Genetic Polymorphisms of UDP-Glucuronosyltransferases and Susceptibility to Antituberculosis Drug-Induced Liver Injury: A Systematic Review and Meta-Analysis

Methods The PRISMA statement was strictly followed, and the protocol was registered in PROSPERO (CRD42022339317). The PICOS framework was used: patients received antituberculosis treatment, UGTs polymorphisms (mutants), UGTs polymorphisms (wild), AT-DILI, and case-control studies. Eligible studies were searched through nine databases up to April 27, 2022. The study's qualities were assessed by the revised Little's recommendations. Meta-analysis was conducted with a random-effects model using odds ratios (ORs) with 95% confidence intervals (95% CIs) as the effect size. Results Twelve case-control studies with 2128 cases and 4338 controls were included, and 32 single nucleotide polymorphisms (SNPs) in the seven UGT genes have been reported in Chinese and Korean. All studies were judged as high quality. The pooled results indicated that UGT1A1 rs3755319 (AC vs. AA, OR = 1.454, 95% CI: 1.100–1.921, P = 0.009), UGT2B7 rs7662029 (G vs. A, OR = 1.547, 95% CI: 1.249–1.917, P < 0.0001; GG + AG vs. AA, OR = 2.371, 95% CI: 1.779–3.160, P < 0.0001; AG vs. AA, OR = 2.686, 95% CI: 1.988–3.627, P < 0.0001), and UGT2B7 rs7439366 (C vs. T, OR = 0.585, 95% CI: 0.477–0.717, P < 0.0001; CC + TC vs. TT, OR = 0.347, 95% CI: 0.238–0.506, P < 0.0001; CC vs. TC + TT, OR = 0.675, 95% CI: 0.507–0.898, P = 0.007) might be associated with the risk of AT-DILI. Conclusions The polymorphisms of UGT1A1 rs3755319, UGT2B7 rs7662029, and UGT2B7 rs7439366 were significantly associated with AT-DILI susceptibility. However, this conclusion should be interpreted with caution due to the low number of studies and the relatively small sample size.


Introduction
Tuberculosis (TB) is a chronic communicable disease caused by Mycobacterium tuberculosis that contributes to high morbidity and mortality worldwide.In 2021, an estimated 10.6 million people fell ill with TB worldwide [1].TB is preventable and, in most cases, treatable.At least 85% of drug-susceptible TB patients are successfully treated [2].However, anti-TB therapy is known to have a hepatotoxicity efect, and anti-TB drug-induced liver injury (DILI) (AT-DILI) has been a long-standing concern in the treatment of TB infection [3].Te reported incidence of AT-DILI varies widely from 2% to 28% in diferent countries, depending on the investigators' defnition of DILI as well as the population being studied [4].A signifcant upward trend in AT-DILI incidence was observed from 1999 to 2020 [5].Te clinical spectrum of AT-DILI includes asymptomatic elevation in liver tests to acute hepatitis and acute liver failure [6].Prompt withdrawal of the anti-TB drugs is the most critical intervention in the management of AT-DILI [4], which could lead to treatment interruption and poor treatment outcomes [7].Terefore, reducing the occurrence of AT-DILI is crucial for the control of TB.
Until now, the specifc mechanisms associated with AT-DILI have been inadequately described.A number of hypotheses on the pathogenesis of AT-DILI have been proposed, such as drug metabolism and transport, immune response, oxidative stress, and mitochondrial dysfunction [8].Many previous mechanistic studies have focused on the isoniazid (INH) metabolic pathway.However, rifampin (RIF) has been reported to cause hemolysis due to the production of drug-dependent antibodies [9].It was believed that hemolysis can generate a larger amount of hemoglobin from destroyed erythrocytes in blood and result in increased levels of free heme [10].In addition, cotreatment with RIF and INH also causes accumulation of the endogenous hepatotoxin protoporphyrin IX in the liver through the alteration of the heme biosynthesis pathway [11].Te principal product of heme catabolism, bilirubin, is eliminated by a conjugation reaction with glucuronic acid, and the glucuronidation reaction is mediated by uridine diphosphate (UDP)-glucuronosyltransferases (UGTs) [12].Te UGT family is a phase II enzyme group responsible for the glucuronidation of numerous endobiotics, xenobiotics, and drugs to facilitate their excretion from the body [13].In humans, 19 functional UGT isoforms comprise two families (UGT1A and UGT2) based on genetic similarity [14].For example, hydrophobic bilirubin is a toxic product of heme metabolism that can be transformed into hydrophilic bilirubin in the liver through conjugation with uridine diphosphate glucuronic acid under the action of UGT1A1 [15].UGT1A1 dysfunction may lead to hepatic vulnerability induced by the accumulation of bilirubin in the liver [16].Tus, the glucuronidation reaction accelerates the elimination of toxic compounds, which plays an important role in the development of AT-DILI.
Human UGT genes have a large number of genetic polymorphisms, which have been confrmed to modulate enzymatic activity or promoter activity [17], further affecting the individual genetic susceptibility to AT-DILI.Recently, a number of studies have investigated the association between single nucleotide polymorphisms (SNPs) in UGTs and the risk of AT-DILI.Among the UGTs, the UGT1A1 gene has been the most extensively studied, but with inconsistent results among diferent populations [18][19][20][21].For example, the SNP rs4148323 AA genotype of UGT1A1 was found to signifcantly reduce the risk of AT-DILI in Chinese patients in one study [21], while it was not associated with AT-DILI in Chinese patients in another study [20].In recent years, the relationships between SNPs in other UGT genes (UGT1A4 [22], UGT2B4 [3], and UGT2B7 [23]) and AT-DILI risk have also been reported.Terefore, it is necessary to conduct a comprehensive systematic review by retrieving all publications reporting the relationship between SNPs in UGT genes and the risk of AT-DILI and clarifying the pooled efects of polymorphisms for AT-DILI.In the present study, we summarize published data to evaluate the relationship between UGT polymorphisms and susceptibility to AT-DILI.

