference and BMI.^{99, 101}

Cell culture experiments support a potential role of developmental phthalate exposures in obesity. For instance MEHP and dicyclohexyl phthalate induced adipogenesis in the adipocyte differentiation 3T3L1 cell culture model.^102-104 Unfortunately, there are no known in vivo studies of developmental phthalate exposure and body mass outcomes in animals, although adult rodents shed some light on the potential role of diethylhexyl phthalate (DEHP) in human obesity. In several studies, adult rodents exposed to DEHP did not increase body and/or fat mass.^{78, 105, 106} Yet when mouse PPARα is replaced by humanized-PPARα, DEHP increased fat mass in adult mice, suggesting that DEHP could increase fat mass in humans.⁷⁸

Bisphenol A

Bisphenol A is used in a variety of hard plastics, can linings, and thermal papers, and is a PPAR and estrogen receptor agonist.¹⁰⁷ The only epidemiology study that examined the association of BPA and obesity in children used cross-sectional data on adolescent girls; being above the national 85^th percentile of BMI for age and sex was associated with less BPA in urine.¹⁰⁰ A similar finding was reported in adults of the 2003-2004 NHANES.¹⁰⁸

There have been numerous rodent studies that reported on the effects of developmental exposures to BPA and body mass (Table 3). Readers are referred to recent reviews for discussion of the obesogenic properties of BPA in experimental models published prior to 2010, however it should be noted here that differences between similar studies have been attributed to estrogenic contamination of feed, cages, and water bottles.^{8, 109} BPA rodent studies published in 2010, as well as in vitro studies are described below.

Table 3.

Immature rodent exposures to peroxisome proliferator activated receptor agonists and obesity. Evidence is presented by chemical class with ascending- ages and doses per chemical.

