You are viewing a javascript disabled version of the site. Please enable Javascript for this site to function properly.
Go to headerGo to navigationGo to searchGo to contentsGo to footer
In content section. Select this link to jump to navigation

What is a Bladder Cancer Molecular Subtype? – Counterpoint


In an accompanying paper, Mattias Höglund discusses on what is a bladder cancer molecular subtype. He emphasizes the need to consider the aim of tumor classification, which is obviously critical to the approach. He also focuses on considering primarily the identity features of the neoplastic cells. Here, we provide a counterpoint. While largely agreeing with his views, we underline that other parameters that may vary in a spatial or temporal scale, and the tumor microenvironment, can also provide relevant information to render tumor classifications clinically useful. Furthermore, tumor heterogeneity and evolution during the disease course - natural or under therapeutic pressure - should be considered.

The “Commentary” article by Mattias Höglund [1], an expert and pioneer in the field of bladder cancer molecular classification [2–4], is certainly instructive and useful. His historical perspective points to the fact that humans have long aimed at classifying “the world around them” in order to better understand it.

Many of the points he raises are important, including the clear distinction between classifications primarily oriented to clinical applications and those oriented to dissect the biology of the tumor. The former are largely based on immunohistochemistry to characterize tumor cells and the latter use so far mainly bulk whole transcriptome data to assign tumors to molecular subtypes without discriminating which cell populations express the subtype-associated gene sets. As he points out, these two types of classifications do not always match. He asserts that molecular classification should be based on the phenotype of the tumor cells themselves, and not include the microenvironment, to define “intrinsic” subtypes (based on features of the cancer cells). He also argues that the molecular classification should also not depend on “state-related” tumor cell features such as proliferation signatures because proliferation is a continuous parameter that can change across time. Similarly, he proposes to avoid the use of tumor microenvironmental features to classify tumors because the extent of intratumoral stromal/inflammatory cell infiltration can be heterogeneous and therefore subject to sampling bias, and it can also change across time and in response to therapeutic pressure. Finally, he defends the idea that classification should encompass both non-muscle-invasive (NMIBC) and muscle-invasive (MIBC) tumors.

All these points merit further thought and many of them are relevant beyond bladder cancer. However, a few of them acquire particular importance in the context of this tumor. More specifically, bladder tumors display wide inter-tumor molecular heterogeneity [5], muscle invasion has a much greater impact on patient management than in other carcinomas, and sampling bias may also be a greater problem. In addition, the impact of intra-tumor heterogeneity is starting to be unraveled [6, 7], and bladder cancer has one of the highest rates of mutations in genes coding for chromatin regulators [8] – pointing to the relevance of epigenetic analyses and “lineage plasticity” – The ongoing revolution in bladder cancer therapy demands urgently better tools for precision medicine.

We would like to discuss some of the ideas that Mattias Hoglund raises as they have important consequences for future bladder cancer research.


The earliest efforts to classify tumors at the molecular level were based on transcriptome profiling because they were the first large-scale datasets that became available, thanks to cDNA arrays and later oligo arrays. This, combined with the fact that the transcriptome is more easily interpretable than other available omics, explains why most molecular classifications performed to date are based on the protein-coding transcriptome. However, many other types of data can be used now, including the non-coding transcriptome, somatic genomic alterations (including mutations and copy number changes), the epigenome (DNA methylome and chromatin accessibility), the proteome and post-translational modifications (such as the phosphoproteome), and the metabolome. The TCGA integrated many of these features to produce “clusters of clusters” to subtype bladder cancers and other solid tumors. Other features, such as variation in germline DNA, including mutations in tumor suppressor genes, could also be considered [9, 10]. The greater difficulties in data generation, biological interpretation (e.g., the non-coding transcriptome and the DNA methylome), and cost (e.g., the metabolome and chromatin accessibility) have made transcriptomic classifications more popular. However, all the other molecular features will provide additional information. The different layers of data may reinforce existing subtypes, introduce further subdivisions, and – in some cases – highlight similarities between tumors that are currently ascribed to different transcriptomic molecular subtypes. They could also point to genes and pathways that are highly relevant for bladder cancer tumorigenesis but are difficult to capture using the transcriptome alone [11]. Thanks to recent advances [12, 13], we envision that the proteome – which is readily interpretable – will become an important data layer in the short term [11, 14].

