| 李琳,张留磊,李自强,段金辉,殷慧佳,韩琳,任继鹏,韩东明.多参数MRI联合临床特征及血清肿瘤标志物预测非小细胞肺癌Ki-67表达水平[J].中国医学影像技术,2026,42(5):722~726 |
| 多参数MRI联合临床特征及血清肿瘤标志物预测非小细胞肺癌Ki-67表达水平 |
| Multiparametric MRI combined with clinical characteristics and serum tumor markers for predicting Ki-67 expression level in non-small cell lung cancer |
| 投稿时间：2025-11-10 修订日期：2026-04-28 |
| DOI：10.13929/j.issn.1003-3289.2026.05.014 |
| 中文关键词: 癌,非小细胞肺 Ki-67抗原 弥散磁共振成像 |
| 英文关键词:carcinoma, non-small-cell lung Ki-67 antigen diffusion magnetic resonance imaging |
| 基金项目: |
|
| 摘要点击次数: 85 |
| 全文下载次数: 43 |
| 中文摘要: |
| 目的 观察基于多参数MRI、临床特征及血清肿瘤标志物预测非小细胞肺癌(NSCLC)Ki-67表达水平的价值。方法 前瞻性纳入67例NSCLC,根据免疫组织化学Ki-67染色结果分为高表达组(＞30%,n=34)与低表达组(≤30%,n=33)。采集多b值弥散加权成像及T1 mapping图像,测量单指数模型[表观弥散系数(ADC)]、双指数模型(真实弥散系数、伪弥散系数和灌注分数)及拉伸指数模型[分布扩散系数(DDC)、时间弥散异质性指数(α)]参数及T1值(T1 mapping),同时收集血清肿瘤标志物等资料。采用组内相关系数(ICC)评估MRI参数测值的观察者间一致性。以弹性网络逻辑回归筛选变量并构建预测模型;绘制受试者工作特征(ROC)曲线,以曲线下面积(AUC)评估模型预测能力,以Hosmer-Lemeshow检验评估模型校准度。结果 MRI参数的观察者间一致性均佳(ICC均＞0.832)。弹性网络逻辑回归模型含吸烟状态、神经元特异性烯醇化酶(NSE)及ADC值共3个独立预测变量(P均＜0.05),其预测NSCLC Ki-67表达水平的AUC为0.890,且校准度良好(P=0.231)。结论 基于吸烟状态、NSE及ADC构建的弹性网络逻辑回归模型用于预测NSCLC Ki-67表达具有一定价值。 |
| 英文摘要: |
| Objective To observe the value of multiparametric MRI, clinical characteristics and serum tumor markers for predicting Ki-67 expression level in non-small cell lung cancer (NSCLC). Methods A total of 67 patients with NSCLC were prospectively enrolled. According to immunohistochemical Ki-67 staining results, the patients were stratified into high-expression group (＞30%, n=34) and low-expression group (≤30%, n=33). Multi-b-value DWI and T1 mapping images were acquired, and parameters including apparent diffusion coefficient (ADC) (monoexponential model), true diffusion coefficient, pseudo-diffusion coefficient and perfusion fraction (biexponential model), distributed diffusion coefficient (DDC) and heterogeneity index (α) (stretched exponential model), as well as T1 value (T1 mapping) were measured. Clinical data including serum tumor markers were collected. Inter-observer agreement for measurements of MRI parameters were assessed using intra-class correlation coefficients (ICC). Elastic net logistic regression was employed for variable selection and model development. The receiver operating characteristic (ROC) curve was plotted, and the area under the curve (AUC) was used to evaluate the predictive performance of the model. Calibration of the model was assessed with Hosmer-Lemeshow test. Results All MRI parameters had good inter-observer agreement (all ICC＞0.832). Three predictive variables, including smoking status, neuron-specific enolase (NSE) and ADC value were retained in the elastic net logistic regression model (all P＜0.05). The AUC of the model for predicting Ki-67 expression level in NSCLC was 0.890, with good calibration (P=0.231). Conclusion The elastic net logistic regression model constructed based on smoking status, NSE and ADC demonstrated certain value for predicting NSCLC Ki-67 expression level. |
| 查看全文 查看/发表评论 下载PDF阅读器 |
|
|
|
