Exploration Of Pathogenic Microbial Detection Model Based On Mass Spectrometry Analysis Of Metabolic Gas VOCs And Its Diagnostic Value For Pulmonary Tuberculosis

Objective:Tuberculosis(TB)is a global health problem that affects millions of people each year.Current methods for TB diagnosis have limitations,including low sensitivity and specificity,time-consuming,requiring invasive procedures and being limited by samples.In recent years,the analysis of volatile organic compounds(VOCs)in metabolic gases has emerged as a promising method for the detection of disease-causing microorganisms.The aim of this study was to explore a model for the detection of pathogenic microorganisms based on metabolic gas VOCs by high pressure photoionization time-of-flight mass spectrometry(HPPI-TOFMS)technology and to assess its potential value for the diagnosis of tuberculosis.Methods:First,we cultured common pathogens of clinical pulmonary infection,including Mycobacterium tuberculosis,Staphylococcus aureus,Klebsiella pneumoniae and Escherichia coli,established the correct culture method and collected the metabolic VOCs produced by the strains.Second,we recruited 500 participants,including 250 patients with pulmonary tuberculosis and 250 healthy controls,and participants provided exhaled breath samples.The spectrum of VOCs in exhaled breath was analyzed by HPPI-TOFMS technology.Multiple statistical methods such as principal component analysis(PCA),logistic regression(LR)model,and machine learning algorithm were used to analyze the data to establish a pathogenic microorganism detection model,and the sensitivity,specificity,and area under the curve(AUC)of the model were calculated to evaluate its diagnostic performance.In addition,we conducted a comparative study on the detection of pathogenic microbial metabolic gas and host exhaled gas spectrum.With the support of HMPITOFMS,characteristic volatile biomarkers were mined from VOCs and applied to the identification of common pathogenic microorganisms,and machine learning algorithm models and validation methods were established based on the detection mass spectrum data.The specific biomarkers of target pathogenic microorganisms were extracted by the characteristics of metabolic gas VOCs,and biochemical analysis and identification were carried out.Finally,the detection model of pathogenic microorganism metabolic gas was transformed and applied to the rapid diagnosis of clinical tuberculosis.Results:We found significant differences in metabolic gas VOCs between Mycobacterium tuberculosis and three other common pulmonary infections,and they could also be clearly distinguished by the established pathogenic microbial detection model.The sensitivity of the differentiation model for Mycobacterium tuberculosis and Staphylococcus aureus was 0.943 with a specificity of 1.000,the sensitivity of the differentiation model for Mycobacterium tuberculosis and Klebsiela pneumoniae was 0.987 with a specificity of 1.000,and the sensitivity and specificity of the differentiation model for Mycobacterium tuberculosis and Escherichia coli was 0.943 with a specificity of 1.000,both with a sensitivity and specificity of more than 90%.In addition,we found significant differences in the exhaled gas VOCs profiles between TB patients and healthy controls,and we analyzed the differential VOCs produced by Mycobacterium tuberculosis against the exhaled gas VOCs of TB patients,and found that three VOCs(i.e.,m/z104,m/z113,and m/z127)were able to distinguish TB patients from healthy controls significantly.A logistic regression model score was further developed,and the results showed that the AUC value of this assay model for TB diagnosis was 0.76(95% confidence interval 0.7263-0.8112),and the sensitivity and specificity of the model reached 63% and 86%,respectively,indicating that m/z104,m/z113,and m/z127 can be used as potential diagnostic biomarkers for TB.Conclusion:The metabolic gas VOCs of Mycobacterium tuberculosis and three other commonly infected bacteria in the lung were similar and different;the exhaled gas VOCs of TB patients and healthy controls were also similar and different;the three VOCs identified(m/z104,m/z113,m/z127)were expected to be diagnostic markers for TB,and the constructed models also demonstrated better diagnostic performance.This indicates that the VOCs-based HPPI-TOFMS assay model is an efficient,non-invasive and rapid pathogenic microbial detection method,which is expected to be an important tool for TB diagnosis.Overall,our study provides new insights into the diagnosis of tuberculosis and may contribute to the development of more accurate and effective tuberculosis detection methods.