The genetic variants in calcium signaling related genes influence anti-tuberculosis drug induced liver injury: A prospective study.

Although many genetic variants related to anti-tuberculosis drug induced liver injury (ATDILI) have been identified, the prediction and personalized treatment of ATDILI have failed to achieve, indicating there remains an area for further exploration. This study aimed to explore the influence of single nucleotide polymorphisms (SNPs) in Bradykinin receptor B2 (BDKRB2), Teneurin transmembrane protein 2 (TENM2), transforming growth factor beta 2 (TGFB2), and solute carrier family 2 member 13 (SLC2A13) on the risk of ATDILI.The subjects comprised 746 Chinese tuberculosis (TB) patients. Custom-by-design 2x48-Plex SNPscanTM kit was employed to genotype 28 selected SNPs. The associations of SNPs with ATDILI risk and clinical phenotypes were analyzed according to the distributions of allelic and genotypic frequencies and different genetic models. The odds ratio (OR) with corresponding 95% confidence interval (CI) was calculated.Among subjects with successfully genotyped, 107 participants suffered from ATDILI during follow-up. In BDKRB2, patients with rs79280755 G allele or rs117806152 C allele were more vulnerable to ATDILI (PBonferronicorrection = .002 and .03, respectively). Rs79280755 increased the risk of ATDILI significantly whether in additive (OR = 3.218, 95% CI: 1.686-6.139, PBonferroni correction = .003) or dominant model (PBonferroni correction = .003), as well as rs117806152 (Additive model: PBonferroni correction = .05; dominant model: PBonferroni correction = .03). For TENM2, rs80003210 G allele contributed to the decreased risk of ATDILI (PBonferroni correction = .02), while rs2617972 A allele conferred susceptibility to ATDILI (PBonferroni correction = .01). Regarding rs2617972, significant findings were also observed in both additive (OR = 3.203, 95% CI: 1.487-6.896, PBonferroni correction = .02) and dominant model (PBonferroni correction = .02). Moreover, rs79280755 and rs117806152 in BDKRB2 significantly affected some laboratory indicators. However, no meaningful SNPs were observed in TGFB2 and SLC2A13.Our study revealed that both BDKRB2 and TENM2 genetic polymorphisms were interrogated in relation to ATDILI susceptibility and some laboratory indicators in the Western Chinese Han population, shedding a new light on exploring novel biomarkers and targets for ATDILI.

Introduction

Anti-tuberculosis (TB) drugs hold the key to thwart the rise and spread of TB. However, the adverse drug reactions (ADRs) caused by these anti-TB drugs have become new problems that cannot be ignored. Among all types of ADRs, anti-tuberculosis drug induced liver injury (ATDILI) has a high prevalence (2–30%[) and mortality (22.7%[), unpredictable course and adverse impact on anti-TB treatment, thereupon is becoming a mainstream topic for researchers.[ ATDILI is defined as a heterogeneous set of responses triggered by anti-TB drugs,[ usually manifesting as a decreased liver function.[ A growing body of evidence implicates calcium signaling in mitochondrial dysfunction, oxidative stress, and ensuing ATDILI.[ In addition, an access key role of calcium signaling in some inflammatory processes also supports the involvement of this signaling pathway in the development of ATDILI.[ Now, risk factors related to calcium signaling have been explored extensively, aiming to identify their value of risk assessment, diagnosis and personalized treatment in ATDILI. Among these factors, genetic factors, especially single nucleotide polymorphisms (SNPs), are considered to play a crucial role due to the unpredictable and non-dose-dependent characteristics of ATDILI.[

