Predicting chemotherapy response in small cell lung cancer using habitat imaging and peritumoral radiomics: a dual-center study
This study developed and validated a CT-based model combining habitat imaging and peritumoral radiomics to predict platinum-based chemotherapy response in small cell lung cancer (SCLC), and determined the optimal peritumoral margin. This dual-center retrospective study enrolled 315 SCLC patients receiving first-line platinum-based chemotherapy, divided into chemotherapy-sensitive and chemotherapy-resistant groups based on RECIST 1.1. Patients from Center 1 ( n = 212) were split into training and internal validation sets (8:2), and those from Center 2 ( n = 103) formed the external validation set. Clinical-imaging features were selected via univariate and multivariate logistic regression to build the Clinical model. Three peritumoral margins (0–3 mm, 0–6 mm, -3 to + 3 mm) and intratumoral habitats were analyzed. Features were selected via LASSO regression. Finally, an integrated model was built by combining habitat features, peritumoral features, and clinical-imaging factors. Model performance was evaluated using receiver operating characteristic (ROC) curves, decision curve analysis (DCA), and the Hosmer-Lemeshow goodness-of-fit test to assess discrimination, clinical utility, and calibration. The 0–3 mm peritumoral margin was optimal, with the model achieving AUCs of 0.851, 0.790, and 0.750 in the training, internal validation, and external validation sets, respectively. The habitat model showed promising performance with AUCs of 0.978, 0.873, and 0.882. The nomogram combined clinical-imaging, 0–3 mm peritumoral, and habitat features, achieving AUCs of 0.987, 0.909, and 0.911 across the three datasets. This nomogram shows promise for predicting SCLC chemotherapy response and may facilitate clinical translation.
Authors
- Meiyan Liao (ORCID: https://orcid.org/0000-0002-9515-6635)
- Jingting Wang (ORCID: https://orcid.org/0000-0002-1028-177X)
- Yonggang Li (ORCID: https://orcid.org/0000-0001-8974-3094)
- Tian Gan (ORCID: https://orcid.org/0000-0001-6235-5278)
- Feng Xiao (ORCID: https://orcid.org/0000-0003-3289-3459)
- Ting Li (ORCID: https://orcid.org/0000-0003-3219-1545)
- Yun Long
Institutions
- Zhongnan Hospital of Wuhan University (CN)
- First Affiliated Hospital of Soochow University (CN)
Publication Details
- Journal
- Scientific Reports
- Published
- 2026-11-13
- DOI
- https://doi.org/10.1038/s41598-026-75634-6
- Primary Topic
- Radiomics and Machine Learning in Medical Imaging
- Type
- article
- Field-Weighted Citation Impact
- 0.00
Funders
- Zhongnan Hospital of Wuhan University