Effect of Prenatal Phthalate Exposure on Childhood Atopic Dermatitis: a Systematic Review and a Meta-analysis of Birth Cohort Studies

Background: The association between prenatal exposure to phthalate and childhood atopic dermatitis (AD) has been previously investigated; however, the results are inconsistent. We aimed to perform a systematic review and meta-analysis of birth cohort studies to investigate whether prenatal exposure to phthalate increases the risk of developing AD in children. Methods: We performed an electronic search of the MEDLINE, Embase, and Cochrane library databases. Studies were critically appraised, and meta-analyses were performed. Results: Among 129 articles identied, 11 studies met the eligibility criteria. Included studies originated from Europe (n = 5), USA (n = 4), and Asia (n = 2). The study sample size ranged from 147 to 1024 mother-child pairs. Quality assessment using the Newcastle–Ottawa scale of all studies had scores of six or greater. A meta-analysis of data from eight selected studies suggested that monobenzyl phthalate (MBzP) exposure was signicantly associated with the risk of AD development (odds ratio 1.16, 95% condence interval 1.04-1.31, I 2 = 17.36%). However, AD development was not associated with other phthalate metabolites such as mono-(2-ethylhexyl) phthalate, monoethyl phthalate, mono-isobutyl phthalate, mono-n-butyl phthalate, and the sum of di-[2-ethylhexyl] phthalate on the development of AD (all P-values > 0.05). Conclusions: Our study suggests a positive association between prenatal exposure to MBzP and the development of childhood AD.


Introduction
Atopic dermatitis (AD) is a common childhood chronic in ammatory skin disease with a prevalence of 15-20% in children. [1,2] AD is associated with multiple comorbidities, such as asthma and allergic rhinitis, causes sleep disturbance, and results in an impaired quality of life. [3][4][5] The pathogenesis of AD is heterogeneous and multifactorial. Recent studies have demonstrated that immune dysregulation, microbial dysbiosis, and epidermal barrier defects contribute to the pathophysiology of AD. [6] In addition to genetic predisposition, environmental factors or interactions between them are known to affect the development of AD. [7] Phthalates are diesters of phthalic acid (1,2-benzenedicarboxylic acid) and are commonly used as stabilizers and plasticizers in personal care products, plastics, toys, hair sprays, shampoos, food packing, home furnishings, and building materials. [8,9] Because of their widespread use and property to leach, human exposure to phthalates arises mainly from ingestion, inhalation, and skin absorption. [10,11] Several studies have shown that higher concentrations of urinary metabolites were found in women than in men [12], and pregnant women as well as fetuses, could be highly sensitive to the potential harmful effects of toxic metabolites of phthalates. [13,14] Phthalates can cross the placental barrier and have been measured in amniotic uid in studies on humans. [15,16] Furthermore, phthalates have potential immunomodulatory properties in animal models. [17][18][19] Because the prenatal period is critical in the development of the immune system, several studies have been performed to demonstrate that the exposure to phthalates during pregnancy is associated with the development AD in childhood. [20,21] However, epidemiological data on the development of AD in relation to phthalate exposure during the fetal period remain inconsistent. Therefore, we aimed to identify the association between prenatal phthalate exposure and childhood AD by performing a systematic review with a meta-analysis. Two reviewers (YMP and MJ) independently and in duplicate, screened titles and abstracts using a standardized protocol. Disagreements were resolved by discussion with a third researcher (KA). Titles and abstracts were initially reviewed, and then, the nal selection was based on the full text according to the following inclusion criteria: (1) birth cohort studies, (2) studies in which prenatal phthalate exposure was measured, and (3) studies providing the incidence of AD data. Exclusion criteria were (1) animal or laboratory studies, (2) non-English studies, (3) studies not presenting original data (conference, review articles, editorials, guidelines, and reports), and (4) studies measuring phthalate not from humans. The primary outcome was the incidence of AD in children. A protocol of this study is registered online at Prospero (registration number CRD42020158654).

Data extraction
Two reviewers (MJ and JK) independently extracted data and the discrepancy was resolved in consultation with an expert investigator (KA). The following data were extracted from all articles using a data-extraction sheet: rst author, year of publication, study design, country, sample characteristics (age, sex, and sample size), exposure characteristics (phthalate metabolite and its specimen, units of concentration in phthalate, timing at measurement of phthalate), and outcome characteristics (assessment timing to diagnose AD). Relative risk, odds ratio (OR), and 95% con dence interval (CI) of the association between prenatal phthalate exposure and the incidence of AD in children were also extracted. If the estimated data were not found in those articles, we contacted the corresponding authors to ask for the detailed data.

