Prenatal exposure to phthalate and decreased body mass index of children: a systematic review and meta-analysis

Phthalates are well-known endocrine-disrupting chemicals. Many detrimental health effects of phthalates were investigated, but studies on the association of phthalates with obesity in children showed inconsistent results. Thus, this systematic review and meta-analysis were performed to clarify whether prenatal and postnatal exposures to phthalates are associated with physical growth disturbances in children. We performed the systematic review and meta-analysis following the PRISMA 2020 statement guidelines, and found 39 studies that met our inclusion criteria, including 22 longitudinal and 17 cross-sectional studies. We observed a significant negative association between the prenatal exposure to DEHP and the body mass index (BMI) z-score of the offspring (β = − 0.05; 95% CI: − 0.10, − 0.001) in the meta-analysis, while no significant association between the prenatal exposure to DEHP and the body fat percentage of the offspring was observed (β = 0.01; 95% CI: − 0.41, 0.44). In the systematic review, studies on the association between phthalates exposure in childhood and obesity were inconsistent. Prenatal exposure to phthalates was found to be associated with decreased BMI z-score in children, but not associated with body fat percentage. Our findings suggest that phthalates disturb the normal muscle growth of children, rather than induce obesity, as previous studies have hypothesized.


Results
shows the process to include the relevant studies for the systematic review. We screened 1561 records, and excluded 1425 studies based on their titles. After the abstracts were reviewed, 74 irrelevant studies were excluded. The full texts of 62 studies were assessed, and we found 39 studies that met our inclusion criteria. The reference lists of the included studies were manually checked, and no additional studies were searched in this step. Table 1 summarizes the a total of 39 observational studies. The study size varied between 72 and 2,884 participants. Studies were conducted in the United States (n = 12), China (n = 6), South Korea (n = 4), Taiwan (n = 3), Spain (n = 2), Netherlands (n = 1), Australia (n = 1), France (n = 1), Greece (n = 1), Italy (n = 1), Germany (n = 1), Canada (n = 1), Sweden (n = 1), Thailand (n = 1), Iran (n = 1), Mexico (n = 1), and in multiple European countries (n = 1). Supplementary Table S2 shows reasons for exclusion in full-text review. The quality of the studies as assessed by the NOS is presented in Supplementary Table S3-4. The scores of the included longitudinal studies (n = 22) ranged from 8 to 9, and all were classified as good quality. Cross-sectional studies (n = 17) ranged from 5 to 8 and included 12 high-quality studies and 5 moderate-quality studies. Table S5 describes the studies investigating the association between prenatal exposure to phthalates and BMI. Among the 39 studies, 17 studies investigated the association between prenatal exposure to phthalates and BMI. The statistical significance of associations between phthalate metabolites and BMI of children is summarized in Supplementary Table S6. Figure 2 shows the results of a meta-analysis on the association between prenatal DEHP exposure and BMI z-score in children. Ten studies presented eligible results for the meta-analysis. Data from Agay-Shay et al. were not included because they were derived from the same study population (INMA cohort) as that from Shoaff et al. K. Data from Berger et al. were not included because they presented an unadjusted beta coefficient from the Bayesian hierarchical model, and they were derived from the same study population (CHAMACOS cohort) with Harley et al. Heterogeneity among these studies was also not significant (P = 0.338). In the random effect model, there was a significant negative association between prenatal DEHP exposure and BMI z-score index (β = − 0.05; 95% CI: − 0.10, − 0.001). Visual inspection of the funnel plot revealed no asymmetry ( Supplementary  Fig. S1), and the Egger test showed no publication bias (P = 0.542). Figure 3 shows the meta-analysis results on the association between prenatal DBP exposure and BMI z-scores in children. Seven studies presented data on BMI z-scores, and these were selected for the meta-analysis. Heterogeneity among these studies was suspected, but it was not statistically significant (P = 0.161). In the randomeffects model, there was no significant association between prenatal DBP exposure and BMI z-score (β = − 0.02; 95% CI: − 0.10, 0.06). Visual inspection of the funnel plot revealed no asymmetry ( Supplementary Fig. S2), and the Egger test showed no publication bias (P = 0.271).

