Beyond the buzzwords: How AI and advancing diagnostic technologies will transform early bladder cancer care
Over the past few years, artificial intelligence (AI) has gone from sci-fi prophecies to real-life application. And it’s accelerating fast. But as with any novel technology, it’s important to look beyond extreme examples and instead focus on the realistic practicalities of AI – and where its application might lead.
Healthcare – with its need for advanced resources and efficiency – is seemingly a perfect home for AI. But the sector has been slow on digitalizing, let alone bringing in such game-changing technologies. Yet gradually, as healthcare facilities around the world digitalize their processes, we’re seeing not just a new way of getting results, but a new way of thinking.
This is now being demonstrated in bladder cancer care with advanced tools emerging that soon expect to become the new norm. In uro-oncology, AI can have a variety of utilities, such as aiding pre-surgical planning through 3D-imaging, and improving the quality of procedures through augmented reality guidance.1,2 In addition, utilizing AI in diagnostics can provide support through real-time tumor detection, histological categorization, risk stratification, and treatment planning.3,4
So, how can AI and diagnostic technologies help when it comes to the care pathway in non-muscle-invasive bladder cancer (NMIBC)? To see the potential in this field, it’s important to first understand the context of the disease itself.
Why is it crucial to address challenges in NMIBC?
NMIBC remains a major health concern worldwide. In fact, recurrences and progression to muscle invasive bladder cancer (MIBC) affect nearly half of the diagnosed NMIBC population, potentially due to misdiagnosis, delay of diagnosis and incomplete resection of tumors.5–7 In NMIBC, missing early recurrence of high-risk lesions may lead to substantial delay of appropriate treatments, which may have a detrimental impact on disease prognosis.8–10
So what can be done to address this? Accurate, timely diagnosis and prediction of recurrence and progression is essential in NMIBC management,12 and advanced tools that can help us to achieve this are key4 – this is where the use of AI is beginning to show promise.4,12
The need for early, accurate diagnosis and improved risk stratification
Expert consensus and NMIBC guidelines are aligned on there being a need for improving the overall quality of NMIBC care and diagnostics in particular.15,16
Identifying patients’ risk factors, or ‘risk stratification’, is vital when planning, predicting prognosis and outcomes, and managing treatment for NMIBC.17 However, current models to support risk stratification have limitations. There’s no universally accepted standard, and the significance of various risk factors used in such models isn’t always clear.18,19 Current models also still largely exclude molecular and genomic profiling of bladder tumors17,20 – factors which can be crucial for identifying a targeted treatment approach. This can lead to delayed or suboptimal treatment – resulting in worse outcomes for NMIBC patients and higher costs of care.18
Early and precise diagnosis of NMIBC is essential for accurate risk stratification, decision making and treatment planning. This in turn may enhance the effectiveness of treatment, by identifying patients who are unlikely to benefit from specific treatments and avoiding unnecessary procedures – leading to more cost-effective care.21,22
Shifting to a
precision-based approach in NMIBC
Improving diagnostic precision is becoming increasingly important in NMIBC. NMIBC care is just starting to tap into modern technologies like advanced genetic testing and is expected to further shift with the emergence of new targeted treatment options and technological advancements like AI.23
precision-based approach in NMIBC
Improving diagnostic precision is becoming increasingly important in NMIBC. NMIBC care is just starting to tap into modern technologies like advanced genetic testing and is expected to further shift with the emergence of new targeted treatment options and technological advancements like AI.23
What is precision diagnostics in NMIBC?
Precision diagnostics doesn’t just consider the clinical presentation and traditional risk factors, it allows a more thorough understanding of the genetic and molecular characteristics of tumors and identifies abnormalities that might be causing it to be more aggressive.24
How? By utilizing current conventional diagnostics such as radiology and cystoscopy, in combination with emerging tools including next-generation imaging techniques, biomarkers based on immunohistochemistry, next-generation sequencing (NGS), multi-omics, and leveraging artificial intelligence (AI)/machine learning (ML).24
A precision-based approach aims to get greater effectiveness from treatment and less off-target effects. The rapid advancement of technologies, the shift towards precision medicine and emerging novel targeted treatments in NMIBC all drive the renewed emphasis of the importance of the diagnostic process.
What are the benefits?
