Predicting biochemical recurrence of prostate cancer with artificial intelligence

Abstract

Background The first sign of metastatic prostate cancer after radical prostatectomy is rising PSA levels in the blood, termed biochemical recurrence. The prediction of recurrence relies mainly on the morphological assessment of prostate cancer using the Gleason grading system. However, in this system, within-grade morphological patterns and subtle histopathological features are currently omitted, leaving a significant amount of prognostic potential unexplored. Methods To discover additional prognostic information using artificial intelligence, we trained a deep learning system to predict biochemical recurrence from tissue in H&E-stained microarray cores directly. We developed a morphological biomarker using convolutional neural networks leveraging a nested case-control study of 685 patients and validated on an independent cohort of 204 patients. We use concept-based explainability methods to interpret the learned tissue patterns. Results The biomarker provides a strong correlation with biochemical recurrence in two sets (n = 182 and n = 204) from separate institutions. Concept-based explanations provided tissue patterns interpretable by pathologists. Conclusions These results show that the model finds predictive power in the tissue beyond the morphological ISUP grading.

Tumor Location and a Tumor Volume over 2.8 cc Predict the Prognosis for Japanese Localized Prostate Cancer

“…The unique biological characteristics of tumor types in different prostate regions can help guide individualized treatment and patient risk stratification. Finally, further validation of our clinical parameters using the latest imaging system PSMA/PET [ 65 ] or artificial intelligence system (deep learning) [ 66 ] may enhance the clinical importance of this study.…”

Harnessing artificial intelligence for prostate cancer management

“…AI has achieved promising results in directly predicting BCR from WSIs, outperforming traditional clinical indicators such as GS in several studies. 83 , 84 , 85 , 86 , 87 , 88 , 89 , 90 For instance, Yamamoto et al. applied an unsupervised deep neural network (DNN) model to extract features from unannotated post-RP H&E slides to predict 1-year BCR, achieving an AUC of 0.845 in an external validation set, which was better than using only GS (AUC = 0.721).…”

“… 96 However, relying solely on morphological analysis may be insufficient for long-term outcome predictions, as randomness of mutations and the potential for escaped cells to acquire additional mutations can affect recurrence. 68 , 86 …”

“…used automatic concept explanations (ACEs) to explain which image features in their model were used to make decisions for predicting BCR. 86 Therefore, enhancing the interpretability of AI models may discover new histopathological features and increase our understanding of this disease.…”

Recurrent Prostate Cancer Diagnostics with 18F-PSMA-1007 PET/CT: A Systematic Review of the Current State

“…Despite that, recurrence of the PCa is frequent and patients with clinically intermediate or high-risk prostate cancer after initial radical treatment need clinical follow-up. Biochemical recurrence is when PSA levels rise in the blood to a certain threshold after prostate cancer treatment—radical prostatectomy or radiation therapy [ 4 ]. The biochemical persistence is defined as a PSA level persistence/recurrence after radical prostatectomy such that the PSA fails to fall to undetectable levels [ 5 ].…”