Prenatal exposure to phthalates and BMI z-scores. Supplementary
Prenatal exposure to phthalates and body fat percentage. Supplementary Table S7 describes studies that investigated the association between prenatal exposure to phthalates and body fat percentage. Among the 39 studies, seven were included. The results for this association were inconsistent, and only a limited number   Figure 4 shows the results of the meta-analysis for the association between prenatal DEHP exposure and body fat percentage. Six studies presented body fat percentage data, which were chosen for the meta-analysis. Heterogeneity among these studies was not found (P = 0.358). In the random-effect model, no significant associations between prenatal DEHP exposure and body fat percentage were found (β = 0.01; 95% CI: − 0.41, 0.44). In addition, visual inspection of the funnel plot revealed no asymmetry ( Supplementary Fig. S3), and the Egger test showed no publication bias (P = 0.287). www.nature.com/scientificreports/ Figure 5 shows the results of a meta-analysis on the association between prenatal DBP exposure and body fat percentage. Five studies presented data regarding body fat percentage, and these were selected for the metaanalysis. Heterogeneity among these studies was not found (P = 0.184), and there were no significant associations between prenatal DBP exposure and body fat percentage (β = − 0.42; 95% CI: − 1.04, 0.19). Furthermore, visual inspection of the funnel plot revealed no asymmetry ( Supplementary Fig. S4), and the Egger test showed no publication bias (P = 0.601). Table S9 describes studies evaluating the association between the prenatal exposure to phthalates and body composition indices other than BMI or body fat percentage. Berman et al. assessed height, BMI, and body composition as measured by dual-energy X-ray absorptiometry (total fat percentage, total fat mass, and total lean mass), and the reported MiNP and MEHP were associated with decreased total lean mass 25 . Meanwhile, Buckley et al. used overweight/obesity defined by BMI z-score as the outcome variable 27 . Lee et al. also reported the association between phthalate metabolites and BMI z-score, fat mass percentage, fat mass index (FMI), skeletal muscle mass percentage, and skeletal muscle index (SMI) and reported that high levels of prenatal exposure to phthalates were significantly associated with decreased SMI among girls 35 . Maresca et al. reported that prenatal non-DEHP phthalate exposure was associated with lower BMI z-score, WC, and fat mass in boys during early childhood, contrary to their hypothesis 36 . Valvi et al. reported that weight gain Z-score was significantly associated with prenatal exposure to DEHP among boys 43 . Nidens et al. investigated the association between phthalate metabolites in prenatal maternal urine and weight gain (%) first 2 years of life, but it was not significant 54 . Table S10 summarizes the studies assessing the association between the postnatal exposure to phthalates and the BMI. The results of the included studies were inconsistent, and there were limited studies that reported the association of BMI with phthalate metabolites as continuous variables. Chang et al. studied 152 children in Taiwan and reported non-significant associations of BMI with DEHP metabolites, MnBP, and MiBP 29 . Shaoff et al. also analyzed the data of 219 children from the HOME study and reported associations between BMI z-score at 8 years and DEHP metabolites at prenatal, 1, 2, 3, 4, 5, and 8 years of age were not statistically significant 39 . The only significant association was between a ten-fold increase in DEHP metabolites at 5 years of age and a 0.04-unit increase in BMI z-score. Trasande et al. reported that a unit increase in the natural log-transformed sum of LMWP was associated with a 0.07-unit increase in BMI z-score using the data of children surveyed at Estimates were standardized as β and 95% confidence intervals as one unit increase of natural log of DEHP metabolites.  Table S12 shows the studies on the association between the postnatal exposure to phthalates and the BMI. Chang et al. cross-sectionally studied 132 children and reported no association between phthalate metabolites and body fat percentage 29 , and Hou et al. studied 308 Taiwanese children and reported a significant association between the MnBP and MiBP and the waist-to-hip ratio 32 . Shaoff et al. analyzed the data of 219 children from the HOME study and showed significant associations between the waist circumference at 8 years of age and the sum of DEHP metabolites at 5 years of age 39 . There were significant associations between the body fat percentage at 8 years of age and the sum of DEHP metabolites at 1 and 5 years of age. In China, a case-control study on 57 boys with constitutional delay of growth and puberty and 110 controls reported that higher urinary phthalate metabolites were associated with constitutional delay of growth and puberty 47 . Another cohort study with 100 children reported non-significant associations between DEHP and DBP metabolites at 4 years of age and body indices until 24, including waist circumference, body fat percentage, and trunk fat percentage) 48   www.nature.com/scientificreports/

Discussion
Main findings of the study. The systematic review and meta-analysis were performed to investigate the association between phthalates and physical growth in children. In the systematic literature review, a significant and negative association was found between the prenatal exposure to DEHP and the BMI z-score of the offspring, but there was no significant association between the prenatal exposure to DEHP and DBP and the body fat mass percentage of the offspring. Additionally, previous studies on the association between phthalates exposure in childhood and obesity were inconsistent in the systematic review.
Prenatal exposure to phthalates and growth disturbance. We found that prenatal phthalate exposure and decreased offspring's BMI were significantly associated. It implies that phthalates could act as disrupting chemicals on normal development instead of obesogens. Previous researches have focused on obesity, and found inconsistent results. Among children aged 5-12 years in the U.S., prenatal exposures to DEHP and DBP were associated with increased obesity 37 . However, Vafeiadi et al. investigated five-hundred mother-child dyads, and found that prenatal phthalate exposure was not significantly associated with overweight at ages 4-6 years 42 . Buckley et al. studied 707 children in the U.S. and found that BMI z-scores in girls aged 4-7 years were negatively associated with prenatal exposure to DEHP 27 . These inconsistent results lead to the idea that phthalates could not be obesogen. Our recent study suggested the selective association of phthalate exposure with the development of muscle mass than fat mass could explain the inconsistent associations between prenatal exposure to phthalates and BMI in children 35 . A cross-sectional study in the U.S. also showed that an increased urinary concentration of phthalate metabolites is associated with decreased lean mass 61 . If phthalate exposure could disturb the growth of muscle mass rather than induce obesity, it could explain the inconsistencies reported in previous studies regarding the association between prenatal exposure to phthalates and BMI during childhood.

