Life-Time Environmental Chemical Exposure and Obesity: Review of Epidemiological Studies Using Human Biomonitoring Methods

The exponential global increase in the incidence of obesity may be partly attributable to environmental chemical (EC) exposure. Humans are constantly exposed to ECs, primarily through environmental components. This review compiled human epidemiological study findings of associations between blood and/or urinary exposure levels of ECs and anthropometric overweight and obesity indices. The findings reveal research gaps that should be addressed. We searched MEDLINE (PubMed) for full text English articles published in 2006–2020 using the keywords “environmental exposure” and “obesity”. A total of 821 articles were retrieved; 102 reported relationships between environmental exposure and obesity indices. ECs were the predominantly studied environmental exposure compounds. The ECs were grouped into phenols, phthalates, and persistent organic pollutants (POPs) to evaluate obesogenic roles. In total, 106 articles meeting the inclusion criteria were summarized after an additional search by each group of EC combined with obesity in the PubMed and Scopus databases. Dose-dependent positive associations between bisphenol A (BPA) and various obesity indices were revealed. Both individual and summed di(2-ethylhexyl) phthalate (DEHP) and non-DEHP metabolites showed inconsistent associations with overweight and obesity indices, although mono-butyl phthalate (MBP), mono-ethyl phthalate (MEP), and mono-benzyl phthalate (MBzP) seem to have obesogenic roles in adolescents, adults, and the elderly. Maternal exposure levels of individual POP metabolites or congeners showed inconsistent associations, whereas dichlorodiphenyldichloroethylene (DDE) and perfluorooctanoic acid (PFOA) were positively associated with obesity indices. There was insufficient evidence of associations between early childhood EC exposure and the subsequent development of overweight and obesity in late childhood. Overall, human evidence explicitly reveals the consistent obesogenic roles of BPA, DDE, and PFOA, but inconsistent roles of phthalate metabolites and other POPs. Further prospective studies may yield deeper insights into the overall scenario.

Over time, the use of synthetic chemicals has grown exponentially with the development of commerce and industry (33). Excessive usage results in environmental contamination. Humans are exposed to these ECs through environmental media by ingestion, inhalation, absorption, and even through transplacental transfer and breast milk (34)(35)(36)(37)(38)(39)(40)(41)(42). The human exposure levels of these ECs are generally estimated by biomonitoring of their metabolites or parent compounds in human urine or blood (cord blood or peripheral blood) as exposure biomarkers worldwide (43)(44)(45)(46)(47).
Recently, there has been increased interest in epidemiological studies of EC biomonitoring and subsequent evaluation of their obesogenic effects (4,8,34,(48)(49)(50)(51). A concise view of the overall epidemiological findings is required to clarify whether obesogenic evidence of ECs is sufficient or consistent for the advancement of future research. Some previous reviews have explored the obesogenic role of ECs. However, most of these considered only a single group of ECs, and/or selected ECs based on their endocrine-disrupting properties, and/or considered limited exposure and outcome assessment period or age, and even not focused on epidemiological studies, and/or focused on a mechanism (52)(53)(54)(55)(56)(57)(58)(59).
A further review addressing the current epidemiological evidence of the obesogenic effects of ECs at all stages of life from a public health perspective is needed. Accordingly, the objectives of the present review are to illuminate epidemiological study findings of the associations between EC exposure and anthropometric overweight and obesity indices, uncover the current research gap, and contemplate future research.

Selection and Grouping of EOs
Research articles that demonstrated the associations between environmental exposure and obesity in MEDLINE of PubMed were searched for using "environmental exposure" AND "obesity" as keywords to select EOs ( Figure 1). After additional filtering for full text, journal articles, inclusion of humans, English, and publication year (2006-2020), a total of 821 articles were retrieved. Of these, 719 articles were excluded owing to the following reasons: abstract not available (n=10); involved clinical trials (n=7), review/systematic review/metaanalysis (n=299); cell line studies (n=12); animal studies and statistical/computational models (n=21); editorial/commentary/ protocol and approach (n=19); investigated associations of EC exposure with other adverse outcomes, including hypertension, puberty, diabetes, polycystic ovary syndrome, cardiovascular diseases, cancer risk, and others, and simple biomonitoring and ecological studies (n=351). In the remaining 102 articles, the ECs were predominantly studied environmental exposure (ECs = 62 and others = 40). Also, the production and uses of agricultural, industrials, and other synthetic chemicals are increasing, and recognized as major environmental pollutants over other environmental exposures namely heavy metals, noise or sound, green space and particulate matters. Therefore, we selected ECs as the major EOs apart from other environmental exposure and grouped them as follows: (i) phenols [bisphenol A (BPA), bisphenol S (BPS), bisphenol F (BPF), and others], (ii) phthalates (all phthalates and their metabolites), and (iii) persistent organic pollutants (POPs) [organochlorine compounds (OCs), polybrominated diphenyl ethers (PBDEs) and per-and polyfluoroalkyl substances (PFASs), and their metabolites or congeners] (Figure 1).