Bradykinin receptor B2 (BDKRB2) encodes a G-protein coupled receptor of bradykinin (BK).[ Accumulating reports have confirmed that BDKRB2 is capable of regulating calcium signaling. Through binding to BDKRB2, BK allows calcium to enter, leads to calcium-induced calcium release and evokes calcium signaling via upregulating the expression of transient receptor potential melastatin 7 (TRPM7).[ Teneurin transmembrane protein 2 (TENM2) encodes a type 2 membrane protein, consisting of a cytosolic N-terminus, a single transmembrane region and an extracellular C-terminal domain.[ Existing researches imply that TENM2 is inclined to a better interaction with latrophilin-1, and this interaction elicits intracellular calcium signaling.[ Transforming growth factor beta 2 (TGFB2) encodes the transforming growth factor beta family of cytokines which functions in proliferation, differentiation, adhesion, and migration in many cell types.[ Close relationship between TGFB2 and calcium signaling has been recognized. TGFB2 enables to transmit signals via calcium signaling,[ and thus play roles in some diseases such as cardiomyopathy in mouse models.[ Solute carrier family 2 member 13 (SLC2A13) is responsible for encoding GLUT13, an H+/myoinositol cotransporter.[ Ongoing evidence shows that SLC2A13 participates indirectly in calcium signaling. SLC2A13 is closely associated with the transport of inositol, while inositol is the key molecule in regulating calcium signaling.[ Clearly, these 4 genes, BDKRB2, TENM2, TGFB2, and SLC2A13, are correlated with calcium signaling. Therefore, it seems that these 4 genes influence the individual susceptibility to ATDILI via calcium signaling.

Although the exploration of genetic variants related to ATDILI have never been stopped, it is far from to predict and individualize the treatment of ATDILI based on existing findings. More novel genetic variants in different genes and different populations should be identified to facilitate our understanding of ATDILI. Considering the heavy burden of ATDILI in Southwest China,[ we conducted this prospective study in Western Chinese Han population to investigate the relationship between ATDILI and genetic variants in BDKRB2, TENM2, TGFB2, and SLC2A13, aiming to evaluate the potential value of these 4 genes polymorphisms in the risk assessment, pathogenesis, and personalized treatment of ATDILI.

Materials and methods

Study population

From December 2016 and April 2018, this prospective study consecutively recruited TB participants registering in the West China Hospital of Sichuan University. Blood and other specimens were collected from all participants for TB diagnosis and liver function examination. The clear TB evidence and normal liver function before anti-TB treatment were need for all included patients. Once participants suffered from HIV, immunodeficiency diseases or other lung or liver disorders, they would be excluded. After recruitment, all subjects would be treated with a 6-month 4-drug standard treatment (2 months of rifampicin, isoniazid, pyrazinamide, and ethambutol, followed by rifampicin and isoniazid for 4 months) and received liver function examination regularly. Patients would also be excluded if they were treated with analgesics and antipyretics including acetaminophen, hypoglycemic drugs including glitazones, anticonvulsants, and herbal medicines during the 6-month follow-up.

The diagnostic criteria of ATDILI was described by Watkins et al.[ Specifically, ATDILI was identified based on serum alanine aminotransferase (ALT) > 2 times upper limit of the normal (ULN) or aspartate aminotransferase (AST) > 2 times ULN combined with total bilirubin > 2 times ULN during anti-TB therapy.

This trial was approved by the Ethics Committee of West China Hospital of Sichuan University. The signed written informed consents were collected from all included TB patients.

Genes genotyping

Peripheral whole blood of each patient was collected for extracting genomic DNA by QIAamp DNA blood mini kit (Qiagen, Germany). After considering minor allele frequency (MAF) (≥0.02) in both Southern Han Chinese and Han Chinese in Beijing, locations, linkage disequilibrium (LD) constant (r2 < 0.8) and others, 28 SNPs: 8 SNPs in BDKRB2, 7 SNPs in TENM2, 8 SNPs in TGFB2, and 5 SNPs in SLC2A13 were selected by Haploview version 4.1 (The Broad Institute, Cambridge, MA, USA). All SNPs were genotyped by the custom-by-design 2x48-Plex SNPscanTM kit (Genesky Biotechnologies Inc., Shanghai, China). Approximately 10% samples would be redetected to calculate the concordance for quality assessment.

Statistical analysis

Continuous variables and categorical variables were compared by Mann–Whitney's U test and chi-square test or Fisher's exact test, respectively. While Hardy–Weinberg equilibrium (HWE), and allelic and genotypic frequencies were evaluated by chi-square analysis or Fisher's exact test. PLINK version 1.07 was applied for identify the relationship between selected SNPs and ATDILI by logistic regression analysis, while SHEsis was employed to perform Linage analysis and haplotype construction (MAF ≥ 0.01). Odds ratio (OR) with corresponding 95% confidence interval (CI) was calculated for measuring of relationships. Significance was set at P ≤ .05. Power and Sample Size Program was used to calculate the power based on the sample size of this work. Furthermore, some online tools were applied to predictive the functions of candidate SNPs.