Possible mechanism.
In the meta-analysis, prenatal exposure to phthalates was significantly associated with decreased BMI z-score but not with FMI. A possible explanation for this association is the antiandrogenic effects of phthalates on muscle development 5,62 . A murine study reported that androgen withdrawal mice showed decreased myofibrillar protein synthesis, and anabolic steroid administration reversed the effect 63 .
Another study using mice also reported that testosterone had positive effects on muscle mass and the ultras- www.nature.com/scientificreports/ tructure of muscles 64 . In an animal study, prenatal DEHP exposure led to decreased testosterone production in the offspring both in the fetal and postnatal period 65 . Several epidemiologic studies support that androgen is associated with muscle growth. A study with 50 boys and 50 girls aged 8-17 years reported that muscle strength was positively associated with testosterone levels 66 . Another study reported that testosterone is related with muscle mass and strength with a dose-response manner among hysterectomized women 67 . Furthermore, prenatal phthalate exposure is associated with decreased anogenital distance, which is positively related with antiandrogenic properties 68,69 . Increased phthalate metabolites were associated with decreased levels of serum testosterone in another human study 70 . Among children, the positive association between serum testosterone and SMI has been investigated 71 . Therefore, the antiandrogenic properties of phthalates could be an important link between prenatal exposure to phthalates and decreased SMI. Inflammation is a possible mediator of disruption of muscle development following phthalate exposure. Phthalates exacerbate inflammatory response by increasing inflammatory cytokines 72 . A human study reported that DEHP exposure could induce interleukin-1β production in neonatal neutrophils 73 . In vitro study also reported that increased gene expression of inflammatory cytokines could be induced by DEHP 74 . Inflammatory cytokines are also associated with the inhibition of expression of myogenic miRNA in myoblasts and promoting muscle protein degradation 75,76 . Therefore, it could be inferred that inflammation due to phthalates could be associated with decreased SMIs.
Insulin-like growth factor-1 (IGF-1) could be another possible link of the association of phthalates with muscle mass. IGF-1 pathway is an important regulator of muscle growth processes in children 77 . Several epidemiologic studies have reported that urinary phthalate metabolites are negatively associated with IGF-1 41,[78][79][80] . These studies support that phthalates could lead to decreased muscle growth in children via IGF-1.

