Meningiomas are the most prevalent primary brain tumors. Due to their increasing burden on healthcare, meningiomas have become a pivot of translational research globally. Despite many studies in the field of discovery proteomics, the identification of grade-specific markers for meningioma is still a paradox and requires thorough investigation. The potential of the reported markers in different studies needs further verification in large and independent sample cohorts to identify the best set of markers with a better clinical perspective. A total of 53 fresh frozen tumor tissue and 51 serum samples were acquired from meningioma patients respectively along with healthy controls, to validate the prospect of reported differentially expressed proteins and claimed markers of Meningioma mined from numerous manuscripts and knowledgebases. A small subset of Glioma/Glioblastoma samples were also included to investigate inter-tumor segregation. Furthermore, a simple Machine Learning (ML) based analysis was performed to evaluate the classification accuracy of the list of proteins. A list of 15 proteins from tissue and 12 proteins from serum were found to be the best segregator using a feature selection-based machine learning strategy with an accuracy of around 80% in predicting low grade (WHO grade I) and high grade (WHO grade II and WHO grade III) meningiomas. In addition, the discriminant analysis could also unveil the complexity of meningioma grading from a segregation pattern, which leads to the understanding of transition phases between the grades. The identified list of validated markers could play an instrumental role in the classification of meningioma as well as provide novel clinical perspectives in regard to prognosis and therapeutic targets.

中文翻译：

---

血清和组织的大规模靶向蛋白质组学显示了对高级别和低级别脑膜瘤进行分类的实用性

脑膜瘤是最常见的原发性脑肿瘤。由于医疗负担日益增加，脑膜瘤已成为全球转化研究的中心。尽管发现蛋白质组学领域有许多研究，但脑膜瘤级别特异性标记物的鉴定仍然是一个悖论，需要彻底的研究。不同研究中报告的标记物的潜力需要在大型且独立的样本队列中进一步验证，以从更好的临床角度确定最佳标记物集。分别从脑膜瘤患者和健康对照中采集了总共 53 个新鲜冷冻肿瘤组织和 51 个血清样本，以验证从大量手稿和知识库中挖掘的报告的差异表达蛋白和声称的脑膜瘤标记物的前景。还包括一小部分胶质瘤/胶质母细胞瘤样本，以研究肿瘤间分离。此外，还进行了基于机器学习 (ML) 的简单分析，以评估蛋白质列表的分类准确性。使用基于特征选择的机器学习策略，发现来自组织的 15 种蛋白质和来自血清的 12 种蛋白质的列表是最好的分离器，在预测低等级（WHO I 级）和高等级（WHO 等级）方面的准确度约为 80% II 级和 WHO III 级）脑膜瘤。此外，判别分析还可以从分离模式揭示脑膜瘤分级的复杂性，从而有助于理解分级之间的过渡阶段。确定的经过验证的标记物列表可以在脑膜瘤的分类中发挥重要作用，并为预后和治疗目标提供新的临床观点。