Introduction

Non muscle invasive bladder cancer (NMIBC) has a highly variable clinical behaviour not adequately predicted by their histological grade and clinical parameters. Some are indolent; others quickly progress to muscle-invasive disease. The discrepancy between phenotype and genotype is compounded further by interobserver variability in pathological grading.

Methods

Whole transcriptomic (WT) analysis of 178 bladder tumours (158 NMIBC and 20 MIBC or metastatic) was performed from formalin fixed paraffin embedded (FFPE) tissues incorporating messenger RNA expression, splice variants, gene fusion and mutation detection. In NMIBC, we used a discovery (n=38) and 2 validation cohorts (n= 40 and 80). These data were integrated and tested for correlations with both pathological grading and clinical outcomes. Conventional pathological grading for both WHO 1973 (grade 1, 2 and 3) and 2004 (low grade-LG vs high grade-HG) classifications was reviewed by 3 different expert uro-pathologists and kappa statistic for interobserver variability was calculated.

Results

Unsupervised clustering of data from RNA sequencing revealed classification of three robust-non-overlapping molecular subtypes of NMIBC termed Grade Related Index (GRI) 1, GRI2 and GRI3. GRI1 comprised of almost exclusively LG tumours, while GRI3 clustered with HG MIBC tumours. After assessment by expert pathologists, kappa for interobserver variability in 1973 WHO histological grading was 0.40 whereas it reached 0.78 for the 2004 classification. Most discrepant cases clustered in molecular subtype GRI2. GRI subclassification independently predicted disease progression in NMIBC (p=0.004, gray test). FGFR3 mutations, FGFR3::TACC3 fusion events and Hedgehog were strongly enriched in GRI1. GRI3 disease was associated with a germ stem cell-like phenotype and upregulation in APOBEC3B.

Conclusions

WT sequencing data delineated three molecular classes of NMIBC, and improved prediction of disease progression from NMIBC to MI compared to conventional histologic grading. WT analysis could be integrated to a new WHO classification.

Funding

Canadian Cancer Society - Impact Grant