Rapid diagnosis of drug-resistant and drug-sensitive tuberculosis using mass spectrometry and machine learning

*Important Notices: The Lancet / SSRN preprint We publish a non-peer-reviewed, preliminary scientific report and should not be taken as conclusive, to guide clinical practice/health-related actions, or to be treated as established information.

A recent study published in Preprints of The Lancet found that SSRN* A team of Chinese researchers developed a machine learning-based diagnostic tool to detect tuberculosis and drug-resistant tuberculosis using nanoparticle-enhanced laser desorption/ionization mass spectrometry (NPELDI MS) and metabolic fingerprinting from serum samples. has been identified.

study: Machine learning for serum metabolic fingerprinting for the diagnosis of tuberculosis and drug-resistant tuberculosis: an observational study.Image Credit: Kateryna Kon / Shutterstock

Background

After coronavirus disease 2019 (COVID-19), tuberculosis is the most common cause of death from infectious agents and the emergence of drug-resistant bacteria Mycobacterium tuberculosis exacerbated public health concerns. Unfortunately, statistics show that nearly 40% of his tuberculosis cases are not diagnosed in time for early treatment.

Current tuberculosis detection methods include immunological tests, pathogen detection from sputum smears, and molecular biological methods. The sputum smear test is rapid but has low specificity and sensitivity in detecting tuberculosis, mycobacterial culture is accurate but has a long processing time, and is difficult for large-scale same-day testing of tuberculosis patients in general hospitals. . Furthermore, molecular diagnosis of tuberculosis and rifampicin-resistant tuberculosis from sputum samples presents challenges due to its high cost and variable susceptibility to extrapulmonary tuberculosis. Therefore, a highly sensitive and specific viable test method that can be used to rapidly detect TB and drug-resistant TB is essential.

About research

In this study, researchers designed an NPELDI MS platform equipped with machine learning algorithms to simultaneously detect TB and drug-resistant TB based on metabolome fingerprints of serum samples.when Mycobacterium tuberculosis Metabolites that parasitize host macrophages, influence host metabolism, and are formed from a variety of metabolic reactions exhibit responses to environmental, proteomic, and genomic changes. Identification and quantification of these metabolic fingerprints can be used for detection and diagnosis of various diseases.

For this observational study, researchers recruited 110 pulmonary tuberculosis patients and 118 healthy individuals between 2020 and 2021. Mycobacterium tuberculosis culture, Mycobacterium tuberculosis nucleic acid Detection, chest radiography, and lung histopathologic diagnosis.Drug susceptibility testing or Gene Xpert testing for rifampicin resistance Mycobacterium tuberculosis It was used to classify tuberculosis patients.

Serum samples were collected for NPELDI MS analysis and metabolic fingerprints were determined. This was processed using machine learning algorithms to identify biomarkers for drug-susceptible and drug-resistant tuberculosis.

result

As a result, it was reported that the NPELDI MS-based machine learning method could distinguish tuberculosis patients from healthy subjects with 85% sensitivity and 100% specificity. The method was also able to distinguish between rifampicin-sensitive and rifampicin-resistant TB patients with a sensitivity of 87.5% and a specificity of 85.7%.

Metabolic biomarkers used to detect tuberculosis and drug-resistant tuberculosis consisted of lipids such as monoglycerides, phosphatidylcholines, ceramides, triglycerides, cholesteryl esters, amino acids, phosphates, octacosanoic acid, and other basic compounds. rice field. Through biomarker analysis, Mycobacterium tuberculosis Infections dysregulate lipid metabolic pathways such as sphingolipid and glycerophospholipid metabolism. Phosphates such as nicotinamide adenine dinucleotide phosphate and all-trans-heptaprenyl diphosphate were aberrantly expressed, and glutathione levels were low in tuberculosis patients.

Biomarkers used to distinguish between rifampicin-susceptible and rifampicin-resistant TB patients included uric acid, taurine, ascorbic acid, and homocysteine, which were elevated in rifampicin-resistant TB patients. Researchers believe that sulfur amino acids, such as homocysteine ​​and taurine, may be indicative of drug-resistant tuberculosis because they are associated with antioxidant and membrane-stabilizing effects. , is thought to protect the liver from the toxic effects of antituberculous drugs such as rifampicin and isoniazid. Homocysteinemia, elevated homocysteine ​​levels in serum or plasma, is often reported during antituberculous therapy.

This study had some limitations, including the small sample size of patients with rifampicin-resistant tuberculosis. This is because only 58 serum samples were included in the analysis to determine biomarkers for distinguishing between drug-resistant and drug-sensitive tuberculosis patients. It was from a rifampicin-sensitive patient. This may affect diagnostic accuracy.

Conclusion

In summary, our findings identified a biomarker panel and a serum metabolic fingerprint to diagnose tuberculosis from serum samples with high specificity and sensitivity, and to distinguish between rifampicin-resistant and rifampicin-sensitive patients. A machine learning algorithm using optimized iron particles and the NPELDI MS method can help detect tuberculosis early and accurately, improving prognosis and care for tuberculosis patients.

*Important Notices: The Lancet / SSRN preprint We publish a non-peer-reviewed, preliminary scientific report and should not be taken as conclusive, to guide clinical practice/health-related actions, or to be treated as established information.