Importantly, the clinical relevance of each data type will mandate which classifiers become standard beyond biology. In lung cancer, for example, DNA analyses have major impact in patient management and therefore dominate the genomic taxonomy. By contrast, the management of patients with breast or colorectal cancer responds to a combination of DNA-based, transcriptomic, and protein expression features.


In most cancers, the key event required for tumor progression/dissemination is the invasion of the basement membrane. In bladder cancer, an additional key event is the invasion of the muscularis propria which has a major impact on management. Invasion of the bladder wall can be viewed as a continuous (progressively invading) or discontinuous phenomenon, whereby additional properties must be acquired by the tumor cells, or even possibly in certain cases only by a phenotype switch (i.e., epithelial-to-mesenchymal transition) [15]. The rather unique binary classification of bladder tumors according to muscle-invasion has strong clinical prognostic value and currently serves as the basis for deciding whether a patient needs definitive therapy (radical cystectomy or trimodal therapy). Therefore, one can argue – with good reason – that NMIBC and MIBC are distinct disease entities. One can go even further by studying Ta and T1 tumors separately [16, 17].

In contrast, to acquire a global view of bladder cancer progression, the joint analysis of NMIBC and MIBC is a tempting way to go. For example, this strategy has recently been applied to the molecular classification of upper tract urothelial tumors [18]. Several issues have hampered this effort in bladder cancer. One is that a large fraction of MIBC are diagnosed de novo, i.e., in patients without a clinical history of NMIBC. This implies that most of the precursors of MIBC are understudied. A second issue is that the tissue micro-environment (TME) of NMIBC and MIBC are substantially different. Therefore, in simple unsupervised analyses, one of the first divisions that will become apparent will be the division between MIBC and NMIBC. To circumvent this problem, the group of Mattias Höglund has pioneered the approach of focusing on tumor cells [1].


As stressed by Mattias Höglund, the aim of classifying is key to the outcome: one can perform an unsupervised classification mainly based on the biology of the tumors, with a subsequent view to the relevance to the diagnosis and response to treatment. Another way to go is to focus on a classification aimed at predicting such clinical parameters using supervised analyses. The clinically oriented classification is, by its construction, highly dependent of the treatments used, whereas the first one is not. The often-protracted clinical course of patients with bladder cancer introduces major confounding factors in the classifications, especially when considering incident and prevalent tumors together. However, the clinically oriented and the biologically oriented classifications need not be considered as mutually exclusive. Indeed, diagnostic or prognostic markers can be specific for a given biological subtype (for example, CDKN2A loss has been reported to be a marker of tumor progression in FGFR3-mutated NMIBC) and a biological subtype could be, in itself, a clinical marker (for example, the basal subtype is associated with poor prognosis) [19, 20].


Focusing on the tumor cells enables to jointly study NMIBC and MIBC and may avoid sampling biases. However, recent single-cell RNAseq data indicate that all bladder cancers contain mixtures of tumor epithelial cells that can be assigned to different molecular subtypes [21]. These studies have also revealed that cells within the TME are also highly heterogeneous. The diversity of the TME, with CAFs, nerve cells, T-cells, macrophages, and other hematopoietic cells that can also be assigned to different molecular subtypes [22] adds complexity but may also allow for refinement in classifiers. Overall, it is unclear how tumor cell and stromal cell heterogeneity will affect our ability to develop robust prognostic and predictive biomarkers. With that said, bulk gene expression signatures remain among the best predictive biomarkers for immune checkpoint inhibitors – cytotoxic T cell and interferon signatures are associated with benefit, whereas TGF-beta signatures are associated with resistance [23]. There are different types of TME, each with different dialogues with tumor cells; therefore, considering the TME in the molecular classification can be of great importance. The stromal content could result from sampling bias, in which case one could question the existence of the stroma-rich subtype of the consensus classification [24] and the largely overlapping “P53-like” subtype of the MDA classification [25] as well as the “luminal-infiltrated” subtype of the TCGA classification [8]. On the other hand, it is also conceivable that the stroma reflects features that are dictated by the tumor cells, indeed reflecting crucial aspects of the tumor biology that may impact on specific tumor – stroma paracrine interactions and specific clinical features.