Literature Search and Inclusion Criteria
A primary search in PubMed and Scopus databases for each group of EO used the keywords "bisphenols" AND "obesity," "phthalate" AND "obesity", and "persistent organic pollutants" AND "obesity" to identify original research articles of human epidemiological studies. Additional PubMed filtering and Scopus refining were performed to select relevant articles ( Figure 2). Articles were considered relevant when they investigated the associations of selected EOs with anthropometric overweight and obesity indices. The references of the selected primary research articles were also searched for relevant publications. A secondary search was also performed for each group of POPs combined with obesity ( Figure 2).
All full-length articles, short communications, and brief reports of original research work from all over the world, irrespective of sex, religion, and race/ethnicity, were included in this review ( Figure 1). Inclusion criteria included (i) epidemiological study (cohort, cross-sectional, and casecontrol); (ii) all ages and/or life-stage at exposure or outcome assessment; (iii) primary outcomes of overweight and/or obesity, or at least one anthropometric index of overweight or obesity; (iv) EO concentrations measured in urine or blood as human biomonitoring; (v) assessment of only non-occupational exposure levels of EOs; (vi) published after postulating "obesogen hypothesis"; and (vii) written in English. All other articles were excluded ( Figure 2). Finally, 106 original research articles were included in this review.

Environmental Phenols and Obesity
We summarized a total of 33 human epidemiological studies, including 13 cohort studies and 20 cross-sectional studies that explored the association between prenatal and early-to later-life urinary phenols, especially bisphenol exposure levels with anthropometric overweight and obesity indices ( Table 1 and Matrix Table 1). Most of the cohort studies were birth cohorts. The study subjects enrolled in the birth cohorts ranged from 173 to 1128 mother-child pairs. Among the 20 cross-sectional studies, 9 involved children and adolescents between the ages of 3 and 19 years, and 11 involved adults and elderly participants >18 years. Both the cohort and cross-sectional studies measured BPA, BPS, and BPF in spot urine other than the first morning void urine, or 24 h urine.
Maternal urinary BPA levels showed null or positive associations with one or more anthropometric obesity indices   in infants and toddlers (61,73,80). Similar associations were also found between maternal BPA exposure levels and obesity measures in preschoolers and school-aged children. These associations were sex-specific (8, 50, 61, 70-73, 78, 80). Only one study reported negative associations between prenatal BPA exposure and BMI z-score and %BF (79). Toddler and preschooler exposure levels of BPA reported null or positive associations with overweight or obesity indices in toddlers, preschoolers, and school-aged children (8,70,72,73,78,79). Associations were mostly null in children 5 to 9 years of age (72,79). Urinary BPA concentrations among school-aged children showed inconsistent relationships with one or more obesity n, number; y, year; m, month; T r , trimester; Q, quartile/quantile; T, tercile/tertile; NA, not available; GO, general obesity; AO, abdominal/central obesity; BPA, bisphenol A; BPS, bisphenol S; BPF, bisphenol F; DP, dichlorophenol; BeP-3, benzophenone-3; IQR, interquartile range; BMI, body mass index; BMIZ, BMI z-score; WC, waist circumference; BF, body fat; FM, fat mass; FMI, fat mass index; LBMI, lean body mass index; ST, skinfold thickness; OR, odds ratio; RR, relative risk; GW, weeks of gestation; ∑DEHP, sum of di-2-ethylhexyl phthalate. 1 age ( a child age, b maternal age); 2 sex ( a child sex); 3 education level ( a maternal/paternal education, b caregivers education, c mother years of schooling); 4 socioeconomic status ( a household/family income, b poverty to income ratio, c insurance status); 5 physique ( a pre-pregnancy/maternal BMI, b height, c BMI, d gestational weight gain, e maternal height, f prepregnancy obesity); 6 food ( a breast feeding, b total energy/calorie intake, c maternal diet quality score, d alternative healthy eating index, e eating junk food, vegetables or fruit, f sugersweetend beverage consumption, g fatty acid intake, h food security during pregnancy, i soda consumption during pregnancy, j child fast food and sweet consumption at 9 y, k unbalanced diet); 7 smoking ( a child's passive smoking, b smoking during pregnancy, c serum/urinary cotinine); 8 alcohol consumption; 9 activity ( a regular exercise, b regular exercise, c TV/video watching time, d computer use, e sports/activities, f playing video games); 10 race ( a maternal/paternal/child race/ethnicity, b maternal country of origin); 11 information of pregnancy ( a parity, b gestational age, c time of day of urine collection in the 1st and 3rd trimester); 12 maternal marital status; 13 location of participants or study ( a study Centre, b years of USA residence, c residence); 14 past history ( a depression score, b diabetes, c lipid profile, d health status, e depressive symptoms, f systolic blood pressure, g C-reactive protein, fasting plasma glucose, insulin, alanine aminotransferase and gamma-glutamyl transferase, h hypertension, i TC); 15 work status during pregnancy; 16 others ( a urinary creatinine level, b survey year, c maternal/prenatal sum of DEHP, d urinary specific gravity).
