A novel radiomics model combining GTVp, GTVnd, and clinical data for chemoradiotherapy response prediction in patients with advanced NSCLC

TL;DR

This study developed a new radiomics model combining tumor volume data and clinical information to better predict treatment response in advanced lung cancer patients.

Contribution

A novel multimodal radiomics model integrating GTVp, GTVnd, and clinical data for improved chemoradiotherapy response prediction in NSCLC.

Findings

The GTVp-based model had an AUC of 0.855 in training and 0.775 in validation.

The multimodal model achieved a higher validation AUC of 0.863.

Combining GTVp, GTVnd, and clinical data improved predictive performance over GTVp-only models.

Abstract

Numerous radiomic models have been developed to predict treatment outcomes in patients with NSCLC receiving chemotherapy and radiation therapy. However, computed tomography (CT) radiomic models that integrate the Gross Tumour Volume of the primary lesion (GTVp), the Gross Tumour Volume of nodal disease (GTVnd), and clinical information are relatively scarce and may offer greater predictive accuracy than models focusing on GTVp alone. This study aimed to evaluate the efficacy of a CT radiomic model combining GTVp, GTVnd, and clinical data for predicting treatment response in unresectable stage III–IV NSCLC patients undergoing concurrent chemoradiotherapy.

A total of 101 patients with unresectable stage III–IV NSCLC were included. GTVp was delineated using lung windows, and GTVnd was delineated using mediastinal windows. Radiological features were extracted using Python 3.6, then subjected to F-test and Lasso regression for feature selection. Logistic regression was performed on the selected radiological features. Clinical information was analysed with univariate and multivariate logistic regression to identify significant clinical variables. Five models were developed and evaluated, incorporating GTVp, GTVnd, and clinical data.

The GTVp-based radiomics model achieved an area under the curve (AUC) of 0.855 in the training cohort and 0.775 in the validation cohort. The multimodal composite model (integrating GTVp, GTVnd, and clinical parameters) significantly outperformed the GTVp-only model, with a training AUC of 0.862 and validation AUC of 0.863, demonstrating superior predictive performance for concurrent chemoradiotherapy response in this patient population.

Figures