Radiomics Signature Predicts the Recurrence-Free Survival in Stage I Non-Small Cell Lung Cancer.

BACKGROUND

We aimed to explore the predictive value of radiomics signature for the recurrence-free survival (RFS) in patients with resected stage I non-small cell lung cancer.

METHODS

From January 2009 to December 2011, patients with resected stage I non-small cell lung cancer were divided into sub-solid and pure-solid groups according to presence of ground glass opacity in computed tomography. A total of 107 extracted radiomics features were reduced to 8 features by using LASSO Cox analysis to develop a radiomics signature for RFS prediction. Univariate and multivariate survival analyses were applied to identify independent prognostic variables, the Harrell concordance index (C-index) was measured to assess their prediction performance.

RESULTS

Our study included 378 patients with a median follow-up time of 63.2 months. The radiomics signature could stratify all patients into high-risk (180 of 378) and low-risk group (198 of 378) with different RFS (P < .001). In the sub-solid group (n = 115), 3 patients who occurred relapse were categorized into the high-risk group by the radiomics signature. In the pure-solid group, patients with low risk (141 of 263) had a better outcome than those with high risk (122 of 263) (P < .001). Multivariate analyses revealed that the histology (P < .001) and the developed radiomics signature (P < .001) remained independent prognostic factors for RFS.

CONCLUSIONS

Radiomics signature may be an independent imaging biomarker for predicting the survival, which may guide for personalizing treatment option in patients with stage I non-small cell lung cancer.