The subtyping of bladder cancers is a complicated business. It is, however, necessary as it has important implications for prognosis and therapeutic decisions. CLI caught up with Asst. Prof. Chelsea L. Jackson (University of Manitoba, Winnipeg, MB, Canada) and Prof. David Berman (Queen’s University, Kingston, ON, Canada) to find out more about how the current thinking is simplifying the process.

For those of us who are unfamiliar with the details of bladder cancer, can you give us a brief overview of/introduction to the different subtypes, please?

Several research groups have revealed that similar to breast cancer, bladder cancers can be clustered into molecular subtypes with distinct biological features. These molecular subtypes are broadly classified as ‘basal-like’ or ‘luminal-like,’ corresponding to basal stem cells that reside at the basement membrane or their more differentiated progeny at the epithelial surface, respectively. Accordingly, bladder cancer subtypes represent the expression of specific urothelial differentiation programmes and tumour-cell intrinsic genomic features. The basal subtype expresses basal cell markers such as basal keratins (KRT5, KRT14), epidermal growth factor receptor (EGFR), cellular adhesion molecules (CD44, P-cadherin) and squamous epithelial programmes (DSC2/3, DSP, KRT6). Conversely, the luminal subtype expresses transcription factors that maintain a differentiated urothelial state, including transcription factors such as GATA-binding protein 3 (GATA3) and Forkhead box A1 (FOXA1). These luminal subtypes have been further divided based on their expression of specific cell-cycle regulation pathways. A ‘luminal-papillary’ or ‘urothelial-like’ (Uro) subtype is defined by active fibroblast growth factor receptor 3 (FGFR3) signalling, and losses of early cell-cycle genes such as CDKN2A. Other luminal subtypes include the ‘luminal unstable’ or ‘genomically unstable’ (GU) subtype which is defined by genomic instability, including loss of expression of tumour suppressors such as retinoblastoma (RB1), loss of cell-cycle regulation genes (e.g. Cyclin D) and FGFR3 signalling. The inclusion of other more rare subtypes may vary by study and methodology, but can include a neuroendocrine (NE) subtype and mesenchymal-like (Mes-like) subtype, which demonstrate non-urothelial histology and gene expression patterns. Subtyping schemes using RNA-based profiling techniques often report a ‘stromal-rich’ subtype and show signatures of adjacent cell types, including fibroblasts, muscle fibres, and immune cells. The luminal subtypes are well represented in the very earliest forms of bladder cancer, carcinoma in situ and low grade papillary urothelial carcinoma. Basal, NE, and Mes-like subtypes begin to manifest in high-grade invasive cancers, and herald progression of bladder cancer from non-muscle-invasive bladder cancer (NMIBC) to muscle-invasive bladder cancer (MIBC) and subsequent meta-static disease. Given their differences in cancer biology and molecular features, subtypes are believed by many to provide insights into molecular mechanisms of tumour recurrence, progression or therapy resistance.

Why is it important to diagnose bladder cancer in that detail?

As a part of an International Society for Urologic Pathology initiative, we recently worked with a group of expert pathologists, urologists, and oncologists to address this question. The resulting paper “Working Group 4: molecular subtypes of bladder cancer – principles of classification and emerging clinical utility” may be  of interest to your readers as it provides a more in depth review  of subtyping, the science behind it, and its potential uses [See item 4 in the Bibliography for further details]. In that paper, which also provides a number of useful references on this topic, we emphasized that despite a great deal of elegant science and progress, there are currently no clearly beneficial ways to use molecular subtyping in clinical decision-making. One of the major barriers is that studies to date have shown mixed and variable correlations between subtypes, response to therapy, and patient outcomes. As discussed below, these problems likely relate, at least in part to vagaries of mRNA profiling, where changes in
the ratio of cancer cells and benign cells can yield unstable or incorrect subtyping information. So the same cancer may be called luminal by one group and basal by another. We and others have addressed this issue by implementing subtype classification schemes based on immunohistochemistry (IHC). Studies using single cell sequencing are starting to show new  and important insights into molecular subtypes.

