Introduction:

While lower tract urothelial carcinoma has undergone multiple large gene expression profiling initiatives including the TCGA, upper tract urothelial carcinoma (UTUC) has had relatively little evaluation. Only two prior studies have utilized gene expression profiling to identify that UTUC are mostly luminal papillary with Tcell depleted immune environments. Moss et al (EU, 2017) performed an unsupervised consensus clustering of RNA-seq gene expression data with 4 clusters identified on 31 specimens and attempted to correlate with survival, but limited due to small numbers. We sought to leverage the large volume of UTUC cases at MSKCC and determine if gene expression in UTUC can correlate with outcomes of these patients and specifically recurrence of these often aggressive tumors.

Methods:

Leveraging the large volume of UTUC cases at MSKCC we identified 83 nephroureterectomy specimens that were suitable for RNA-sequencing and passed quality metrics. These were patients who were not metastatic at time of surgery and no prior muscle invasive bladder cancer. Unsupervised consensus clusters of the specimens were generated utilizing RSEM for alignment and DESeq2 for analysis. Clinical characteristics including smoking status, gender, tumor grade, stage, tumor location, prior therapy and survival and recurrence were evaluated between the generated clusters. Kaplan-Meier curves were generated to evaluate survival between the clusters as well.

Results:

83 patients with sequenced nephroureterectomy primary specimens. Utilizing the best fit for analysis 3 consensus clusters (Figure 1) were identified with no difference in age, gender, smoking status, prior history of non-muscle invasive bladder cancer or neoadjuvant/adjuvant chemotherapy rates. The main difference is seen by cluster 2 of which all 18 patients were >pT1 (invasive), harbor variant histology and had increased nodal involvement. This translated to significantly worse recurrence-free survival, cancer-specific survival and overall survival for Cluster 2 patients (Figure 2). Cluster 2 is characterized with high TP53 mutations and low FGFR3 mutations and a higher immune cell infiltration. When we utilize prior consensus classifiers previously validated for use in lower tract urothelial carcinoma, we can see that majority of the patients are luminal papillary. The only cluster that sees basal gene expression is within Cluster 2.

Conclusion:

Upper tract urothelial carcinoma can be classified by three clusters based on whole gene expression profiling. The cluster 2 patients have significantly higher recurrence rates with worse survival. This cluster was associated with high TP53 and low FGFR3 mutations with higher immune cell infiltration. While not clinically validated, genomic expression classification of these tumors provides clusters that may contribute to future signatures to identify those that may benefit from multimodality treatments to decrease the rates of recurrence and cancer-specific death.

Funding: This work was supported by the Sidney Kimmel Center for Prostate and Urologic Cancers, the Michael and Zena Wiener for Therapeutics Program in Bladder Cancer, Pin Down Bladder Cancer, Cycle for Survival, the Marie-Josee and Henry R. Kravis Center for Molecular Oncology, NIH/NCATS Grant Number UL1-TR002384, the National Cancer Institute Cancer Center Core Grant Number P30-CA008748 and by SPORE in Bladder Cancer P50-CA221745.