Chemical	Exposure Window	Species and gender	Duration of Follow- up	Dose	Outcome
Perfluoroalkyls
PFOA24	GD 1-17	Mus females	18 mo old	0.01, 0.1, or 0.3 mg/kg BW dailya	Increased BW 20-37 wk oldb, increased WAT 18 mo old, decreased BAT 18 mo old
PFOA24	GD 1-17	Mus females	18 mo old	1, 3, or 5 mg/kg BW dailya	Decreased BW various ages, decreased WAT 18 mo old, increased BAT 18 mo old
PFOA85	GD 7-17, GD 10-17, GD 13- 17, or GD 15- 17	Mus males	35 wk old	3, 5, 10 or 20 mg/kg BW dailya	Decreased BW in males through PND71
PFOA85	GD 7-17, GD 10-17, GD 13- 17, or GD 15- 17	Mus females	35 wk old	3, 5, 10 or 20 mg/kg BW dailya	Decreased BW until weaning, increased BW 23 wk old and older if exposed GD 13-17
PFOA84	3-7 wk old	Mus females	7 wk old	10 mg/kg BW daily, 5 days/wka	Decreased BW 7 wk old
PFOA84	3-7 wk old	Mus females	7 wk old	1 or 5 mg/kg daily, 5 days/wka	NS BW
PFOA24	8-10.5 wk old	Mus females	18 mo old	0.01, 0.1, 0.3, 1, 3, or 5 mg/kg dailya	NS BW, NS WAT, NS BAT
Short lived, but ubiquitous pollutants
BPA182	1 wk preconception- weaning	Mus males	28 d old	5, or 10 μg/mlc ≈ 1.2 mg total	NS BW
BPA115	Pre- implanted embryos	Mus males and females	PND 21d	1 nM, or 100 μMe	Increased BW PND 21
BPA111	GD0-PND 21d	Mus males and females	14 wk old	1 μg/kgfg ≈ 0.25 μg/kg BW	Increased BW 3 wk old; NS BW 4-14 wk old; decreased fat if fed HFD 14 wk old
BPA112	GD 0- PND 21d	Rattus male	PND 21	0.15 ppmf	Increased BW PND 21
BPA112	GD 0- PND 21d	Rattus male	PND 72	1.5 ppmf	NS BW
BPA112	GD 0- PND 21d	Rattus females	PND 72	0.15, 1.5, 75 ppmf	NS BW
BPA112	GD 0- PND 21d	Rattus male	PND 72	75, 750, or 2250 ppmf	Decreased BW various days PND 4- 21
BPA112	GD 0- PND 21d	Rattus females	PND 72	750, or 2250 ppmf	Decreased BW various days PND 4- 21
BPA183	GD 3- PND 21d	Mus males and females	21 d old	2, or 200 μg/kg BW/dayf	NS BW
BPA110	GD 9-16	Mus males	6 mo old	10 μg/kg BWh	Increased BW at birth & weaning
BPA110	GD 9-16	Mus males	6 mo old	100 μg/kg BWh	Decreased BW at birth- weaning
BPA110	GD 9-16	Mus females	6 mo old	10, or 100 μg/kg BWh	Decreased BW 3-6 mo old
BPA184	GD 9- 20 (partiution)	Mus females	12 mo old	25, or 250 μg/kg BWi	Increased BW 9 mo old
BPA150	GD 10- PND 30	Mus males	PND 31	1 μg/mLc ≈ 0.26 mg/kg BW	NS BW, NS WAT PND 31
BPA150	GD 10- PND 30	Mus males and females	PND 31	10 μg/mLc ≈ 2.72 mg/kg BW	Increased BW, increased WAT PND 31
BPA150	GD 10- PND 30	Mus females	PND 31	1 μg/mLc ≈ 0.26 mg/kg BW	Increased BW, increased WAT PND 31
BPA150	GD 10- PND 30	Mus females	PND 31	10 μg/mLc ≈ 2.72 mg/kg BW	Increased BW PND 31
BPA146	GD 11- 17	Mus males	PND 60	2 μg/kg BWh	NS BW
BPA185	GD 11- 17	Mus males and females	PND 22d	2.4 μg/kg BWa	Increased BW PND 22
BPA146	GD 11- 17	Mus females	60 d old	2, or 20 μg/kg BWh	Decreased BW various days
BPA147	GD 11- 17	Mus females	PND 310	2, or 20 μg/kg BWa	NW BW
BPA146	GD 11- 17	Mus males	60 d old	20 μg/kg BWh	Decreased BW various days
BPA186	GD 12- PND 21d	Rattus males	PND 90	2.4 μg/kg BWa	Increased BW PND 90
BPA149	GD 15- 19	Mus	16 wk old	0.5, or 10 mg/kg BWh	Increased BW
BPA187	GD 21- PND 21d	Rattus females	16 mo old	1, or 10 mg/Lc ≈ 0.1, or 1.2 mg/kg BW	Increased BW various days
BPA187	GD 21- PND 21d	Rattus males	3 mo old	1, or 10 mg/Lc ≈ 0.1, or 1.2 mg/kg BW	Increased BW various days
BPA188	PND 5	Rattus males	8 wk old	0.02, 0.2, 2, or 20 μg totalj	NS BW
BPA186	PND 21- 90	Rattus males	PND 90	2.4 μg/kg BWa	NS BW
BPA.113	4- 10 wk	Rattus males	10 wk old	20 mg/kg BWh	NS BW
BPA.113	4- 10 wk	Rattus males	10 wk old	100, or 200 mg/kg BWh	Decreased BW 10 wk old
Metals
TBT123	“pregnancy”- PND 21d	Mus males	PND 21	15, or 50 ppmc	Decreased BW PND 7
TBT122	GD 0- PND 21d	Rattus females	15 wk old	125 ppmfk	Decreased BW 9-15 wk old
TBT70	GD 0- PND 21d	Rattus females	6 wk old	25 μg TBT/gf	Decreased BW PND 28 and 6 wk old
TBT120	GD 4- PND 21d	Rattus males and females	PND 23	6 mg/kg BWa	Decreased BW PND 1- 2
TBT120	GD 4- PND 21d	Rattus males and females	PND 23	2 mg/kg BWa	NS BW
TBT121	GD 6-17	Mus males and females	PND 55	7.5 mg/kg BWa	NS BW
TBT121	GD 6-17	Mus males and females	PND 55	15 mg/kg BWa	Decreased BW PND 1
TBT119	GD 12- 18	Mus males and females	10 wk old	0.05, or 0.5 mg/kg BWl	NS BW, increased WAT 10 wk old
TBT125	PND 24- 45	Mus males	PND 84	0.05 mg/kg BW every 3 daysa	Increased BW PND 56- 84
TBT125	PND 24- 45	Mus males	PND 84	0.5 mg/kg BW every 3 daysa	NS BW
TBT126	“after quarantine” (which began PND 21)- 45 days later	Mus males	45 days after treatment	0.5, or 50 μg/kg BW every 3 daysa	NS BW, NS fat mass
TBT126	“after quarantine” (which began PND 21)- 45 days later	Mus males	45 days after treatment	5 μg/kg BW every 3 daysa	Increased BW gain, increased fat mass
TBT69	6- 12 wk old	Rattus males	12 wk old	0.04 ng/gf	Increased BW 8- 12 wk old
TBT122	9- 15 wk	Rattus females	15 wk old	125 ppm dietfk	Decreased BW 9-15 wk old
Thiazolidinediones
Pioglitazone165	7- 12.5 wk	Rattus males	12.5 wk old	12 mg/kg BWa	Increased BW 7.5- 12.5 wk old
Pioglitazone165	10.5- 15 wk	Rattus males	15 wk old	12 mg/kg BWa	Increased BW 12-15 wk old
Rosiglitazone.166	PND 21- 60	Rattus females	PND 60	11 μmolf	Increased WAT PND 60, increased BAT PND 60
Englitazone168	GD 16- 21	Rattus males and females	PND 0	50 mg/kg BWa	Decreased BW PND 0