[All outcome ranges within the first bracket indicate the 95% CI].
indices (8,63,68,79,81). However, several studies recruited children with ages ranging from 6 to 19 years and investigated the associations of urinary BPA, BPS, and BPF exposure levels with overweight and obesity indices. All these studies found positive associations with one or more anthropometric parameters of obesity (7,34,62,66,82,83,85). Adult exposure levels of BPA, BPS, and BPF were also positively associated with at least one anthropometric index of obesity in adults and elderly individuals (2, 41, 48, 64, 65, 67, 69, 75-77, 84, 86) with the exception of inconsistent associations in one study (63). One panel study (cohort) investigated the association between urinary BPA concentrations and overweight. The authors reported a positive association in the case of overall and female study participants, but not in male participants (74). Some other studies also observed a sex-stratified relationship between prenatal bisphenol exposure and overweight and obesity indices (8,61,72,79). A few studies reported sex-dependent associations between childhood bisphenol exposure levels and obesity or adiposity measures (7,66,81). Race-or ethnicityspecific associations of urinary BPA concentrations with obesity indices were also reported, with a significant association of BPA levels only in non-Hispanic white subjects (83). Pubertal status was reported as a confounder of the associations between BPA concentrations and BMI, WC, and ST, especially in girls (70,79). Maternal exposure levels of 2, 5-dichlorophenol, benzophenone-3, and triclosan showed null associations with %FM in children aged 4-9 years (71). In contrast, one study reported positive associations between urinary 2, 5-dichlorophenol levels in children aged 6-8 years and BMI, WC, and %BF in later childhood, which consistently increased up to 13 years of age (68). BPA levels in urine varied among the studies and ranged from non-detectable to >2594 ng/ml ( Table 1). Children and adolescents (6-19 years) with urinary BPA, BPS, and BPF concentrations of ≥2, ≥1.30, and ≥0.2 ng/ml, are susceptible to developing overweight or obesity (7,34,62,81). In adults, BPA, BPS, and BPF showed obesogenic effects at concentrations ≥0.71, ≥1, and 1 ng/ml, respectively (41,64,67,69,77,86). In addition, BPA concentrations ≥0.39 ng/ml may be responsible for subsequent development of overweight or obesity in elderly people (74).

Environmental Phthalates and Obesity
A total of 32 studies (11 birth cohort, 19 cross-sectional, and 2 case-control studies) explored the association of both prenatal and postnatal urinary exposure levels of phthalate metabolites with overweight and obesity measures in human populations of different ages ( Table 2, and Matrix Tables 2, 3). In the birth cohort studies, urine samples were collected from both the pregnant mother and their children aged 1-14 years. The study subjects ranged from 128 to 1128 mother-child pairs in the birth cohorts. Among the 19 cross-sectional studies, 11 involved children and adolescents, 8 involved only adults and elderly people (male and/or female) of different ages. Almost all the studies determined phthalate metabolites in the spot urine of the study participants.
Associations between maternal 1 st trimester DEHP exposure levels and obesity measures in preschoolers, school-aged children, and adolescents were inconsistent (50,70,87,93). Similarly, the individual or sum of maternal 2 nd trimester urinary DEHP metabolites showed both positive and null associations with different obesity indices in preschoolers, school-aged children, and adolescents (50,87,95,96). However, negative or null associations were found between the 2 nd and 3 rd trimester DEHP exposure levels and anthropometric obesity indices in infants, preschoolers, school-aged children, and adolescents (50,87,(98)(99)(100)(101). Infant (1 year) exposure to DEHP was negatively associated with obesity indices at 8 years of age. In contrast, preschoolers exposed to DEHP (4 and 5 years) were negatively or positively associated (93,96). Associations of DEHP exposure levels at 6-19 years of age (individual metabolite levels or sum of levels) with overweight or obesity indices in school-aged children and adolescents were very inconsistent (89-92, 94, 96, 102, 105-111). Most of the studies that recruited adults and elderly people reported positive or null associations between one or more DEHP metabolites or the sum of DEHP and different overweight and obesity indices in overall adult and elderly populations or after sex stratification (2,48,49,65,88,97,104,106,111,112).
