Background
The value of plasma threonine, cysteine, and piperonamide as diagnostic biomarkers for non-small cell lung cancer (NSCLC) has been rarely explored. The lack of a validation set containing confounders is common to most previous metabolomics studies. The
Conclusions
Plasma amino acids and piperonamide have potential as diagnostic biomarkers in NSCLC. This metabolic biomarker panel appears useful for the diagnosis and screening of NSCLC. In addition, metabolomic and transcriptomic integration pathway analysis may help elucidate the mechanism of NSCLC occurrence and development and even reveal new treatment vulnerabilities.
Methods
A total of 250 participants were included in this study, including 167 patients with pathologically confirmed NSCLC and 83 healthy controls (HCs). These participants were divided into training set, validation set 1, and validation set 2 in chronological order and in a certain proportion. The plasma levels of 22 amino acids and 1 piperonamide in these pre-treatment NSCLC patients and HCs were measured by LC-MS/MS. Metabolic biomarkers were identified after multivariate analysis, univariate analysis, receiver operating characteristic (ROC) analysis. Furthermore, these biomarkers and transcriptomic data were subjected to joint pathway analysis.
Results
The area under the ROC curve (AUC) values for threonine, piperonamide, arginine, alanine, cysteine, methionine, and histidine in the integrated data set were 0.911, 0.848, 0.909, 0.869, 0.786, 0.597 and 0.637, respectively. This panel composed of these 7 metabolites showed good diagnostic capability for NSCLC (the AUC of this diagnostic panel in each data set was greater than 0.9). The specificity of this diagnostic panel in validation set 2, which included confounders, was 0.970, similar to that of the other datasets. The presence of confounding factors had little effect on the diagnostic accuracy of this panel. The ROC analysis of this diagnostic panel between all stage I NSCLC patients and HCs showed AUC, sensitivity, and specificity of 1.000, 1.000, and 0.988, respectively. Moreover, PSAT1, SHMT2, AOC3, and MAOB were found to be involved in the metabolism of threonine and cysteine. Conclusions: Plasma amino acids and piperonamide have potential as diagnostic biomarkers in NSCLC. This metabolic biomarker panel appears useful for the diagnosis and screening of NSCLC. In addition, metabolomic and transcriptomic integration pathway analysis may help elucidate the mechanism of NSCLC occurrence and development and even reveal new treatment vulnerabilities.