Abstract

Background. Many studies have assessed the relationship between gene polymorphisms in multidrug resistance protein 3 (MDR3) and the risk of intrahepatic cholestasis of pregnancy (ICP); however, there are many conflicting narratives.

Objectives. This meta-analysis was conducted to assess the association between MDR3 gene polymorphisms and ICP.

Materials and methods. A multi-database search was conducted using Web of Science, Embase, PubMed, and Chinese Biomedical Literature (CBM) databases. Eleven eligible studies focusing on 4 single nucleotide polymorphisms (SNPs) in the MDR3 gene were selected for analysis. A fixed- or random-effects model was utilized for allelic, dominant, recessive, and superdominant genes.

Results. The pooled results indicated a statistically significant association between MDR3 polymorphism rs2109505 and an increased risk of ICP in both the general population and the Caucasian population. No statistically significant associations were found between MDR3 polymorphism rs2109505 and ICP in Italian or Asian populations for the 4 genetic models. The MDR3 polymorphism rs1202283 was associated with susceptibility to ICP in both the general and Italian populations.

Conclusions. The MDR3 rs2109505 and rs1202283 polymorphisms are associated with ICP susceptibility: however, they displayed no correlation with an increased risk of ICP.

Key words: meta-analysis, intrahepatic cholestasis of pregnancy, SNP, susceptibility, MDR3 gene polymorphism

Introduction

Intrahepatic cholestasis of pregnancy (ICP), a pregnancy-related liver disorder, is characterized by maternal pruritus without a skin rash in the 3^rd trimester. The disorder is recognized as a reversible form of cholestasis in late pregnancy. In addition, patients have varying degrees of elevated total serum bile acids, with or without elevated serum transaminases.¹ The ICP is associated with an increased risk of adverse perinatal outcomes, including stillbirth, preterm delivery, fetal staining of amniotic fluid, fetal distress, or asphyxia.² Elevated total serum bile acids caused by ICP are associated with adverse perinatal outcomes.

It has been acknowledged that genetic factors are closely related to the pathogenesis of ICP.³ For instance, multidrug resistance protein 3 (MDR3), serving as ATP-binding cassette subfamily B member 4 (ABCB4),² was reportedly associated with the pathogenesis of ICP. The MDR3 protein serves as a phosphatidylcholine flippase, transporting phosphatidylcholine from the lumen of the hepatocyte to the lumen of the biliary canaliculi.⁴ Moreover, MDR3 dysfunction and failure of phosphatidylcholine secretion triggered by ABCB4 mutations lead to luminal epithelial cell damage and subsequent biochemical abnormalities, which play a key role in the pathogenesis of ICP.² However, a study based on ICP patients from west Sweden showed no association between MDR3 gene mutations and the pathogenesis of ICP.^{5, 6} On this basis, we assume that there are conflicting views regarding the roles of MDR3 gene mutations in the pathogenesis and patients’ susceptibility to ICP.

Objectives

This meta-analysis was conducted to assess the relationship between MDR3 gene polymorphisms and ICP susceptibility.

Materials and methods

Search strategy

The protocols used in the search strategy were described according to the previous methods (PROSPERO ID: CRD42020164230). A systematic search was conducted and focused on the studies regarding variants of the MDR3 gene in ICP patients published in PubMed, Web of Science, Embase, and Chinese Biomedical Literature (CBM) databases until January 2021. There were no language restrictions in the search strategy. Following keywords were used in the search: “intrahepatic cholestasis of pregnancy” OR “recurrent intrahepatic cholestasis of pregnancy” OR “obstetric cholestasis” OR “cholestasis”, AND “multidrug resistance protein 3” OR “ABCB4 protein, human” OR “MDR3 protein.”

Studies meeting the following inclusion criteria were eligible to be included in this meta-analysis: 1) studies which examined MDR3 gene polymorphisms in patients with or without ICP; 2) case-control studies in which the controls had experienced physiological pregnancies; 3) in cases of 2 or more studies from the same cohort, the most integrated study was included; 4) studies which reported the number of genotype variants related to ICP susceptibility. Studies with the following conditions were excluded: 1) the literature revealed familial ICP or other cholesteric liver diseases; 2) studies enrolling nonpregnant women; 3) research or experimental studies that were not conducted on animals; 4) case reports or meeting abstracts; 5) studies lacking particular genotypic data or sequence analysis of MDR3 polymorphisms.

Data extraction

Data were extracted by 2 investigators independently, including author’s name, year of publication, nationality, race, sample size, patient characteristics, risk gene polymorphisms, and method of genotyping. In controversial cases, a thorough discussion was conducted with a 3^rd investigator until a consensus was reached.