Similar inconsistent associations were also found among non-DEHP metabolites [mono-butyl phthalate (MBP), mono-ethyl phthalate (MEP), mono-methyl phthalate (MMP), mono-benzyl phthalate (MBzP), mono-isobutyl phthalate (MiBP), mono-(carboxylnonyl) phthalate (MCNP), mono-isononyl phthalate (MINP), and others], and obesity indices at different stages of life. First to 3 rd -trimester maternal urinary concentrations of non-DEHP metabolites (except MCPP) displayed null or negative associations with anthropometric parameters of obesity in preschoolers, school-aged children, and adolescents (50,70,93,98,99,101,103). In contrast, one study found positive associations between maternal urinary concentrations of MEP, MBP, MBzP, and MiBP and obesity indices among all study participants (95). Another study also found positive associations after sex-stratified analysis in both males (MBzP) and females (MiBP and MBP) (87). Exposure levels of non-DEHP metabolites in toddlers and preschoolers showed null associations with their obesity measures (96,101). However, one study described positive associations between MEP, MBP, and MBP, and obesity indices in girls, with negative associations in boys (93). Exposure levels of school-aged to adolescents to non-DEHP metabolites (MMP, MEP, MBP, MiBP, and MBzP) were mostly positively associated with one or more anthropometric indices in school-aged children or adolescents (89,91,92,94,96,102,(105)(106)(107)(108)(109)(110). In contrast, after sex stratification, inconsistent associations were evident (94,102,107,110). One study recruited subjects 6-80 years old and found inconsistent associations among non-DEHP metabolite concentrations at different exposures (6-11, 12-19, 20-59 and 60-80 y) and corresponding outcome assessment ages (111). Exposure levels of non-DEHP metabolites in adults and the elderly also showed null or positive associations with their overweight and obesity indices (2,48,49,65,88,97,104,106,112). One study evaluated ethnicity-  • Log 2 -transformed prenatal urinary metabolites of DEP, DBP, BzBP, and DEHP were positively associated with BMIZ, WCZ, and %BF at 5, 7, 9,10.5, and 12 y.    age ( a child age, b maternal age, c age squared); 2 sex ( a child sex); 3 education level ( a maternal/paternal education); 4 socioeconomic status ( a household/family income, b poverty to income ratio, c insurance, d maternal receipt of public assistance); 5 physique ( a pre-pregnancy/maternal BMI, b height, c gestational weight gain, d birth weight); 6 food ( a maternal diet quality score, b total energy/calorie intake, c child`s food insecurity and fast-food consumptions at each point, d food security, prenatal fruit/vegetables and fish consumptions, e prenatal vitamin use, f total fat intake, g breast feeding, h dietary factors); 7 smoking ( a smoking during pregnancy, b serum/urinary cotinine); 8 alcohol consumption/drinking status; 9 activity ( a physical activity, b exercise, c recreational activity, d TV/video watching time, e sedentary hours, f metabolic equivalent hours); 10 race ( a maternal/paternal/child race/ethnicity, b maternal country of origin); 11 information of pregnancy ( a parity, b gestational age, c cesarean section and delivery experience); 12 maternal marital status; 13 location of participants or study ( a site/area, b housing type, c years in US prior to delivery, d cohort); 14 past history ( a lipid profile, b SBP/DBP, c blood sugar, d depressive symptoms, e diabetes) 15 job/occupation ( a work status during pregnancy); 16 reproductive factors ( a tanner stages, b puberty onset, c menopausal status/hormone use); 17 using cosmetics, plastic packaging and bottled drinks; 18 others ( a urinary specific gravity, b urinary creatinine level, c phthalic acid, d prenatal BPA, e urine dilution and collection date, f an interaction term between age and phthalate concentration categories, g an interaction term between age squared and phthalate concentration categories, h an interaction term between race/ethnicity and age).
[In all cases in the outcome, ranges within the first bracket indicate the 95% CI.] dependent association and found that higher maternal urinary concentrations of MCPP heightened the odds of being overweight or obese in Hispanics than in non-Hispanic blacks, although null associations were found with BMI (99). Prepubertal girls showed positive associations between % MEHHP and BMI, WC, and %BF, and showed significant odds increase in the 3 rd and 4 th quartiles compared to the 1 st quartile. The relationship was null in pubertal girls (90).