Quality assessment
We used the Newcastle-Ottawa scale (NOS) to assess the quality of nonrandomized studies including case-control and cohort. The system allowed for a maximum of nine points which indicates the highest quality, with scores of six or higher denoting high quality. There are three categories: selection (0-4 points), comparability (0-2 points), and outcome or exposure (0-3 points).

Statistical analysis
Data were analyzed using R 3.6.1 (Vienna, Austria; http://www.R-project.org/), package 'metafor.' Only studies with similar exposure to phthalate metabolites were used for meta-analysis. Studies with clinical or different methodological heterogeneity were excluded when pooling the data to improve comparability among studies. A meta-analysis was performed for every speci c compound whenever three or more compounds estimated with heterogeneity compatible to a meta-analysis were available. Summary estimates based on adjusted OR and 95% CIs were used to assess the association between phthalate exposure and risk of childhood AD and were visualized using forest plots. The OR from a study with categorical exposure was transformed to OR for the continuous scale exposure using weighted linear regression of the central values between the categories of exposure on the corresponding ORs. After pooling ORs from exclusive age groups [22], a meta-analysis was conducted to investigate the association between prenatal phthalate exposure and development of AD regardless of timing of outcome assessment. To determine if prenatal phthalate exposure has a more signi cant association with early-onset AD, we performed subgroup analysis by selecting a study with an outcome assessment within 3 years after birth. If the P-value for the Cochrane's Q test was less than 0.10 and I 2 exceeded 50%, we considered heterogeneity to be substantial. [23] We used the random-effect model in our meta-analysis. The potential for publication bias was assessed using a funnel plot analysis and the weighted Egger's regression test.

Study selection and characteristics
Our searches identi ed 129 potentially relevant studies using a systematic search strategy. After the duplicate records were removed, 78 unique publications were reviewed with titles and abstracts according to our inclusion criteria. Fifteen studies were identi ed for full text-review, and 11 articles were nally included in the systematic review because two studies did not have available eczema data and the other two studies were not birth cohort studies (Fig. 1). The characteristics of the included studies are presented in Table 1. Included studies originated from Europe (n = 5), USA (n = 4), and Asia (n = 2). The study sample size ranged from 147 to 1024 mother-child pairs. All selected articles were from cohort studies which were published between 2012 and 2019. Both boys and girls were included in 10 studies, whereas one study recruited only boys. The diagnosis of AD in the included studies was assessed at age 0-9 years using the International Study of Asthma and Allergies in Childhood (ISSAC) questionnaires and the incidence of AD ranged from 9.7-34.6% during the study period. Quality assessment of all studies using the NOS scale had scores of six or greater ( Table 2). The congeners or metabolites of phthalates in the meta-analyses were as follows: mono-benzyl phthalate (MBzP), mono-(2-ethylhexyl) phthalate (MEHP), mono-ethyl phthalate (MEP), mono-isobutyl phthalate (MiBP), mono-n-butyl phthalate (MnBP), and the sum of di- [2-ethylhexyl] phthalate (ΣDEHP).  The score ranged from 0 to 9 (selection ≤ 4, comparability ≤ 2, outcome or exposure ≤ 3).

Association between prenatal phthalate exposure and childhood AD
Of the 11 studies included in the systematic review, a total of eight studies were included in the meta-analysis, [20,[24][25][26][27][28][29][30] excluding three papers that could not be synthesized (e.g., phthalate measurement from blood, [21,31] and binary estimates. [32] The forest plot addresses the association between prenatal phthalate exposure and AD development until age 7 years (Fig. 2) Fig. 2B-F). Subgroup analyses performed for early-onset AD is presented in Table 3. Only prenatal exposure to MBzP had a signi cant association with the risk of early childhood AD (OR 1.22, 95% CI 1.00-1.49, I 2 = 29.65%).

Publication bias
Publication bias was not shown as judged by no signi cant Egger regression test for any of the above-mentioned outcomes (P ≥ 0.078). The funnel plot is shown in Supplemental Figure S1.