Results

A total of 746 TB patients were enrolled in our study (Fig. 1), nevertheless, 28 selected SNPs were successfully genotyped among 686 participants. Among these 686 subjects, 107 participants suffered from ATDILI during our 6 months follow-up. Significant differences in the incidence of fever (P = .02), ALT levels (P < .001), AST levels (P < .001), alkaline phosphatase (ALP) levels (P = .03), gamma glutamyl transpeptidase (GGT) levels (P = .004) and uric acid levels (P = .03) were identified between the cases and the controls. While there were no meaningful findings in other characteristics (Table 1).

Figure 1

Selection of patients included in this study. ATDILI = anti-tuberculosis drug induced liver injury, TB = tuberculosis.

Table 1

The characteristics of enrolled patients.

The relationship between selected SNPs and ATDILI

All genotypes of 28 SNPs did not deviate from the HWE in controls. In BDKRB2, rs79280755, and rs117806152 were associated with the risk of ATDILI. The mutant G allele of rs79280755 and C allele of rs117806152 increased the risk of ATDILI significantly (PBonferroni correction = .002 and .03, respectively). Furthermore, rs79280755 conferred significantly increased risk of ATDILI in both additive (OR = 3.218, 95% CI: 1.686–6.139, PBonferroni correction = .003) and dominant model (OR = 3.218, 95% CI: 1.686–6.139, PBonferroni correction = .003), as well as rs117806152 (additive model: OR = 2.424, 95% CI: 1.292–4.548, PBonferroni correction = .05; dominant model: OR = 2.613, 95% CI: 1.369–4.988, PBonferroni correction = .03).

In TENM2, both rs80003210 and rs2617972 had significant impacts on susceptibility to ATDILI. For rs80003210, patients carrying G allele had the decreased risk of ATDILI with an OR of 0.156 (95% CI: 0.038–0.642, PBonferroni correction = .02). However, rs80003210 conferred comparable risk of ATDILI based on 3 genetic models. For rs2617972, A allele carriers had 3.083 times (95% CI: 1.455–6.532) higher risk of ATDILI than C allele carriers (PBonferroni correction = .01). An adverse effect was identified in both additive model (OR = 3.203, 95% CI: 1.487–6.896, PBonferroni correction = .02) and dominant model (OR = 3.203, 95% CI: 1.487–6.896, PBonferroni correction = .02).

Whether in TGFB2 or SLC2A13, no meaningful SNPs were found (Tables 2 and 3).

Table 2

The comparison of allelic and genotypic frequency between cases∗ and controls∗.

Table 3

The results of genetic model analyses.

Subgroup analyses

Age (the threshold: 50 years) and sex have been reported as risk factors of ATDILI,[ while TB subtypes were also taken into consideration for subgroup analyses.

A total of 31 ATDILI cases and 217 non-ATDILI controls were classified in the elder subgroup (≥50 years), while the remaining 438 patients were in another subgroup. Older patients carrying A allele of BDKRB2 rs79280755 had 4.671 times (95% CI: 1.477–14.770) higher risk of ATDILI than those with G allele (PBonferroni correction = .03), whereas comparable risk of ATDILI was identified in 3 genetic models. In the younger subgroup, no meaningful findings were observed.

There were 413 males and 273 females in this trial. In BDKRB2, both rs79280755 A allele and rs117806152 C allele conferred susceptibility to ATDILI (PBonferroni correction < .001 and .009, respectively). The genetic model analyses demonstrated that both rs79280755 and rs117806152 increased the risk of ATDILI whether in dominant model (PBonferroni correction < .001 and .009, respectively) or additive model (PBonferroni correction < .001and .02, respectively). While females with C allele of TENM2 rs2617972 were more susceptible to ATDILI with an OR of 4.000 (95% CI: 1.353–11.820, PBonferroni correction = .05).

