Study On The Molecular Mechanism Of Encoding Ectodysplasin α Receptor Associated Death Domain For Anti-tuberculosis Drugs-induced Liver Injury In Tuberculosis Patients

Objective:Since 2000,according to WHO data,the incidence rate of tuberculosis worldwide has dropped by about 1.5% a year.Despite the progress made in global control of tuberculosis,new cases and mortality rates have declined,but the incidence rate and mortality rate of tuberculosis is still a huge burden in the world.The combination of isoniazid(INH),rifampicin(RIF),pyrazinamide(PZA)and ethambutol(EMB)(2HRZE / 4HR therapeutic schedule,HRZE therapeutic schedule for short)is the first choice for the treatment of tuberculosis.However,Adverse reactions caused by antituberculosis drug-induced liver injury(ATB-DILI)are the most serious and common problems in the treatment of tuberculosis.How to early warn and intervene the potential risk of ATB-DILI in tuberculosis patients,so as to reduce the drug-resistant tuberculosis and the failure of antituberculosis treatment caused by the interruption of treatment due to adverse drug reactions,is a hot topic in this field.Previous studies have shown that the occurrence of ATB-DILI is closely related to host genetic factors,and single nucleotide polymorphisms(SNPs)are widely used as a kind of genetic markers in host genetic factors.In recent years,with the wide application of artificial intelligence technology in medicine,a multi-dimensional and systematic disease diagnosis and prediction model based on machine learning algorithm combined with electronic medical records and host genetic factors can make intelligent decision and early risk assessment for clinical medicine,and its effectiveness is better than that based on single factor(such as SNPs).But at the same time,it is one of the challenges faced by clinicians and researchers that how to mine useful clinical information from the existing massive clinical information and biomarker data to build a disease early warning model withhigh diagnostic efficiency.In our previous study,we used the genome wide association study(GWAS)strategy to study the susceptibility genes of ATB-DILI patients in Southwest China.In the first-line antituberculosis drug HRZE group,SNPs of retinoid X receptor alpha(RXRA),encoding(ectodysplasin α receptor,EDAR)associated death domain(EDARADD),zinc finger protein(ZNF385D)and other genes were significantly associated with ATB-DILI.It is necessary to further expand the sample size of these SNPs that have been identified earlier and have important association with the occurrence of ATB-DILI,to verify their clinical association with ATB-DILI in another independent sample population,and to further explore the specific mechanism of target genes with clinical application potential leading to the occurrence and development of ATB-DILI through in vitro experiments.Therefore,this research mainly starts from two aspects to carry on the continuation research.First,based on GWAS research in the early stage of the research group,the sample size was further expanded,and the validity of SNPs sites was verified in another independent sample population.The biological function of positive association sites was preliminarily predicted by the analysis of the biological information.The systematic component stratification analysis was conducted according to different populations,and further explored the more detailed and comprehensive susceptibility SNPs sites related to ATB-DILI Point,The machine learning algorithm was used to combine the electronic medical record with the ATB-DILI association sites selected by verification to construct the early warning model related to the occurrence risk of ATB-DILI.On the other hand,the specific molecular mechanism of ATB-DILI related candidate genes regulating the occurrence and development of ATB-DILI was further explored through in vitro experiments,which provided theoretical basis for clinical risk prevention and disease management of ATB-DILI.Part Ⅰ Using machine learning algorithm to integrate differential SNPs screened by GWAS and electronic medical record information to construct early warning model of ATB-DILI1.Materials and Methods:1.1 GWAS screening of SNPs and study cohort1.1.1 Candidate SNPs The research group used Illumina human Omni express gene chip(7000677 tag SNPs)to conduct GWAS research in the previous research cohort,and screened SNPs that were related to ATB-DILI.According to our previous research standard of differential screening,30 differential SNPs were selected as the target SNPs in this study.1.1.2 Study cohort The study included Chinese Han population who were diagnosed with active tuberculosis(TB)in public health clinical center of Chengdu from November 2019 to June 2020 and received anti TB drug chemotherapy for June months.According to the hospital information management system(HIS),the diagnosis