Precision diagnosis, along with increasing support from AI, may bring us closer to addressing the challenges faced in NMIBC care. It can identify patients who are more likely to respond, or unlikely to benefit, to particular treatment. Plus, more individualized and targeted therapies can increase the chance of a response to treatment while reducing side effects as opposed to more general approaches like chemotherapy.24
A shift towards precision diagnostics and personalized treatment is clearly already underway, with approvals of targeted drugs such as erdafitinib and pembrolizumab in NMIBC,27 as well as AI-based diagnostic tools being explored in several areas of uro-oncology, including prostate and bladder cancer.23,28
An example of precision diagnostics in practice: targeting CIS lesions for patient selection and treatment monitoring
Tools to enhance cystoscopy can improve detection and monitoring of ‘marker lesions’ to evaluate response to treatments, giving patients who may have been treatment-unresponsive the opportunity to be reassessed and offered other novel treatment options as part of a bladder sparing approach.21,22 Evaluating the recurrence or lack of change in such lesions, as part of defining complete response, can be used as an ‘optical biomarker’ for treatment effect.10,29,30
For example, it is increasingly common to target lesions like carcinoma in-situ (CIS) to select patients for new treatments indicated for BCG-unresponsive NMIBC with CIS, and to monitor these patients for response to treatment.10,29,30
Due to their small and flat nature, CIS lesions represent a challenge for urologists,31 and account for about 10% of NMIBC. These lesions can be hard to detect with current diagnostic tools, and its presence can indicate a high risk of recurrence and worse outcomes for patients.8,31 When CIS is suspected, and if the required equipment is available, NMIBC guidelines recommend using fluorescence-guided biopsies to improve detection of these lesions. This more targeted approach improves the ‘hit rate’ and can lead to more accurate classification, risk stratification and decision making.21,22,32
The emergence of novel diagnostic technologies in NMIBC
The accuracy of current diagnostic tools can vary widely. However, there has been remarkable innovation in NMIBC over the last decade with the emergence of novel technologies – potentially providing clinicians with more efficient tools.33,34
These newer tools will not replace current established methods, but can be used to work together in parallel, each bringing something valuable to the table and enhancing the overall diagnostic performance. We explore examples of these tools and how they can be combined or enhanced in further detail below – click through each of the sections to find out more about each tool.
What are the tools, and how can they support a precision-based
approach in NMIBC?
Visual enhancement techniques (optical wide view)
Visual enhancement technologies, which aim to improve the performance of standard white light cystoscopy, are well established in clinical practice and have been used for NMIBC for almost two decades.35 Photodynamic technology using light-sensitive and fluorescent substances, such as blue-light cystoscopy (BLC), also have increasingly important utility for NMIBC and use imaging agents, allowing for improved tumor detection, aiding diagnosis, risk stratification and treatment selection.35
These technologies are now increasingly leveraging improvements in image resolution, combined imaging technologies, digital processing algorithms and AI to further enhance their performance.36 For example, a number of algorithms are being developed to improve the quality of imaging by reducing artifacts and noise, such as in BLC imaging. The corrections and enhancements made to BLC images result in an improved view for clinicians, and restore critical features that may normally be lost in traditional imaging to support accurate diagnosis.37
Advancements in sensors, or 'chip-on-the-tip', and software also bring new possibilities, and can benefit from ML, and advanced data processing techniques. Using these different image modalities allow detailed analysis via multiple images, providing a comprehensive overview of patients’ characteristics.38
Micro- and molecular enhancement technologies (optical biopsies)
Novel, high-resolution microscopic imaging techniques can provide real-time diagnosis of the bladder lining during cystoscopy and transurethral resection of bladder tumors (TURBT), and can support next generation pathology (NGP) approaches.39,40
These include:39
- Optical coherence tomography (OCT)
- Confocal laser endomicroscopy (CLE)
- Raman spectroscopy
- Multiphoton microscopy
Real-time assessment with such tools may help reduce false-positive findings and unnecessary follow-up procedures.41,42
In the future, these technologies could be combined with other imaging tools, or enhanced with AI, to allow clear visualization of the entire bladder. For example, it has been demonstrated that a combination of micro-enhancement technology with fluorescence imaging technologies, such as BLC, could further improve accuracy in diagnosis,43 and a combination of raman spectroscopy with molecular surface-enhanced targeting nanoparticles can improve bladder tumor detection.44
Interpreting microscopic information can be complex and can often lead to variability between observers. Yet, these concerns might be addressed by using advanced AI/ML support.33 Optical enhancement technologies already function as a cornerstone in NMIBC diagnosis, and with further enhancement through AI, it’s likely this will continue into the future.
