**2. Early bladder cancer subtyping systems**

The earliest representative studies on the molecular classification of bladder cancer were conducted by Sjödahl et al. at the University of Lund, initially proposed in 2010 and finalized in 2012 and the tumor differentiation classification

by Chan et al., of the Baylor University group, 2012 [19]. The Baylor classification focused on tumor biology. They proposed a KRT14/Thy-1/CD44 expressing cancer stem cells as the bladder cancer precursor. This cancer stem cell evolved into a partially differentiated KRT5/KRT17/CD44-positive progeny, which in turn acquired KRT8/18 expression and eventually differentiated into luminal cells expressing uroplakins and KRT20. Tumors were classified into basal, intermediate and differentiated classes based on their resemblance to normal urothelial differentiation. Based on the above classification, the KRT14 group (basal subtype) showed the poorest prognosis and were also found to be resistant to neoadjuvant cisplatin-based chemotherapy.

The Lund University group initially defined two intrinsic molecular subtypes named as MS1 and MS2 based on gene expression, genomic, and gene mutation levels by whole genome comparative genomic hybridization and mutation studies [7]. By combining molecular and pathologic data, it was possible to divide tumors into grade 1 or 2 (MS1) and grade 3 (MS2) (WHO1999), and into Ta (MS1) and ≥ T2 (MS2) stages based on the MS1 and MS2 subtypes.

Subsequently, an extensive biological interpretation of gene expression data identified that biological themes including immune, late cell cycle, keratin, receptor tyrosine kinases and FGFR3 signatures determined their data structure. Based on tumor histopathology, gene signatures, and status of FGFR3, PIK3CA, and TP53 mutations, three major subtypes of UC were defined: urobasal (Uro) (further subdivided into MS1a, MS1b, and MS2b2.1), genomically unstable (GU) (MS2a1 and MS2a2), and SCC-like (SCCL) (MS2b2.2) [8]. Subsequent studies also identified an "infiltrated" group in which the stromal inflammatory transcripts were prominently expressed. Among the urobasal tumors (MS2b2.1), a subset showed "progressed phenotype" with aberrant expression of basal keratins in suprabasal cell layers and upregulation of late cell cycle activity. These tumors were mostly large and invasive, and were named urobasal B to distinguish it from urobasal A tumors that were non-muscle invasive in almost all cases. The molecular subtypes thus defined, transcended pathological staging and all four subtypes (UroA, UroB, GU, and SCCL) were detected among T1 tumors. The initial Lund taxonomic studies included both NMIBC and MIBC and subsequent studies focussed predominantly on MIBC classification.

#### **3. Molecular subtyping of early-stage bladder cancer**

Variability in terminology has created a challenge in the molecular classification of early-stage bladder cancer. Treating clinicians emphasize the dichotomous division of BCs into NMIBC and MIBC, and often lump all NMIBCs together when planning molecular studies. In contrast pathologists tend to see a stark difference between non-invasive tumors and invasive tumors limited to the lamina propria and classify accordingly.

Molecular diversity in non-invasive BC differs from that of MIBC. Non-invasive BC histologically includes papillary UC and flat CIS, although both may co-exist in the same patient. As we have discussed earlier, low-grade non-invasive papillary UC has a high frequency of FGFR3 mutation. LGUC progresses to HGUC and invasive carcinoma through the acquisition of TP53 mutations and 9p21 loss involving the gene encoding CDKN2A. In contrast, most CIS lesions have TP53 mutations early in evolution and do not acquire FGFR3 mutations [4, 5].

In the Lund system [8], majority of non-invasive UCs are urothelial-like, while CIS may be either urothelial-like or genomically unstable. They identified a "CIS signature" by utilizing a 16-gene classifier which was specifically expressed in flat

#### *Molecular Classification of Bladder Cancer DOI: http://dx.doi.org/10.5772/intechopen.97393*

CIS, as well as in early-stage invasive carcinoma with associated CIS, and a large proportion of MIBCs of the basal-squamous subtype. In addition, MIBCs with concurrent CIS had greater genomic instability compared with those without it. Although majority of non-invasive papillary UCs were of luminal subtype, there is substantial molecular diversity among cases. The most clinically relevant diversity was related to cell cycle regulatory genes. Tumors with greater activation of the cell cycle had higher rates of recurrence and progression to MIBCs.

The UROMOL study of 2016 [20] evaluated expression profiles of NMIBCs including non-invasive papillary UC and invasive UC limited to the lamina propria (stage T1). A few cases of CIS and a small group of MIBCs were also included for comparison. Tumors were classified into 3 subtypes based on relative expression of luminal and basal-squamous markers and cell cycle activity. Three subtypes were proposed, named as Type 1 (early cell cycle activation and higher luminal gene expression), Type 3 (early cell cycle activation with lower luminal gene expression) and Type 2 (late cell cycle activation). Type 2 tumors, which included the highest proportion of T1 samples, had the greatest propensity to progress to muscle invasion. On the other hand, expression of luminal genes did not significantly affect patient outcome. In addition, non-invasive papillary tumors also varied in the degree of chromosomal instability. The unstable group had tumors with higher proliferation, greater mutational burden, and high-grade histology.

Von Kessel et al., in 2018 [21], determined that methylation status of GATA2, TBX2, TBX3 and ZIC4 and mutations in FGFR3, TERT, PIK3CA, and RAS correlated with progression rates of NMIBC. Wild-type FGFR3 and GATA2 and TBX3 methylation were significantly correlated with NMIBC progression. Thus, high risk NMIBC group was reclassified into good, moderate and poor prognostic classes with low, medium and high risk of progression.

However, because molecular subtyping of non–muscle-invasive bladder cancer has not demonstrated clear value in clinical decision making, it is not currently recommended to incorporate it on a routine basis, as per the ISUP recommendations published in 2020.
