For this study, the authors used whole transcriptomic analysis of 178 bladder tumors (158 NMIBC, 20 MIBC/metastatic) from FFPE tissue, incorporating messenger RNA expression, splice variants, gene fusion, mutation detection and immune checkpoint inhibitor cascades. The Molecular Grade Related Index (MGRI) was assessed based on the pathologist’s identification of necrosis, hemorrhage, and or bladder cancer. The CTLA, PD-1, LAG3, TIM3, TIGIT and B7 were compiled as an index including all major cascade genes. Data were integrated and tested for correlations with pathological grading and clinical outcomes. Conventional pathological grading for WHO 1973 (Grade 1, 2, 3) and 2004 (low grade vs high grade) classifications was reviewed by three expert uro-pathologists. Kappa statistic for interobserver variability was calculated. For model validation, the authors used an independent RNA-seq dataset (n = 209 .
Unsupervised clustering of RNA-Seq data distinguished three molecular subtypes of NMIBC: MGRI1, MGRI2, and MGRI3. MGRI1 was comprised almost exclusively of low grade tumors, and MGRI3 clustered with high grade MIBC. Kappa for interobserver variability of expert pathologists was 0.46 and 0.77 in 1973 and 2004 WHO classification, respectively; almost all discrepant tumors were MGRI2. FGFR3 mutations, FGFR3:TACC3 fusion events and MGRI1 genes were associated with components of xenobiotic metabolism (p = 2.51x10-09) signaling systems, in particular, GTPase regulation (p = 0.002), respiratory cycle genes (p = 0.004), HOX cluster (p = 0.005). Eighty-eight percent of MGRI2-associated differentially expressed genes were shared within the MGRI1/3 cluster; MGRI2 is not a distinct molecular subtype but rather a blended display of MCGI1 and MGRI3 genes. MGRI independently predicted progression to MIBC (n = 138, HR = 2.96, 95%CI 1.70-5.13). 5-year PFS in a combined data set (n = 347) differed significantly for MGRI1 (100%) vs MGRI2 (92.2%) vs MGRI3 (73.5%, p = 1.99x10-05, Gray’s test). The PD-1 pathway immune checkpoint cascade independently predicted for recurrence, however there were insufficient events at this time point to evaluate progression.
Dr. Chandrasekar noted several take home points for his presentation, highlighting that RNA-seq delineates three molecular classes of NMIBC and that the current WHO classification does not correspond to biological reality. Importantly, molecular grading predicts progression to MIBC and overcomes pathological interobserver variability. Further evaluation of the immune checkpoint landscape is warranted in NMIBC.
Speaker: Thanappan Chandrasekar, University of Toronto, Princess Margaret Cancer Center, Toronto, ON, Canada
Co-Authors: Alexandre Zlotta, Jess Shen, Aidan Noon, Haiyan Jiang, Annette Erlich, Cynthia Kuk, Ruoyu Ni, Balram Sukhu, Kin Chan, Morgan Roupret, Thomas Seisen, Eva Comperat, Joan Sweet, Girish S. Kulkarni, Neil Eric Fleshner, Azar Azad, Theodorus van der Kwast, Jeff L Wrana
Written by: Zachary Klaassen, MD, Urologic Oncology Fellow, University of Toronto, Princess Margaret Cancer Centre, Twitter: @zklaassen_md, at the 2018 American Society of Clinical Oncology Genitourinary (ASCO GU) Cancers Symposium, February 8-10, 2018 - San Francisco, CA
1. Hedegaard J, Lamy P, Nordentoft, I, et al. Comprehensive Transcriptional Analysis of Early-Stage Urothelial Carcinoma. Cancer Cell 2016;30(1):27-42.