Analysis and prediction of pathological risk factors for lung cancer by plasma metabolomics

Abstract

Background

Common pathological risk factors for lung cancer include lymphovascular invasion (LVI), spread through air space (STAS) and perineural invasion (PNI). They are important factors that cause tumor recurrence after surgery, and also reference indices to determine the scope of surgical resection. At present, there is a lack of effective methods for preoperative judgment of the type and number of risk factors.

Methods

In this study, lung cancer was divided into three groups according to the number of pathological risk factors. Metabolites that can identify the number of risk factors were selected. Their differential diagnosis ability was evaluated by the area under ROC curve, sensitivity, specificity, Cohen's Kappa and F1 index.

Results

Among 160 lung cancer, 119 had no risk factors, 24 had one risk factor, and 17 had two or more risk factors. After the comparisons of these three groups with healthy volunteers, five metabolites with differential diagnosis ability were screened out. The indicators to assess their discriminative ability were satisfactory.

Conclusion

Metabolites discovered by metabolomics can identify the number of pathological risk factors associated with lung cancer. It is helpful to determine the extent of lung resection before surgery.