Abbreviations: BAT, brown adipose tissue; BMI, body mass index; BPA, bisphenol A; BW, body weight; GD, gestation day; mo, month; NS, not-significant; PFOA, perfluorooctanoic acid; PND, postnatal day; OR, odds ratio; SD, standard deviation; SDS, standard deviation score; TBT, tributyltin; WAT, white adipose tissue; WC, waist circumference; wk, week

^aOral gavage

^bLast measurement of study

^cWater

^dWeaning

^eIn vitro

^fDiet

^gDams ate more diet when it contained BPA PND 14-21

^hSubcutaneous

ⁱOsmotic pump

^jIntracisternal

^kDams ate less diet when it contained TBT

^lIntraperitoneal

Mice exposed to low BPA levels from mid- to late- gestation had significantly increased body mass at birth and weaning, while mice exposed to relatively high (100 μg/kg) BPA from mid- to late- gestation had significantly decreased body mass from birth through weaning.¹¹⁰ Thereafter, the body weights of male offspring exposed to BPA in utero did not differ from control until 6 months of age when they were no longer followed. However female mice exposed to either BPA dose weighed significantly less than control mice at 3 months of age.¹¹⁰

In another study pups from dams that ate food containing ecologically relevant levels of BPA weighed more at weaning. Male and female offspring had similar bodyweights as did controls from weaning to 9 wks old when the study ended, however females had lower fat mass compared to controls at 9 wks.¹¹¹ Rats exposed to BPA in utero experienced little effect of BPA on body weights through post natal day 72, although there were some transient decreases in body weights of males and females in the highest BPA dose groups.¹¹² Adolescent male rats that were exposed to high doses of BPA had decreased body weights at the end of the BPA treatment period, when they were 10 weeks old.¹¹³

In vitro studies support the influence of early life BPA exposure on obesity. For instance, 48 hours after two-cell mouse embryos were cultured with BPA, a greater number became blastocysts if exposed to 1 nM or 3 nM BPA, while fewer became blastocysts if exposed to 100 uM BPA.^{114, 115} Further when 2 cell mouse embryos were exposed to either 1 nM or 100 uM BPA, they were heavier than controls at weaning (end of study).¹¹⁵ In another series of experiments, BPA induced the differentiation of 3T3-L1 fibroblast into adipocytes.^{104, 116, 117}

Metals

Organotins

Organotins are used in plasticizers, slimicides, fungicides, anti-foulants, catalysts, and stabilizers.⁹ While organotins are found in humans, there are no known studies of their relationship with body mass or adiposity in humans to date.⁹ The experimental evidence for tributyltin (TBT) and triphenyltin (TPT) as potential obesogens has been reviewed recently and readers are referred to that review for additional discussion of organotins research.⁹

TBT and TPT are RXR and PPARγ agonists.^{118, 119} Acute pre- and post- natal exposure to TBT (sufficient to increase mortality) decreases post- pubertal growth in mice (Table 3).^120-122 At lower doses, prenatal TBT exposure appears to have little to no impact on body mass while increasing adipose mass. Peripubertal exposure to TBT seems to increase the likelihood of increases in both body and fat mass.

