Related Experiment Video
Updated: Nov 2, 2025
Introduction of an Integrated Pathology Image Management, Artificial Intelligence, and Reporting System
Published on: July 11, 2025
Using deep learning to identify bladder cancers with FGFR-activating mutations from histology images
Constantine S Velmahos1, Marcus Badgeley2, Ying-Chun Lo3
1University of Massachusetts Medical School, Worcester, MA, USA.
Background:
In recent years, the fibroblast growth factor receptor (FGFR) pathway has been proven to be an important therapeutic target in bladder cancer. FGFR-targeted therapies are effective for patients with FGFR mutation, which can be discovered through genetic sequencing. However, genetic sequencing is not commonly performed at diagnosis, whereas a histologic assessment of the tumor is. We aim to computationally extract imaging biomarkers from existing tumor diagnostic slides in order to predict FGFR alterations in bladder cancer.
Methods:
This study analyzed genomic profiles and H&E-stained tumor diagnostic slides of bladder cancer cases from The Cancer Genome Atlas (n = 418 cases). A convolutional neural network (CNN) identified tumor-infiltrating lymphocytes (TIL). The percentage of the tissue containing TIL ("TIL percentage") was then used to predict FGFR activation status with a logistic regression model.
Results:
This predictive model could proficiently identify patients with any type of FGFR gene aberration using the CNN-based TIL percentage (sensitivity = 0.89, specificity = 0.42, AUROC = 0.76). A similar model which focused on predicting patients with only FGFR2/FGFR3 mutation was also found to be highly sensitive, but also specific (sensitivity = 0.82, specificity = 0.85, AUROC = 0.86).
Conclusion:
TIL percentage is a computationally derived image biomarker from routine tumor histology that can predict whether a tumor has FGFR mutations. CNNs and other digital pathology methods may complement genome sequencing and provide earlier screening options for candidates of targeted therapies.
Insights
Tumor histology images can predict fibroblast growth factor receptor (FGFR) mutations in bladder cancer. This method uses tumor-infiltrating lymphocyte (TIL) percentage, identified by convolutional neural networks (CNNs), to screen patients for targeted therapies.
Area of Science:
- Oncology
- Computational Pathology
- Genomics
Background:
- The fibroblast growth factor receptor (FGFR) pathway is a key therapeutic target in bladder cancer.
- FGFR-targeted therapies benefit patients with specific FGFR mutations, typically identified via genetic sequencing.
- Genetic sequencing is not standard at diagnosis, while tumor histology is.
Purpose of the Study:
- To computationally extract imaging biomarkers from diagnostic tumor slides to predict FGFR alterations in bladder cancer.
- To assess the feasibility of using routine histology to identify patients eligible for FGFR-targeted therapies.
Main Methods:
- Analysis of genomic profiles and H&E-stained bladder cancer slides from The Cancer Genome Atlas (n=418).
- Utilized a convolutional neural network (CNN) to identify tumor-infiltrating lymphocytes (TIL).
- Employed logistic regression to predict FGFR activation status using TIL percentage.
Main Results:
- A CNN-based TIL percentage model accurately predicted any FGFR gene aberration (AUROC=0.76).
- A refined model specifically for FGFR2/FGFR3 mutations demonstrated high sensitivity and specificity (AUROC=0.86).
Conclusions:
- TIL percentage, a derived image biomarker from histology, can predict FGFR mutations in bladder cancer.
- Digital pathology methods like CNNs can complement genomic sequencing for early screening of targeted therapy candidates.