Data regarding the lowest threshold levels of phthalate metabolites for overweight or obesity outcomes in humans are limited. Low molecular weight phthalate (LMWP) metabolite concentrations ≥0.27 µmol/ml were associated with significantly increased overweight or obesity indices in male children and adolescents (106). Another study reported increased BMI and WC for median urinary MEP concentrations ≥131 and ≥948 µg/g creatinine, respectively (110).

Environmental POPs and Obesity
A total of 41 human epidemiological studies (33 cohort and 8 cross-sectional studies) explored the relationships between in utero and early life exposure to POPs and anthropometric indices of overweight and obesity among infants, children, adults, and elderly populations ( Table 3). The studies assessed POP levels in blood (serum/plasma) or umbilical cord blood (whole blood, serum/plasma).
A total of 8 epidemiological studies (7 cohort and one crosssectional study) investigated the associations of several PBDEs with anthropometric measures of obesity along with other POPs in children, adults, and elderly individuals. Inconsistent associations were documented ( Table 3 and Matrix Table 4). PBDE congeners, including BDE28, BDE47, BDE99, BDE100, BDE153, and BDE154, were mainly associated with obesity indices. In most of the included studies, the BDE153 congener was negatively associated with one or more overweight or obesity indices in children and adults (4,114,117,123,127). All other PBDE congeners (BDE28, BDE47, BDE99, BDE100, BDE154, BDE209, and sum of PBDE), except BDE154, showed null associations with obesity indices (4,103,114,117,127,141,145). One study instead showed a positive association between BDE47 and BMI in adults ≥18 years of age (117). Associations of PBDE congeners with obesity in elderly people aged ≥70 years were null in two separate studies (141,145). Early childhood exposure to PBDEs was negatively associated (BDE153) or inconsistent (others) with obesity indices, especially BMI and WC, at 7 years of age (114,127).
Eleven birth cohort studies investigated the associations between in utero or maternal and childhood exposure to PFAS and obesity indices. Associations between maternal exposure levels of PFAS metabolites and obesity indices were inconsistent ( Table 3 and Matrix Table 5). First-to 2 nd -trimester exposure levels of PFOS and PFOA showed inconsistent associations with the obesity indices of infants and toddlers (51,146). In contrast, almost all studies found positive associations between maternal PFOA concentrations and different obesity indices in overall and/or sex-stratified populations of preschoolers and school-aged children (3,115,121,122,124), with two exceptions: in utero PFOA exposure showed a null association with BMI, WC, and overweight in school-aged children (120,137). In contrast, PFOS and other PFAS levels were inconsistently associated with anthropometric measures of overweight and obesity in preschoolers and school-aged children (113). Prenatal exposure levels of PFOA were positively associated only in adult females, but the associations were null for all PFASs when considering the overall population (138). However, exposure levels of PFAS in preschoolers and school-aged children mostly showed null or negative associations with overweight or obesity indices (3,120,121). Gestational exposure levels of PFOA ≥4.3-6.4 ng/ml were associated with increased WC in the children 2-8 years of age (122). In contrast, 1 st and 2 nd trimester exposure levels of PFOA (0.5-≤7.10 ng/ml in boys and 1.10 to ≤6.70 ng/ml in girls) showed null associations with BMI or overweight at 7 years of age (137).
Positive, negative, or null associations have also been reported between in utero or prenatal and postnatal, and between early childhood to the elderly concerning exposure to OCs and overweight and obesity indices ( Table 3 and Matrix Table 6) (103, 115-119, 121, 125, 126, 128-136, 139-145, 147-149). Maternal 1 st to 3 rd trimester blood and/or umbilical cord blood levels of OC metabolites, especially DDE and HCB levels, were positively associated with different anthropometric indices of obesity, whereas associations of PCBs, DDT metabolites, and b-HCH concentrations were null or positive in toddlers and preschoolers (115,116,126,132,142,148). Inconsistent associations (positive and null) were also found between PCBs, DDT metabolites, DDE, HCB, and b-HCH levels in the 1 st to 3 rd trimester maternal blood or umbilical cord blood and obesity indices in school-aged children (103, 118, 130, 133-136, 139, 149). One study found positive associations of 2-week postpartum HCB levels, but not other OCs, with anthropometric indices in 18month-old and 5-year-old children (121). Among the OCs, DDT and its metabolite DDE showed potent positive associations with obesity indices in the overall population or in school-aged boys and girls (103,118,130,134,135). Only one study investigated the relationship between prenatal exposure levels of DDE and adult obesity measures, and subsequently addressed the positive associations of adults aged 20-50 years. PCBs showed null associations in the same study participants (147). Again, associations sometimes varied among the countries within the study. A prospective cohort study of 412 Norwegian and Swedish mother-child pairs observed a non-monotonic dose-response relationship between PCB-153 concentrations and child overweight/obesity among Swedish children at 5 years of age, but not in Norwegian children (115). Exposure levels of PCB153 and DDE metabolites in infants were not associated with obesity measures in preschool and school-aged children (133). Early childhood or preschooler exposure levels of HCB, DDE, and PCBs were negatively associated with anthropometric parameters in preschoolers (121). School-aged exposure levels of PCBs, DDE, and HCB showed inconsistent associations with obesity indices in school-aged children, adolescents, and adults (125,129). Exposure levels of DDE and b-HCH in adults (≥18 years) showed positive associations, PCBs showed inconsistent associations (positive and negative), and other OCs also showed null associations with different anthropometric indices of overweight and obesity (117,119,128,131,143,144). OC exposure in elderly people also showed       contradictory findings. DDE exposure levels showed positive or null associations, whereas PCBs showed very inconsistent associations (positive, negative, and null) with anthropometric indices in elderly people aged 50 to 75 years (140,141,145). Furthermore, cord blood HCB levels >1.03 ng/ml were associated with increased BMI in children at 6.5 years of age (149).