Neonatal mice exposed to TBT in utero had greater Oil Red O staining (indicative of lipid droplets) in their livers, testis, and adipose tissues.¹¹⁹ After cross-fostering with untreated lactating mice, these mice had similar body masses but males exposed to TBT in utero had 20% increased adipose mass over controls in adulthood (Table 3).¹¹⁹ Similar TBT exposure levels in utero decreased body weights of male mouse pups in the first week after birth, but not in the second or third weeks after birth (Table 3).¹²³ Likewise female rat pups exposed to TBT throughout their gestation and nursing had significantly reduced body weights at four and six weeks old (Table 3).⁷⁰ Unfortunately these later two studies did not evaluate fat mass or monitor their rodents into adulthood.^{70, 123}

In another set of experiments where mice were exposed to TBT in utero, more of their multipotent stem cells differentiated into adipocytes when collected from adult mice compared to similarly collected ex vivo cells from vehicle treated mice.¹²⁴ This resulted in a greater lipid accumulation within stem cells turned-adipocytes that were from mice prenatally exposed to TBT compared to vehicle treated mice.¹²⁴ Further, the stem cells from mice exposed to TBT in utero had a greater propensity to become lipid-filled adipocytes when exposed to more TBT or the diabetic drug rosiglitizone, another PPARγ agonist. This increased adipogenic capacity may have resulted in a TBT-induced shift in cell population; prior to experimentally induced differentiation of adipocytes, there were 6% more preadipocytes detected among ex vivo cells from mice prenatally exposed to TBT compared to mice exposed to vehicle.¹²⁴ TBT also appears to act as a developmental obesogen at lower doses in non-mammal animal models. TBT cause a dose-dependent increase in ectopic adipocyte formation around the gonads of male and female Xenopus that were exposed as tadpoles.¹¹⁹

Male mice exposed to TBT during puberty had increased body mass, associated with increased relative fat mass (Table 3)^{125, 126} Similar observations were also seen in male rats (Table 3).⁶⁹ In rats exposed to TBT in utero through adulthood, the trend of body mass and fat is less consistent than seen in other TBT developmental exposure studies; male rats had a small decrease in body mass while two other studies found opposite effects of TBT exposure on the body weights of female rats.¹²⁷

Cell culture models support a role of developmental exposure to organotins in obesity. TBT also induces adipogenesis through PPARγ in multipotent stem cells of mice and humans.¹²⁴ TBT and TPT induce differentiation of 3T3-L1 adipocytes.^{104, 118, 119, 128}

Lead

Lead poisoning is associated with developmental neurotoxicity. There is some evidence that lead exposure may also influence the risk of obesity, but most human studies do not support a positive association between developmental lead exposure and obesity. Lead levels in the teeth of male and female children (mean age 7.4) in the US were positively associated with their BMI measured at the same time.¹²⁹ These childhood dentin lead levels increased as BMI increased from the beginning of the study period to BMI in their young adulthood (mean age 20.5).¹²⁹ However, the lead levels in patella and in tibia, which reflect recent lead exposure and long-term cumulative lead exposure respectively, of these first and second grade children (mean age 7.4 yrs) were not associated with change in BMI measured in young adults.^{129, 130} A cross-sectional study found no association between blood lead levels and obesity in 11 year olds,¹³¹ which are not associated in adult women either.¹³² Another study of adults showed a marginally significant inverse dose response relationship between age-adjusted patella lead levels, which reflect recent exposures, during adulthood and abdominal obesity (p = 0.07).^{130, 133} Animal research is consistent with an early-life susceptibility to lead-associated adiposity and suggests there is an effect of gender: gestational exposure to lead increased the body mass of male but not female middle-aged mice.¹³⁴