Altogether 474/686, 68/686, and 144/686 subjects were classified into pulmonary TB (PTB) subgroup, extra PTB (EPTB) subgroup and PTB combined with EPTB (PTB with EPTB) subgroup, respectively. The susceptibility to ATDILI for PTB patients were potentially endowed to the mutant alleles of BDKRB2 rs79280755 and BDKRB2 rs117806152 (PBonferroni correction = .04 and .04, respectively). In PTB with EPTB subgroup, the meaningful relationship was identified between TENM2 rs2617972 and the risk of ATDILI (PBonferroni correction = .009) (Table 4).

Table 4

The results of subgroup analyses.

LD analysis and haplotype construction

Based on the cut-off value of pairwise r2 > 0.80, 2 SNPs of BDKRB2 (rs76192091 and rs4900312), as well as 2 SNPs of BDKRB2 (rs4905469 and rs8012552) and 3 SNPs of TGFB2 (rs6657275, rs10482796, and rs6684205) were in a LD block, respectively (Fig. 2). Nevertheless, no haplotypes, which were constructed based on these SNPs, reached statistically significant (Table 5).

Figure 2

Linkage disequilibrium plots. The threshold was set at pairwise r2 > 0.80. The percentages in diamonds and color of diamonds represent pairwise r2 values for all pairs of SNPs and the intensity of pairwise r2, respectively. (A) Linkage disequilibrium plots of 8 single-nucleotide polymorphisms (SNPs) in BDKRB2; (B) linkage disequilibrium plots of 8 single-nucleotide polymorphisms (SNPs) in TGFB2.

Table 5

Haplotype constructions of BDKRB2a and TGFB2b variants related to the risk of ATDILI.c.

The association of SNPs and clinical phenotypes

Based on dominant or recessive model, the potential influence of meaningful SNPs in BDKRB2 (rs79280755 and rs117806152) and TENM2 (rs80003210 and rs2617972) on clinical characteristics was investigated further. For BDKRB2 rs79280755, G allele-containing genotypes indicated significantly higher platelet counts (P = .003), percentage of monocyte (P = .02) and erythrocyte sedimentation rate (P = .02). Regarding BDKRB2 rs117806152, patients carrying C allele-containing genotypes showed higher platelet counts (P = .009) and erythrocyte sedimentation rate (P = .04) than those with AA genotype. No significant findings on the relationship between TENM2 gene polymorphisms and clinical characteristics were observed (Fig. 3).

Figure 3

The impact of 2 single-nucleotide polymorphisms (SNPs) in BDKRB2 on clinical phenotypes. (A) The impact of rs79280755 in dominant model on platelet counts; (B) the impact of rs79280755 in dominant model on percentage of monocyte; (C) the impact of rs79280755 in dominant model on erythrocyte sedimentation rate; (D) the impact of rs117806152 in dominant model on platelet counts; (E) the impact of rs117806152 in dominant model on erythrocyte sedimentation rate. ESR = erythrocyte sedimentation rate.

Discussion

This present study found BDKRB2 and TENM2 gene polymorphisms, but not TGFB2 and SLC2A13, had influence on the risk of AIDILI. The mutant alleles of BDKRB2 rs79280755, BDKRB2 rs117806152, and TENM2 rs2617972 were the adverse elements of ATDILI, while a decreased risk of ATDILI was associated with the mutant allele of TENM2 rs80003210. Subgroup analyses identified the relationships between 3 SNPs (BDKRB2 rs79280755, BDKRB2 rs117806152, and TENM2 rs2617972) and the risk of ATDILI for patients with different ages, genders, and TB subtypes. Moreover, the influence of these meaningful SNPs on laboratory indicators was also explored. These findings provided experimental evidence for some new ATDILI-related targets, which promoted the development of ATDILI related research to some extent.

As we described above, BDKRB2 acts though participating in calcium signaling pathway, mitogen-activated protein kinase (MAPK), and other signal pathways to affect inflammatory processes, endocrine regulation, and drug response.[ In our study, BDKRB2 rs79280755 and BDKRB2 rs117806152 are intron variants which have not been reported thus far. Online tool, HaploReg, suggests that more than 10 transcription factor binding motifs (TFBMs) are altered by rs79280755, and most of changed motifs contribute their share to regulate transcription (https://pubs.broadinstitute.org/mammals/haploreg/detail_v4.1.php?query=&id=rs79280755). Interestingly, BDKRB2 has been recognized as a transcriptional regulator of specific genes,[ consistent with the functional predictions provided by HaploReg to some extent. Notably, one of the affected transcription factors, formerly hepatic nuclear factor 3alpha (Foxa1), is known as a pioneer transcription factor and responsible for normal development of liver and lung.[ Vatamaniuk et al[ have revealed that Foxa1 involves in calcium influx and regulation of oxidative phosphorylation. Herein, rs79280755 may lead to significantly different risk of ATDILI via mediating calcium metabolism by affecting the Foxa1.