As with the stroma, proliferation can be viewed as a continuous feature. However, different types of proliferation can exist, even in normal cells, such as symmetric or asymmetric division. In bladder cancer, mutations, fusions, and copy number changes in genes involved in the cell cycle (e.g., CCND1, CCNE1, CDKN2A, RB1) are frequent relevant events. Tumors enriched in early vs. late cell cycle genes have been reported [3, 26]. Therefore, proliferation is not only a quantitative trait (percentage of cells cycling) but also a qualitative trait and both types of information may be incorporated in the molecular classification.


In a given bladder cancer patient, the different tumors diagnosed either synchronously or metachronously generally originate from a single somatic cell (clonal origin of tumors). During tumor evolution, different subclones arise [27]. Although the different subclones often belong to the same molecular subtype, differences can be observed. In the most used mouse model of bladder cancer, induced by BBN [28, 29] a switch to the basal subtype is frequent, starting at an early progression stage [15]. In humans, a similar phenotype switch could occur at the CIS stage [30, 31] and different subtypes can be observed in the same tumor (estimated to be present in 7 to 15% of MIBC cases) [6, 21, 32]. This subtype heterogeneity must be also considered in terms of clinical outcomes, including chemotherapy response [6]. How to best approach this point from a technical standpoint in clinical practice remains an important task. Artificial intelligence-based image analyses and multiparameter in situ approaches (as subtype cannot be determined by a set of antibodies alone) are some of the available options. In this regard, single cell and spatial transcriptomics will make significant contributions [21, 33, 34]. Issues related to reproducibility, standardization, and cost need to be considered.


Claude Bernard already pointed out “there are no diseases, there are patients “, in agreement with the fact that each tumor results from the myriad of possible combinations of mutations and genotypes. Therefore, as indicated by Mattias Höglund, ultimately, the number of subtypes could be as high as the number of tumors (personalized medicine). In bladder cancer, the simplest division would be to differentiate between luminal and non-luminal tumors. Further subdivisions of luminal and non-luminal tumors can be considered: among the non-luminal neoplasms, the basal/squamous tumors and the neuroendocrine-like tumors have distinct molecular and clinical features; among the luminal tumors, a subgroup enriched in FGFR3 mutations emerges in many molecular classification systems [24]. As the number of samples analyzed with omics data increases, the finer the subdivision into subtypes can be.

Subtypes are important for the identification of altered genes and pathways specific to a group of tumors and for stratification-based medicine. The future goal of precision medicine is to tailor treatments to individual patients and, therefore, to specifically target the different tumor subclones, the genes and pathways that are altered, and to predict their evolution under treatment pressure. The rapid development of new technologies should make these achievable goals in the foreseeable future: the implementation of liquid biopsy, bioinformatics tools, and artificial intelligence are examples thereof. In the meantime, molecular classification continues to play a valuable role and the establishment of consensus agreement has been a key step to facilitate communication among the scientific community [24, 35]. Through experience with several other tumor types, including breast [36–38] and brain cancers [39], we now know that it is feasible and clinically useful to group tumors into molecular subtypes. A major challenge is to leverage the knowledge acquired in bladder cancer to prospectively test their value in the management of patients [40]. Clinical trials based on molecular subtyping have already been initiated [41].

We might conclude this counterpoint commentary by saying that, “All classifications are wrong (or not completely right) but some are useful”, paraphrasing statistician George Box [42].


The authors have no acknowledgements.


FR wrote the first draft of the manuscript, with subsequent contributions from FXR and DM. All authors agreed the final version of the manuscript.


None to be reported.


FR, FXR and DM are Editorial Board members of this journal, but have not been involved in the peer-review process nor had access to any information regarding its peer-review.



Höglund M . What is a bladder cancer molecular subtype? Bladder Cancer. 2023; (in press).


Lindgren D , Frigyesi A , Gudjonsson S , Sjödahl G , Hallden C , Chebil G et al. Combined gene expression and genomic profiling define two intrinsic molecular subtypes of urothelial carcinoma and gene signatures for molecular grading and outcome. Cancer Res. (2010) ;70: (9):3463–72.


Sjödahl G , Lauss M , Lövgren K , Chebil G , Gudjonsson S , Veerla S et al. A molecular taxonomy for urothelial carcinoma. Clin Cancer Res. (2012) ;18: (12):3377–86.


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):113–25.


Hoadley KA , Yau C , Wolf DM , Cherniack AD , Tamborero D , Ng S et al. Multiplatform analysis of 12 cancer types reveals molecular classification within and across tissues of origin. Cell. (2014) ;158: (4):929–44.


