文献速递神经肿瘤

新诊断CT脑病变何时需分期扫描?多变量逻辑回归分析

When is a staging scan required for newly diagnosed brain lesions on CT? A multivariate logistic regression analysis.
PubMed 文献 · Acta Neurochir (Wien) · 2022-10-08

本研究基于神经肿瘤MDT讨论的疑似新发脑肿瘤患者,利用2020年7-12月222例数据构建多变量逻辑回归方案,在2021年1-6月216例中验证。预测转移性疾病最敏感的变量为既往癌症史、多发病灶、病灶<4 cm及幕下位置;具备任一特征即推荐行胸腹盆CT分期检查的方案灵敏度99.1%(AUC 0.704)。

要点

  • 既往癌症史、多发病灶、病灶<4 cm、幕下位置是预测转移性疾病的敏感变量。
  • 具备上述任一特征即行胸腹盆CT分期的方案灵敏度99.1%(AUC 0.704)。
  • 仅对具备四项预测变量之一的患者行分期检查可减少不必要的胸腹盆CT。

核心结论

  • 若患者仅具备四项已识别预测变量之一时才行胸腹盆CT分期检查,可减少不必要的分期检查。

研究设计与样本

  • 回顾性分析在三级神经外科中心神经肿瘤MDT讨论的疑似新发脑肿瘤患者,方案构建阶段收集2020年7月至12月222例患者数据,方案测试阶段收集2021年1月至6月216例患者数据,采用多变量逻辑回归识别预测变量并计算各方案的灵敏度、特异度及曲线下面积(AUC)。

主要发现

  • 预测转移性疾病最敏感的变量为既往癌症史、多发病灶、病灶<4 cm及幕下位置。
  • 基于具备上述任一特征即推荐行胸腹盆CT分期的方案灵敏度为99.1%(AUC 0.704)。

关键数值与推荐

  • 推荐行胸腹盆CT分期的方案:具备既往癌症史、多发病灶、病灶<4 cm、幕下位置四项特征中任一项,灵敏度99.1%(AUC 0.704)。

临床怎么用

  • 对于CT新发脑病变疑似脑肿瘤患者,可依据是否具备既往癌症史、多发病灶、病灶<4 cm或幕下位置中任一特征来选择是否行胸腹盆CT分期检查。

局限性

  • 原文未报告局限性。

以上解读依据论文摘要(全文未开放获取或许可不允许翻译转载)。

原文

PURPOSE: For patients with a new lesion on CT head (CTH) suspected to be a brain tumor, a staging chest, abdomen, and pelvis CT (CTCAP) is only warranted if a metastatic lesion is suspected. Unnecessary CTCAPs are often performed too early in a patient's journey due to poor patient selection. We sought to create a protocol to guide the selection of patients for CTCAPs based on their CTH findings. METHODS: Patients with suspected new brain tumors discussed at the neuro-oncology MDT at a tertiary neurosurgical center were reviewed. Patient demographics and CTH features were collected. For protocol creation, data was collected from July to December 2020, and predictor variables were identified using multivariate logistic regression. Candidate protocols were assessed in a protocol testing stage using similar data collected from January to June 2021. Sensitivity, specificity, and area under the curve (AUC) were computed for each protocol. RESULTS: Variables from the protocol creation stage (222 patients) were assessed in the protocol testing stage (216 patients). The most sensitive variables predicting metastatic disease were a previous history of cancer, multiple lesions, lesion < 4 cm, and infratentorial location. A protocol recommending a CTCAP based on the presence of one of these features has a sensitivity of 99.1% (AUC 0.704). CONCLUSIONS: Unnecessary CTCAPs are reduced if performed only if a patient has one of the four identified predictor variables.

来源:Acta Neurochir (Wien),查看原文。版权归原作者所有。

引用

  1. Ali AMS, Varasteh AA, Konteas AB, et al. When is a staging scan required for newly diagnosed brain lesions on CT? A multivariate logistic regression analysis. Acta Neurochir (Wien). 2022. PMID: 36208346. doi:10.1007/s00701-022-05374-9

供执业医师工作参考,不替代临床判断;数值与分级以指南原文为准。