Statistical analyses

The association of single nucleotide polymorphism (SNP) in the MDR3 gene with the risk of ICP was estimated by combining pooled odds ratios (ORs), 95% confidence intervals (95% CIs) and Z-test results in the allelic, dominant, recessive, and superdominant gene models. The simplest case was represented by a polymorphism with 2 alleles (A – mutant-type, B – wild-type), one of which (A) is thought to be associated with the disease. Association analysis collects information about the numbers of disease-free and diseased subjects with each of the 3 genotypes (AA, AB and BB). A recessive model compares AA with BB+AB, a dominant model compares AB+AA with BB, and a superdominant model compares AA+BB with AB.

For appropriate continuity correction, a value of 0.5 was substituted in the presence of 0 in the number. Meta-analysis was performed on the 4-gene models using the fixed-effects model. The random-effects model was used to evaluate the OR if there was heterogeneity (I² > 50%, p < 0.05) among the studies. The publication bias was evaluated using Egger’s test. The STATA v. 12.0 software (StataCorp LLC, College Station, USA) was used for all analyses, and all tests were two-sided. A value of p < 0.05 was considered statistically significant.

Results

Characteristics of studies

Initially, a total of 291 articles were extracted from PubMed, Web of Science, Embase, and CBM databases. From these articles, 133 were excluded due to duplication, and 113 were excluded after reading the titles and/or abstracts. Subsequently, full texts of 45 articles were accessed, from which 31 articles were excluded, including 2 case reports, 1 conference article, 13 studies focusing on familial ICP or cholesteric disease, 5 review articles, 5 studies reporting no information on the controls, and 5 articles with no analysis of MDR3 mutations. Afterwards, we identified 14 eligible studies involving 83 SNPs in MDR3, including 4 studies focused on the same SNP (Figure 1). Finally, 11 eligible studies investigating 4 SNPs (i.e., rs2109505, rs1202283, rs2302387, and rs2230028) in MDR3 were included.^{7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17}

Meta-analysis

A meta-analysis was performed based on the rs2109505 (involving 967 cases and 1149 controls), rs1202283 (179 cases and 415 controls), rs2302387 (168 cases and 390 controls), and rs2230028 (542 cases and 714 controls) (Table 1).

The pooled results indicated a relationship between the MDR3 rs2109505 polymorphism and increased risk of ICP in both the general (dominant model: OR = 0.72, 95% CI: 0.34–1.53, Z = 1.99, p = 0.047; recessive model: OR = 0.68, 95% CI: 0.33–1.40, Z = 2.23, p = 0.026; superdominant model: OR = 1.38, 95% CI: 0.66–2.90, Z = 2.10, p = 0.036) and Caucasian (allelic model: OR = 1.87, 95% CI: 1.46–2.40, Z = 6.64, p = 0.000; dominant model: OR = 0.49, 95% CI: 0.38–0.63, Z = 6.56, p = 0.000; recessive model: OR = 0.47, 95% CI: 0.22–0.99, Z = 2.74, p = 0.006; superdominant model: OR = 1.92, 95% CI: 1.52–2.44, Z = 5.72, p = 0.000) populations. However, MDR3 rs2109505 polymorphism was not correlated with the pathogenesis of ICP in Italian or Asian populations under the 4 genetic models (all p > 0.05) (Table 2, Figure 2A–D).

We found limited evidence that MDR3 rs1202283 was associated with ICP susceptibility in general (superdominant model: OR = 0.42, 95% CI: 0.12–1.42, Z = 2.43, p = 0.015), Italian (superdominant model: OR = 0.09, 95% CI: 0.00–5.79, Z = 2.44, p = 0.015) and Asian (allelic model: OR = 0.27, 95% CI: 0.16–0.44, Z = 5.27, p = 0.000) populations. No significant correlation was found between the MDR3 rs1202283 polymorphism and susceptibility to ICP in Greek or Caucasian population (all p > 0.05) (Table 2, Figure 3A–D), and MDR3 rs2302387 or MDR3 rs2230028 showed no correlation with the risk of ICP (all p > 0.05) (Table 2, Figure 4A–D and Figure 5A–D).

Publication bias analysis

Egger’s test showed no evidence of publication bias for MDR3 rs1202283 or rs2302387 and ICP risk (p > 0.05). There was a publication bias for rs2109505 in the allelic (t = −6.95, p < 0.01) and dominant (t = −5.84, p = 0.001) gene models, as well as for rs2230028 in the superdominant model (t = 2.98, p = 0.025) (Table 2).