Groeneveld CDS , Harter V , Culine S , Krucker C , Dixon V , de Reyniès A et al. Pure or mixed basal/squamous tumours present decreased outcomes after neoadjuvant chemotherapy in the GETUG-AFU V05 VESPER. Ann Oncol. (2022) ;33: (suppl_7):S785–S807.


Lindskrog SV , Schmøkel SS , Nordentoft I , Lamy P , Knudsen M , PripF et al. Single-nucleus and Spatially Resolved Intratumor SubtypeHeterogeneity in Bladder Cancer. Eur Urol Open Sci. (2023) ;51: , 78–88.


Robertson AG , Kim J , Al-Ahmadie H , Bellmunt J , Guo G , Cherniack AD et al. Comprehensive Molecular Characterization of Muscle-Invasive Bladder Cancer. Cell. (2017) ;171: (3):540–56.


Carlo MI , Ravichandran V , Srinavasan P , Bandlamudi C , Kemel Y , Ceyhan-Birsoy O et al. Cancer Susceptibility Mutations in Patients With Urothelial Malignancies. J Clin Oncol. (2020) ;38: (5):406–14.


Nassar AH , Abou Alaiwi S , AlDubayan SH , Moore N , Mouw KW , Kwiatkowski DJ et al. Prevalence of pathogenic germline cancer risk variants in high-risk urothelial carcinoma. Genet Med. (2020) ;22: (4):709–18.


Groeneveld CS , Sanchez-Quiles V , Dufour F et al. Proteogenomic Characterization of Bladder Cancer Reveals Sensitivity to Apoptosis Induced by Tumor Necrosis Factorrelated Apoptosis-inducing Ligand in FGFR3-mutated Tumors. Eur Urol. 2023:S0302–2838(23)02885-3.


Rodriguez H , Zenklusen JC , Staudt LM , Doroshow JH , Lowy DR The next horizon in precision oncology: Proteogenomics to inform cancer diagnosis and treatment. Cell. (2021) ;184: (7):1661–70.


Mani DR , Krug K , Zhang B , Satpathy S , Clauser KR , Ding L et al. Cancer proteogenomics: current impact and future prospects. Nat Rev Cancer. (2022) ;22: (5):298–313.


Yao Z , Xu N , Shang G et al. Proteogenomics of different urothelial bladder cancer stages reveals distinct molecular features for papillary cancer and carcinoma in situ. Nat Commun. (2023) ;14: (1):5670.


Fontugne J , Wong J , Cabel L , Neyret-Kahn H , Karboul N , Maillé P et al. Progression-associated molecular changes in basal/squamousand sarcomatoid bladder carcinogenesis. J Pathol. (2023) ;259: (4):455–67.


Hurst CD , Alder O , Platt FM , Droop A , Stead LF , Burns JE et al. Genomic Subtypes of Non-invasive Bladder Cancer with Distinct Metabolic Profile and Female Gender Bias in KDM6A Mutation Frequency. Cancer Cell. (2017) ;32: (5):701–15.


Hurst CD , Cheng G , Platt FM , Castro MAA , Marzouka NS , Eriksson P et al. Stage-stratified molecular profiling of non-muscle-invasive bladder cancer enhances biological, clinical, and therapeutic insight. Cell Rep Med. (2021) ;2: (12):100472.


Fujii Y , Sato Y , Suzuki H , Kakiuchi N , Yoshizato T , Lenis AT et al. Molecular classification and diagnostics of upper urinary tract urothelial carcinoma. Cancer Cell. (2021) ;39: (6):793–809.


Rebouissou S , Hérault A , Letouzé E , Neuzillet Y , Laplanche A , Ofualuka K et al. CDKN2A homozygous deletion is associated with muscle invasion in FGFR3-mutated urothelial bladder carcinoma. J Pathol. (2012) ;227: (3):315–24.


Rebouissou S , Bernard-Pierrot I , de Reyniès A , Lepage ML , Krucker C , Chapeaublanc E et al. EGFR as a potential therapeutic target for a subset of muscle-invasive bladder cancers presenting a basal-like phenotype. Sci Transl Med. (2014) ;6: (244):244ra91.


Sfakianos JP , Daza J , Hu Y , Anastos H , Bryant G , Bareja R et al. Epithelial plasticity can generate multi-lineage phenotypes in human and murine bladder cancers. Nat Commun. (2020) ;11: (1):2540.


