Prognostic value of MRI radiomics in locally advanced nasopharyngeal carcinoma treated with cisplatin chemoradiotherapy

This study evaluates the efficacy of cisplatin-based concurrent chemoradiotherapy (CCRT) for locally advanced nasopharyngeal carcinoma (LA-NPC) and assesses the prognostic value of MRI texture analysis. In this retrospective cohort study, 100 consecutive LA-NPC patients treated between February 2021 and January 2024 were retrospectively enrolled and allocated based on treatment received: radiotherapy alone (control group, n = 50) or cisplatin CCRT (study group, n = 50). We compared treatment efficacy (objective response rate [ORR] per RECIST 1.1), pain scores (Visual Analogue Scale), anxiety/depression scores (SAS/SDS), survival outcomes (Kaplan-Meier with log-rank tests), and adverse reactions (CTCAE v4.0). Additionally, in a separate radiomics cohort of 60 LA-NPC patients, we performed 3D whole-tumor segmentation on pretreatment MRI images, extracted 396 radiomic features using AI-Kit software, applied LASSO regression for feature selection, and evaluated model performance using 5-fold cross-validation with intra- and inter-observer reproducibility assessment (ICC >0.75). The CCRT group showed significantly higher ORR (82.0% vs. 56.0%; P = 0.006). Patients receiving CCRT experienced lower pain scores at 1-, 2-, and 3-weeks post-treatment, and greater reductions in anxiety and depression (P < 0.001). Survival analysis revealed superior outcomes for the CCRT group: median overall survival (223 vs. 177 days; HR = 0.68, 95% CI: 0.46-0.99, P = 0.048), local recurrence-free survival (68 vs. 47 days; HR = 0.62, 95% CI: 0.40-0.96, P = 0.032), and distant metastasis-free survival (184 vs. 107 days; HR = 0.54, 95% CI: 0.32-0.91, P = 0.018). Adverse reactions were comparable between groups (22.0% vs. 32.0%; P = 0.261). MRI texture analysis identified two predictive features: 135-degree cluster shadow-step 4 (AUC = 0.861, 95% CI: 0.786-0.936) and full-angle correlation variance-step 1 (AUC = 0.758, 95% CI: 0.654-0.862). The combined model achieved an AUC of 0.928 (95% CI: 0.876-0.979; sensitivity: 86.2%; specificity: 83.9%) with good calibration (Hosmer-Lemeshow P = 0.742). Cisplatin-based CCRT significantly improves survival and quality-of-life outcomes in LA-NPC without increasing adverse effects. MRI texture features serve as a promising non-invasive biomarker for predicting treatment response, though prospective validation is required.

Authors

Institutions

Publication Details

Journal
Journal of Radiation Research and Applied Sciences
Published
2026-09-16
DOI
https://doi.org/10.1016/j.jrras.2026.102691
Primary Topic
Radiomics and Machine Learning in Medical Imaging
Type
article
Field-Weighted Citation Impact
0.00
Controls
|||
ALL TIME
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
article

Prognostic value of MRI radiomics in locally advanced nasopharyngeal carcinoma treated with cisplatin chemoradiotherapy

Jun Guo
Journal of Radiation Research and Applied Sciences
Radiomics and Machine Learning in Medical Imaging
article

Prognostic value of MRI radiomics in locally advanced nasopharyngeal carcinoma treated with cisplatin chemoradiotherapy

Jun Guo
article en

Abstract

This study evaluates the efficacy of cisplatin-based concurrent chemoradiotherapy (CCRT) for locally advanced nasopharyngeal carcinoma (LA-NPC) and assesses the prognostic value of MRI texture analysis. In this retrospective cohort study, 100 consecutive LA-NPC patients treated between February 2021 and January 2024 were retrospectively enrolled and allocated based on treatment received: radiotherapy alone (control group, n = 50) or cisplatin CCRT (study group, n = 50). We compared treatment efficacy (objective response rate [ORR] per RECIST 1.1), pain scores (Visual Analogue Scale), anxiety/depression scores (SAS/SDS), survival outcomes (Kaplan-Meier with log-rank tests), and adverse reactions (CTCAE v4.0). Additionally, in a separate radiomics cohort of 60 LA-NPC patients, we performed 3D whole-tumor segmentation on pretreatment MRI images, extracted 396 radiomic features using AI-Kit software, applied LASSO regression for feature selection, and evaluated model performance using 5-fold cross-validation with intra- and inter-observer reproducibility assessment (ICC >0.75). The CCRT group showed significantly higher ORR (82.0% vs. 56.0%; P = 0.006). Patients receiving CCRT experienced lower pain scores at 1-, 2-, and 3-weeks post-treatment, and greater reductions in anxiety and depression (P < 0.001). Survival analysis revealed superior outcomes for the CCRT group: median overall survival (223 vs. 177 days; HR = 0.68, 95% CI: 0.46-0.99, P = 0.048), local recurrence-free survival (68 vs. 47 days; HR = 0.62, 95% CI: 0.40-0.96, P = 0.032), and distant metastasis-free survival (184 vs. 107 days; HR = 0.54, 95% CI: 0.32-0.91, P = 0.018). Adverse reactions were comparable between groups (22.0% vs. 32.0%; P = 0.261). MRI texture analysis identified two predictive features: 135-degree cluster shadow-step 4 (AUC = 0.861, 95% CI: 0.786-0.936) and full-angle correlation variance-step 1 (AUC = 0.758, 95% CI: 0.654-0.862). The combined model achieved an AUC of 0.928 (95% CI: 0.876-0.979; sensitivity: 86.2%; specificity: 83.9%) with good calibration (Hosmer-Lemeshow P = 0.742). Cisplatin-based CCRT significantly improves survival and quality-of-life outcomes in LA-NPC without increasing adverse effects. MRI texture features serve as a promising non-invasive biomarker for predicting treatment response, though prospective validation is required.

Journal of Radiation Research and Applied SciencesVol. 19(4)
Northern Jiangsu People's Hospital (CN)
Openalex Percentile: Top 11%
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.