Texture analysis of advanced non-small cell lung cancer (NSCLC) on contrast-enhanced computed tomography: prediction of the response to the first-line chemotherapy.

Abstract

To assess whether tumour heterogeneity, quantified by texture analysis (TA) on contrast-enhanced computed tomography (CECT), can predict response to chemotherapy in advanced non-small cell lung cancer (NSCLC). Fifty-three CECT studies of patients with advanced NSCLC who had undergone first-line chemotherapy were retrospectively reviewed. Response to chemotherapy was evaluated according to RECIST1.1. Tumour uniformity was assessed by a TA method based on Laplacian of Gaussian filtering. The resulting parameters were correlated with treatment response and overall survival by multivariate analysis. Thirty-one out of 53 patients were non-responders and 22 were responders. Average overall survival was 13 months (4-35), minimum follow-up was 12 months. In the adenocarcinoma group (n = 31), the product of tumour uniformity and grey level (GL*U) was the unique independent variable correlating with treatment response. Dividing the GL*U (range 8.5-46.6) into tertiles, lesions belonging to the second and the third tertiles had an 8.3-fold higher probability of treatment response compared with those in the first tertile. No association between texture features and response to treatment was observed in the non-adenocarcinoma group (n = 22). GL*U did not correlate with overall survival. TA on CECT images in advanced lung adenocarcinoma provides an independent predictive indicator of response to first-line chemotherapy.