Screening And Evaluation Of New Biomarkers For Tuberculosis Based On Low-abundance Proteomics In Plasma

Background Tuberculosis(TB)is an ancient disease.Archeologists have found spinal tuberculosis from Neolithic human bone fossils and mummies from Egypt 4,500 years ago.According to the World Health Organization,tuberculosis is still one of the Top ten causes of death globally,and it is also the main single source of infection.In 2018,an estimated 10 million people worldwide had TB——including 5.7 million men,3.2 million women,and 1.1 million children——of which there is a total of 1.5 million people died of TB.The prevention and control of tuberculosis is still a challenging issue for the global health system,and new methods and ideas are urgently needed.At present,there is a clinical need for a diagnostic method that can efficiently,accurately and quickly target patients with tuberculosis.Purpose We want to identify a TB diagnosis biomarker or a group of biomarker combinations by identifying low-abundance proteomics analysis of plasma from TB patients and healthy controls.Methods(1)After removing high-abundance proteins from the plasma of 18 tuberculosis patients and 18 healthy controls,differentially expressed proteins were analyzed using TMT-labeled quantitative proteomics.(2)Relying on bioinformatics methods,we performed protein annotation,functional enrichment analysis and protein interaction network analysis on differentially expressed proteins.(3)According to the technical characteristics of parallel response monitoring(PRM)targeted proteomics,we selected suitable differentially expressed proteins for PRM verification.Through receiver operating characteristic(ROC)curve analysis and leave-one-out cross-validation(LOOCV)method,we analyzed every single biomarkers and potential biomarker combinations.(4)Using enzyme-linked immunosorbent assay(ELISA),we further validated potential biomarkers.Through ROC curve analysis and LOOCV method,we obtained candidate biomarkers and biomarker combinations.Results(1)TMT-labeled proteomics analysis showed that,under conditions of P<0.05,fold change>1.2 or<0.8333,there were 43 up-regulated proteins and 71 down-regulated proteins in the tuberculosis group compared with healthy controls.In bioinformatics analysis,the most significant category was "hsa04610 complement and coagulation cascade pathway",which is enriched for 16 differentially expressed proteins.(2)A set of 7 differentially expressed proteins with diagnostic potential for tuberculosis were verified using PRM-targeted mass spectrometry,namely A2M,APOA4,CFH,CFHR5,FGB,FGG,and MBL2.ROC curve analysis and LOOCV method were used to analyze and verify the performance of each protein and each combination.In ROC analysis,CFHR5+FGB+FGG+MBL2 is the optimal biomarker combination,its area under the curve(AUC)of ROC analysis is 0.868,sensitivity is 0.733,and specificity is 0.889.After LOOCV analysis,the optimal biomarker combination A(A2M+APOA4+CFHR5+FGB+FGG)was obtained.It has 80%of the original grouped cases correctly classified,70%of the cross-validated grouped cases correctly classified,80%Correctly classified original grouped TB cases,and 66.7%correctly classified cross-validated grouped TB cases.Combining the two analysis methods,A2M,APOA4,CFHR5,FGB,FGG,and MBL2 were the most promising candidate protein biomarkers.(3)Nine proteins were involved in the ELISA verification experiment:C4d,C5a,C9,CFH,FBG(composed of two groups of FGA,FGB,FGG polypeptide chains),IP-10,MBL2,SAA2,and SERPINA1.Using ROC curve analysis and LOOCV method,we analyzed and verified the individual performance and random combinations of nine proteins.In the ROC analysis,MBL2 and C5a were significantly different between the tuberculosis group and the non-tuberculous group,the AUCs of MBL2 and C5a were both more than 0.7.The diagnostic sensitivity of MBL2 was 0.579 and the specificity was 0.828;the diagnostic sensitivity of C5a was 0.833 and the specificity was 0.615.In the LOOCV analysis,the best performing single biomarker was still MBL2,with 73.7%correctly classified original grouped cases and 73.7%correctly classified cross-validated grouped cases;Biomarker combination 1(C4d+C5a+CFH+FBG+IP-10+MBL2+SAA2+SERPINA1)had the highest comprehensive score,with 61.4%of the original grouped cases correctly classified,89.5%of the correctly classified cross Validated grouped cases,89.5%of correctly classified original grouped TB cases and 78.9%of correctly classified cross-validated grouped TB cases;Biomarker combination 2(C4d+C5a+C9+CFH+FBG+IP-10+MBL2+SAA2)got the highest scores of all combinations in correctly classified cross-validated grouped cases and correctly classified crossvali-dated grouped tuberculosis cases,they were 93%and 84.2%,respectively.Combining the two analysis methods,C4d,C5a,C9,CFH,FBG,IP-10,MBL2,SAA2,and SERPINA1 were the most promising candidate protein biomarkers.Conclusion(1)Based on TM T-labeled quantitative proteomics technology,we analyzed the low abundance proteins in the plasma of the tuberculosis group and healthy control group.114 differentially proteins were identified in tuberculosis group,including 43 up-regulated proteins and 71 down-regulated proteins.(2)Validated by PRM-targeted proteomics,A2M,APOA4,CFHR5,FGB,FGG,and MBL2 were the most potential candidate proteins for tuberculosis biomarkers;the optimal combination of tuberculosis biomarkers was CFHR5+FGB+FGG+MBL2(AUC=0.868,sensitivity=0.733,specificity=0.889).(3)We used ELISA to further verify the possible biomarkers.C4d,C5a,C9,CFH,FBG,IP-10,MBL2,SAA2,and SERPINA1 were the most potential biomarker candidate proteins;we identified two groups of ideal TB biomarker combinations,which were biomarker combination 1(C4d+C5a+CFH+FBG+IP-10+MBL2+SAA2+SERPINA1)and biomarker combinat-ion 2(C 4d+C5a+C9+CFH+FBG+IP-10+MBL2+SAA2).(4)We have confirmed that MBL2 is a significantly different TB biomarker through PRM-targeted proteomics techology and ELISA verification methods,which provides strong support for subsequent functional exploration and disease progression research.