Urologic Oncology: Seminars and Original Investigations ] (2017) ∎∎∎–∎∎∎
Survey Section Submission 3 Commentary on: “Comprehensive transcriptional analysis of early-stage urothelial carcinoma.” Hedegaard J, Lamy P, Nordentoft I, Algaba F, Høyer S, Ulhøi BP, Vang S, Reinert T, Hermann GG, Mogensen K, Thomsen MB, Nielsen MM, Marquez M, Segersten U, Aine M, Höglund M, Birkenkamp-Demtröder K, Fristrup N, Borre M, Hartmann A, Stöhr R, Wach S, Keck B, Seitz AK, Nawroth R, Maurer T, Tulic C, Simic T, Junker K, Horstmann M, Harving N, Petersen AC, Calle ML, Steyerberg EW, Beukers W, van Kessel KE, Jensen JB, Pedersen JS, Malmström PU, Malats N, Real FX, Zwarthoff EC, Ørntoft TF, Dyrskjøt L. Cancer Cell. 2016 Jul 11;30(1):27–42 Abstract Non–muscle-invasive bladder cancer (NMIBC) is a heterogeneous disease with widely different outcomes. We performed a comprehensive transcriptional analysis of 460 early-stage urothelial carcinomas and showed that NMIBC can be subgrouped into 3 major classes with basal- and luminal-like characteristics and different clinical outcomes. Large differences in biological processes such as the cell cycle, epithelial-mesenchymal transition, and differentiation were observed. Analysis of transcript variants revealed frequent mutations in genes encoding proteins involved in chromatin organization and cytoskeletal functions. Furthermore, mutations in well-known cancer driver genes (e.g., TP53 and ERBB2) were primarily found in high-risk tumors, together with APOBEC-related mutational signatures. The identiﬁcation of subclasses in NMIBC may offer better prognostication and treatment selection based on subclass assignment. Commentary The Cancer Genome Atlas reported on the comprehensive molecular characterization of 131 chemotherapy-naïve high-grade muscleinvasive urothelial carcinomas of the bladder in 2014. Analysis of RNA-seq data revealed 4 clusters characterized by distinct gene expression signatures. Cluster I (“papillary-like”) is associated with papillary tumors and FGFR3 alterations. Clusters I and II were associated with high levels of ERBB2 expression, gene expression signature consistent with increased ESR2 signaling, and increased expression of CDH1, urothelial differentiation markers such as UPK3A, the miR-200 family of miRNAs, and the luminal differentiation markers GATA3 and FOXA1. These features were similar to the expression proﬁles of luminal A breast cancers. In contrast, Clusters III and IV showed similarities to basal-like breast cancers and squamous cell carcinoma of the head and neck and lung with high expression of KRT5, KRT6A, KRT14, and EGFR. Gene expression subtypes in muscle-invasive bladder cancer (MIBC) were also identiﬁed by other groups, and there is currently consensus for division into a basal subset characterized by high KRT5/6, high KRT14, low FOXA1, and low GATA3 expression and a more differentiated subset with urothelial features deﬁned by high GATA3, high FOXA1, high KRT20, and high miR99/miR100 expression and enrichment in FGFR3 alterations. Hedegaard et al. performed a complementary study by analyzing the transcriptomes of 460 non–muscle-invasive bladder cancers and 16 MIBC from 10 European institutions. The clinicopathologic features of this group were diverse, and there was good representation from lesions of different grade (papillary urothelial neoplasm of low malignant potential , low grade, and high grade) and stage (carcinoma in situ [CIS], Ta, and T1). Most lesions showed a papillary growth pattern, had size less than 3 cm, and were Bacillus Calmette-Guérin (BCG)-naïve. Unsupervised clustering of the specimens was performed using 8,074 genes, and the authors noted 3 major classes. Compared with Class 1, Classes 2 and 3 more frequently contained tumors that were of higher stage and grade, associated with CIS, and more likely to progress to MIBC. The authors then analyzed the 3 classes with respect to genes known to control cellular processes such as cell cycle, differentiation, and epithelial-mesenchymal transition. Class 1 tumors showed high expression of early cell cycle genes, whereas Class 2 tumors showed high expression of late cell cycle genes. When cytokeratin and differentiation marker expression were examined, Classes 1 and 2 tumors were similar to the luminal tumors seen in The Cancer Genome Atlas (TCGA) study, with Class 2 tumors showing more aggressive features such as enrichment for cancer stem cell and epithelial-mesenchymal transition markers. On the contrary, class 3 tumors were similar to the basal tumors seen in TCGA study. They were able to successfully reduce the gene set to 117 genes for classiﬁcation into the 3 groups, and validation was performed in a separate dataset consisting of 130 bladder tumors and 8 benign samples. Importantly, the authors showed differences in clinical outcome, with Class 2 tumors showing a decreased progression-free survival. High-risk clinicopathologic features such as higher stage and grade, concomitant CIS, and progression to MIBC were also more common in Classes 2 and 3 when compared with Class 1.Signiﬁcantly, no differences in response to BCG were observed among the 3 classes,
1078-1439/r 2017 Elsevier Inc. All rights reserved.
B.H. Lee / Urologic Oncology: Seminars and Original Investigations ] (2017) ∎∎∎–∎∎∎
suggesting that predictors of BCG response may not be entirely deﬁned by a tumor’s transcriptional proﬁle. Currently, the risk of non–muscle-invasive bladder cancers recurrence and progression can be predicted by using one of several scoring systems that include a number of clinicopathologic features. This study adds molecular information in the form of RNA expression and may be useful in reﬁning or ultimately replacing these scoring systems. http://dx.doi.org/10.1016/j.urolonc.2017.07.020 Byron H. Lee, M.D., Ph.D.