If subtyping makes it into the clinic, it could become useful in a number of ways. The division of bladder cancers into early-stage NMIBC and more aggressive MIBC is based on the pathological stage at diagnosis. NMIBC tumours primarily present as stage pTa or pT1 tumours and represent the majority (75%) of diagnoses. Stage pTa tumours are confined to the urothelium, whereas stage pT1 tumours superficially invade the lamina propria. MIBC tumours represent only 25% of bladder cancer diagnoses, presenting as stages pT2–pT4, which are defined by invasion into the muscle layers of the bladder. Patients with NMIBC typically demonstrate a good prognosis, whereas patients with MIBC often experience a poor prognosis. Thus, the distinction between NMIBC and MIBC is critical for defining clinical management and treatment of the disease.

NMIBC is treated with local resection, sometimes followed by intravesical treatments of mitomycin C (MMC) or Bacillus-Calmette Guerin (BCG). This is followed with constant cystoscopies to monitor for recurrence or progression, which is both time consuming and expensive. Conversely, MIBCs are treated much more aggressively, with surgical removal of the bladder (cystectomy) and/or systemic therapies including radiation, chemotherapies, immunotherapies, and targeted agents. MIBCs are treated more aggressively owing to their poor survival rates. Even with the assistance of current risk classifiers, it is difficult to predict which patients will recur, progress or respond to therapies which may result in the over-treatment or under-treatment of patients. Improving risk stratification and management options for patients may rely on adding biomarkers at meaningful clinical decision-making points. For example, diagnostic biomarkers in NMIBC capable of helping a pathologist decide if a cancer is more likely to be high grade may urge the use of intra-vesical treatments and an increased cystoscopy schedule to detect recurrences or progression events earlier. In MIBC, prognostic or predictive biomarkers suggest a tumour is sensitive to chemo-therapy or may be less aggressive. Such biomarkers could allow for bladder sparing approaches to be used. Consequently, appropriate diagnosis is essential to improving treatment options for patients and also reduces costs and burden on the health care system.

Stages of bladder cancer
(Shutterstock.com)

How are the different bladder cancer subtypes usually identified and what are the challenges?

Currently there are no biomarkers, including molecular subtyping, used clinically in the diagnostic process for bladder cancers. The complexity and variability in profiling techniques and bioinformatics approaches has made it challenging to determine biologically and clinically relevant subtypes. This has resulted in overlapping subtyping schemes that are inconsistent across the spectrum of NMIBC and MIBC tumours and are rarely studied together in large cohort studies. A few key issues affecting the subtype classification include methodology differences, threshold determination and heterogeneity.

Advances in profiling techniques have expanded bladder cancer subtypes to include genomic, transcriptomic, epigenomic and proteomic profiles. Although this has provided greater biological insight, it has presented challenges in determining the appropriate method of subtyping. For example, transcriptomic-based subtypes are prone to artifacts from adjacent stromal tissue, which is particularly challenging to avoid in bladder tumours with papillary architecture. Proteomic-based methods like IHC provide tumour and stromal-specific signals, but rely on specific protein targets and may lack sensitivity to detect small changes in biological state.

Clinically meaningful thresholds for subtype classification have not been easily elucidated, as they are both methodology-specific and are not easily transferred between disease states (NMIBC vs MIBC). RNA-based classifiers are challenging to implement in clinical practice and require expensive independent validation. These classifiers can also be influenced by sample quality and tumour content. Similarly, IHC-based thresholds would also require extensive validation, but are more likely to demonstrate inter-observer scoring variability.

Finally, subtype classification can be challenged by tumour and immune microenvironment heterogeneity, spatial heterogeneity within the bladder and temporal changes including therapy-induced heterogeneity. Despite their potential as biomarkers, tumour-intrinsic subtypes work within the context of the tumour immune microenvironment and evolve over time. Patients with NMIBC are particularly unique in that they have a long treatment history from successive tumour resections arising in different parts of the bladder, sometimes with intervening intravesical therapy. This variability between patients and over time and space makes it challenging to compare between profiling studies.

Photomicrograph of a high grade, invasive papillary transitional (urothelial) carcinoma from a biopsy of a patient with hematuria (blood in urine) obtained during cystoscopy (Shutterstock.com)

What can be done to make subtype assignment more simple?