Search Strategy.
Tis systematic review and meta-analysis was reported according to the PRISMA guidelines [24] and has been registered on PROSPERO with ID number CRD42022339317.A comprehensive literature search was performed in English databases (PubMed, Medline, Web of Science, Embase, and Cochrane Library) and Chinese databases (CNKI, WANFANG, VIP, and SinoMed) up to April 27, 2022.Te Medical Subject Headings (MeSH) terms used in the search were "tuberculosis," "antitubercular agents," and "chemical and drug-induced liver injury."Furthermore, the following MeSH synonyms, related terms, and free terms were also included: "antituberculosis," "tuberculosis treatment," "drug-induced liver injury," "drug-induced hepatotoxicity," "uridine diphosphate glucuronosyltransferase," "UDP-glucuronosyltransferase," "UGTs," "uridine 5′-diphospho-glucuronosyltransferase," "polymorphism," and "variant."Te Boolean operators "and" and "or" were applied to combine these terms.At the same time, the reference lists of selected articles and relevant reviews were manually searched to gather other potentially eligible studies.

Eligibility Screening.
Records identifed through all searches were imported into the EndNote X9 software for screening studies, and duplicate records were removed.Two reviewers independently selected the relevant studies meeting the eligibility criteria by titles and abstracts.Full texts were referred to when the above information was inadequate or unspecifc for the determination of eligibility.Any disagreement was discussed and resolved by consensus or by consulting a third reviewer.
Te studies included in this research met the following eligibility criteria: (1) case-control studies designed to investigate the relationship between UGTs polymorphisms and AT-DILI; (2) all patients received anti-TB treatment, of which the case group had AT-DILI, while the control group did not have AT-DILI; (3) the genotype frequency data could be extracted and analyzed; and (4) the language was restricted to English or Chinese.
Te exclusion criteria were the following: (1) conference abstracts, editorials, letters, case reports, reviews, and metaanalyses; (2) sample size for each group of less than 10; and (3) studies with repetitious data (the studies with the most recent or comprehensive data were selected).
All studies were judged as high quality, and the average score was 6.4 (Supplementary Table 1).

Discussion
Te present study aimed to verify whether current evidence supports the relationship between UGTs polymorphisms and AT-DILI risk.Our meta-analysis included 12 casecontrol studies involving 32 SNPs in the seven UGT genes.Based on two original studies, the pooled results indicated that UGT1A1 rs3755319 (heterozygote model) might be associated with AT-DILI risk.In addition, UGT2B7 rs7662029 (allele model, dominant model, and heterozygote model) and rs7439366 (allele model, dominant model, and recessive model) were also statistically associated with AT-DILI risk.Terefore, genetic variants in UGT1A1 and UGT2B7 may have relationships with susceptibility to AT-DILI; thus, they have potential for use as biomarkers in the anti-TB treatment population.
However, the SNP rs4148323, which is the most studied SNP to date, was found to have no signifcant association with AT-DILI risk under any genetic model.Subgroup analysis also obtained similar negative results in Chinese patients (Figure 2, Tables 2 and 3).Further analysis found     that only one original study indicated that patients with the A allele of rs4148323 in UGT1A1 had a lower risk of AT-DILI (A vs. G, OR = 0.371, 95% CI: 0.161-0.857,P = 0.020) [21].Other original studies did not show any signifcant association between the SNP rs4148323 and AT-DILI risk [15,[18][19][20].Although these studies were all designed as case-control studies, diferences in sample size, diagnostic criteria, and adjusted covariates cannot be Test for overall effect: Z = 2.63 (P = 0.009) Heterogeneity: tau 2 = 0.00; chi 2 = 0.06, df = 1 (P = 0.81); I 2 = 0% Favours (AT-DILI) 0.2 0.5 1 2 5 Favours (control)