TENM2, locating on chromosome 5, elicits heterophilic cell–cell adhesion via plasma membrane cell adhesion molecules, calcium signaling, axon guidance, and other pathophysiological processes.[ Our study testified that TENM2 rs2617972 and TENM2 rs80003210 might be the potential pharmacogenetic biomarkers for ATDILI in the Western Chinese Han population. Of the 2 candidate SNPs, rs80003210 is likely to influence the functions of a transcription factor, hepatocyte nuclear factor 4 (HNF4) (https://pubs.broadinstitute.org/mammals/haploreg/detail_v4.1.php?query=&id=rs80003210). Growing investigators have confirmed the relationship between HNF4 and calcium metabolism. Through the animal trials, Niehof et al[ have demonstrated that the HNF4 acts as a master transcriptional regulator for key genes in calcium signaling. Furthermore, HNF4 is also able to function in the preservation of calcium homeostasis via controlling the expression of hepatocyte nuclear factor 1 alpha (HNF1A).[ Obviously, rs80003210 participates in calcium signaling by various ways, and the relationship between rs8000321 and calcium metabolism may explain the role of this variant in the occurrence of ATDILI to some extent.

We first investigated the roles of variants in 4 genes related to calcium signaling in ATDILI, facilitating our understanding of ATDILI etiology and contributing to develop personalized treatment strategies. Unfortunately, our study still suffered from the limitations of sample size and singleness of ethnicity although the power calculation was performed to assess the reliability of our results. Based on some online bioinformatic tools and our results, we predicted the functions of candidate variants in BDKRB2 and TENM2, and functional trials to verify these predictions are warranted urgently.

Conclusion

In summary, we explored the roles of some calcium signaling-related genes and their variants played in ATDILI and first demonstrated that BDKRB2 rs79280755, BDKRB2 rs117806152, TENM2 rs80003210, and TENM2 rs2617972 were in reference to the susceptibility to ATDILI in Western Chinese Han population. The novel biomarkers of ATDILI founded in this work could contribute their share to plot complete genetic map of ATDILI, which could bring benefits to more accurately predict and diagnose the ATDILI. In addition, these new targets may also help researchers to explore the underlying mechanism of this severe disease and develop the effective vaccines or drugs, reducing the heavy disease burden on multiple levels.

Acknowledgments

We are very grateful to Hong Xie for her suggestions on editing English articles.

Author contributions

Conceptualization: Mengyuan Lyu, Binwu Ying.

Data curation: Mengyuan Lyu, Jian Zhou, Hao Chen, Hao Bai, Jiajia Song, Tangyuheng Liu, Yuhui Cheng.

Formal analysis: Mengyuan Lyu, Jian Zhou, Hao Chen, Hao Bai, Jiajia Song, Tangyuheng Liu, Yuhui Cheng.

Funding acquisition: Binwu Ying.

Investigation: Mengyuan Lyu, Jian Zhou, Hao Chen, Hao Bai, Jiajia Song, Tangyuheng Liu, Yuhui Cheng.

Methodology: Mengyuan Lyu, Jian Zhou, Hao Chen, Hao Bai.

Software: Mengyuan Lyu, Jian Zhou.

Supervision: Binwu Ying.

Validation: Binwu Ying.

Visualization: Binwu Ying.

Writing – original draft: Mengyuan Lyu, Jian Zhou, Hao Chen, Hao Bai, Jiajia Song, Tangyuheng Liu, Yuhui Cheng, Binwu Ying.

Writing – review & editing: Mengyuan Lyu, Jian Zhou, Hao Chen, Hao Bai, Jiajia Song, Tangyuheng Liu, Yuhui Cheng, Binwu Ying.