Discussion

The MDR3 polymorphisms have been reported to play a role in the pathogenesis of ICP. As a transporter for phospholipids, the MDR3 protein mediates the transmission of phosphatidylcholine into the bile capillary, which deactivates toxic bile salts to protect the epithelial lining.¹⁸ The MDR3 dysfunction has been linked to hormonal and environmental factors, which may contribute to the pathogenesis of ICP.¹⁹ This meta-analysis was conducted to further explore the relationship between MDR3 gene polymorphisms and the pathogenesis of ICP.

Many studies have previously investigated the relationship between ICP and MDR3 dysfunction, but there are still disputes regarding the conclusions. In a previous study, Anzivino et al. explored ABCB4 (MDR3) and ABCB11 mutations in ICP patients in an Italian population. The authors concluded that ABCB4 mutations were closely involved in the onset of ICP.¹⁰ In addition, 6 MDR3 polymorphisms, namely rs2097937, rs31676, rs1149222, rs4148826, rs2109505, and rs2302386, were significantly associated with an altered risk of ICP, especially rs2109505 (p = 4.6×10⁻⁷).⁹ In a case report, Kamimura et al. presented an ICP patient with the rs1202283 polymorphism, although no strict link between this polymorphism and ICP was observed.²¹ Our data confirmed that MDR3 rs2109505 and rs1202283 polymorphisms were closely associated with the pathogenesis of ICP. Furthermore, susceptibility to ICP was related to racial and territorial factors, as revealed in our meta-analysis. In contrast, there were no significant associations between MDR3 rs2302387 or rs2230028 and the risk of ICP. This makes our data consistent with the findings of a previous study.¹³

The rs1202283 polymorphism was closely related to the pathogenesis of ICP among the Caucasian population in France. This polymorphism may affect the splicing and stability of ABCB4 mRNA, as well as the linkage disequilibrium of other SNPs.¹³ Additionally, Dixon et al. indicated that there was a linkage disequilibrium between rs2109505 (c.711A>T, p.I237I) polymorphism and ABCB4 SNPs,⁹ which was considered to be associated with the pathogenesis of ICP in a Swedish population. For the ABCB4 gene, MDR3 rs2109505 and rs1202283 were the most common SNPs. They may affect the special conformation and substrate specificity of the transporter, which then affect the transmission of phospholipids through the MDR3 protein.²¹ Furthermore, MDR3 rs2109505 and rs1202283 mutations could trigger the dysfunction of the MDR3 transporter, which leads to the decline of phospholipid and phosphatidylcholine in the bile. This would result in the proportion of toxic bile acid, which then leads to injuries of the hepatobiliary cells and an increase in circulating bile acids.

To date, many studies have focused on the association between severe ICP and perinatal outcomes. In an excellent study by Ovadia et al., the authors reported an increased risk of stillbirth for women with serum total bile acids of 100 µmol or more.²² Cui et al. indicated that women with ICP who also showed serum total bile acids of 40 µmol or more were reported to present an increased risk of preterm birth compared to their counterparts with lower bile acids.²³ Moreover, those with severe ICP were associated with fetal cardiac arrhythmia and placental vessel spasms.^{24, 25} Based on these data, together with our analysis, we speculate that severe ICP may be associated with MDR3 polymorphisms.

Limitations

There are some limitations to our study. First, there was a publication bias for the rs2109505 polymorphism in both allelic and dominant gene models, as well as a bias for rs2230028 in the superdominant model. Second, only 4 studies (40%) on rs2109505 and 3 (37.5%) on rs2230028 calculated the Hardy–Weinberg equilibrium (HWE). Moreover, the case groups from 2 gene loci consisted of ICP patients with bile acid levels higher than 40 μmol/L or had elevated levels of gamma-glutamyl transferase (GGT). These 2 conditions were determined as the specific subgroups of ICP, and the rate of MDR3 mutations was higher than in common ICP patients. We suspect that population bias present in this population, in combination with the violation of the HWE contributed to the publication bias in our meta-analysis. In addition, only 9% of the included studies elucidated the MDR3 rs1202283 polymorphism and its impact on ICP risk in the Greek population. In the future, more studies are needed to further investigate the impact of racial and regional factors on the relationship between MDR3 polymorphisms and ICP risk. Furthermore, there was no exploration of the influence of nongenetic factors on ICP susceptibility.

Conclusions

Our meta-analysis indicated that MDR3 rs2109505 and rs1202283 gene polymorphisms were associated with ICP susceptibility. Further studies are required to explore alternative etiologies for ICP susceptibility, such as ethnic, racial, regional, and nongenetic factors.