Advanced radiologic imaging techniques
There is a substantial need for non-invasive diagnostic tools, allowing for better disease staging and pre-operative performance predictions – making the use of novel radiologic techniques a key area of interest. Various AI algorithms are being introduced and validated to help analyze radiologic images, and assist radiologists with diagnostics.45,46
- Multiparametric magnetic resonance imaging (mpMRI) and positron emission tomography (PET) has led to advances when assessing the potential for MIBC, along with advanced data-reporting systems which have shown to improve diagnostic performance in high-risk NMIBC22,47
- In prostate cancer, new tracers like prostate specific membrane antigen (PSMA) are also being used with imaging technologies like PET to aid with diagnosis and treatment
- Advancements in computed tomography (CT) technology, combined with powerful imaging reconstruction software, have led to development of virtual reality cystoscopies (VC)48
- Contrast-enhanced ultrasonography (CEUS) is currently being explored for the differentiation of low- and high-grade NMIBC49
- Diffusion-weighted imaging (DWI) can distinguish between normal and tumor tissue and is used to support histological grading33
Biomarkers (liquid and synthetic biopsies)
The use of biomarkers has the potential to transform NMIBC care, becoming useful for screening, diagnosis, and treatment monitoring. In particular, biomarkers with predictive capabilities may be able to improve the accuracy of risk stratification models and improve the overall management of NMIBC.50
Approaches such as mass spectrometry, multidimensional ‘omics’ technology and novel genomics sequencing are promising and allow the creation of a comprehensive ‘liquid biopsy’ at a low cost.51
Furthermore, different biomarker technologies may be combined with other tools, like photodynamic diagnosis (PDD), radiological imaging and AI, to increase the accuracy in predicting treatment response.52
Biosensors and microdevices
Point of care (POC) tests, also referred to as ‘lab-on-a-chip’ platforms, are portable, analytical biosensor technologies being developed for real-time bedside monitoring of biomarkers, which hold promise for NMIBC detection and diagnosis.53–55
Biosensors are microdevices that can detect the presence of various biomarkers in blood or urine, even at very low detection limits. The ability to monitor changes in these markers during disease progression can support early disease monitoring and intervention in NMIBC.56
Microdevices can also be used to predict tumor responses to treatments. These are implanted into the bladder and release small doses of various drugs, allowing the drug to interact with the tumor in its native environment. The tumor tissue surrounding the device is then analyzed following surgical procedure and can provide a detailed assessment of the drug’s local effectiveness on the tumor.57
Transforming NMIBC care with AI
Advancement in technology, with the use AI and ML, has the potential to dramatically shape medical procedures in the near future. AI and ML have the potential to revolutionize NMIBC management and are increasingly being used to analyze images and other data, and can help identify tumors more accurately, highlighting high-risk patients, enhancing clinical decision making, and fast-tracking patients to more effective targeted treatments.3,4
AI-supported software can enhance current surgical procedures. It can integrate findings from imaging, such as blue light cystoscopy (BLC), with other patient data, such as medical history and pathology grading, to identify patients at risk and predict outcomes.58 There are also increasing advancements in AI-based tools to support clinical decision making, with one tool already available and recommended by NCCN guidelines for prostate cancer. This tool, ArteraAI Prostate Test, analyzes digital pathology images to help identify patients who will benefit from therapy and guide treatment decisions.23
In NMIBC specifically, AI is being increasingly explored in clinical trials, with one tool evaluated in a recent clinical study that uses more comprehensive molecular profiling to evaluate pre- and post-treatment responses to BCG immunotherapy. It aims to create a personalized model to show presence of NMIBC and predict response to BCG.28 Another tool, PROGRxN-BCa (PROGression Risk assessment in NMIBC), presented at the 2024 AUA congress exceeded current tools in NMIBC when predicting disease progression, demonstrating a benefit in avoiding unnecessary treatment escalation.59 In addition, a recent systematic review of AI studies in NMIBC found AI models to generally outperform non-AI models overall when predicting NMIBC outcomes.12
As we look to the future, there are still hurdles to overcome when it comes to broader implementation of these technologies and shifting to precision diagnostics. This includes limited-quality clinical data, lack of guideline recommendations, high equipment costs and lack of regulatory approvals.60
Despite rapid increases in evidence around the use of these tools, there is still room for improvement, and there is a clear need for collaboration between healthcare and AI communities to develop higher quality models, allowing us to reach the next step in NMIBC care.12
Conclusion
With NMIBC remaining a significant health concern globally, the integration of novel diagnostic technologies and AI promises to revolutionize diagnosis and treatment. By enhancing accuracy, integration of data, aiding risk stratification, and guiding personalized treatment approaches, these advancements offer hope for improved patient outcomes and more efficient disease management overall. While challenges such as data quality, cost, utility and regulatory hurdles exist, the trajectory is clear: the combination of AI and precision diagnostics procedures holds immense promise for transforming the landscape of NMIBC care in the years to come.