Purpose: Small cell lung cancer (SCLC) is an aggressive and rapidly progressive malignant tumor characterized by a poor prognosis. Chemotherapy remains the primary treatment in clinical practice; however, reliable biomarkers for predicting chemotherapy outcomes are scarce.
Methods: In this study, 78 SCLC patients were stratified into "good" or "poor" prognosis cohorts based on their overall survival (OS) following surgery and chemotherapeutic treatment. Next-generation sequencing was employed to analyze the mutation status of 315 tumorigenesis-associated genes in tumor tissues obtained from the patients. The random forest (RF) method, validated by the support vector machine (SVM), was utilized to identify single nucleotide mutations (SNVs) with predictive power. To verify the prognosis effect of SNVs, samples from the cbioportal database were utilized.
Results: The SVM and RF methods confirmed that 20 genes positively contributed to prognosis prediction, displaying an area under the validation curve with a value of 0.89. In the corresponding OS analysis, all patients with SDH, STAT3 and PDCD1LG2 mutations were in the poor prognosis cohort (15/15, 100%). Analysis of public databases further confirms that SDH mutations are significantly associated with worse OS.
Conclusion: Our results provide a potential stratification of chemotherapy prognosis in SCLC patients, and have certain guiding significance for subsequent precise targeted therapy.
Keywords: Chemotherapy prognostic; Next generation sequencing; Random forest; SCLC; SDH mutations.
© 2023. The Author(s).