Homologous repair deficiency-associated genes in invasive breast cancer revealed by WGCNA co-expression network analysis and genetic perturbation similarity analysis

Background: Homologous repair deficiency (HRD) causes double-strand break repair to be impeded, which is a common driver of carcinogenesis. However, the therapeutic and prognostic potential of HRD in invasive breast cancer (BRCA) has not been fully explored using comprehensive bioinformatics analysis. Materials and Methods: HRD score was defined as the unweighted sum of LOH, TAI, and LST scores and obtained from the previous study according to Theo A et al. To characterize BRCA tumor microenvironment (TME) subtypes, “ConsensusClusterPlus” R package was used to conduct unsupervised clustering. The xCell algorithm was utilized for tumor composition analysis to estimate the TME in TCGA-BRCA. A WGCNA analysis was conducted to uncover the gene coexpression modules and hub genes in the HRD-related gene module of BRCA. The functional enrichment study was carried out using Metascape. A novel analysis pipeline, Genetic Perturbation Similarity Analysis (GPSA), was used to explore the single-gene perturbation closely related to HRD based on 3048 stable knockdown/knockout cell lines. The prognostic variables were evaluated using univariate COX analysis. Kaplan-Meier (KM) survival analysis was performed to assess the prognostic potential of HRD score. Receiver operator characteristic (ROC) curve was utilized to judge the diagnostic utility. Drug sensitivity was estimated through the R package “oncoPredict” and Genomics of Drug Sensitivity in Cancer (GDSC) database. XSum algorithm was performed to screen the candidate small molecule drugs based on the connectivity map (CMAP) database. Results: Low HRD score suggested a better prognosis in BRCA patients. The tumor with low HRD score had considerably greater degree of infiltration of stromal cells and infiltration of immunocytes was significantly enhanced in the high HRD score group. Using WGCNA, ten co-expression modules were obtained. The turquoise module and 25 hub genes were identified as the most correlated with HRD in BRCA. Functional enrichment analysis revealed that the turquoise gene module was mainly concentrated in the “cell cycle” pathways. Candidate HRD-related gene signatures (MELK) were screened out through WGCNA and GPSA analysis pipeline and then validated on independent validation sets. A small molecule drug (Clofibrate) that has the potential to reverse the increase of high HRD score was screened out to improve oncological outcomes in BRCA. Molecular docking suggested MELK to be one of possible molecular targets in the Clofibrate treatment of BRCA. Conclusion: Based on bioinformatic analysis, we fully explored the therapeutic and prognostic potential of HRD in BRCA. A novel HRD-related gene signature (MELK) were identified through the combination of WGCNA and GPSA analysis. In addition, we detailed the TME landscape in BRCA and identified four unique TME subtypes in group with high or low HRD score group. Moreover, Clofibrate were screened out to improve oncological outcomes in BRCA by reversing the increase of high HRD score. Thus, our study contributes to the development of personalized clinical management and treatment regimens of BRCA.

研究背景：同源重组修复缺陷（Homologous Repair Deficiency, HRD）会阻碍双链断裂修复，这是致癌作用的常见驱动因素。然而，目前尚未通过全面生物信息学分析充分探究同源重组修复缺陷在浸润性乳腺癌（Invasive Breast Cancer, BRCA）中的治疗与预后价值。材料与方法：HRD评分定义为杂合性缺失（Loss of Heterozygosity, LOH）、端粒等位基因不平衡（Telomeric Allelic Imbalance, TAI）以及大片段基因组不稳定（Large-scale State Transition, LST）评分的未加权总和，其数据来源于Theo A等人的既往研究。为表征乳腺癌肿瘤微环境（Tumor Microenvironment, TME）亚型，本研究采用“ConsensusClusterPlus”R包进行无监督聚类。使用xCell算法开展肿瘤组成分析，以估算TCGA-BRCA队列中的肿瘤微环境。通过加权基因共表达网络分析（Weighted Gene Co-expression Network Analysis, WGCNA）挖掘乳腺癌HRD相关基因模块中的基因共表达模块与核心基因。采用Metascape进行功能富集分析。本研究采用全新分析流程——遗传扰动相似性分析（Genetic Perturbation Similarity Analysis, GPSA），基于3048株稳定敲低/敲除细胞系，探索与HRD密切相关的单基因扰动事件。采用单因素COX分析评估预后变量。通过Kaplan-Meier（KM）生存分析评估HRD评分的预后价值。采用受试者工作特征（Receiver Operator Characteristic, ROC）曲线判断其诊断效能。通过R包“oncoPredict”与癌症药物敏感性基因组学（Genomics of Drug Sensitivity in Cancer, GDSC）数据库估算药物敏感性。基于连接图谱（Connectivity Map, CMAP）数据库，采用XSum算法筛选候选小分子药物。研究结果：低HRD评分提示乳腺癌患者预后更佳。低HRD评分肿瘤的基质细胞浸润程度更高，而高HRD评分组的免疫细胞浸润程度显著增强。通过WGCNA分析共获得10个共表达模块，其中绿松石模块与25个核心基因被鉴定为与乳腺癌HRD相关性最高的模块与基因集。功能富集分析显示，绿松石基因模块主要富集于“细胞周期”通路。通过WGCNA与GPSA分析流程筛选出候选HRD相关基因标志物（MELK），并在独立验证集中完成验证。筛选出一种可逆转高HRD评分升高的小分子药物——氯贝丁酯（Clofibrate），其有望改善乳腺癌患者的肿瘤学结局。分子对接实验显示，MELK可能是氯贝丁酯治疗乳腺癌的潜在分子靶点之一。研究结论：本研究通过生物信息学分析全面探究了HRD在乳腺癌中的治疗与预后价值。结合WGCNA与GPSA分析，本研究鉴定出一种全新的HRD相关基因标志物（MELK）。此外，本研究详细描绘了乳腺癌的肿瘤微环境图谱，并在高低HRD评分组中鉴定出4种独特的肿瘤微环境亚型。同时，本研究筛选出氯贝丁酯可通过逆转高HRD评分升高以改善乳腺癌患者的肿瘤学结局。综上，本研究可为乳腺癌的个性化临床管理与治疗方案开发提供理论依据。