An interpretable [18F]FDG PET/CT-based multimodal model for predicting pathological grade and stratifying prognosis in clear cell renal cell carcinoma: a multicenter study

The assessment of the World Health Organization/International Society of Urological Pathology (WHO/ISUP) grade is crucial for the management and treatment of clear cell renal cell carcinoma (ccRCC). This study aimed to develop and validate an interpretable [ 18 F]FDG PET/CT-based multimodal data fusion model (MDFM) for predicting WHO/ISUP grade and stratifying prognosis in patients with ccRCC. This retrospective multicenter study included 238 ccRCC patients (mean age, 58 ± 10 years; 153 male) from three independent centers (training, n = 111; test, n = 47; external validation, n = 80). A clinical model, three imaging-based models, two fusion models, and the MDFM integrating clinical variables, radiomic features, and deep learning features were developed to predict pathological grade. The MDFM achieved improved performance in predicting high-grade ccRCC, with area under the receiver operating characteristic curve of 0.884 (95% CI: 0.821–0.948) in the training set, 0.861 (95% CI: 0.737–0.985) in the test set, and 0.864 (95% CI: 0.787–0.942) in the external validation set. SHapley Additive exPlanations analysis identified the top three predictive features, each originating from a different data modality. The MDFM also demonstrated effective prognostic stratification, with a training Harrell’s C-index of 0.766 and significantly poorer progression-free survival in the MDFM-predicted high-grade group (HR = 2.781, 95% CI: 1.458–5.306, p = 0.001). This interpretable [ 18 F]FDG PET/CT-based multimodal model accurately predicts WHO/ISUP grade and stratifies progression risk in ccRCC. As the first multicenter study to establish such a framework, it offers a promising tool for personalized clinical management and decision-making.

Authors

Institutions

Publication Details

Journal
EJNMMI Research
Published
2026-11-20
DOI
https://doi.org/10.1186/s13550-026-01547-0
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

An interpretable [18F]FDG PET/CT-based multimodal model for predicting pathological grade and stratifying prognosis in clear cell renal cell carcinoma: a multicenter study

Wenyi Wei, Jiatie Li, Kezheng Wang, 庆春 庄 et al.
1 links
EJNMMI Research
Radiomics and Machine Learning in Medical Imaging
article

An interpretable [18F]FDG PET/CT-based multimodal model for predicting pathological grade and stratifying prognosis in clear cell renal cell carcinoma: a multicenter study

Wenyi Wei, Jiatie Li, Kezheng Wang, 庆春 庄, Hao Jiang, Yuchao Xu, Xinglu Zhou, Chunan Wang, Ying Wang, Wei Guo, Liang Cheng, Yanmei Wang
article en

Abstract

The assessment of the World Health Organization/International Society of Urological Pathology (WHO/ISUP) grade is crucial for the management and treatment of clear cell renal cell carcinoma (ccRCC). This study aimed to develop and validate an interpretable [ 18 F]FDG PET/CT-based multimodal data fusion model (MDFM) for predicting WHO/ISUP grade and stratifying prognosis in patients with ccRCC. This retrospective multicenter study included 238 ccRCC patients (mean age, 58 ± 10 years; 153 male) from three independent centers (training, n = 111; test, n = 47; external validation, n = 80). A clinical model, three imaging-based models, two fusion models, and the MDFM integrating clinical variables, radiomic features, and deep learning features were developed to predict pathological grade. The MDFM achieved improved performance in predicting high-grade ccRCC, with area under the receiver operating characteristic curve of 0.884 (95% CI: 0.821–0.948) in the training set, 0.861 (95% CI: 0.737–0.985) in the test set, and 0.864 (95% CI: 0.787–0.942) in the external validation set. SHapley Additive exPlanations analysis identified the top three predictive features, each originating from a different data modality. The MDFM also demonstrated effective prognostic stratification, with a training Harrell’s C-index of 0.766 and significantly poorer progression-free survival in the MDFM-predicted high-grade group (HR = 2.781, 95% CI: 1.458–5.306, p = 0.001). This interpretable [ 18 F]FDG PET/CT-based multimodal model accurately predicts WHO/ISUP grade and stratifies progression risk in ccRCC. As the first multicenter study to establish such a framework, it offers a promising tool for personalized clinical management and decision-making.

EJNMMI Research
Harbin Medical University (CN), General Electric (Spain) (ES), Third Affiliated Hospital of Harbin Medical University (CN), Heilongjiang Provincial Hospital (CN), Second Affiliated Hospital of Harbin Medical University (CN), First Affiliated Hospital of Harbin Medical University (CN), University of South China (CN)
Harbin Medical University, National Natural Science Foundation of China
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