前哨淋巴结阳性乳腺癌非前哨淋巴结状态的多因素分析

Multivariate Analysis of Non-sentinel Lymph Node Status in Breast Cancer Patients with Positive Sentinel Lymph Node

目的探讨影响前哨淋巴结阳性乳腺癌非前哨淋巴结状态的因素,建立判断有否转移的预测模型。方法回顾性分析我院自2003年-2010年共285例前哨淋巴结阳性乳腺癌患者临床病理资料。采用Logistic回归方法分析13种影响前哨淋巴结阳性乳腺癌非前哨淋巴结状态的因素,建立判断有否转移的预测模型,并验证模型的准确度、敏感度、特异性。结果单因素Logistic回归分析结果提示,有6个因素与NSLN转移具有密切相关性,分别为肿瘤大小(OR=1.45,P<0.01)、阳性SLN大小(OR=2 078.49,P<0.01)、阳性SLN数量(OR=2.44,P<0.01)、阴性SLN数量(OR=0.19,P<0.01)、脉管侵犯(OR=11.45,P<0.01)、阳性SLN包膜外扩散(OR=74.34,P<0.01)。Logistic多因素回归分析表明:肿瘤大小、脉管侵犯、阴性SLN数量、阳性SLN大小及阳性SLN包膜外扩散与NSLN转移密切相关(P<0.05)。Logistic回归模型预测前哨淋巴结阳性乳腺癌非前哨淋巴结状态的敏感度为92.62%(138/149),特异性为89.15%(115/129),总符合率91.01%(253/278)。结论 Logistic回归预测模型能较好的判断前哨淋巴结阳性乳腺癌非前哨淋巴结的状态,有助于乳腺肿瘤外科医师选择最佳治疗方案。

Objective To study the factors influencing non-sentinel lymph node （NSLN） status in patients with SLN-positive breast cancer and to establish a predictive model. Methods Clinicopathological data of 285 breast cancer cases with positive sentinel lymph node were collected. The 13 factors were analyzed by Logistic regression. Predictive model was established for judging matastasis. The sensitivity, specificity and accuracy of the Logistic model were calculated. Results Univariate logistic regression revealed signif icant correlation between six variables and NSLN metastasis. The six variables were tumor size （OR = 1.45, P〈0. 01 ）, size of positive SLNs（OR = 2 （178. 49, P〈0. 01 ）, number of positive SLNs （OR = 2. 44, P〈0. 01 ）, number of negative SLNs （OR = 0. 19, P〈0. 01 ）, vascular invasion （OR = 11.45, P〈0. 01 ）, and positive SLN membrane invasion （OR = 74. 34, P〈0. 01 ）. The size of primary tumor, vascular invasion, number of negative SLNs and size of positive SLNs as well as its membrane invasion were significantly related to NSLN metastasis （P〈0. 05）. The rate of sensitivity, specificity and correction classified of the logistic model were 92. 62%,89. 15% and 91.01 %, respectively. Conclusion The model of 5 pre dictive factors are reliable in predicting the NSLN status and guiding clinical treatment in SLN-positive breast cancer.