Development and validation of a risk-stratification model for individualized management of leptomeningeal metastases in EGFR- mutant NSCLC (LM-Index).

PURPOSE: Prognosis for patients with EGFR- mutant non-small-cell lung cancer and leptomeningeal metastasis is highly uncertain, complicating treatment decisions. We aimed to develop and validate a multimodal risk score for predicting overall survival to enable risk-stratified management.

METHODS

In this retrospective, multicenter study, a derivation cohort (n = 350) and an independent external validation cohort (n = 302) were used. Independent prognostic factors were identified via Cox regression and integrated into an integer-based risk score. Model performance was evaluated by the C-index, calibration, and decision curve analysis.

RESULTS

Multivariate analysis identified five independent predictors of poorer OS: ECOG PS ≥ 3, brain metastasis, meningeal enhancement on MRI, positive Cerebrospinal Fluid cytology (CSFC) and intracranial pressure > 220 mmH2O. The resulting risk score stratified patients into low-risk and high-risk groups, with median OS of 22.6 months versus 9.9 months, respectively (p < 0.001). The model demonstrated good discrimination, with a bias-corrected C-index of 0.73. In the external validation cohort, all factors remained significant, and the model maintained consistent performance (C-index = 0.71). The model showed good calibration and positive net benefit.

CONCLUSION

We developed and validated a robust, clinically accessible risk score that accurately stratifies OS in patients with EGFR- mutant NSCLC and LM. This practical tool facilitates risk‑stratified patient counseling and may aid in the design of future clinical trials.