Air Pollution: Cigarette Smoke and Diesel Exhaust

Prenatal maternal smoking is associated with increased occurrence of overweight among children and early adolescents.^{21, 135} There is also evidence that prenatal maternal smoking increases the odds of obesity in children.¹³⁶ The BMI SDS of toddlers was also associated with ever smoking among their mothers.⁴¹ In another study, prenatal maternal smoking was not associated with adiposity measured by MRI during early adolescence in males and females.¹³⁷ However, during late puberty, adolescents exposed to maternal prenatal smoking had 26% and 33% higher subcutaneous- and intra-abdominal fat, respectively, than did their unexposed peers.¹³⁷ These results were independent of sex, age and height.¹³⁷ Parental smoking was also associated with increased overweight and obesity among their children in a cross-sectional study.¹³⁵

During late puberty, adolescents exposed to maternal prenatal smoking had 26% and 33% higher subcutaneous- and intra-abdominal fat, respectively, than did their unexposed peers.

When mice were exposed to cigarette smoke while pregnant, the influence of the cigarette smoke on the body weights of their offspring was gender- and diet- dependent.^{138, 139} Adult female offspring fed a normal diet had significantly increased body weights if exposed to cigarette smoke in utero compared to unexposed females, but cigarette smoke did not impact body weights of females fed a high fat diet for 2 weeks.¹³⁸ Adult male offspring exposed to cigarette smoke in utero had a higher body weight than control-treated males if fed a high fat diet, but there was no cigarette smoking effect on male body weight if males ate a normal diet.¹³⁸ Male rats exposed to nicotine in utero had significantly increased body- and white adipose tissue- mass at weaning and through adulthood.¹³⁹ There was also evidence of adipocyte hypertrophy in the white adipose tissue mass at weaning. In utero nicotine exposure did not change food intake, or energy expenditure. However, nicotine exposure was associated with higher food efficiency (food intake relative to body weight increase), decreased physical activity, decreased brown adipose tissue mass, and decreased thermogenesis.¹³⁹ None of these in utero nicotine effects were evident in female offspring.¹³⁹ Adult male (females not tested) mice exposed to the cigarette smoke- and diesel exhaust- constituent benzo[a]pyrene had increased body weights and weight gain compared to unexposed mice.¹⁴⁰ In another study, the longer that adult male rats were exposed to diesel exhaust, the greater the increase in their body weights.¹⁴¹ These effects have not been reproduced in cell culture, where differentiation of 3T3-L1 preadipocytes, as well as their lipid accumulation, was decreased dose-dependently up to the equivalent exposure to one pack of cigarettes.¹⁴²

Pharmaceuticals

Diethylstilbestrol

Diethylstilbestrol (DES) is a synthetic estrogen. Much of the evidence for DES as an obesogen has been produced in one laboratory.¹⁴³ Female mice exposed to 1 μg/kg/day of DES during PND 1-5 had increased body weight and fat mass as adults.¹⁴⁴ This was not seen in males under the same conditions. Mice exposed to 1 mg DES/kg/day during PND 1- 5 lost weight during treatment, but as a result of rapid compensatory growth during peripuberty, also had increased body weight and fat mass as adults. The increased body mass due to neonatal DES exposure persisted throughout adulthood but was no longer statistically significance when mice were 18 months old.¹⁴³

Shorter studies of perinatal DES exposure have found increased, decreased or no change in body masses of rodents in doses ranging from 0.02- 10 μg/kg/day and found consistently decreased body mass at 200 μg/kg/day. Offspring exposed to 0.2 μg DES/kg maternal body mass/day GD 11-17 had increased body mass during the first week of life in one study but not another.^68,145 However, in other studies, maternal exposure to 0.02 or 0.2 μg/kg during the same prenatal window decreased body weights of male and female offspring at various age until PND 60 (measurements stopped).^{146, 147} Later prenatal (GD16-18) exposure to 0.1 μg DES/kg had no impact on offspring at day 21 or 60.¹⁴⁸ Male mice exposed to about 1 μg DES/kg daily from conception to weaning had lower body- and fat- mass than did controls, while their female littermates had lower fat mass compared to controls (monitored till 14 wk old).¹¹¹ Maternal exposure to 2 μg DES/kg prenatally increased birth weights of female neonates and had no impact on the body mass of female or male offspring thereafter (monitored till 60 days old).¹⁴⁶ Yet female mice whose mothers were exposed to 10 μg DES/kg daily during late gestation had increased body weights through 16 weeks of age when the study ended.¹⁴⁹ Offspring exposed to 200 μg DES/kg maternal body mass/day during either GD 11-17 or GD 16- 18 had decreased body mass as neonates and at PND 60 respectively.^{68, 148} Some of the apparent inconsistencies in DES effects have been previously attributed to feeding a diet with estrogenic properties to both controls and DES treated animals.^{109, 150}

