The Cancer Genome Atlas Expression Subtypes Stratify Response to Checkpoint Inhibition in Advanced Urothelial Cancer and Identify a Subset of Patients with High Survival Probability

Analysis of the IMvigor 210 trials involving patients with platinum-refractory or cisplatin-ineligible urothelial carcinoma who were treated with the PD-L1 inhibitor atezolizumab identified a resistance signature as an immune biomarker. Transcriptome profiling of 368 tumor samples from this trial revealed that the “genomically unstable” Lund subtype classification was associated with the best response. We developed and applied a novel single-patient subtype classifier based on The Cancer Genome Atlas 2017 expression-based molecular subtypes. We identified 11 patients with a neuronal subtype, with a 100% response rate in eight confirmed cases (2 complete response, 6 partial response), and 72% overall, including 3/11 patients with an unconfirmed response. The survival probability was extraordinarily high for the neuronal subtype, which represents a high-risk cohort with advanced disease, and may be secondary to low levels of TGFβ expression and high mutation/neoantigen burden. Patient summary: We describe a methodology for genomic classification of an individual patient's bladder cancer tumor and have identified a subtype that is associated with a high response rate to immunotherapy. This is an important step forward in identifying the right treatment for the right patient, which is the goal of personalized precision medicine. We describe a novel single-patient classifier based on The Cancer Genome Atlas 2017 scheme that identified the neuronal subtype of urothelial carcinoma as an extreme responder to anti-PD-L1 therapy. In the future, trials targeting subtype-based therapy may improve precision delivery of care for urothelial carcinoma.