Background: Renal papillary cell carcinoma (KIRP) is a relatively rare renal malignancy. Although KIRP subtyping about clinical relevance has been defined, there have been scarce number of studies on the molecular characteristics of KIRP subtypes.
Method: In this study, a independent samples t-test was used to identify differentially expressed (DE) miRNAs between tumor and normal samples of KIRP. Meanwhile, we performed unsupervised clustering using DE miRNA expression data to analyze molecular characteristics of KIRP. The Partitioning Around Medoids clustering approach was used to identify molecular subtypes. The cumulative distribution function (CDF), proportion of ambiguously clustered pairs (PAC), principal component analysis (PCA) and consensus heatmaps were used to assess the optimal subtypes. In the differential molecular subtypes, we performed an integrated analysis of survival, DE genes, biological function and somatic mutations on the cohort of KIRP patients from The Cancer Genome Atlas.
Results: From solutions with 2, 3, 4, 5, 6 and 7 clusters we selected three KIRP molecular subtypes after assessing PCA, PAC, CDF and consensus heatmaps. We found that the three subtypes are associated with different overall survival and molecular characteristics. Compared with subtype1 and subtype3, subtype2 had a better prognosis in KIRP patients. After exploring their signaling pathways and biological characteristics, we identified the significantly enriched KEGG pathways and Gene Ontology terms for the three subtypes. The distribution of PARD6B, SETD2, STAG2, CUL3, TNRC18, LRBA, IGSF9B and DUNC1H1 mutations differed between the subtypes.
Conclusion: We performed unsupervised clustering using differentially expressed miRNA expression data and described the three KIRP molecular subtypes. The three subtypes differed in overall survival, molecular characteristics and gene mutation frequency.
Keywords: renal papillary cell carcinoma, unsupervised clustering, molecular subtype, prognosis