One way to simplify subtype assignment is to use a methodology that is faster, cheaper and more resistant to changes in cellular composition, such as IHC. As a widely used methodology in surgical pathology laboratories, there are clinically validated antibodies that can be used to identify specific molecular subtypes. Research from the Lund group validated their RNA-based subtypes using IHC, providing a critical framework from which to subtype bladder cancers. We further simplified this model into a three-antibody algorithm (GATA3, KRT5 and p16) that demonstrates 78% accuracy (95% CI: 67–86%) in identifying the three most prevalent molecular subtypes of MIBC (basal, Uro and GU). This three-antibody algorithm was further validated in NMIBC, demonstrating diagnostic, prognostic and predictive associations for these IHC-based subtypes. Importantly, using the Lund data we demonstrated the importance of a parsimonious model, whereby an eight-antibody model did not outperform a three-antibody model. A two-antibody model with KRT14 and RB1 showed improved accuracy (85%, 95% CI: 71–94%) compared to the three-antibody model, but these antibodies are not readily available in many clinical labs and therefore require additional validation. IHC-based subtyping algorithms provide additional flexibility in that they can be easily scored visually or using automated scoring software. The use of GATA3 and KRT5 IHC has also been previously shown to stain the most consistently across whole slides in addition to tissue microarrays, further supporting their use in a diagnostic capacity.

What do you envisage for future developments in bladder cancer diagnostics?

Currently, the grade and stage of bladder cancers at diagnosis drive clinical management and treatment options for patients, with little to no attention given to biomarker status or subtype assignment. As part of the ongoing switch from reviewing glass slides to digital images, there are rapidly advancing improvements in computer vision technology that will almost certainly improve diagnosis and grading. Future developments in bladder cancer diagnostics are likely incorporate these pathological data with tumour subtype and immune biomarkers into personalized management plans for patients. With respect to subtyping, our work demonstrated that although rare, basal NMIBC tumours present with high-grade high-stage tumours with a statistically significant risk for progression and recurrence and may warrant treatment intensification compared to other subtypes. Furthermore, when applying the subtyping algorithm to luminal NMIBC tumours, we found two widely expected luminal subtypes, Uro and GU and we were surprised to find for the first time a third luminal subtype, which we called UroKRT5+, that was strongly positive for KRT5 and GATA3. As the UroKRT5+ subtype was distinctively less aggressive and the basal subtype more aggressive than the others, classifying first into basal and luminal and further distinguishing luminal Uro, GU and UroKRT5+ subtypes identified significantly different risks for recurrence and progression.

Despite the breadth of work in NMIBC and MIBC to identify subtypes and their prognostic and predictive significance, additional work is needed to harmonize the subtyping schemes and profiling approaches to create a unified subtyping classification for NMIBC and MIBC. Large prospective and retrospective cohorts are required to begin validating an IHC subtyping approach and its clinical significance. Advances in AI-driven digital pathology platforms may also encourage the use of automated IHC scoring that could standardize staining quantification and expedite IHC-based subtyping in clinical practice. Importantly, these challenges for bladder cancer subtyping are documented within the research community and much effort has been placed towards creating resources, including larger cohorts to support this continued work.

The experts

Asst. Prof. Chelsea L. Jackson PhD
Department of Pathology, University of Manitoba, Winnipeg, MB Canada

Email: cjackson5@sharedhealthmb.ca

Prof. David M. Berman MD, PhD
Pathology and Molecular Medicine, Queen’s University, Kingston, ON Canada