Controversies and Elucidation
We present evidence of the relationship between urinary/blood levels of selected EOs and their metabolites or congeners, and anthropometric overweight and obesity indices. These relationships are contentious. Prenatal or in utero, newborn, and early childhood to elder life exposure to selected EOs might contribute to the development of adiposity at different stages of life, although the findings were inconsistent depending on exposure and outcome assessment periods. Some studies have clarified positive associations, whereas other studies described negative or null associations for the same EO exposure levels and the subsequent anthropometric indices of obesity (Tables 1-3).
A representative example is two separate birth cohort studies from China and the United States (8,72) with almost the same number of children (430 and 408). The studies indicated contradictory associations of BPA concentrations at age 3 years with anthropometric obesity indices at age 7 years. The study from China found positive associations, whereas the US study age ( a maternal age/maternal age at delivery, b child age, c offspring age at follow up); 2 sex ( a child sex); 3 education level ( a maternal/paternal education); 4 socioeconomic status ( a household/ family income); 5 physique ( a pre-pregnancy/maternal BMI, b gestational weight gain, c birth weight, d baseline BMI/obesity, e paternal BMI, f parental overweight/obesity, g rapid growth status, h height/maternal height, i lean mass, j pre-pregnancy obesity); 6 food ( a prenatal vitamin use, b breast feeding/breast fed, c duration of breastfeeding, d maternal fish intake during pregnancy, e maternal diet, f child diet, g total energy/calorie intake); 7 smoking ( a active/passive smoking during pregnancy, b paternal smoking during pregnancy, c maternal serum cotinine, d cigarette smoking); 8 alcohol consumption/drinking status; 9 activity ( a time playing outside, b TV/video watching time, c time playing video games, d child physical activity/physical activity, e regular exercise); 10 race ( a maternal/paternal/child race/ethnicity, b maternal country of origin/birth); 11 information of pregnancy ( a parity/interpregnancy interval, b type of delivery, c child gestational age, d child birth order, e number of live births, f numbers of offspring pregnancies); 12 maternal marital status; 13 location of participants or study ( a years of USA residence/time in the US at birth, b history of lactation, c study center/study sub-cohort); 14 past history ( a depressive symptoms, b lipid profile, c HIV status) 15 occupation/job ( a work status during pregnancy, b social class); 16 reproductive factors; 17 others ( a gestational age at blood drawing, b visit, and interaction between visit and PFAS, c time of blood draw, d prenatal PCBs).
[In all cases in the outcome, ranges within the first bracket indicate the 95% CI].
found null associations, despite the same exposures and outcome ages (8,72). Many other studies have also reported contradictory findings among the same exposure and outcome age groups (Tables 1-3). In contrast, some studies conducted in different countries recruiting different populations reported similar associations between the same or different EO and obesity outcomes (7,34,62,82,83,89,92). These conflicting findings across studies might be explained by methodological variations, particularly the characteristics of the study populations. Other potential reasons are exposure levels (low, medium, or high) and the timing and duration of EO exposure. Associations seem to differ between boys and girls, adult males, and females (Tables 1-3). Some studies reported ethnicity-specific associations between EO exposure and obesity indices (83,99,112). The reasons for racial and ethnic differences in overweight and obesity are largely unknown. Possible reasons might be the different patterns of calorie intake or energy consumption, physical activity, metabolic activity, endocrine, and genetic susceptibility among racial and ethnic groups (150,151).
