Clear cell renal cell carcinomas (ccRCC) show a broad range of clinical behavior, and prognostic biomarkers are needed to stratify patients for appropriate management. Stanford University researchers sought to determine whether long intergenic non-coding RNAs (lincRNAs) might predict patient survival. Candidate prognostic lincRNAs were identified by mining The Cancer Genome Atlas (TCGA) transcriptome (RNA-seq) data on 466 ccRCC cases (randomized into discovery and validation sets) annotated for ~21,000 lncRNAs. A previously uncharacterized lincRNA, SLINKY (Survival-predictive LINcRNA in KidneY cancer), was the top-ranked prognostic lincRNA, and validated in an independent University of Tokyo cohort (P=0.004). In multivariable analysis, SLINKY expression predicted overall survival independent of tumor stage and grade [TCGA HR=3.5 (CI, 2.2-5.7), P < 0.001; Tokyo HR=8.4 (CI, 1.8-40.2), P = 0.007], and by decision tree, ROC and decision curve analysis, added independent prognostic value. In ccRCC cell lines, SLINKY knockdown reduced cancer cell proliferation (with cell-cycle G1 arrest) and induced transcriptome changes enriched for cell proliferation and survival processes. Notably, the genes affected by SLINKY knockdown in cell lines were themselves prognostic and correlated with SLINKY expression in the ccRCC patient samples. From a screen for binding partners, the researchers identified direct binding of SLINKY to Heterogeneous Nuclear Ribonucleoprotein K (HNRNPK), whose knockdown recapitulated SLINKY knockdown phenotypes. Thus, SLINKY is a robust prognostic biomarker in ccRCC, where it functions possibly together with HNRNPK in cancer cell proliferation.
A. Flowchart summarizing the approach for discovery and validation phases; see text for details. B.–E. Kaplan-Meier plots of overall survival for the top prognostic lincRNA SLINKY, shown for (B, C) ten randomly-selected iterations (of 1,000 total) from the discovery and validation phases; (D) the complete TCGA dataset; and (E) the independent Tokyo dataset. P-values (log-rank test) are indicated. F. SLINKY expression (log2 RPKM; ordered from lowest to highest) in normal kidney and matched ccRCC samples from the TCGA dataset; P-value (Mann-Whitney U-test) indicated. Inset (100% stacked column chart) summarizes SLINKY expression categorized as low (RPKM < 0.01), medium (0.01≤RPKM < 1), or high (RPKM≥1). G. SLINKY expression in primary ccRCC samples either without or with subsequent metastasis, from the Tokyo dataset. Inset (100% stacked column chart), as above. H. Box plots (median, quartiles, 2 stdev, and outliers) showing SLINKY expression across 21 different cancer types and the matched normal tissues from the TCGA dataset. N, normal; T; tumor; cancer-type acronyms are from TCGA.