8
Journal of Tropical Medicine ignored.For example, one 1 : 1 matched case-control study was conducted in China with the largest sample size to date (461 cases and 466 controls) [21].Tat study employed 5 ULN of ALT as a diagnostic criterion and did not adjust for covariates in the analysis, while others employed a 2 ULN criterion and adjusted for some covariates.Previous studies have found that the rs4148323 homozygous mutation and heterozygous mutation caused the enzymatic activity of UGT1A1 to decrease by 30-40% and 60-70%, respectively, and then signifcantly increased total bilirubin levels in vivo [33].Among patients who received anti-hepatitis C virus drug treatment, the A allele of rs4148323 in UGT1A1 could be considered as a risk factor for drug-induced ALT elevation and liver injury [34].Terefore, further studies are needed to confrm the association between SNP rs4148323 and AT-DILI risk, although our meta-analysis found no association based on the present studies.For UGT1A1 rs3755319, in silico analysis indicated that the C allele might induce transcription binding changes and reduce UGT1A1 expression [35].However, a signifcant association between SNP rs3755319 and AT-DILI risk was found under the heterozygote model.Te haplotype TGG (rs3755319-rs2003569-rs4148323) in UGT1A1 was discovered to be associated with a marginally higher risk of ATLI (OR = 5.071, 95% CI: 1.007-25.531,P = 0.049) [15], and no association was observed between rs3755319 and RIF pharmacokinetics in South African patients with TB [36].Terefore, SNP rs3755319 as a genetic risk marker was not robust enough according to our results, and more original studies are needed to confrm the above conclusion.For the other two statistically signifcant SNPs (UGT2B7 rs7662029 and rs7439366), the present metaanalysis under multiple genetic models and two original case-control studies observed that the AG genotype of rs7662029 and the TT genotype of rs7439366 in UGT2B7 increased the risk of AT-DILI [31,32].A previous study indicated that genetic polymorphisms in the coding and promoter regions of UGT2B7 had important clinical  implications for pharmacology and toxicology and could induce AT-DILI through clinically signifcant changes in drug clearance [37].Tis phenomenon has also been observed in other forms of DILI; for example, the T allele of rs7439366 in UGT2B7 was more common in diclofenac hepatotoxicity patients (OR = 8.5, 95% CI: 1.1-69.9,P = 0.026) [38].Of course, determining the potential application value of rs7662029 and rs7439366 in UGT2B7 requires further research.
Identifcation of a genetic predisposition to AT-DILI is of paramount importance.Tese meta-analysis results indicated that UGT1A1 rs3755319, UGT2B7 rs7662029, and rs7439366 might be associated with the risk of AT-DILI, which would help to identify susceptible populations for liver injury in patients with anti-TB treatment.If used as a test prior to prescription, genotyping of these genes would prevent potential AT-DILI.However, although various genetic polymorphisms have been identifed to be associated with DILI susceptibility, few prospective genetic screening tests have met the threshold for clinical application [39,40].Te main reason is that the low incidence rate of DILI leads to a low positive predictive value for currently identifed genetic variations, making them unsuitable for preprescription screening [41].As described above, the reported incidence of AT-DILI is relatively low [4].So, the low DILI incidence could not warrant the cost and efort associated with genetic testing [42].Genetic polymorphisms of UGTs may be not useful in preemptive tests to reduce DILI incidence, but they can aid DILI diagnosis and clinical decision-making [40].
Tis study was the frst to summarize all relevant studies investigating the relationships of UGTs polymorphisms with AT-DILI risk under diferent genetic models and to perform a meta-analysis of the data reported in those studies.Te quality of the included studies was high.Nevertheless, the study had several limitations.First, the number of included studies was small, and the sample size was relatively small for determining genetic association, which made it difcult to draw a robust conclusion.Second, the study subjects were only Chinese and Koreans (limited to Asian countries), which minimized the possibility of discovering meaningful genetic associations.Because fewer than ten studies were included, a publication bias test was not performed.Finally, there existed a high heterogeneity for UGT1A1 rs4148328, and I-squares were larger than 60% under diferent genetic models.Te diagnosis of AT-DILI, causality assessment, and adjustment for covariates were not uniform in those studies, which may be sources of potential heterogeneity.

Conclusion
Te current meta-analysis indicated that UGT1A1 rs3755319, UGT2B7 rs7662029, and UGT2B7 rs7439366 were signifcantly associated with AT-DILI risk, and these three SNPs may be used as potential genetic risk markers in anti-TB treatment patients.However, this conclusion should be interpreted with caution due to the low number of studies and the relatively small sample size.
Figure 5(c), Tables2 and 3).Sensitivity analysis confrmed these signifcant relationships after excluding the low-quality study.Another two SNPs (rs10028494 and rs7668282) reported in a single study were not signifcantly associated with AT-DILI risk[23].

Table 1 :
Characteristics of the included studies.

Table 2 :
Meta-analysis results of the association between SNPs in UGT1A1/UGT2B7 and AT-DILI risk under the allele model.

Table 3 :
Meta-analysis results of the association between SNPs in UGT1A1/UGT2B7 and AT-DILI risk under diferent genetic models.