Among the environmental phenols, BPA has been widely investigated and has been positively associated with anthropometric overweight and obesity indices, mainly in school-aged children, adolescents, and adults. The use of BPA has been decreasing to reduce its negative health impact. This has led to increased use of BPS and BPF. Several studies investigated the association of BPS and BPF with obesity measures and described inconsistent relationships (7,34,48,50,67). These few studies might be insufficient to conclusively determine the reason for the contradictory associations. Both DEHP and non-DEHP metabolites showed inconsistent associations with overweight and obesity indices at different stages of life. Among the non-DEHP metabolites, MEP, MMP, and MBzP seem to have obesogenic roles in adult and elderly humans. Among the POPs, DDE and PFOA showed almost consistent positive associations with obesity. PFOS also seems to be positively associated with obesity measures, but the associations were sometimes inconsistent. Compared with DDE, DDT showed a weaker association with obesity indices. Although DDT and DDE have already been banned in many countries, the long half-lives of these EOs (7 and 10 years for DDT and DDE, respectively) in both the environment and humans might be responsible for the adverse effects (152)(153)(154)(155)(156)(157). Similarly, PFAS metabolites are also very persistent in the environment (half-lives of 3-10 years) and humans are exposed through ingestion of contaminated food, drinking water, and ingestion or inhalation of PFAS from contaminated dust and soil, and even via transplacental and breast milk passage from mother to child (158)(159)(160)(161)(162)(163).
Usually, a single EO or a group of similar EOs was included in previous studies, making the results straightforward and easily interpretable. The rising concern is that generalized linear regression can provide a simple relationship between a single chemical or a group of similar chemicals and outcomes, but cannot explore the joint effect of mixed exposure (48). In addition, to study causality, researchers need to consider mixed environmental exposures and their complex nonlinear interactions. Eventually, ignoring the joint effects of other chemicals could contribute to false-positive or false-negative results (164). We found only a limited number of studies that investigated the associations between cumulative exposure to EOs and overweight and obesity indices using a multipollutant approach. Findings were inconsistent (48,103,116,117). In one study, the associations of phthalate metabolites and bisphenols with obesity indices varied when considering single and cumulative exposure levels using three different statistical models (48). Thus, the application of a multipollutant statistical model to clarify the joint effects of mixed EOs should be accepted and utilized to explore the effect of a cumulative exposure burden on the outcomes in one direction per occasion, and the exposure-response function of each chemical, while controlling other chemicals at certain levels.
Some EOs (e.g., bisphenols and phthalates) are lipophilic. They can accumulate in adipose tissue of obese women and can influence the development of obesity in their offspring. A recent population-based prospective cohort of 1396 mothers showed that women in highest group of pre-pregnancy BMI (>30kg/m 2 ) had significant higher concentrations of BPS [OR=0. 15 (165). Another study found that prenatal exposure to PCBs (>1.95 µg/g lipid) was associated with increased BMI in girls from overweight mothers, but not in normal-weight mothers (130). Thus, pre-pregnancy BMI is an important confounder that must be considered when investigating obesity outcomes in a birth cohort. Three studies considered pre-pregnancy BMI as a confounder in birth cohorts (103,132,142). Adjustment of pre-pregnancy BMI might shed light on the relationship between EO exposure and obesity.
Daily consumable items (diet or foods and personal care products) are an important route of exposure to several EOs and are intrinsically related to energy balance. BPA and phthalate exposures occur primarily through ingestion and dermal absorption, as these compounds are found in common consumer goods, such as food containers, children's toys, and personal care products (166)(167)(168)(169). Thus, it can be predicted that those who consume or use more of these products are more likely to have higher exposure levels and, perhaps, are more likely to be obese. Several studies reported a direct link between dietary exposure to EOs and obesity (41,83,85). Most of the studies in the current review considered diet, calorie intake, energy consumption, and physical activity as potential confounders to address the relationships that strengthen the findings (Tables 1-3).