Anti-psychotics

The number of office visits made by children and adolescents that included antipsychotic drug treatment increased 6-fold from 1993 to 2002; over 90% of these prescriptions are atypical antipsychotics (AAPs).¹⁵¹ AAPs increase body weight and waist circumference, and children and adolescents have a higher risk than do adults in developing these adverse effects.^152-154 According to a Medicaid database review, children utilizing AAP therapy had greater than double the odds of being diagnosed with obesity.¹⁵⁵ In a prospective study of children, all four AAP treatments examined were significantly associated with increased body weight, fat mass, BMI, and waist circumference.¹⁵⁶ For instance, in just 12 weeks, the mean increase in fat mass of children taking aripiprazole, olanzapine, quetiapine, and risperidone was highly significant, at 2.4, 4.1, 2.8 and 2.4 kg, respectively, while the change in fat mass of untreated children was a mere 0.4 kg.¹⁵⁶ Olanzapine was also associated with the greatest gain in body weight, BMI, and waist circumference.¹⁵⁶ Olanzapine and risperidone were associated with extreme weight gain, in over 90% and 40% respectively, of adolescents in a small 12 week study.¹⁵⁴

In just 12 weeks, the mean increase in fat mass of children taking aripiprazole, olanzapine, quetiapine, and risperidone was highly significant, at 2.4, 4.1, 2.8 and 2.4 kg, respectively, while the change in fat mass of untreated children was a mere 0.4 kg.

Surprisingly few studies of AAP effects on adiposity in immature rodents exist. Lactational exposure to olanzapine increased body mass and increased waist to hip ratios in male and female mouse offspring during the third and fourth weeks of life (end of study).¹⁵⁷ Similarly, both males and females exposed to risperidone via lactation had higher waist to hip ratios.¹⁵⁷ Yet only female mice exposed to risperidone had increased body mass during the fourth week of life.¹⁵⁷ Rats exposed to either olanzapine, risperidone, sulpiride, or haloperidol during puberty had significantly increased body weights and percent intra-abdominal fat.¹⁵⁸ However, pubertal exposure to ziprasidone did not cause these effects.¹⁵⁸ The adult rodent literature on how AAPs cause weight gain has been recently reviewed.¹⁵⁹

Thiazolidinediones

Thiazolidinediones (TZDs), e.g. rosiglitazone, troglitazone, and pioglitazone, are PPARγ agonists used to treat type 2 diabetes. TZDs decrease insulin resistance, circulating triglycerides and free fatty acids.¹⁶⁰ Paradoxically, the PPARγ agonist activity of TZD is also associated with increased body weight gain in clinical trials of adults.^{161, 162} Few clinical trials of TZDs exist in children. In one clinical trial, pioglitazone increased BMI standard deviation scores in adolescents with type I diabetes (Table 1).¹⁶³ Another clinical trial reported that rosiglitazone treatment did not alter the BMI standard deviation scores, waist circumference or skin folds of adolescents (Table 1).¹⁶⁴

Animal studies favor a positive association between TZDs and obesity. Body weight gain was substantially greater in rats that were exposed to pioglitazone in late puberty compared to unexposed rats.¹⁶⁵ Pioglitazone exposures beginning in late puberty nearly doubled fat pad weight in these rats, while causing a much more modest increase in fat pad mass in older rats.¹⁶⁵ These effects on adiposity may have been mediated by increased food intake among rats exposed to pioglitazone.¹⁶⁵ Pubertal exposure to rosiglitazone increased brown and white adipose tissue mass of rats.¹⁶⁶ Adult chickens exposed to troglitazone also have increased fat mass, however, when the troglitazone exposure was confined to their first day of life, they had significantly less fat pad mass and ate less at one- and two- months of age.¹⁶⁷ Consistent with this developmental finding, rats exposed to englitazone during late gestation gave birth to smaller pups.¹⁶⁸

