Abstract:Abstract Background Renal cell carcinoma with sarcomatoid features (sRCC) is a unique kidney cancer subtype associated with aggressive biology and poor clinical outcomes. Intriguingly, sRCC has recently exhibited preferential responsiveness to immune checkpoint blockade (ICB) therapies. Within the RCC population, however, there currently is a paucity of biomarkers predictive for ICB response, representing a critical unmet clinical need for optimal therapy selection. Therefore, we sought to derive a transcriptomic signature from sRCC samples encapsulating this paradoxical hyper-aggression and ICB-responsiveness to identify RCC patients most likely to benefit from ICB irrespective of sarcomatoid feature presence. Methods Nephrectomy specimens from patients with RCC were processed for single cell RNA sequencing (scRNAseq). Clustering was performed and annotated tumor cells were computationally isolated. Tumor cell counts were aggregated and differential expression between sRCC and non-sRCC cases was performed. Genes significantly upregulated in sRCC were next filtered against differential gene expression data of sRCC vs non-sRCC from three clinical RCC datasets (TCGA, CheckMate, & Javelin101) to identify genes enriched in sRCC across scRNAseq and bulk RNA sequencing. The prognostic and predictive features of this gene signature were then assessed in cohorts of patients with RCC receiving ICB. Results In total, 73,123 unique cells from 18 RCC patients (10 sRCC; 8 clear cell RCC) were analyzed by scRNAseq, including 5,386 tumor cells. Differential gene expression of tumor clones revealed 20 genes significantly enriched in sRCC relative to ccRCC. Filtering against public differential gene expression data resulted in 10 genes included within the final Sarcomatoid Signature (SS). SS expression scores for clinical specimens were calculated by single-sample gene set enrichment analysis. SS scores were significantly enriched in sRCC patient tumors across TCGA KIRC, CheckMate, and IMmotion151 datasets (p<0.001 for all cohorts). Within TCGA KIRC, SS scores were significantly increased in patients with nuclear grade 4 (p<0.001) and metastatic (p<0.001) disease, while stratification by median SS score revealed worsened overall (HR=2.19, p<0.001) and disease-free (HR=2.08, p<0.001) survival among SS-high patients. Amongst the CheckMate cohort (including patient samples from the CheckMate-010 & CheckMate-025 trials), SS-high patients experienced improved progression free survival (PFS) when treated with nivolumab relative to everolimus (HR=0.70, p=0.055), whereas no difference was seen between therapies in the SS-low patient population (HR=0.98, p=0.9). In the IMmotion151 trial, which was not utilized for gene filtering and signature refinement and thus represents an independent validation cohort, SS-high patients experienced reduced PFS duration (HR=1.39, p<0.001) irrespective of treatment arm. However, SS-high patients experienced improved objective response rates with the ICB-containing atezolizumab/bevacizumab regimen compared to sunitinib (p<0.001) and relative to SS-low patients receiving atezolizumab/bevacizumab (p=0.028) (Figure 1). Within the SS-high population, patients experienced prolonged PFS with atezolizumab/bevacizumab (HR=0.70, p=0.003) relative to sunitinib whereas no PFS difference between arms was seen in the SS-low population (HR=1.01, p>0.99). Conclusions This novel SS derived from scRNAseq demonstrates negative prognostic yet positive predictive value for ICB response. In a domain currently void of efficacious biomarkers, the SS, upon prospective validation, may assist in optimal therapy selection for patients with RCC.

Sequential reinforcement active feature learning for gene signature identification in renal cell carcinoma

Identification of expression signatures for non-small-cell lung carcinoma subtype classification

Identifying potential risk genes for clear cell renal cell carcinoma with deep reinforcement learning

An ensemble learning-based feature selection algorithm for identification of biomarkers of renal cell carcinoma

Identification and Validation of an 8-Gene Expression Signature for Predicting High-Fuhrman Grade Renal Cell Carcinoma.

Abstract 3500: Refining diagnosis of renal cell carcinoma subtypes through single-cell resolution transcriptomic signatures

Construction and Validation of a 9-Gene Signature for Predicting Prognosis in Stage III Clear Cell Renal Cell Carcinoma.

Identification and Validation of a 44-Gene Expression Signature for the Classification of Renal Cell Carcinomas.

An improved clear cell renal cell carcinoma stage prediction model based on gene sets

Machine Learning Gene Signature to Metastatic ccRCC Based on ceRNA Network

Integration of bulk RNA-seq and single-cell RNA-seq constructs: a cancer-associated fibroblasts-related signature to predict prognosis and therapeutic response in clear cell renal cell carcinoma

Artificial intelligence for diagnosing and predicting survival of patients with renal cell carcinoma: Retrospective multi-center study

26 Applying single cell RNA sequencing of sarcomatoid renal cell carcinoma for the development of a novel transcriptomic biomarker to predict immunotherapy response

Single-cell sequencing reveals novel proliferative cell type: a key player in renal cell carcinoma prognosis and therapeutic response

Identification of a novel prognostic gene signature in a clear cell renal cell carcinoma (ccRCC) population using an integrated multi-study single-cell RNAseq dataset.

Development and Validation of a Robust Multigene Signature As an Aid to Predict Early Relapse in Stage I-III Clear Cell and Papillary Renal Cell Cancer

Single-nucleus sequencing unveils heterogeneity in renal cell carcinomas microenvironment: Insights into pathogenic origins and treatment-responsive cellular subgroups

Identification of a competing endogenous RNA network related to immune signature in clear cell renal cell carcinoma

The Ion Channel-Related Gene Signatures Correlated with Diagnosis, Prognosis, and Individualized Treatment in Patients with Clear Cell Renal Cell Carcinoma.

Lncrnas Expression Signatures Of Renal Clear Cell Carcinoma Revealed By Microarray

Bioinformatics profiling integrating a four immune-related long non-coding RNAs signature as a prognostic model for papillary renal cell carcinoma