Discussion
To the best of our knowledge, the present study is the rst systematic review and meta-analysis on the association between prenatal exposure and childhood AD. Six types of urinary phthalate metabolites including MBzP, MEHP, MEP, MiBP, MnBP, and ΣDEHP were investigated in this meta-analysis.
The forest plot of MBzP with low heterogeneity (I 2 = 17.36%) suggests that prenatal MBzP exposure was signi cantly positively associated with the development of childhood AD (OR 1.16, 95% CI 1.04-1.31). In contrast, there were no signi cant associations between other urinary phthalate metabolites (MEHP, MEP, MiBP, MnBP, and ΣDEHP) and the development of childhood AD. Our result indicates that fetal exposure to phthalate may act as one of the environmental triggers that increase the risk of developing AD after birth.
Phthalates have often been classi ed into two groups based on their molecular weight. Among phthalate metabolites included in our study, MEP (urinary metabolites of diethyl phthalate) and MnBP (urinary metabolites of dibutyl phthalate) are low molecular weight phthalates which are used as solvents in the manufacture of personal care products (e.g., cosmetics, shampoos, perfumes, and nail polish), paints, and adhesives. [33,34] MEHP (urinary metabolites of DEHP), MiBP (urinary metabolites of diisobutyl phthalate), and MBzP (urinary metabolites of benzylbutyl phthalate) are high molecular weight phthalates (ester side-chain lengths, ve or more carbon) which are used as plasticizers in polyvinyl chloride products for building materials, medical devices, and food packaging. [35] Ingestion is believed to be the major route of exposure to DEHP, [36] whereas dermal exposure to personal care products is known to be an important source of exposure to diethyl phthalate. [37] A Swedish study of 1,674 pregnant women showed that polyvinyl chloride ooring in the kitchen and the parents' bedrooms was associated with higher levels of urinary MBzP in the rst trimester. [38] In the present study, a positive correlation was only found between urinary MBzP concentrations and the development of AD in a meta-analysis. During the process of merging data for meta-analysis, the categorical scale of prenatal urinary phthalate exposure levels was converted to a continuous scale following Soomro et al. [26] and Wang et al. [30] The transformed OR might affect the overall results. However, the transformed OR from those two studies were signi cant not only for MBzP but also for MEHP, MEP, and MiBP; nevertheless, only MBzP was signi cant in a meta-analysis. Indeed, the ndings from a birth cohort study by Just et al. [32] also support our results by showing signi cant association between prenatal exposure to MBzP and AD development at age 2 years, although this study was not included in the present meta-analysis because it was not possible to convert their data to a continuous scale. It remains unclear why only MBzP was found to be signi cant, but ethnic differences may be a reason because only Taiwan and French cohort studies showed signi cant associations. [26,39] Other factors such as housing type, lifestyle, or socioeconomic status of the patients may contribute to a statistical signi cance if these factors cause more frequent and continuous exposure to benzylbutyl phthalate in their homes during pregnancy than to other phthalates. Unfortunately, we could not con rm whether these risk factors are different among the studies included in the present meta-analysis.
The mechanism of AD development by prenatal phthalate exposure has been investigated in several studies. Prenatal and neonatal exposure of mice to DEHP induced AD-like skin lesions in dust mite allergen-sensitized offspring and upregulated the T helper 2-dominant expression of eosinophilic in ammation and mast cell degranulation. [40] Lanson et al. [41] reported dibutyl phthalate could drive T helper 2 responses following skin exposure via induction of thymic stromal lymphopoietin (TSLP) gene expression. Diisononyl phthalate aggravated AD-like skin lesion induced by house dust mites in atopic-prone NC/Nga mice, which involves eosinophilic in ltration, mast cell degranulation, TSLP expression, activation of surface markers on bone marrow-derived dendritic cells, and enhanced production of thymus-and activation-regulated chemokine (TARC/CCL17) and macrophage-derived chemokine (MDC/CCL22) [42] In contrast, MBzP might increase the risk of eczema via a nonallergenic mechanism. Just et al. [32] showed that there was no association between prenatal urinary MBzP and sensitization to indoor allergens or total IgE in children with AD. A Canadian cohort study of pregnant women showed that urinary concentration of MBzP was not signi cantly associated with the levels of cord blood IgE, IL33, and TSLP expression which were known to be associated with allergic immune response. [43] The strength of this study is that it is the rst systematic review and meta-analysis of all relevant birth cohort studies to investigate the effect of prenatal exposure to phthalate on the development of childhood AD. Nonetheless, our review had several limitations. There was a lack of papers reported till date; therefore, the number of studies in the present meta-analysis was small. Furthermore, it was hard to pool the data because they used different exposure units and different measured samples of exposure. If more studies are reported in the future, more objective results can be observed through a systematic review and meta-analysis. Another limitation is that most studies included in our meta-analysis evaluated phthalate metabolites using a single spot urine sample except Gascon et al. [29] in which the exposure levels were measured during the rst and third trimesters of pregnancy. Measurement of urinary phthalate metabolites at a single time point does not re ect overall phthalate exposure during pregnancy, because phthalates have short biologic half-lives that are rapidly excreted by urine. [44] Nevertheless, our observation that MBzP showed statistical signi cance could further provide strong support to the premise that prenatal MBzP exposure is associated with the occurrence of AD in children.

Conclusion
This meta-analysis showed that prenatal MBzP exposure is associated with the development of childhood AD. Our results suggest that minimizing the exposure of MBzP during pregnancy may be needed to prevent the development of AD.

Declarations
Ethical Approval and Consent to participate Not applicable

Consent publication
Not applicable