Hong Lin, Zheng Lan, He Yu-Feng, Fang Ya-Xing, Chen Hui, Chen Kang-Jia, Zhou Shu-Guang
Department of Gynecology, Anhui Women and Children's Medical Center Hefei, Maternal and Child Health Center of Anhui Medical University, The Fifth Affiliated Clinical College of Anhui Medical University, Hefei, 230001, Anhui, China.
Department of Gynecology, Linquan Maternity and Child Healthcare Hospital, Fuyang, 236400, Anhui, China.
Biochem Genet. 2025 Jun 4. doi: 10.1007/s10528-025-11151-x.
Endometriosis is one of the most common benign gynecological disorders, characterized by persistent pain and challenges with fertility. Anoikis, a specific form of apoptosis, occurs when cells detach from the extracellular matrix. Recent researches have revealed that anoikis resistance is the most important prerequisite for endometriosis development, but it's still not entirely apparent, yet, what part anoikis plays in endometriosis pathogenesis. We obtained the GSE141549 dataset from the GEO database, which served as our training cohort. Anoikis-related genes (ANRGs), sourced from the GeneCards database, were integrated with differentially expressed genes (DEGs) identified in GSE141549. Consequently, we obtained a collection of differentially expressed anoikis-related genes (DE-ANRGs). Subsequently, we employed machine learning algorithms to screen for key diagnostic DE-ANRGs and developed a nomogram model to predict and diagnose endometriosis. Ultimately, we validated our findings through in vitro experiments and an online endometriosis database. Within the training cohort, a total of 47 DE-ANRGs were identified. Furthermore, three machine learning methods pinpointed four diagnostic genes (CAV1, PDK4, CSPG4, SERPINE1). Based on these genes, we constructed a nomogram to facilitate the prediction and clinical diagnosis of endometriosis. To assess model's predictive accuracy, clinical adaptability, and discriminative ability. We performed calibration curves, Receiver Operating Characteristic curves and decision curve analysis. All assessments demonstrated our model's outstanding performance. Ultimately, consistent expression trends of four genes were observed in GSE7305 test cohort, clinical specimens, and the Turku database when compared to the training cohort. In addition, we also have revealed the immune landscape differences, which may offer new promising immunotherapeutic targets for endometriosis patients in the future. In addition to offering novel perspectives on the role of anoikis in endometriosis pathogenesis, our analysis also identified a panel of distinctive biomarkers with significant diagnostic potential.
子宫内膜异位症是最常见的良性妇科疾病之一，其特征为持续性疼痛和生育难题。失巢凋亡是细胞从细胞外基质脱离时发生的一种特定形式的细胞凋亡。最近的研究表明，失巢凋亡抗性是子宫内膜异位症发展的最重要前提，但失巢凋亡在子宫内膜异位症发病机制中所起的作用仍不完全清楚。我们从基因表达综合数据库（GEO数据库）获取了GSE141549数据集，将其作为我们的训练队列。从基因卡片数据库获取的失巢凋亡相关基因（ANRGs）与在GSE141549中鉴定出的差异表达基因（DEGs）进行整合。因此，我们获得了一组差异表达的失巢凋亡相关基因（DE - ANRGs）。随后，我们使用机器学习算法筛选关键的诊断性DE - ANRGs，并开发了一个列线图模型来预测和诊断子宫内膜异位症。最终，我们通过体外实验和一个在线子宫内膜异位症数据库验证了我们的研究结果。在训练队列中，共鉴定出47个DE - ANRGs。此外，三种机器学习方法确定了四个诊断基因（CAV1、PDK4、CSPG4、SERPINE1）。基于这些基因，我们构建了一个列线图以促进子宫内膜异位症的预测和临床诊断。为了评估模型的预测准确性、临床适应性和鉴别能力，我们进行了校准曲线、受试者工作特征曲线和决策曲线分析。所有评估均表明我们的模型表现出色。最终，与训练队列相比，在GSE7305测试队列、临床标本和图尔库数据库中观察到这四个基因一致的表达趋势。此外，我们还揭示了免疫格局差异，这可能为未来的子宫内膜异位症患者提供新的有前景的免疫治疗靶点。除了为失巢凋亡在子宫内膜异位症发病机制中的作用提供新的视角外，我们的分析还鉴定出一组具有显著诊断潜力的独特生物标志物。
