Enhancing prognostic accuracy in unresectable esophageal squamous cell carcinoma: Integrating CT radiomics with clinical data

Background Esophageal squamous cell carcinoma (ESCC) presents significant management challenges, particularly in patients who do not undergo esophagectomy, due to high rates of locoregional failure, distant metastases, and resulting mortality. Purpose To predict overall survival (OS), locoregional failure (LRF), and distant metastasis-free survival (DMFS) in non-surgically treated ESCC patients. Methods We retrospectively analyzed 125 patients with ESCC treated with radiotherapy from July 2010 to July 2019. Radiomic features were extracted from pre-treatment CT simulator images, which were selected through intraclass correlation and univariate logistic regression. The top 20% of features with the highest area under the receiver operating characteristic curve (AUC) underwent 20 repetitions of multivariate analysis with recursive feature elimination and cross-validation (RFECV). This analysis also integrated clinical data including age, sex, body weight, tumor location, tumor length, histologic tumor grade, T-stage and N-stage, along with treatment data such as chemotherapy status, prescribed radiation dose, number of treatment fractions, and dose per fraction to improve predictive accuracy. Results Radiomic models demonstrated superior prognostic performance, especially when combined with clinical or treatment data. The best model, which incorporated radiomics and clinical characteristics, produced AUC values ranging from 0.746 (±0.103) to 0.833 (±0.090) in validation set. This significantly exceeded the performances of traditional models, whose AUC values ranged from 0.532 (±0.137) to 0.659 (±0.099). Conclusions CT-based radiomic features, particularly when combined with clinical features, significantly improved prognostic performance over traditional clinical or treatment features. These findings support the potential of integrated radiomic approaches for risk stratification in ESCC and warrant external validation before clinical implementation.

Authors

Institutions

Publication Details

Journal
European Journal of Radiology Open
Published
2026-11-19
DOI
https://doi.org/10.1016/j.ejro.2026.100828
Primary Topic
Radiomics and Machine Learning in Medical Imaging
Type
article
Field-Weighted Citation Impact
0.00

Funders

Controls
|||
ALL TIME
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
OCT
NOV
article

Enhancing prognostic accuracy in unresectable esophageal squamous cell carcinoma: Integrating CT radiomics with clinical data

Chawalit Lertbutsayanukul, Anussara Prayongrat, Sira Sriswasdi, Chakkapong Chakkabat et al.
European Journal of Radiology Open
Radiomics and Machine Learning in Medical Imaging
article

Enhancing prognostic accuracy in unresectable esophageal squamous cell carcinoma: Integrating CT radiomics with clinical data

Chawalit Lertbutsayanukul, Anussara Prayongrat, Sira Sriswasdi, Chakkapong Chakkabat, Danita Kannarunimit, Yothin Rakvongthai, Sarin Kitpanit, Sararas Khongwirotphan, Thanakrit Chanchayanon
article en

Abstract

Background Esophageal squamous cell carcinoma (ESCC) presents significant management challenges, particularly in patients who do not undergo esophagectomy, due to high rates of locoregional failure, distant metastases, and resulting mortality. Purpose To predict overall survival (OS), locoregional failure (LRF), and distant metastasis-free survival (DMFS) in non-surgically treated ESCC patients. Methods We retrospectively analyzed 125 patients with ESCC treated with radiotherapy from July 2010 to July 2019. Radiomic features were extracted from pre-treatment CT simulator images, which were selected through intraclass correlation and univariate logistic regression. The top 20% of features with the highest area under the receiver operating characteristic curve (AUC) underwent 20 repetitions of multivariate analysis with recursive feature elimination and cross-validation (RFECV). This analysis also integrated clinical data including age, sex, body weight, tumor location, tumor length, histologic tumor grade, T-stage and N-stage, along with treatment data such as chemotherapy status, prescribed radiation dose, number of treatment fractions, and dose per fraction to improve predictive accuracy. Results Radiomic models demonstrated superior prognostic performance, especially when combined with clinical or treatment data. The best model, which incorporated radiomics and clinical characteristics, produced AUC values ranging from 0.746 (±0.103) to 0.833 (±0.090) in validation set. This significantly exceeded the performances of traditional models, whose AUC values ranged from 0.532 (±0.137) to 0.659 (±0.099). Conclusions CT-based radiomic features, particularly when combined with clinical features, significantly improved prognostic performance over traditional clinical or treatment features. These findings support the potential of integrated radiomic approaches for risk stratification in ESCC and warrant external validation before clinical implementation.

European Journal of Radiology OpenVol. 17
Chulalongkorn University (TH), King Chulalongkorn Memorial Hospital (TH)
Chulalongkorn University, Faculty of Medicine, Chulalongkorn University
Good health and well-being
Radiomics and Machine Learning in Medical Imaging
AI Navigator

Ask Laika to Summarize, Analyze, and Connect papers live on the map.

Summarize Papers & Methodologies

Extract key findings, datasets, and comparative methods across publications.

Benchmark Rankings & Visual Analytics

Rank top research institutions, authors, funders, topics, and journals by Field-Weighted Citation Impact (FWCI) and paper volume with instant charts.

Connect Distant Disciplines

Bridge topological clusters on the map to find hidden collaborative intersections.

Laika access coming soon