@article {Song557470, author = {Qingyuan Song and John D. Seigne and Alan R. Schned and Karl T. Kelsey and Margaret R. Karagas and Saeed Hassanpour}, title = {A Machine Learning Approach for Long-Term Prognosis of Bladder Cancer based on Clinical and Molecular Features}, elocation-id = {557470}, year = {2019}, doi = {10.1101/557470}, publisher = {Cold Spring Harbor Laboratory}, abstract = {PURPOSE Improving the consistency and reproducibility of bladder cancer prognoses necessitates the development of accurate, predictive prognostic models. Current methods of determining the prognosis of bladder cancer patients relies on manual decision-making including factors with high intra- and inter-observer variability, such as tumor grade. To advance the long-term prediction of bladder cancer prognoses, we developed and tested a computational model to predict the 10-year overall survival for bladder cancer patients without considering tumor grade classification.MATERIALS AND METHODS We utilized a population-based dataset from the New Hampshire Cancer Registry with 1,225 bladder cancer patients diagnosed between 1994 and 2004. A weighted logistic regression model was trained using features including pre-treatment factors with high reproducibility including demographic characteristics, risk factors such as history of cigarette smoking, clinical information such as muscle invasiveness and tumor histology, and molecular features such as p53 immunohistochemical (IHC) positivity, while excluding less reliable measures such as tumor grade.RESULT Our model predictor of 10-year survival (F1 score = 0.78) was largely driven by age, muscle invasiveness and p53 IHC positivity and strongly related to patient survival in Cox models (p = 0.0013) even after adjustment for tumor grade and treatment. These results suggest that bladder cancer prognosis can be improved by machine learning methods and avoiding factors with high intra- and inter-observer variability.CONCLUSION Our study demonstrated a machine learning approach using a combination of clinical and molecular features could provide a better long-term prognosis for bladder cancer patients in comparison to tumor grade that suffers from low intra- and inter-observer variability. If validated in clinical trials, this automated approach can guide personalized management and treatment for bladder cancer patients.}, URL = {https://www.biorxiv.org/content/early/2019/02/21/557470}, eprint = {https://www.biorxiv.org/content/early/2019/02/21/557470.full.pdf}, journal = {bioRxiv} }