Using deep learning to identify bladder cancers with FGFR ‐activating mutations from histology images
| Autor: | Constantine S. Velmahos, Ying‐Chun Lo, Marcus A. Badgeley |
|---|---|
| Rok vydání: | 2021 |
| Předmět: |
Male
0301 basic medicine Cancer Research Databases Factual Gene Expression medicine.disease_cause Sensitivity and Specificity DNA sequencing tumor‐infiltrating lymphocytes 03 medical and health sciences Lymphocytes Tumor-Infiltrating 0302 clinical medicine convolutional neural networks medicine Humans Receptor Fibroblast Growth Factor Type 3 Radiology Nuclear Medicine and imaging Molecular Targeted Therapy Receptor Fibroblast Growth Factor Type 2 RC254-282 Original Research Mutation Bladder cancer Tumor-infiltrating lymphocytes business.industry Neoplasms. Tumors. Oncology. Including cancer and carcinogens Clinical Cancer Research deep learning Digital pathology Histology medicine.disease Receptors Fibroblast Growth Factor Logistic Models 030104 developmental biology Urinary Bladder Neoplasms Oncology Fibroblast growth factor receptor fibroblast growth factor receptors 030220 oncology & carcinogenesis Cancer research Biomarker (medicine) Female Neural Networks Computer business |
| Zdroj: | Cancer Medicine Cancer Medicine, Vol 10, Iss 14, Pp 4805-4813 (2021) |
| ISSN: | 2045-7634 |
| DOI: | 10.1002/cam4.4044 |
| Popis: | 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. We found that a convolutional neural network, trained on widely accessible tumor histology slides, is highly sensitive at predicting which bladder cancers have a fibroblast growth factor receptor activating mutation. The significance of our finding is that deep learning models, trained on routinely collected histology images, can be a powerful tool to screen for patients who could benefit from targeted therapies. |
| Databáze: | OpenAIRE |
| Externí odkaz: |
|