and treatment information of all the patients in the electronic medical records were collected,mainly including demographic indicators,clinical symptoms and signs,laboratory examination information and imaging data.1.2 Target SNPs typing method DNA was extracted from peripheral blood mononuclear cells of tuberculosis patients.The 30 SNPs were genotyped by 48-Plex SNP scan(?) Typing technique.The SNPs with correlation with the risk of ATB-DILI were verified.1.3 Association between candidate SNPs and susceptibility to ATB-DILI The association between target SNPs and ATB-DILI was investigated by Plink software at the levels of allele,genotype and different genetic models.Haploview 4.2 software was used for linkage disequilibrium detection,haplotype construction and association analysis of haplotype frequency and susceptibility to ATB-DILI.GMDR software was used to analyze gene-gene interaction.SPSS was used to analyze the correlation between target SNPs and EMR information of ATB-DILI patients.Through bioinformatics analysis,the potential functions of susceptible SNPs of ATB-DILI were annotated on the e QTL website.1.4 Machine learning algorithm integrates SNPs markers and electronic medical records to build ATB-DILI early warning model The early warning model of ATB-DILI was constructed by machine learning method.The final outcome variable was whether ATB-DILI occurred,including the clinical features and laboratory test indexes of electronic medical record information,combined with the ATB-DILI susceptibility sites which were screened through verification,The first stage risk warning model of ATB-DILI was constructed by using lasso regression and single factor Logistic regression.The model with the best fit and the least number of predictors was selected by using the Akaike information criterion(AIC).ROC curve and calibration curve were used to evaluate the predictive effectiveness,discrimination and consistency of the model,and clinical decision curve was used to evaluate the clinical application value and applicability of the model.Results:2.1 General situation of study population The study included 1694 patients who received anti tuberculosis treatment(500patients with ATB-DILI and 1194 patients with ATB-DILI tolerance).Among them,113 patients in ATB-DILI group were excluded from the treatment scheme that did not meet the requirements and the liver function was abnormal before treatment.221 patients were excluded from the control group,2 patients with non-conforming SNP classification were excluded,and 1358 cases(385 were finally included patients with ATB-DILI,973 patients with ATB-DILI tolerance)were the final cohort of this study.2.2 Association between candidate SNPs and susceptibility to ATB-DILI The analysis of 30 target SNPs showed that: compared with G allele,T allele at rs1253618 of EDARADD gene had higher risk of ATB-DILI(OR = 1.260;95% CI:1.066-1.489,P = 0.007),and T allele at rs627341 of ZNF385 D gene had lower risk of ATB-DILI(OR = 0.790;95% CI: 0.623-1.003,P = 0.007),P values of the above loci were corrected by Bonfferoni.There was also significant difference between the two loci and ATB-DILI in genetic model analysis.The additive model(or = 1.256;95% CI: 1.063-1.483,P = 0.007)and dominance model(OR = 1.512;95% CI:1.149-1.988,P = 0.007)of EDARADD rs1253618 locus showed significant difference,the rs627341 locus of ZNF385 D gene showed additive model(OR =0.788;95% CI: 0.620-1.002,P = 0.052)and dominant model(OR = 0.742;95% CI:0.567-0.970,P = 0.029).The TT and GT genotypes at rs1253618 of EDARADD gene had higher alanine aminotransferase(ALT)level(P = 0.010)compared with GG genotype,and the TT genotype at rs627341 of ZNF385 D gene had higher C-reactive protein level(P = 0.025)and lower blood calcium level(P = 0.048)compared with CT and TT genotype.There was no significant difference between the other SNPs in the cohort(all P > 0.05).There was no significant difference in the distribution frequency of the above two SNPs among ATB-DILI patients with different severity and injury types(all P > 0.05).2.3 Machine learning algorithm integrates genome information and electronic medical records to build ATB-DILI early warning model In this study,electronic medical records combined with information characteristics of genomic SNPs were used as predictors of candidate variables.A total of 21 clinical features,45 laboratory test indexes and 30 SNPs were included as candidate variables.Using machine learning method,Lasso regression and single-factor logistic regression were used to screen out 18 candidate variables.Confounding factor analysis of candidate variables showed that there was no multicollinearity or interaction between candidate variables.1358 subjects were randomly divided into test set and validation set according to the ratio of 7:3,and 11 models were established by multivariate logistic regression method