FUTURE RESEARCH DIRECTIONS

Epidemiologic Research

In order to strengthen the hypothesis that certain chemicals cause obesity, one would ideally want a study with four key properties. 1) If obesity modifies the pharmacokinetics and/or pharmacodynamics of lipophilic chemicals, the direction of causal association between obesity and chemicals may be particularly unclear in cross-sectional studies. Ideal obesogens studies would be prospective longitudinal studies with clear separation between exposure assessment and case ascertainment in order to delineate the causal direction of association. 2) Further, the odds or risk of an association between toxicant and obesity due only to the pharmacokintic effects of obesity should be quantified. This would allow epidemiologists to identify how strong an association between toxicants and obesity must be in order to dismiss the conclusion that the association is an artifact of obesity's influence on the toxicant's pharmacokinetic behavior. 3) If chemicals are banned and/or human exposure levels are declining, it would be desirable to have exposures measured in samples collected prior to the time at which chemicals were banned and/or human exposure levels peaked, in order to increase power. 4) Whenever possible, exposures should be measured in utero and in infancy, because this period appears to be the most environmentally sensitive window for setting the lifetime metabolic trajectory.

Experimental Research

Given the relevance of differentiation of adipocytes in childhood obesity, cellular screens of compounds that cause differentiation of multipotent human stem cells or 3T3-L1 cells at environmentally relevant concentrations would be helpful to prioritize in vivo characterization of obesogens. Because a substantial breadth of obesogen evidence relates to chemicals associated with PPAR binding and/or expression, in vitro screens that determine PPAR binding and expression by environmentally relevant levels of candidate obesogens would also be useful to identify chemicals used for in vivo obesogen research. Such a screen would evaluate whether a cumulative addition model of additivity would be useful in testing mixtures of PPAR ligands in a modified version of the TEQ model of additivity, the PPAR equivalence (PPEQ) model. In vivo experimental research to examine mechanisms of obesogens would ideally encompass 4 characteristics. 1) Experimentalists should use whole animals, or at least validate their cell culture models in vivo, because metabolism is a multi-organ dynamic process. Lipid homeostasis involves constant interaction between adipose tissue, liver, muscle, and the hypothalamus. All these tissue are potential targets of metabolic perturbation. 2) Obesogens must be identified by examining energy balance. Only a change in energy balance can change body mass. Caloric intake should be quantified and controlled if divergent across treatment groups. “Energy in” is burned, stored, or excreted. Experiments should identify activity, body temperature, and caloric content of feces precisely. Animals should be placed in metabolic chambers before they reach a steady state of obesity in order to identify initiating metabolic defects leading to obesity. It is not sufficient to merely measure body weight. Adipose tissue mass and distribution should be measured both in vivo using MRI and at study termination. 3) Chemicals should be administered according to the route of human exposure, or as closely as possible. In deciding how to model exposure to ingested chemicals, the stress of oral gavage should be weighed against the potential drawbacks of mixing the exposure into food, such as controlling for potential changes in appetite. 4) Finally, administered doses as well as internal measures of tissue dose, often whole blood- and serum- levels, must be reported to facilitate comparison with humans. For certain rapidly eliminated compounds, urinary concentrations may be appropriate to measure for animal/human comparisons. 5) Positive, as well as negative controls, should be included to be sure that the experiment has the power to detect a response.

CONCLUSION

Although the epidemiology data on developmental obesogens is not yet clear, animal studies indicate that developmental obesogens do exist and numerous chemicals that are candidate obesogens are identified here. While very few studies could singly address all of our recommended research directions, prospective longitudinal cohorts are a powerful approach to obesogens research and would lead to translational studies that integrated some of our epidemiologic research suggestions with our experimental research suggestions.

