Construction and validation of renal cell carcinoma tumor cell differentiation-related prognostic classification (RCC-TCDC): an integrated bioinformatic analysis and clinical study

Renal cell carcinoma (RCC) is a heterogeneous malignancy with diverse gene expression patterns, molecular landscapes, and differentiation characteristics of tumor cells. It is imperative to develop molecular RCC classification based on tumor cell differentiation for precise risk stratification and personalized therapy. We obtained scRNA-seq profiles from GSE159115 and bulk RNA-seq profiles from TCGA-KIRC cohort. We then performed scRNA-seq cluster analysis, monocle2 pseudotime analysis, and prognostic analysis to obtain tumor cell differentiation-related prognostic genes (TCDGs). Subsequently, we conducted consensus clustering to construct the RCC tumor cell differentiation-related prognostic classification (RCC-TCDC) and implemented prognostic and multi-omics analyses. Moreover, we utilized Lasso regression to help develop a multivariable prognostic model. In addition, we performed correlation analysis and Cmap algorithm for regulatory network establishment and candidate inhibitor prediction. We eventually included 370 kidney neoplasm patients in Xinhua cohort to undergo immunohistochemical staining and scoring for classification and comprehensive statistical analyses, including Chi-square tests, Kaplan-Meier survival analyses, and multivariable Cox regression analysis . 32 TCDGs were identifiedand RCC-TCDC was constructed to classify TCGA-KIRC patients into RCC-low differentiation (RCC-LD) (S100A11+ SH3BGRL3+, high risk), RCC-moderate differentiation (TSPAN7+, medium risk), and RCC-high differentiation (RCC-HD) (AQP1+ NPR3+, low risk). Notably, RCC-LD was validated as anindependent risk factor for both OS (p = 0.015, HR = 14.0, 95%CI = 1.67–117.8) and PFS (p = 0.010, HR = 4.0, 95%CI = 1.39–11.7) of RCC patients in Xinhua cohort, taking RCC-HD as reference. We constructed and validated a robust molecular classification system, RCC-TCDC, elucidating three distinct RCC subtypes.肾细胞癌（Renal cell carcinoma, RCC）是一类异质性恶性肿瘤，其肿瘤细胞具有多样的基因表达模式、分子图谱及分化特征。亟需开发基于肿瘤细胞分化的肾细胞癌分子分型方法，以实现精准风险分层与个性化治疗。本研究从GSE159115数据集获取单细胞RNA测序（scRNA-seq）表达谱，从TCGA-KIRC队列获取批量RNA测序（bulk RNA-seq）表达谱。随后通过单细胞RNA测序聚类分析、Monocle2拟时序分析及预后分析，筛选得到肿瘤细胞分化相关预后基因（TCDGs）。后续采用共识聚类方法构建肾细胞癌肿瘤细胞分化相关预后分型（RCC-TCDC），并开展预后及多组学分析。此外，利用Lasso回归构建多变量预后模型。同时，通过相关性分析与CMap算法完成调控网络构建与候选抑制剂预测。本研究最终纳入新华队列中的370例肾肿瘤患者，对其进行免疫组化染色与评分以完成分型，并开展包括卡方检验、Kaplan-Meier生存分析及多变量Cox回归分析在内的综合统计学分析。本研究共鉴定出32个TCDGs，并构建RCC-TCDC将TCGA-KIRC队列患者分为三类：肾细胞癌低分化型（RCC-LD，特征为S100A11+、SH3BGRL3+，高风险）、肾细胞癌中分化型（特征为TSPAN7+，中风险）以及肾细胞癌高分化型（RCC-HD，特征为AQP1+、NPR3+，低风险）。值得注意的是，以RCC-HD为参照，在新华队列中RCC-LD被证实为肾细胞癌患者总生存期（overall survival, OS）与无进展生存期（progression-free survival, PFS）的独立危险因素（OS：p=0.015，风险比HR=14.0，95%置信区间CI=1.67–117.8；PFS：p=0.010，HR=4.0，95%CI=1.39–11.7）。本研究构建并验证了一套稳健的分子分型系统RCC-TCDC，可明确区分三种不同的肾细胞癌亚型。