Phthalates exposure in children and body composition indices.
The results of searched studies in the systematic review were inconsistent for the associations between the phthalates exposure in children and their body composition. Several researchers reported that phthalate exposure in children could be related with obesity, although obesity was inconsistently associated with phthalate metabolites, and the number of studies was limited to perform the meta-analysis. As one of the results with a significant association, a cross-sectional study involving 845 Danish children aged 4-9 years reported that children's height and weight are negatively www.nature.com/scientificreports/ associated with urinary phthalate metabolites 79 . However, several studies showed a positive association between phthalates and obesity. A study that used NHANES data reported that LMWP could be associated with increased BMI z-score 13 , and a longitudinal study in the U.S. also noted that obesity at 8 years of age was associated with phthalate exposure at 5 years of age 39 . These studies suggested that the role of peroxisome proliferator activated receptors (PPARs), nuclear hormone receptors that have regulatory roles in adipogenesis and lipid storage, is important to induce adipogenesis and obesity [81][82][83] . Because phthalate exposure is associated with decreased thyroid hormone 84 , hormonal homeostasis can be disturbed due to phthalates, leading to fat accumulation and obesity. A Chinese metabolome study investigated 69 overweight/obese children and 80 normal-weight children. It was reported that urinary MnBP concentration differed between the two groups and was associated with arginine, proline, and butyraldehyde 46 . However, several studies had no significant associations between phthalates, obesity, and BMI 13,28,29,38,39,44 . Some researchers argued that the association between urinary phthalates metabolites and obesity was not derived from the causal association between phthalates exposure in children and obesity. For instance, the recent study that explained the mechanism for cross-sectional studies for the association between phthalates and higher BMI demonstrated that the higher energy intake in the overweight and obese could result in the concomitant higher phthalates exposure 85 . Additionally, ultra-processed food consumption is associated with overweight and weight gain 86 , and is also associated with urinary phthalates metabolites 87 . Therefore, cross-sectionally observed association between phthalates metabolites and obesity might reflect the association of the dietary pattern and the amount of consumption with obesity. Additionally, urinary phthalate metabolite may be measured higher among children with more adipose and muscle mass. Given the absorption, distribution, metabolism, and excretion of phthalates, absorbed phthalates in the human body distribute mainly in the intestine and liver, and they are rapidly excreted.
On the other hand, a relatively small portion of absorbed phthalates is distributed in fat and muscle tissue. Still, they are excreted slower than those in the intestine and liver, resulting in a relatively higher proportion of phthalates in the human body 88 . Therefore, observed cross-sectional associations between phthalates and obesity in children might not be causal. Inconsistent results and related factors make it difficult to conclude the association between phthalates exposure in childhood and weight gain. Studies with longitudinal design and studies suggesting plausible mechanism, such as hormonal, epigenetic and/or metabolomic changes, are needed in the future.
Exposure assessment for phthalates. It has been assumed that a single measure of phthalate metabolites can adequately reflect exposure across the studies. All studies included in the meta-analysis also had the same assumption. Assessing DEHP exposures may be inconclusive because various metabolites of DEHP are rapidly metabolized in vivo and quickly excreted. As the excretion half-lives of DEHP metabolites are 0.5-3.0 days 20 , urine biomarkers can only reflect recent exposure. However, all studies included in the meta-analysis considered DEHP metabolites in the mothers' or children's urine. In all longitudinal studies, DEHP metabolites were assessed at a one-time point, rather than repeated measurements in a few-day interval. Included crosssectional studies were also measured DEHP metabolites only once from children's urine.
The temporal stability of DEHP metabolites over weeks to months has been studied. The daily variation of phthalates' urinary metabolites was investigated using urine samples of fifty participants on eight consecutive days, and reported intra-class coefficients of urinary DEHP metabolites as 0.20-0.34 89 . Another study reported one spot urine sample could predict the three-month average concentration of DEHP metabolites with sensitivity and specificity as 0.56 and 0.83, respectively 90 . It has been suggested that DEHP metabolites measured in the spot urine showed reasonable temporal stability for weeks to months, although it has limitations on stability [91][92][93][94][95] . In addition, a recent study investigated 805 urine samples of 16 volunteers for 6 months and suggested that adequately classifying the exposure level of participants requires several samples per subject 96 . In this systematic review, no studies measured phthalates repeatedly in a short time period to measure phthalates exposure more accurately. Therefore, all studies included in our meta-analyses assumed implicitly or explicitly that a single measurement could reflect exposure over a considerable period.

Strengths and limitations.
To overcome the inconclusive results on the association between phthalate exposure and children's growth 17 , we rationally and preferentially selected the estimate for the association between phthalates (exposure) and body composition indices (outcome). We used the sum of phthalate metabolites to assess the total exposure amount because the molar sum of several metabolites of DEHP is currently considered the best estimate of exposure rather than a simple mass sum of DEHP metabolites. Furthermore, the time points of measurement for phthalate exposure (prenatal or postnatal) and the methods for assessing body composition indices, including BMI, BMI z-score, and body fat percentage, differ across studies. In the present study, we attempted to collectively analyze the results in a meta-analysis with the abovementioned methods, which was also described in a previous meta-analysis study 97 . Therefore, we estimated the up-to-date summarized results for the association of prenatal phthalates exposure and body composition indices in children.
This study has several limitations. First, the calibration of the amount of exposure to phthalates considering the duration of exposure is not assessed in the systematic review and meta-analysis, because it is practically impossible. Second, the included studies had limited information and had methodological differences 98 , although standardized values from β estimates and 95% CIs were used to perform meta-analyses. If raw data can be obtained and pooled analysis is performed, more robust results may be expected. In the studies we reviewed, phthalate metabolites were measured from spot urine samples of participants. There is no study with repetitive measurement for accurate phthalates measurement for the association between phthalates and body composition www.nature.com/scientificreports/ indices. In the future, more repetitive methods such as using mean levels of various phthalate metabolites assessed at multiple time points could increase the precision and accuracy of predicting phthalate exposure 99 .

Conclusion
This systematic review and meta-analysis showed that prenatal exposure to phthalates is significantly associated with low BMI in children, but not with body fat mass. In addition, prenatal phthalate exposure may affect the disturbance of normal growth of children rather than act as an obesogen. Future studies on the health effects of phthalates should consider their detrimental effects on the expected growth of children. Furthermore, it is necessary to administer stricter and broader regulations on phthalates in living environments.