Email: bermand@queensu.ca

Bibliograpy
1. Jackson CL, Chen L, Hardy CS et al. Diagnostic and prognostic implications of a three-antibody molecular subtyping algorithm for non-muscle invasive bladder cancer. J Pathol Clin Res 2022;8(2):143–154. doi: 10.1002/cjp2.245.
2. Hardy CS, Ghaedi H, Slotman A et al. Immunohistochemical assays for bladder cancer molecular subtyping: optimizing parsimony and performance of Lund taxonomy classifiers. J Histochem Cytochem 2022;70(5):357–375. doi: 10.1369/00221554221095530.
3. Sjödahl G, Jackson Chelsea L et al. Molecular profiling in muscle invasive bladder cancer: more than the sum of its parts. J Pathol 2019;247(5):563–573. doi: 10.1002/path.5230.
4. Warrick JI, Al-Ahmadie H, Berman DM et al. International Society of Urological Pathology Consensus Conference on Current Issues in Bladder Cancer. Working Group 4: molecular subtypes of bladder cancer – principles of classification and emerging clinical utility. Am J Surg Pathol 2023;10:1097. doi: 10.1097/PAS.0000000000002053.
5. Slotman A, Xu M, Lindale K et al. Quantitative nuclear grading: an objective, artificial intelligence-facilitated foundation for grading noninvasive papillary urothelial carcinoma. Lab Invest 2023;103(7):100155. doi: 10.1016/j.labinv.2023.100155.
6. Sjödahl G, Lauss M, Lövgren K et al. A molecular taxonomy for urothelial carcinoma. Clin Cancer Res 2012;18:3377–3386. doi: 10.1158/1078-0432.CCR-12-0077-T.
7. Sjödahl G, Eriksson P, Liedberg F, Höglund M. Molecular classification of urothelial carcinoma: global mRNA classification versus tumour-cell phenotype classification. J Pathol 2017;242(1):113–125. doi: 10.1002/path.4886.
8. Damrauer JS, Hoadley KA, Chism DD et al. Intrinsic subtypes of high-grade bladder cancer reflect the hallmarks of breast cancer biology. PNAS 2014;111(8):3110–3115. doi: 10.1073/pnas.1318376111.
9. Choi W, Porten S, Kim S et al. Identification of distinct basal and luminal subtypes of muscle-invasive bladder cancer with different sensitivities to frontline chemotherapy. Cancer Cell 2014;25(2):152–165. doi: 10.1016/j.ccr.2014.01.009.
10. Robertson AG, Kim J, Al-Ahmadie H et al. Comprehensive molecular characterization of muscle-invasive bladder cancer. Cell 2017;171(3):540-556.e25. doi: 10.1016/j.cell.2017.09.007.
11. Kamoun A, de Reyniès A, Allory Y et al. A consensus molecular classification of muscle-invasive bladder cancer. Eur Urol 2020;77(4):420–433. doi: 10.1016/j.eururo.2019.09.006.
12. Hedegaard J, Lamy P, Nordentoft I et al. Comprehensive transcriptional analysis of early-stage urothelial carcinoma. Cancer Cell 2016;30(1):27–42. doi: 10.1016/j.ccell.2016.05.004.
13. Lindskrog SV, Prip F, Lamy P et al. An integrated multi-omics analysis identifies prognostic molecular subtypes of non-muscle-invasive bladder cancer. Nat Commun 2021;12(1):2301. doi: 10.1038/s41467-021-22465-w.
14. Dadhania V, Zhang M, Zhang L et al. Meta-analysis of the luminal and basal subtypes of bladder cancer and the identification of signature immunohistochemical markers for clinical use. EBioMedicine 2016;12:105–117. doi: 10.1016/j.ebiom.2016.08.036.
15. Seiler R, Ashab HAD, Erho N et al. Impact of molecular subtypes in muscle-invasive bladder cancer on predicting response and survival after neoadjuvant chemotherapy. Eur Urol 2017;72(4):544–554. doi: 10.1016/j.eururo.2017.03.030.
16. Sjödahl G, Eriksson P, Patschan O et al. Molecular changes during progression from nonmuscle invasive to advanced urothelial carcinoma. Int J Cancer 2020;146(9):2636–2647. doi: 10.1002/ijc.32737.
17. Sjödahl G. Molecular subtype profiling of urothelial carcinoma using a subtype-specific immunohistochemistry panel. Methods Mol Biol 2018;1655:53–64. doi: 10.1007/978-1-4939-7234-0_5.
18. Guo CC, Bondaruk J, Yao H et al. Assessment of luminal and basal phenotypes in bladder cancer. Sci Rep 2020;10(1):9743. doi: 10.1038/s41598-020-66747-7.