53Microenvironmental characteristics of disease relapse after nephrectomy in clear cell renal cell carcinoma

Background Localized renal cell carcinoma (RCC) is primarily treated with nephrectomy, but approximately 30% of patients with localized RCC eventually develop local recurrence or distant metastasis. The KEYNOTE-564 trial demonstrated overall survival benefit with adjuvant pembrolizumab for high-risk RCC after nephrectomy. However, the survival at 48 months in patients without adjuvant therapy was 86.0%, suggesting that a large proportion of high-risk patients can be spared adjuvant immunotherapy. However, factors at a molecular level to predict disease relapse, and ultimately, who requires adjuvant immunotherapy, in RCC remain undetermined. Therefore, we sought to elucidate the microenvironmental characteristics of disease relapse after nephrectomy in patients with clear cell RCC (ccRCC). Methods Gene expression matrices and associated metadata from The Oncology Research Information Exchange Network (ORIEN) as a discovery cohort, and The Cancer Genome Atlas Kidney Renal Clear Cell Carcinoma (TCGA-KIRC) and Clinical Proteomic Tumor Analysis Consortium (CPTAC) ccRCC as validation cohorts, were obtained. Patients with no adjuvant immunotherapy exposure were classified into one of two groups: Relapse (R) – patients who had disease relapse within 5 years after nephrectomy—or Non-Relapse (NR) – those without relapse for at least 5 years (3 years for CPTAC ccRCC) after nephrectomy. Patients with metastatic disease at the time of ccRCC diagnosis/nephrectomy were excluded. Clinical characteristics were summarized and compared between the two groups using Fisher’s exact test. Single sample gene set enrichment analysis (ssGSEA) scores of Hallmark, IMmotion, and tertiary lymphoid structure signatures were assigned to cases utilizing TPM-normalized gene expression. GSEA of Hallmark, Reactome, and Gene Ontology pathways was performed on transcriptomic datasets. xCell immune deconvolution was also performed on normalized counts. Enrichment of gene set signature and deconvolution scores was compared via the Wilcoxon rank sum test between R versus NR groups. All analyses were performed with R version 4.4.2. Results In total, 152 (R: 66; NR: 86) patients in the ORIEN cohort, 166 (R: 91; NR 75) in TCGA-KIRC and 55 (R: 19; NR: 36) in CPTAC ccRCC were included. Higher T stage was more frequently seen in R than NR in all cohorts (P < .05). When assessed by ssGSEA in the ORIEN cohort, the epithelial mesenchymal transition (EMT) (P < .05) Hallmark pathway and IMmotion-derived complement cascade (P < .01), and FAS pentose phosphate (P < .05) molecular subset signatures were enriched in R, and these findings were all validated in the TCGA and CPTAC datasets. Immune-related (such as acute inflammatory response, macrophage differentiation, neutrophil activation, humoral immune response, and complement pathways), proliferation-related, and extracellular matrix/cell motility-related gene sets, including EMT, were significantly enriched in R, whereas oxidative metabolism-related gene sets were more enriched in NR (Figure). No difference in tertiary lymphoid structure signature expression was observed between R and NR. Immune deconvolution analysis revealed enrichment of NK T cells in R and eosinophils in NR in the ORIEN and TCGA datasets (P < .05). Conclusions These findings suggest that programs associated with tumor aggression such as EMT and proliferation, IMmotion-derived complement cascade and FAS pentose phosphate pathways, alongside numerous inflammatory signals are contributing processes to disease relapse after nephrectomy in ccRCC. Such characteristics may have use as biomarkers for the identification of disease relapse, as well as potential therapeutic targets for (neo)adjuvant strategies in patients with high-risk localized ccRCC. Ongoing work comparing immune cell compositions will further elucidate the immunological determinants of ccRCC disease fate following nephrectomy, creating avenues for the tailoring of perioperative therapy strategies.