Scholarlink

Objectives <p>We developed and validated a combined CT-based model to predict microvascular invasion (MVI) in patients with intrahepatic cholangiocarcinoma (ICC) for surgical resection decision-making.</p> Materials and methods <p>This retrospective study included 292 patients with pathologically confirmed ICC between January 2012 and May 2023 at four institutions. The patients were divided into training (<i>n</i> = 141), test (<i>n</i> = 60), and external validation (<i>n</i> = 91) cohorts. The radiomics features were extracted from arterial and portal venous phases, and a combined model incorporating a CT radiomics signature and clinicoradiological features was developed and validated. The 2-year recurrence-free survival (RFS) of the MVI risk stratification was analyzed using the Kaplan–Meier method with a log-rank test and compared between anatomical and non-anatomical, or major and minor resections.</p> Results <p>The combined model incorporating 20 radiomics features and 3 clinicoradiological features achieved areas under the curve of 0.938, 0.889, and 0.844 in the training, test, and external validation cohorts, respectively. In the high MVI risk group, the 2-year RFS was significantly higher in patients who underwent major hepatectomy than in those who underwent minor hepatectomy (log-rank <i>p</i> = 0.007), especially for tumors located subcapsularly (log-rank <i>p</i> = 0.005). In the low MVI risk group, the 2-year RFS was significantly lower in patients who underwent major hepatectomy than in those who underwent minor hepatectomy (log-rank <i>p</i> = 0.033), particularly for tumors ≤ 5 cm (log-rank <i>p</i> = 0.011).</p> Conclusions <p>We developed and validated a robust combined CT-based model that accurately identified the MVI status of ICC and assisted in surgical decision-making.</p> Key Points <p><Emphasis Type="BoldItalic">Question</Emphasis> <i>An effective preoperative MVI prediction model for ICC patients should consider its implications for different surgical resection strategies</i>.</p> <p><Emphasis Type="BoldItalic">Findings</Emphasis> <i>We developed and validated a combined model integrating CT radiomics and clinicoradiological features to accurately identify preoperative MVI status for surgical guidance</i>.</p> <p><Emphasis Type="BoldItalic">Clinical relevance</Emphasis> <i>For ICC patients, minor hepatectomy is recommended for those with low-risk MVI, particularly for tumors</i> ≤ <i>5</i> <i>cm, while major hepatectomy is advised for those with high-risk MVI, especially for tumors located subcapsularly</i>.</p> Graphical Abstract <p></p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Preoperative prediction model of microvascular invasion in intrahepatic cholangiocarcinoma patients based on CT radiomics can assist clinical surgical decision-making: a multicenter study

  • 期刊论文

  • 发表日期：

    2025年8月14日

查看全文
  • Ziwei Liu,
  • Haixiong Chen,
  • Shaomin Yang,
  • Rong Zhang,
  • Fan Yang,
  • Jieying Feng,
  • Ke Xu,
  • Zhilin Tan,
  • Yanqiu Feng,
  • Fusheng Ouyang,
  • Xiaohong Li,
  • Xinjie Chen,
  • Yifan Chen,
  • Zhiping Cai,
  • Xingqun Guan,
  • Chun Luo,
  • Baoliang Guo,
  • Qiugen Hu

摘要

Objectives

We developed and validated a combined CT-based model to predict microvascular invasion (MVI) in patients with intrahepatic cholangiocarcinoma (ICC) for surgical resection decision-making.

Materials and methods

This retrospective study included 292 patients with pathologically confirmed ICC between January 2012 and May 2023 at four institutions. The patients were divided into training (n = 141), test (n = 60), and external validation (n = 91) cohorts. The radiomics features were extracted from arterial and portal venous phases, and a combined model incorporating a CT radiomics signature and clinicoradiological features was developed and validated. The 2-year recurrence-free survival (RFS) of the MVI risk stratification was analyzed using the Kaplan–Meier method with a log-rank test and compared between anatomical and non-anatomical, or major and minor resections.

Results

The combined model incorporating 20 radiomics features and 3 clinicoradiological features achieved areas under the curve of 0.938, 0.889, and 0.844 in the training, test, and external validation cohorts, respectively. In the high MVI risk group, the 2-year RFS was significantly higher in patients who underwent major hepatectomy than in those who underwent minor hepatectomy (log-rank p = 0.007), especially for tumors located subcapsularly (log-rank p = 0.005). In the low MVI risk group, the 2-year RFS was significantly lower in patients who underwent major hepatectomy than in those who underwent minor hepatectomy (log-rank p = 0.033), particularly for tumors ≤ 5 cm (log-rank p = 0.011).

Conclusions

We developed and validated a robust combined CT-based model that accurately identified the MVI status of ICC and assisted in surgical decision-making.

Key Points

Question An effective preoperative MVI prediction model for ICC patients should consider its implications for different surgical resection strategies.

Findings We developed and validated a combined model integrating CT radiomics and clinicoradiological features to accurately identify preoperative MVI status for surgical guidance.

Clinical relevance For ICC patients, minor hepatectomy is recommended for those with low-risk MVI, particularly for tumors ≤ 5 cm, while major hepatectomy is advised for those with high-risk MVI, especially for tumors located subcapsularly.

Graphical Abstract