Department of Surgical Oncology, The University of Texas M. D. Anderson Cancer Center, Houston, TX 77030, USA.
Ann Surg. 2010 Oct;252(4):611-7. doi: 10.1097/SLA.0b013e3181f56419.
Esophageal cancer patients with pathologic lymph-node involvement (pN1) generally have a poor prognosis with surgery alone. We, therefore, constructed a nomogram to predict the risk of pN1 prior to surgical resection and externally validated the clinical utility of the model.
A total of 273 esophageal adenocarcinoma patients treated with surgery alone were reviewed from 2 different institutions (University of Texas M. D. Anderson Cancer Center = 164, training set; University of Rochester School of Medicine and Dentistry = 109, validation set). Pretreatment clinical parameters were used to construct a nomogram for predicting the risk of pN1. Internal and external validation of the nomogram was performed to assess clinical utility.
Of the 164 patients in the training set, 56 patients (34%) had lymph-node involvement (pN1). Significant factors associated with pN1 on univariable logistic regression analysis (using a P < 0.05) included endoscopically determined clinical tumor depth (cT), clinical nodal (cN) status, and clinical tumor length (cL). Multivariable analysis suggested the significant independent factors were cT (odds ratio, 5.6; 95% confidence interval, 1.7-18.6; P < 0.01) and cL >2 cm (odds ratio, 7.0; 95% confidence interval, 2.7-18.1; P < 0.001). Regression tree analysis was used to determine the best cutoff for cL. A nomogram was created for pN1 using these clinical parameters and was internally validated by bootstrapping with a predicted accuracy of 85.1%. External validation performed on the validation set demonstrated an original C-index of 0.777 suggesting good clinical utility.
Our analyses demonstrate that the risk of pathologic nodal involvement in esophageal adenocarcinoma patients can be estimated by this clinical nomogram, which will allow the identification of patients at high-risk of harboring positive lymph-nodes, who may be candidates for en bloc resection and/or neoadjuvant treatment.
单纯手术治疗的病理淋巴结受累(pN1)食管癌患者预后通常较差。因此,我们构建了一个列线图来预测手术前 pN1 的风险,并对模型的临床实用性进行了外部验证。
回顾了来自 2 个不同机构(德克萨斯大学 M.D.安德森癌症中心=164 例,训练集;罗切斯特大学医学院和牙科学院=109 例,验证集)的 273 例单纯手术治疗的食管腺癌患者。使用术前临床参数构建预测 pN1 风险的列线图。通过内部和外部验证来评估列线图的临床实用性。
在训练集中的 164 例患者中,56 例(34%)有淋巴结受累(pN1)。单变量逻辑回归分析显示与 pN1 相关的显著因素包括内镜确定的临床肿瘤深度(cT)、临床淋巴结(cN)状态和临床肿瘤长度(cL)。多变量分析表明,显著的独立因素是 cT(优势比,5.6;95%置信区间,1.7-18.6;P<0.01)和 cL>2cm(优势比,7.0;95%置信区间,2.7-18.1;P<0.001)。回归树分析用于确定 cL 的最佳截断值。使用这些临床参数创建了一个预测 pN1 的列线图,并通过 bootstrap 进行了内部验证,预测准确率为 85.1%。在验证集中进行的外部验证显示原始 C 指数为 0.777,表明具有良好的临床实用性。
我们的分析表明,该临床列线图可以估计食管腺癌患者病理淋巴结受累的风险,这将有助于识别具有阳性淋巴结高风险的患者,这些患者可能是整块切除和/或新辅助治疗的候选者。