文献速递腹部技术与安全CT

CT预测胃癌隐匿性腹膜转移或细胞学阳性

Prediction of occult peritoneal metastases or positive cytology using CT in gastric cancer.
PubMed 文献 · Eur Radiol · 2023-07-06

本研究回顾性纳入520例接受分期腹腔镜探查或腹腔灌洗细胞学检查的胃癌患者,构建并内部验证整合CT图像特征与临床病理参数的列线图,用于术前个体化预测隐匿性腹膜转移。模型纳入胃壁分层2级、肿瘤厚度和Lauren弥漫型三个预测因子,AUC为0.830(95%CI 0.788-0.873),内部验证AUC为0.826(95%CI 0.756-0.870),敏感度76.0%、特异度78.8%、准确率78.3%。

要点

  • CT示胃壁分层2级、肿瘤厚度及Lauren弥漫型是隐匿性腹膜转移的独立预测因子。
  • 基于上述三因素的列线图预测效能优于仅用临床病理参数的原始模型(p<0.001)。
  • 该列线图便于临床医生术前个体化评估隐匿性腹膜转移风险,不限于专业放射科医生使用。

核心结论

  • 基于CT表型的列线图对胃癌隐匿性腹膜转移具有良好的区分度和校准度,可用于术前个体化风险分级。

研究设计与样本

  • 回顾性研究纳入520例接受分期腹腔镜探查或腹腔灌洗细胞学检查的胃癌患者,以探查或细胞学结果作为参考标准。
  • 采用单变量和多变量逻辑回归筛选预测因子并构建列线图,通过ROC、准确率和C-index评估模型性能,采用1000次bootstrap重采样进行内部验证,Delong检验比较模型间AUC差异。

关键影像表现/征象

  • 胃壁分层2级、肿瘤厚度和Lauren弥漫型是隐匿性腹膜转移的显著预测因子(p<0.05)。
  • 包含上述三因素的列线图预测效能显著高于原始模型(p<0.001)。

诊断标准与关键数值

  • 列线图模型预测隐匿性腹膜转移的AUC为0.830(95%CI 0.788-0.873),1000次bootstrap内部验证AUC为0.826(95%CI 0.756-0.870)。
  • 模型敏感度为76.0%,特异度为78.8%,准确率为78.3%。

临床应用与阅片要点

  • 阅片时可关注胃壁分层2级和肿瘤厚度,结合Lauren弥漫型病理结果,使用列线图术前评估隐匿性腹膜转移风险。
  • 该模型便于临床医生使用,有助于选择合适治疗方案,但需注意其基于单中心回顾性数据,外部适用性待验证。

局限性

  • 回顾性研究设计,仅采用内部验证(bootstrap重采样),未进行外部验证。
  • 样本来自单中心,模型泛化能力需进一步多中心研究证实。

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)。

英文原文(摘要)· 点击展开

OBJECTIVE: Accurate prediction of preoperative occult peritoneal metastasis (OPM) is critical to selecting appropriate therapeutic regimen for gastric cancer (GC). Considering the clinical practicability, we develop and validate a visible nomogram that integrates the CT images and clinicopathological parameters for the individual preoperative prediction of OPM in GC.

METHODS: This retrospective study included 520 patients who underwent staged laparoscopic exploration or peritoneal lavage cytology (PLC) examination. Univariate and multivariate logistic regression results were used to screen model predictors and construct nomograms of OPM risk. The performance of the model was detected by using ROC, accuracy, and C-index. The bootstrap resampling method was considered internal validation of the model. The Delong test was used to evaluate the difference in AUC between the two models.

RESULTS: Grade 2 mural stratification, tumor thickness, and the Lauren classification diffuse were significant predictors of OPM (p < 0.05). The nomogram of these three factors (compared with the original model) showed a higher predictive effect (p < 0.001). The area under the curve (AUC) of the model was 0.830 (95% CI 0.788-0.873), and the internally validated AUC of 1000 bootstrap samples was 0.826 (95% CI 0.756-0.870). The sensitivity, specificity, and accuracy were 76.0%, 78.8%, and 78.3%, respectively.

CONCLUSIONS: CT phenotype-based nomogram demonstrates favorable discrimination and calibration, and it can be conveniently used for preoperative individual risk rating of OPM in GC.

CLINICAL RELEVANCE STATEMENT: In this study, the preoperative OPM prediction model based on CT images (mural stratification, tumor thickness) combined with pathological parameters (the Lauren classification) showed excellent predictive ability in GC, and it is also suitable for clinicians to use rather than limited to professional radiologists.

KEY POINTS: • Nomogram based on CT image analysis can effectively predict occult peritoneal metastasis in gastric cancer (training area under the curve (AUC) = 0.830 and bootstrap AUC = 0.826). • Nomogram model combined with CT features performed better than the original model (established using only clinicopathological parameters) in differentiating occult peritoneal metastasis of gastric cancer.

来源:Eur Radiol,查看原文。版权归原作者所有。

引用

  1. Liu, Ding, Wu, et al. Prediction of occult peritoneal metastases or positive cytology using CT in gastric cancer. Eur Radiol. 2023. PMID: 37414883. doi:10.1007/s00330-023-09854-z

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。