5Integrative clinical, genomic, transcriptomic, and immunopathologic characterization of circulating KIM-1 in metastatic RCC

Abstract Background Kidney injury molecule-1 (KIM-1) is a transmembrane protein that is overexpressed in renal cell carcinoma (RCC). The JAVELIN Renal 101 trial established avelumab+axitinib as a standard treatment regimen for advanced renal cell carcinoma (RCC). We evaluated the prognostic value of KIM-1 and explored the association between KIM-1 levels and underlying tumor biology in RCC, leveraging translational analyses including genomic, transcriptomic, and immunohistochemistry data. Methods KIM-1 was measured in plasma at baseline (C1D1) and at 12 weeks (C3D1) using an electrochemiluminescence-based assay. Cox regression analyses were performed for progression-free survival (PFS) and overall survival (OS) and adjusted for baseline clinical characteristics. The predictive performance of our models for 18-month OS was evaluated using time-dependent receiver operating characteristic (ROC) curves. Associations between KIM-1 levels and clinical and translational data were evaluated using the Wilcoxon rank-sum test for categorical groups. Differential gene expression (DGE) and gene set enrichment analysis (GSEA) were performed using DESeq2, with KIM-1 treated as a continuous variable. Results Plasma for analysis was available from 612 patients (69% of the ITT population), including 323 treated with avelumab+axitinib and 289 with sunitinib. Elevated baseline KIM-1 levels were correlated with higher tumor burden as assessed by the sum of tumor diameters (Spearman’s ρ = 0.55; P < .0001). Lower KIM-1 levels were found at baseline in patients with IMDC favorable versus intermediate (185.09 vs. 330.95 pg/ml; P < .001) and poor (185.09 vs. 1285.51 pg/ml; P < .001) risk, and in intermediate versus poor (330.95 vs. 1285.51 pg/ml; P < .001) risk groups. Loss-of-function (LOF) BAP1 mutations were associated with higher tumor KIM-1 RNA expression (P < .0001) and plasma protein expression (742.29 vs. 280.47 pg/ml; P = .038) which remained significant after adjustment for tumor burden, as well as greater decreases in KIM-1 on avelumab+axitinib (-578.4 vs. -75.8 pg/ml; P = .01) but not on sunitinib. Transcriptomic analysis showed that the expression of HAVCR1, the gene encoding KIM-1, showed the strongest association with KIM-1 protein levels (Spearman’s ρ = 0.31; P < .0001), and that higher KIM-1 levels were associated with interferon gamma response whereas lower KIM-1 levels were associated with a hypoxia transcriptional program. Higher KIM-1 levels were also associated with enrichment for proliferative versus angiogenic gene expression signatures (P = .013). At 12 weeks, a greater decrease in KIM-1 was observed on avelumab+axitinib versus sunitinib (-65.55% vs. -31.02%; P < .001). Patients with PD-L1 positive tumors had a greater decrease in KIM-1 (-56.80% vs. -33.49%; P = .001) across arms. Tumors of patients who had a decrease in KIM-1 had a higher baseline total CD8+ infiltration (1.54% vs. 1.05%; P = .004), including within the tumor center (1.48% vs. 0.92%; P = .004). In all patients, lower KIM-1 at baseline or at 12 weeks was associated with longer PFS and OS (Table). Landmark analyses showed that increases in KIM-1 from baseline to 12-week follow-up were associated with shorter PFS and OS (Table). Lower KIM-1 levels were found among patients with partial response (PR) versus progressive disease (327.84 vs. 759.82 pg/ml; P = .04), complete response versus PR (94.22 vs. 327.84 pg/ml; P < .001), and exceptional versus intermediate response (117.13 vs. 325.81 pg/ml; P = .003). The 18-month AUC was 0.67 for baseline KIM-1 and 0.71 for 12-week KIM-1 for predicting OS. Adding baseline KIM-1 to IMDC improved the 18-month AUC from 0.69 to 0.75. The highest predictive performance was observed when IMDC was combined with 12-week KIM-1, corresponding to AUC of 0.78. Conclusions We present the first integrative clinical, transcriptomic, genomic, and immunopathologic evaluation of circulating KIM-1. High KIM-1 is correlated with poor prognosis in RCC and aggressive biology as characterized by mutational and transcriptomic analyses. Prospective studies are needed for the clinical implementation of KIM-1 as a biomarker in RCC.