| 蔡雨秦,罗良平,杨志荷,陈芳,艾竹,陈伊镕,向之明.肺浸润性非黏液型腺癌多层螺旋CT特征回归模型列线图术前预测Ki-67表达[J].中国医学影像技术,2023,39(4):530~535 |
| 肺浸润性非黏液型腺癌多层螺旋CT特征回归模型列线图术前预测Ki-67表达 |
| Regression model nomogram based on multi-slice spiral CT characteristics of pulmonary invasive non-mucinous adenocarcinoma for preoperative prediction of tumor Ki-67 expression |
| 投稿时间：2022-11-29 修订日期：2023-01-13 |
| DOI：10.13929/j.issn.1003-3289.2023.04.010 |
| 中文关键词: 肺肿瘤 体层摄影术,X线计算机 腺癌 Ki-67抗原 |
| 英文关键词:lung neoplasms tomography, X-ray computed adenocarcinoma Ki-67 antigen |
| 基金项目:国家自然科学基金(82171931)、广州市科技计划(202102080572)、广州市番禺区科技计划(2019-Z04-01)。 |
|
| 摘要点击次数: 2062 |
| 全文下载次数: 787 |
| 中文摘要: |
| 目的 基于肺浸润性非黏液型腺癌(INMA)多层螺旋CT(MSCT)特征构建回归模型列线图,观察其术前预测肿瘤Ki-67表达的价值。方法 回顾性分析283例经术后病理证实的肺INMA患者的临床、病理及MSCT资料,按7 ∶ 3比例将其分为训练集(n=198)和验证集(n=85);根据病理结果分为高级别组与非高级别组;以受试者工作特征(ROC)曲线法获得Ki-67表达区分高级别与非高级别肺INMA的最佳诊断阈值,并分为高、低Ki-67表达组。基于训练集临床、病理及MSCT资料,单因素分析及多因素二元logistic回归分析,筛选肺INMA Ki-67表达的独立影响因素,构建回归模型及列线图,以验证集评估其预测肺INMA Ki-67表达的效能及其临床价值。结果 283例肺INMA中,高级别组106例,非高级别组177例。以Ki-67表达区分高级别与非高级别肺INMA的最佳诊断阈值为12.50%。高级别与非高级别组间Ki-67表达差异有统计学意义(P＜0.001)。训练集含高、低Ki-67表达肿瘤 66及132例,验证集含29及56例。MSCT所示肺INMA实性成分占比(CTR)是其Ki-67表达的独立影响因素。将CTR及病灶类型纳入回归模型并构建列线图,其在验证集预测肺INMA Ki-67高表达的曲下线面积为0.741;决策曲线分析显示,阈值取0.14~0.41和0.55~0.65时,列线图临床净获益较高。结论 肺INMA MSCT特征回归模型列线图可于术前有效预测其Ki-67表达。 |
| 英文摘要: |
| Objective To establish a regression model nomogram based on multi-slice spiral CT (MSCT) characteristics of pulmonary invasive non-mucinous adenocarcinoma (INMA), and to observe its value for preoperative prediction of tumor Ki-67 expression. Methods Clinical, pathological and MSCT data of 283 patients with postoperative pathologically confirmed pulmonary INMA were retrospectively analyzed. The patients were divided into training set (n=198) and validation set (n=85) at the ratio of 7 ∶ 3. According to pathological results, the patients were divided into high-grade group and non-high-grade group. Receiver operating curve (ROC) method was used to obtain the optimal diagnostic threshold of Ki-67 expression to distinguish high-grade from non-high-grade pulmonary INMA, then the patients were classified into high or low Ki-67 expression group. The independent influencing factors of Ki-67 expression of pulmonary INMA in training set were screened based on clinical, pathological and MSCT using univariate analysis and multivariate binary logistic regression analysis. A regression model was constructed and visualized as nomogram, and its efficacy for predicting Ki-67 expression of pulmonary INMA in validation set, as well as its clinical value were analyzed. Results Among 283 patients, high-grade INMA was found in 106 cases, while non-high-grade INMA was found in 177 cases. The optimal diagnostic threshold of Ki-67 expression for distinguishing high-grade and non-high-grade pulmonary INMA was 12.50%. Significant difference of Ki-67 expression was detected between high-grade and non-high-grade groups (P＜0.001). There were 66 cases with high and 132 with low Ki-67 expression in training set, while 29 and 56 cases in validation set, respectively. Consolidation-to-tumor ratio (CTR) showed on MSCT was the independent influencing factor for high Ki-67 expression of pulmonary INMA. CTR and lesion type were included in the regression model, and a nomogram was constructed, with the area under the curve for predicting high Ki-67 expression pulmonary INMA in validation set of 0.741. Decision curve analysis showed that this nomogram might bring better net clinical benefit when the threshold was 0.14 to 0.41 and 0.55 to 0.65. Conclusion Regression model nomogram based on MSCT characteristics of pulmonary INMA could be used to effectively predict tumor Ki-67 expression before operation. |
| 查看全文 查看/发表评论 下载PDF阅读器 |
|
|
|
