1272P - Postoperative survival prediction in non-small cell lung cancer patients based on driver genotypes

Background

Non-small cell lung cancer (NSCLC) is a heterogeneous disease with various driver gene mutations. Currently, it remains unknown the differences in postoperative survival among patients with different genotypes. We previously constructed a clinical nomogram that effectively predicts postoperative survival in NSCLC patients (C-index 0.71), but whether integration of gene classification improves its efficacy remains unclear.

Methods

We collected data from patients representing six distinct driver genotypes at the First Affiliated Hospital of Guangzhou Medical University. Clinical characteristics and prognosis data were collected. We compared the overall survival, disease-free survival, and recurrence-free survival among different gene types. The previous prediction model was externally validated using the above data, and its predictive performance was assessed in populations with different genotypes. Furthermore, we constructed a novel survival prediction model by integrating gene subtyping and compared its predictive performance with that of the original model.

Results

1397 patients were enrolled in the analysis. Significant differences were observed in overall survival (P<0.05) and disease-free survival (P<0.05) and recurrence-free survival(P<0.05) among patients with different driver genotypes. The predictive model demonstrated good generalizability within our cohort (C-index: 0.71, 95% CI 0.68 to 0.74). However, the performance of the model varied across different genotypes. The clinical factor prediction model exhibited favorable applicability in patients with EGFR L858R mutation, EGFR 19 deletion mutation, and wild-type, but were not found in patients with rare EGFR mutations, RAS/BRAF mutations and fusion mutations. The new prediction model integrating genotypes slightly improves the performance of the previous clinical model (C-index: 0.74 vs 0.71).

Conclusions

Patients with different genotypes display diverse prognostic outcomes. The predictive model maintains practice predictive performance in certain driver gene subtypes. Furthermore, the integration of genotyping with clinical risk factors yields slight improvements in prediction performance.

Clinical trial identification

Legal entity responsible for the study

The authors.

Funding

Has not received any funding.

Disclosure

All authors have declared no conflicts of interest.