Prediction of non-muscle invasive bladder cancer recurrence using machine learning of quantitative nuclear features

Naoto Tokuyama,Akira Saito,Ryu Muraoka,Shuya Matsubara,Takeshi Hashimoto,Naoya Satake,Jun Matsubayashi,Toshitaka Nagao,Aashiq H Mirza,Hans-Peter Graf,Eric Cosatto,Chin-Lee Wu,Masahiko Kuroda,Yoshio Ohno

doi:10.1038/s41379-021-00955-y

Naoto Tokuyama, Akira Saito + Show 12 more

Open Access

https://doi.org/10.1038/s41379-021-00955-y

Copy DOI

Abstract

Non-muscle invasive bladder cancer (NMIBC) generally has a good prognosis; however, recurrence after transurethral resection (TUR), the standard primary treatment, is a major problem. Clinical management after TUR has been based on risk classification using clinicopathological factors, but these classifications are not complete. In this study, we attempted to predict early recurrence of NMIBC based on machine learning of quantitative morphological features. In general, structural, cellular, and nuclear atypia are evaluated to determine cancer atypia. However, since it is difficult to accurately quantify structural atypia from TUR specimens, in this study, we used only nuclear atypia and analyzed it using feature extraction followed by classification using Support Vector Machine and Random Forest machine learning algorithms. For the analysis, 125 patients diagnosed with NMIBC were used; data from 95 patients were randomly selected for the training set, and data from 30 patients were randomly selected for the test set. The results showed that the support vector machine-based model predicted recurrence within 2 years after TUR with a probability of 90% and the random forest-based model with probability of 86.7%. In the future, the system can be used to objectively predict NMIBC recurrence after TUR.

Highlights

Bladder cancer is the ninth common malignant tumor worldwide[1]; it is clinically classified into non-muscle invasive bladder cancer (NMIBC) and muscle invasive bladder cancer (MIBC)
Currently, artificial intelligence (AI) based on digital pathology images is being used for diagnosis, morphological classification, and prognosis prediction of various cancers[16–18]
It is very promising to create objective and accurate predictions from digital imaging data in various cancers, no previous studies have used AI on digital pathology images to predict the prognosis of NMIBC

Summary

Introduction

Bladder cancer is the ninth common malignant tumor worldwide[1]; it is clinically classified into non-muscle invasive bladder cancer (NMIBC) and muscle invasive bladder cancer (MIBC). 70% of bladder cancers are reported to be NMIBC at the time of initial diagnosis[2]. An important point is that the treatment strategy depends on the presence or absence of muscle layer invasion. NMIBC is considered to have a favorable prognosis. The rate of intravesical recurrence of NMIBC after transurethral resection of the bladder tumor (TURBT) is still as high as 30–50%3. To reduce the recurrence risk, bacillus Calmette-Guerin (BCG) therapy is recommended for high- and intermediate-risk categories. Since BCG therapy is often associated with side effects such as hematuria, fever, and pain, its indications must be fully considered[4]. Accurate evaluation of the recurrence risk is the most important factor in the management of NMIBC. The current risk classification system provided by the American

Methods

Results

Conclusion