Aksoy Yagiz, Chou Angela, Mahjoub Mahiar, Sheen Amy, Sioson Loretta, Ahadi Mahsa S, Gill Anthony J, Fuchs Talia L
Cancer Diagnosis and Pathology Group, Kolling Institute of Medical Research, Royal North Shore Hospital, St Leonards, NSW, Australia; NSW Health Pathology, Department of Anatomical Pathology, Royal North Shore Hospital, Sydney, NSW, Australia; Sydney Medical School, University of Sydney, Sydney, NSW, Australia.
Cancer Diagnosis and Pathology Group, Kolling Institute of Medical Research, Royal North Shore Hospital, St Leonards, NSW, Australia; NSW Health Pathology, Department of Anatomical Pathology, Royal North Shore Hospital, Sydney, NSW, Australia.
Pathology. 2023 Jun;55(4):449-455. doi: 10.1016/j.pathol.2022.11.009. Epub 2023 Feb 4.
Recent advances in the management of diffuse pleural mesothelioma (DPM) have increased interest in prognostication and risk stratification on the basis that maximum benefit of combination immunotherapy appears to be seen in patients who otherwise would have the worst prognosis. Various grading schemes have been proposed, including the recently published Mesothelioma Weighted Grading Scheme (MWGS). However, predictive modelling using deep learning algorithms is increasingly regarded as the gold standard in prognostication. We therefore sought to develop and validate a prognostic nomogram for DPM. Data from 369 consecutive patients with DPM were used as independent training and validation cohorts to develop a prognostic tool that included the following variables: age, sex, histological type, nuclear atypia, mitotic count, necrosis, and BAP1 immunohistochemistry. Patients were stratified into four risk groups to assess model discrimination and calibration. To assess discrimination, the area-under-the-curve (AUC) of a receiver-operator-curve (ROC), concordance-index (C-index), and dissimilarity index (D-index) were calculated. Based on the 5-year ROC analysis, the AUC for our model was 0.75. Our model had a C-index of 0.67 (95% CI 0.53-0.79) and a D-index of 2.40 (95% CI 1.69-3.43). Our prognostic nomogram for DPM is the first of its kind, incorporates well established prognostic markers, and demonstrates excellent predictive capability. As these factors are routinely assessed in most pathology laboratories, it is hoped that this model will help inform prognostication and difficult management decisions, such as patient selection for novel therapies. This nomogram is now freely available online at: https://nomograms.shinyapps.io/Meso_Cox_ML/.
弥漫性胸膜间皮瘤（DPM）管理方面的最新进展，使人们对预后评估和风险分层的兴趣增加，因为联合免疫疗法似乎在预后最差的患者中能带来最大益处。已经提出了各种分级方案，包括最近发布的间皮瘤加权分级方案（MWGS）。然而，使用深度学习算法的预测模型越来越被视为预后评估的金标准。因此，我们试图开发并验证一种DPM的预后列线图。来自369例连续DPM患者的数据被用作独立的训练和验证队列，以开发一种预后工具，该工具纳入了以下变量：年龄、性别、组织学类型、核异型性、有丝分裂计数、坏死和BAP1免疫组化。患者被分为四个风险组，以评估模型的区分度和校准度。为了评估区分度，计算了受试者工作特征曲线（ROC）的曲线下面积（AUC）、一致性指数（C指数）和差异指数（D指数）。基于5年ROC分析，我们模型的AUC为0.75。我们的模型C指数为0.67（95%CI 0.53 - 0.79），D指数为2.40（95%CI 1.69 - 3.43）。我们的DPM预后列线图是同类中的首个，纳入了成熟的预后标志物，并显示出出色的预测能力。由于这些因素在大多数病理实验室中都是常规评估的，希望这个模型将有助于为预后评估和困难的管理决策提供信息，例如为新疗法选择患者。该列线图现在可在以下网址免费在线获取：https://nomograms.shinyapps.io/Meso_Cox_ML/ 。
