Reviewing AI's Role in Non-Muscle-Invasive Bladder Cancer Recurrence Prediction

TL;DR

NMIBC has a recurrence rate of $70-80\%$ and high management costs, challenging current prognostic methods that rely on standard TNM-based scoring. The review surveys 25 ML-based NMIBC recurrence studies across imaging, pathology, genomics, and clinical data, classifying them into four modeling categories and assessing methodological robustness and generalizability. Multimodal models integrating radiomics/pathology/genomics with clinical data frequently achieve superior performance (e.g., AUROCs up to $0.91$ and c-indices up to around $0.88$), but widespread adoption is hindered by small sample sizes, retrospective designs, and interpretability concerns. The paper calls for large multi-center validation, explainable AI approaches, and seamless clinical-workflow integration to realize potential reductions in costs and improvements in patient outcomes.

Abstract

Notorious for its 70-80% recurrence rate, Non-muscle-invasive Bladder Cancer (NMIBC) imposes a significant human burden and is one of the costliest cancers to manage. Current tools for predicting NMIBC recurrence rely on scoring systems that often overestimate risk and have poor accuracy. This is where Machine learning (ML)-based techniques have emerged as a promising approach for predicting NMIBC recurrence by leveraging molecular and clinical data. This comprehensive review paper critically analyses ML-based frameworks for predicting NMIBC recurrence, focusing on their statistical robustness and algorithmic efficacy. We meticulously examine the strengths and weaknesses of each study, by focusing on various prediction tasks, data modalities, and ML models, highlighting their remarkable performance alongside inherent limitations. A diverse array of ML algorithms that leverage multimodal data spanning radiomics, clinical, histopathological, and genomic data, exhibit significant promise in accurately predicting NMIBC recurrence. However, the path to widespread adoption faces challenges concerning the generalisability and interpretability of models, emphasising the need for collaborative efforts, robust datasets, and the incorporation of cost-effectiveness. Our detailed categorisation and in-depth analysis illuminate the nuances, complexities, and contexts that influence real-world advancement and adoption of these AI-based techniques. This rigorous analysis equips researchers with a deeper understanding of the intricacies of the ML algorithms employed. Researchers can use these insights to refine approaches, address limitations, and boost generalisability of their ML models, ultimately leading to reduced healthcare costs and improved patient outcomes.

Figures (6)

• Figure 1: Stages of Tumour metastasis illustrated in AJCC TNM staging system. Carcinoma-in-situ (CIS), Ta and T1 are non-muscle invasive stages and T2 - T4 are muscle invasive stages. CIS: primary stage where tumour is confined to inner bladder lining. Ta: tumour limited to epithelium. T1: tumour reaches the lamina propria. Stage III (T2): tumour invades into bladder wall muscle. T3: tumour spreads to the fat around the bladder. Stage IV (T4): tumour spreads to nearby pelvic organs/tissues.
• Figure 2: Distribution of major risk factors for bladder cancer. Tobacco smoking accounts for approximately 50% of cases, occupational exposures contribute to 18%, and other factors (including exposure to arsenic, chronic bladder inflammation, previous radiation or chemotherapy, diet, and genetic predisposition) comprise the remaining 32%. Data synthesized from multiple sources burgerEpidemiologyRiskFactors2013letasiovaBladderCancerReview2012farlingBladderCancerRisk2017chenBladderCancerScreening2005cumberbatchEpidemiologyBladderCancer2018.
• Figure 3: This graph shows the overall growth in bladder cancer recurrence studies and the exponential rise in ML-based approaches. While the total number of bladder cancer studies has increased linearly over the past two decades, the adoption of ML methods has grown at a much faster, exponential rate. This trend suggests that ML approaches are becoming increasingly relevant in the field, with future research likely to continue emphasizing advanced computational methods.
• Figure 4: Machine learning workflow model for bladder cancer prediction. (a) pre-processing steps –Data stored in a secure database goes through image segmentation or filtration, depending on the data type. Then feature selection is applied to identify useful features in the data while discarding the redundant and unimportant features. (b) application of ML algorithms - ML algorithms are selected and trained, prediction is made, and performance is evaluated using the most suitable metrics. Then a robust comparison is made to deduce added benefit and the superiority over the existing frameworks.
• Figure 5: PRISMA flow diagram showing search methodology, inclusion and exclusion criteria.