An emerging unifying theme of obesogen effects reviewed here is the evidence of non-linear effects on body weight and adiposity. Specifically, many chemicals caused cachexia at high doses, but increased body mass at doses closer to the range seen in humans. Historically, bench toxicology research has employed doses of substances log folds higher than observed human exposures in order to detect mortality and severe morbidities, e.g. cachexia, as endpoints using relatively few animals.¹⁶⁹ More obesogens will likely emerge as more ‘subtle’ effects are detected at doses that more closely emulate human exposure levels.

Across different chemical classes reviewed here, numerous studies reported sex-specific effects. It is not clear whether these are true biological phenomena, spurious associations due to multiple testing, or artifacts of study design and publication bias (e.g. only examining/publishing data on one sex). Two primary facts support a biological underpinning to the gender specificity of obesogens. 1) Many candidate obesogens cause sex-specific toxic effects on sexual maturation, reproduction, and cancers. 2) There are well-known sex effects on fat regulation, including the regulation of adipose tissue distribution and leptin signaling by estrogen.¹⁷⁰

Relative to body weight, children ingest more toxicants than do adults eating the same toxicant-contaminated diet.¹⁷¹ Some obese children are likely at an even greater risk of toxic exposures than are lean children because of the content of their diet; higher consumption of fatty animal-based foods is correlated with higher POPs levels in human serum and milk.^172-174 Given the number of POPs that are candidate obesogens, reducing consumption of animal-based fatty foods is likely a sound anti-obesity life style choice both because of nutrition benefit and reduction of exposure to chemical contamination.

There likely are other chemicals in the environment that increase risk of obesity and have yet to be recognized. These obesogens will be discovered among the tens of thousands of new synthetic chemicals invented and produced in the past half century. The majority of these materials have not been tested for toxicity despite public health interventions to reduce exposures to known toxic substances. At present, there are too few mechanistic animal studies and too many cross-sectional epidemiology studies to prove with certainty that any chemicals cause obesity. However, future research needs in both fields are warranted.

The potential effect of chemicals on the developmental programming of obesity is great, yet there are very few studies of chemical exposures during pre- and peri- natal development designed to assess later obesity. Given this, every physician is advised to obtain a brief history of occupational and environmental exposure from every patient and to ask more detailed follow-up questions or seek consultation with a specialist in occupational and environmental medicine if the initial screen raises suspicion of toxic exposure. While outside the scope of this review, the clinician is also alerted that the dysfunctional metabolic, immune, endocrine, and reproductive systems of obese children may also reduce their capacity to defend against toxic insults.^{50, 106, 133, 139, 175-178} Thus the clinician must also seek to identify exposures that cause damage to obese children that is not seen in lean children, in addition to seeking to identify obesogens.

How else can obesogens be identified, given the inherent budget and time limitations relative to the abundant numbers of chemicals in manufacturing, commerce, and waste streams? Existing pharmaceutical research may be an excellent screen. There are a number of anti-psychotic and anti-diabetic drugs that also increase obesity in children and adolescents, and in most cases the molecular mechanisms of these drugs are understood, even if the exact reason for the obesity side-effects are not known.^{152, 179-181} There are many chemicals that are already known to influence people through the same molecules on which these drug classes act. These candidate obesogens should be prioritized for obesity research. Cell based screens to identify new chemicals that act on these receptors, such as the PPAR receptors, are also available to help identify candidate obesogens.¹⁰⁷

The use of adipocyte differentiation as a cell culture model to identify obesogens also has important implications for childhood obesogen research in particular. The total number of adipocytes increases during development. However, the number of adipocytes in an adult are approximately constant whether they are lean or obese; significant weight gain or loss in adults is not accompanied by respective increases or decreases in adipocyte numbers, instead adipocyte size is correlated to adult adiposity (for further discussion see recent reviews^{54, 55}). These observations support the notion that the number of adipocytes a person will have is determined during childhood and adolescence. If the total body fat mass is a function of # adipocytes x adipocyte size, any chemical that increases adipocyte numbers in developing organisms has the potential to greatly increase the total body fat mass. While the field of research on the chemical origins of childhood obesity is in its infancy, tools exist to strengthen its depth and breadth.