Transcriptomic and Protein Analysis of Small Cell Bladder Cancer Identifies Prognostic Biomarkers and DLL3 as a Relevant Therapeutic Target - Beyond the Abstract

Small cell bladder cancer (SCBC), also known as neuroendocrine bladder cancer, is a rare histological variant that is difficult to treat and frequently presents with more advanced disease as compared to the more common urothelial carcinoma. It shares similarities with both urothelial bladder cancer and small cell lung cancer, and treatment patterns in both of these malignancies inform treatment options in SCBC in the absence of prospective data. The genomics of this disease is also poorly understood as small cell bladder cancer samples were not included in the initial version of the Cancer Genome Atlas (TCGA) published in 2014 and only a few samples morphologically identified as small cell bladder cancer were included in the subsequent update in 2017. 

The general lack of data regarding this rare histological variant of bladder cancer, and in particular a dearth of understanding of its genomics and the mechanisms underlying pathophysiology of SCBC led us to undertake this investigation. We started with the identification of all small cell and neuroendocrine bladder cancer cases that were seen at Cleveland Clinic over the previous 30 years through the review of pathology records. This review yielded clinical, demographic and treatment characteristics of 63 identified patient cases. Importantly, significant subsets of these patients had archived tumor tissue available for immunohistochemistry analysis of tumor markers (n = 53) as well as analysis of mRNA expression (n = 39) using EdgeSeq oncology biomarker panel. This allowed for the correlation of gene and protein expression with relevant pathologic and clinical characteristics including overall survival (OS). Overall among the 39 patients who had tissue available for gene expression analysis, gene expression profiling was performed on 39 primary tumor samples as well as 6 adjacent normal urothelium samples and 1 metastatic sample from the same patients. Differential gene expression analyses comparing tumor and adjacent “normal” urothelial samples nominated several potential therapeutic targets including DLL3, PD-L1, ASCL1, and NCAM1/CD56. The protein expression of these tumor markers was further confirmed by immunohistochemistry analyses. Finally, an SCBC PDX model was also used to assess the in vivo efficacy of a DLL3-targeting antibody-drug conjugate (ADC) that is currently in early phase clinical trials in other DLL3-expressing malignancies.

Our transcriptomic analysis identified mRNA expression patterns that correlated with distinct clinical phenotypes including more aggressive and more indolent disease. Tumors with a more “normal-like” pattern of gene expression came from patients who clinically did better and had a longer OS compared to patients whose tumors had a “metastasis-like” pattern of gene expression. Furthermore, unsupervised hierarchical clustering revealed that the gene expression pattern of any small cell tumor was more closely clustered with other small cell tumors than to adjacent “normal” urothelial tissue from the individual in which that tumor arose; this suggests a convergent evolution towards a similar pathologic phenotype. Similarly, gene expression patterns associated with metastatic disease clustered most closely with gene expression of the primary tumor sample from which that metastasis arose. Altogether these data support a model where a clonal event in the development of this aggressive variant is captured in the gene expression pattern of localized disease and persists through the process of metastatic progression. This is consistent with the early aggressive phenotype of small cell tumors and has implications for biomarker development and risk stratification.

We found the expression of certain biomarkers including DLL3, ASCL1, and NCAM1/CD56 to be common in SCBC, and also found DLL3 and CD56 protein expression to be prognostic of clinical outcomes, including a shorter OS in a multivariate analysis. The strong correlation of gene and protein expression of both DLL3 and NCAM1/CD56 suggested that this expression is regulated at the transcriptional level. Finally, we tested the hypothesis that DLL3 could serve as a predictive biomarker and therapeutic target for a DLL3-targeting ADC. Indeed, in a PDX model of SCBC, this agent showed durable efficacy as compared to standard of care chemotherapy, with preferential targeting of DLL3-expressing tumor-initiating cells.

To our knowledge, this was the largest study to investigate tumor markers and genomics in small cell bladder cancer. In this multifaceted analysis, we were able to show that gene expression patterns in SCBC associate with clinical phenotypes that range from more indolent to very aggressive disease and may have a potential impact on treatment selection. We found that overexpression of DLL3 mRNA and protein is common in SCBC and expression levels were prognostic of worse outcomes. Moreover, DLL3 may be an important predictive biomarker and therapeutic target in SCBC. We showed preclinical data demonstrating in vivo efficacy of a DLL3-targeted antibody-drug conjugate in a PDX model of SCBC. Altogether, these data enhance our understanding of small cell bladder cancer biology and may lead to improvements in treatment selection, clinical trial design, and patient care for this rare but aggressive disease. 

Written by: Vadim S Koshkin, MD, Assistant Professor, Hematology and Oncology, Hematology and Oncology, University of California San Francisco, UCSF Health, Petros Grivas, MD, PhD, medical oncologist, Seattle Cancer Care Alliance,  Clinical Director, Genitourinary Cancers Program, University of Washington Medicine, Associate Professor, Department of Medicine, Division of Oncology, University of Washington School of Medicine, and Omar Y Mian, MD, Radiation Oncology, General Radiation Oncology, Urologic Cancer, Cleveland, OH

Read the Abstract