Puberty features hormonal transition. Both girls and boys undergo physical changes. Puberty has been associated with the development of obesity (170). Several studies evaluated the relationship between EOs and anthropometric measures of obesity in an age-and sex-specific manner before and after puberty (7,61,66,72,73,78,79,81,96). A sex-stratified analyses found that increased exposure to urinary concentrations of BPA was positively associated with the sum of skinfold thickness (ST) in girls, while exposure to MEHP, MEHHP, MECPP, and MEOHP were inversely related to BMI z-score, WC, and the sum of ST in boys (70). However, when the analyze was restricted to children who had not yet begun the pubertal transition, the results shifted and showed positive relationships between BPA in girls and MEHP in boys with the sum of ST. In the prenatal exposure period, the authors observed an inverse relationship between MBzP and a child's BMI z-score, but this finding did not persist when the analyses were restricted to children prior to puberty. In a case-control study, prepubertal girls showed positive associations between %MEHHP and BMI, WC, and % BF, and showed significantly increased odds in the 3 rd and 4 th quartiles compare to the 1 st quartile, whereas the relationship was null in pubertal girls (90). How the associations differ before and after puberty is not yet clearly understood. Knowledge of hormone levels related to pubertal growth, including thyroid hormones, leptin, adiponectin, and others, might provide more insight into the potential mechanisms of EO-mediated s adiposity (171).

Research Gap
One of the limitations of the birth cohort studies outlined here is the use of single spot urine during the 1 st , 2 nd, or 3 rd trimester to estimate EO exposure. The biological half-lives of some of these chemicals are short and they are quickly excreted in urine (e.g., phthalates and bisphenols). Epidemiologists ideally prefer to use 24 h urine and repeated urine sampling when assessing these chemicals in relation to obesity, which occurs incrementally over time and has a multifactorial etiology (172,173). The time of the day or season could account for some intrapersonal or interpersonal variations in urinary concentrations of analytes in single spot urine samples (174)(175)(176)(177). However, single spot is the conventional test, despite these methodological limitations. In most of the included studies, biomonitoring EOs were done using methods lacking validated external quality assurance. Maintaining internal and external quality control and quality assurance might make study findings comparable and could strengthen the findings. Some studies had very limited information on pre-pregnancy BMI due to the availability of self-reported weight and the timing of recruitment in their original birth cohorts. These studies relied on maternal BMI. However, most of the studies collected data using self-reported questionnaires or home visits. Therefore, under-or over-estimated data could not be avoided. There was little or no data of phthalate metabolite (both DEHP and non-DEHP) levels and subsequent obesity assessment in infants and toddlers (Matrix Tables 2, 3). In addition, the assessment of obesity in adolescents related to OC (DDT, DDE, b-HCH, HCB, and PCBs) exposure was insufficient (Matrix Table 6). Very few studies investigated the relationship between cumulative EO exposure levels and overweight or obesity indices. Therefore, the possibility that prenatal and/or postnatal exposure to other unmeasured chemicals correlated with measured chemicals may have confounded the associations under study cannot be excluded. Finally, there are scant data concerning the trajectories of exposure and outcome assessments.

Future Contemplations and Research Design
Environmental epidemiologists should clearly infer whether exposure to ECs might influence weight gain or obesity, or whether obese study participants might have greater exposure to, or excretion of, ECs by conducting long-term follow-up studies in child and adult populations. Further prospective studies should aim to collect data with repeated measures over extended periods to improve exposure classification, increase general understanding of the timing of exposure, and address the temporal relationship between ECs and obesity. Given the gradual decrease of some ECs and increase exposure to some alternate ECs in human populations, continued biomonitoring of these alternate ECs and further investigations on their obesogenic effects in humans could be undertaken. Researchers should target study participants at all stages of life to assess exposure and obesity outcomes at each age. Many other chemicals, including pesticides, heavy metals, and particulate matter, have been reportedly associated with obesity outcomes in vitro and in vivo in animal studies. However, their obesogenic effects have not yet been completely evaluated in humans (178)(179)(180)(181)(182)(183)(184)(185)(186). Thus, an exposome-based approach needs to be developed to investigate the possible obesogenic effects of chemicals, xenobiotics, and pollutants in humans to explore the overall scenario of cumulative exposure. Studies of the obesogenic effects of ECs in the context of diet, stress, ethnicity, gender, and other factors, using sophisticated statistical models to assess complex exposures should be done.

CONCLUSIONS
The collective data indicate that BPA, DDE, and PFOA have consistent obesogenic effects in humans. Other bisphenols, phthalates, and POP metabolites or congeners have contradictory relationships with obesity at different outcome assessment times. Further prospective cohort studies with cumulative exposure assessments are required. The findings of this review will increase the awareness of the obesogenic effects of ECs among the general population.

AUTHOR CONTRIBUTIONS
NCM: investigation, data curation, analysis, and writing-original draft. SK: data curation and writing-review and editing. YI and MK: conceptualization, project administration, funding acquisition, supervision, and writing-review and editing. All authors contributed to the article and approved the submitted version.

ACKNOWLEDGMENTS
The first author is grateful to the Department of Biochemistry and Molecular Biology, Shahjalal University of Science and Technology, Sylhet, Bangladesh, for approval of his study leave.