with test set data.The optimal model was selected according to the principle of goodness of fit and least predictor,including 8 predictors of fever,ALT,aspartate aminotransferase(AST),rs1253618 and hematocrit.The C-index of the prediction model in the test set is 0.836,the sensitivity is 59.90%,the specificity is 90.20%,the consistency test Sp =0.817,the maximum offset Emax = 0.094,and the average offset Eave = 0.012.In the validation set,C-index = 0.802,Sensitivity 69.50%,Specificity 91.00%,consistency test Sp = 0.830,maximum deviation Emax = 0.105,average deviation Eave = 0.014.The clinical decision-making curve shows that the prediction model has moderate clinical prediction applicability when the prediction risk threshold is between 0.1 and0.8.Part Ⅱ Study on the molecular mechanism of encoding ectodysplasin receptor associated death domain for anti-tuberculosis drugs-induced liver injury in tuberculosis patientsMaterials and Methods:1.The concentrations of INH(600 μ m),RIF(200 μ m)and INH + RIF(600 μ m +200 μ m)were divided into three groups: single group and combined group.The ATB-DILI cell model was established by treating Hep G2 and L-02 cells with INH(600 μ m),RIF(200 μ m)and INH + RIF(600 μ m + 200 μ m).2.CCK8 assay was used to detect the cell activity at different time phases,and the m RNA and protein expression levels of apoptosis related genes,including Caspase-3 and Caspase-9,were detected Objective to investigate the effect of EDARADD gene expression on the proliferation and apoptosis of ATB-DILI cells.3.EDARADD overexpression plasmid and si RNA were transfected to detect the inflammatory pathway of nuclear factor-65(NF-κ b).The expression levels of IL-6,IL-1 β,IL-10,IL-22,TNF-α and interferon-γ were measured γ.Objective to investigate the effect of EDARADD gene expression on the expression of NF-κ B p65 and inflammatory factors.4.To investigate the effect of EDARADD gene expression on oxidative stress pathway in ATB-DILI cells model,we transfected EDARADD overexpression plasmid and si RNA to detect the m RNA and protein expression levels of Nrf2 and antioxidant enzymes(SOD,CAT,GSH).Results:1.After INH + RIF was used on Hep G2 and L-02 cells for 24 hours,the expression levels of EDARADD m RNA and protein were detected.The results showed that EDARADD m RNA and protein were significantly high in the TB drug treatment group(P < 0.01).2.After transfection of EDARADD overexpression plasmid into ATB-DILI cells,the activity of ATB-DILI cells with EDARADD overexpression was significantly lower than that of the control group(P < 0.01);The activity of ATB-DILI cells with EDARADD si RNA was significantly higher than that of the control group(P < 0.01).3.EDARADD overexpression of ATB-DILI cells could enhance the expression levels of NF-κ B p65,IFN-γ,IL-1 β,TNF-α and IL-6,and down regulate the expression levels of IL-22 and IL-10(P < 0.01),while si RNA EDARADD overexpression of ATB-DILI cells could reduce the expression levels of NF-κ B The expression levels of p65,IFN-γ,IL-1 β,TNF-α and IL-6,and the expression levels of IL-22 and IL-10 were up-regulated(P < 0.01).4.Overexpression of EDARADD in ATB-DILI cells decreased the expression levels of Nrf2,SOD,cat and GSH,while si RNA EDARADD gene enhanced the expression levels of Nrf2,SOD,cat and GSH(P < 0.01).Conclusion:Based on the susceptibility SNPs of ATB-DILI screened by GWAS in the early stage of our research group,we further expanded the sample to carry out verification research in another independent sample population,deeply mined more detailed and comprehensive susceptibility loci related to ATB-DILI,and constructed an early warning model of ATB-DILI occurrence by using machine learning algorithm combined with electronic medical records and verified ATB-DILI susceptibility loci :1.EDARADD and ZNF385 D genes are associated with the occurrence of ATB-DILI in Chinese Han nationality.T allele at rs1253618 of EDARADD gene increases the risk of ATB-DILI while T allele at rs627341 of ZNF385 D gene decreases the relative risk of ATB-DILI.2.The early warning model of ATB-DILI,which was constructed by the laboratory index and EDARADD gene rs1253618,has good fitting and medium efficiency prediction value and potential for clinical application.3.In vitro experiments preliminarily confirmed the regulatory function of EDARADD in the occurrence and development of ATB-DILI: 1).Anti tuberculosis drugs INH and RIF can participate in the occurrence and development of ATB-DILI by up regulating the expression of EDARADD.2).EDARADD is involved in the occurrence and development of ATB-DILI by reducing cell activity and increasing apoptosis.3).EDARADD participates in the development of ATB-DILI by enhancing the expression of NF-κ B p65,IFN-γ,IL-1 β,TNF-α and IL-6,and down regulating the expression of IL-22 and IL-10.4).Cat and GSH are involved in the occurrence and development of ATB-DILI.