Department of Reproductive Medicine Center, Binzhou Medical University Hospital, No. 661 Huanghe 2nd Road, Binzhou, 256603, China.
Department of Radiology, Binzhou Medical University Hospital, Binzhou, 256603, China.
Sci Rep. 2023 Jan 18;13(1):980. doi: 10.1038/s41598-022-27326-0.
Polycystic ovary syndrome (PCOS), a common reproductive endocrine disease, has clinically heterogeneous characteristics. Recently, cuproptosis causes several diseases by killing cells. Hence, we aimed to explore cuproptosis-related molecular clusters in PCOS and construct a prediction model. Based on the GSE5090, GSE43264, GSE98421, and GSE124226 datasets, an analysis of cuproptosis regulators and immune features in PCOS was conducted. In 25 cases of PCOS, the molecular clusters of cuproptosis-related genes and the immune cell infiltration associated with PCOS were investigated. Weighted gene co-expression network analysis was used to identify differentially expressed genes within clusters. Next, we compared the performance of the random forest model, support vector machine model, generalized linear model, and eXtreme Gradient Boosting for deciding the optimum machine model. Validation of the predictive effectiveness was accomplished through nomogram, calibration curve, decision curve analysis, and using other two datasets. PCOS and non-PCOS controls differed in the dysregulation of cuproptosis-related genes and the activation of immunoreaction. Two cuproptosis-related molecular clusters associated with PCOS were identified. Significant heterogeneity was noted in immunity between the two clusters based on the analysis of immune infiltration. The immune-related pathways related to cluster-specific differentially expressed genes in Cluster1 were revealed by functional analysis. With a relatively low residual error and root mean square error and a higher area under the curve (1.000), the support vector machine model demonstrated optimal discriminative performance. An ultimate 5-gene-based support vector machine model was noted to perform satisfactorily in the other two validation datasets (area under the curve = 1.000 for both). Moreover, the nomogram, calibration curve, and decision curve analysis showed that PCOS subtypes can be accurately predicted. Our study results helped demonstrate a comprehensive understanding of the complex relationship between cuproptosis and PCOS and establish a promising prediction model for assessing the risk of cuproptosis in patients with PCOS.
多囊卵巢综合征（PCOS）是一种常见的生殖内分泌疾病，具有临床异质性特征。最近，铜死亡通过杀死细胞引起多种疾病。因此，我们旨在探讨 PCOS 中与铜死亡相关的分子簇，并构建预测模型。基于 GSE5090、GSE43264、GSE98421 和 GSE124226 数据集，对 PCOS 中铜死亡调节剂和免疫特征进行了分析。在 25 例 PCOS 中，研究了与铜死亡相关基因的分子簇和与 PCOS 相关的免疫细胞浸润。使用加权基因共表达网络分析识别簇内差异表达基因。接下来，我们比较了随机森林模型、支持向量机模型、广义线性模型和极端梯度提升在确定最佳机器模型方面的性能。通过列线图、校准曲线、决策曲线分析和使用另外两个数据集来验证预测效果的有效性。PCOS 和非 PCOS 对照组在铜死亡相关基因的失调和免疫反应的激活方面存在差异。确定了与 PCOS 相关的两个铜死亡相关分子簇。基于免疫浸润分析，两个簇之间的免疫存在显著异质性。通过功能分析揭示了与 Cluster1 中簇特异性差异表达基因相关的免疫相关途径。支持向量机模型具有相对较低的残差和均方根误差以及较高的曲线下面积（1.000），表现出最佳的判别性能。最终的基于 5 个基因的支持向量机模型在另外两个验证数据集中表现良好（曲线下面积均为 1.000）。此外，列线图、校准曲线和决策曲线分析表明，可以准确预测 PCOS 亚型。我们的研究结果有助于全面了解铜死亡与 PCOS 之间的复杂关系，并建立一种有前途的预测模型，用于评估 PCOS 患者铜死亡的风险。