Chen Z , Zhou L , Liu L , Hou Y , Xiong M , Yang Y et al. Single-cell RNA sequencing highlights the role of inflammatory cancer-associated fibroblasts in bladder urothelial carcinoma. Nat Commun. (2020) ;11: (1):5077.


Mariathasan S, Turley SJ, Nickles D, Castiglioni A, Yuen K, Wang Y, et al. TGFß attenuates tumour response to PDL1 blockade by contributing to exclusion of T cells. Nature. (2018) ;554: (7693):544–548.


Kamoun A , de Reyniès A , Allory Y , Sjödahl G , Robertson AG , Seiler R et al. Bladder Cancer Molecular Taxonomy GrouA Consensus Molecular Classification of Muscle-invasive Bladder Cancer. Eur Urol. (2020) ;77: (4):420–33.


Choi W , Porten S , Kim S , Willis D , Plimack ER , Hoffman-Censits J 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):152–65.


Hedegaard J , Lamy P , Nordentoft I , Algaba F , Høyer S , Ulhøi BP et al. Comprehensive Transcriptional Analysis of Early-StageUrothelial Carcinoma. Cancer Cell. (2016) ;30: (1):27–42.


van Tilborg AA , de Vries A , de Bont M , Groenfeld LE , van der Kwast TH , Zwarthoff EC Molecular evolution of multiple recurrent cancers of the bladder. Hum Mol Genet. (2000) ;9: (20):2973–80.


Shin K , Lim A , Odegaard JI et al. Cellular origin of bladder neoplasia and tissue dynamics of its progression to invasive carcinoma. Nat Cell Biol. (2014) ;16: (5):469–78.


Van Batavia J , Yamany T , Molotkov A et al. Bladder cancers arise from distinct urothelial sub-populations. Nat Cell Biol. (2014) ;16: (10):982–91.


Wullweber A , Strick R , Lange F , Sikic D , Taubert H , Wach S et al. Bladder Tumor Subtype Commitment Occurs in Carcinoma In Situ Driven by Key Signaling Pathways Including ECM Remodeling. Cancer Res. (2021) ;81: (6):1552–66.


Bondaruk J , Jaksik R , Wang Z , Cogdell D , Lee S , Chen Y et al. The origin of bladder cancer from mucosal field effects. IScience. (2022) ;25: (7):104551.


Sirab N , Drubay D , Maillé P , Popova T , Ngo C , Gentien D et al. Multilayer spectrum of intratumoral heterogeneity in basal bladder cancer. J Pathol. (2022) ;256: (1):108–18.


Gouin KH 3rd , Ing N , Plummer JT , Rosser CJ , Ben Cheikh B , Oh C et al. An N-Cadherin 2 expressing epithelial cell subpopulation predicts response to surgery, chemotherapy and immunotherapy in bladder cancer. Nat Commun. (2021) ;12: (1):4906.


Lindskrog SV , Schmøkel SS , Nordentoft I et al. Single-nucleusand Spatially Resolved Intratumor Subtype Heterogeneity in BladderCancer. Eur Urol Open Sci. (2023) ;51: , 78–88.


Lerner SP , McConkey DJ , Hoadley KA , Chan KS , Kim WY , Radvanyi F et al. Bladder Cancer Molecular Taxonomy: Summary from a Consensus Meeting. Bladder Cancer. (2016) ;2: (1):37–47.


Prat A , Pineda E , Adamo B et al. Clinical implications of the intrinsic molecular subtypes of breast cancer. Breast. (2015) ;24: (Suppl 2):S26–35.


Cardoso F , Kyriakides S , Ohno S et al. Early breast cancer: ESMO Clinical Practice Guidelines for diagnosis, treatment and follow-up†. Ann Oncol. (2019) ;30: (8):1194–1220.


Burstein HJ , Curigliano G , Thürlimann B et al. Customizing local and systemic therapies for women with early breast cancer: the St. Gallen International Consensus Guidelines for treatment of early breast cancer 2021. Ann Oncol. (2021) ;32: (10):1216–1235.


Louis DN , Perry A , Wesseling P et al. The WHO Classification of Tumors of the Central Nervous System: a summary. Neuro Oncol. (2021) ;23: (8):1231–1251.


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. doi: 10.1097/PAS.0000000000002053. Epub ahead of print.



Box , George E.P. Science and statistics. Journal of the American Statistical Association. (1976) ;71: (356):791–9.