中国人群非前哨淋巴结转移预测模型研究

Nomogram for predicting the likelihood of additional nodal metastases in breast cancer patients with a positive sentinel node biopsy

目的腋窝淋巴结清除(axillary lymph node dissection,ALND)是乳腺癌前哨淋巴结(sentinel lymph node,SLN)阳性患者的标准术式,但很多患者进行ALND后并未发现腋窝淋巴结转移。为了避免患者进行不必要的非前哨淋巴结活检(sentinel lymph node biopsy,SLNB),提高患者的生活质量,本研究建立适合中国人群的预测非SLN阳性率模型。方法收集2010-01-01-2017-09-30中国临床肿瘤学会(Chinese Society of Clinical Oncology,CSCO)数据(研究号:CSCO BC RWS18005),其中行SLNB并行淋巴结清除患者1 036例,数据完整患者共235例。建立2个模型,模型1根据多因素分析将P<0.05变量纳入模型,模型2根据临床、病理资料建立,并将2个模型进行对比研究。结果多因素分析结果示,肿瘤直径、脉管瘤栓、肿瘤类型、SLN阳性数与SLN总数比值(the ratio of sentinel lymph node positive number and total sentinel lymph node ratio,SLN.POS.RATIO)差异有统计学意义,P<0.05。将上述4个变量纳入模型1,将模型1变量加上年龄、雌激素受体、孕激素受体、人表皮生长因子受体2和Ki-67纳入模型2,模型1受试者工作特征(receiver operating characteristic,ROC)曲线下面积(area under curve,AUC)为0.739,模型2ROC AUC为0.747,2个模型均具有良好的预测效能,2个模型对于非SLN阳性转移率预测差异无统计学意义,P=0.428 5。结论依据临床病理等参数建立的模型1,对于预测乳腺癌个体患者非SLN转移概率方便、快捷,可能对乳腺癌ALND范围的合理选择有一定的指导意义。

OBJECTIVE The intraoperative ALN status in patients with breast cancer is determined by the sentinel lymph node biopsy(SLNB).However,many patients have not found axillary lymph node metastasis after axillary lymph node dissection.We established a non-sentinel lymph node metastasis model for the Chinese population to avoid unnecessary non-sentinel lymph node biopsy and improving the quality of life of patients.METHODS Chinese Society of Clinical Oncology(CSCO)(Research number is CSCO BC RWS 18005)data from January 1,2010 to September 1,2017 were collected,including 1 036 patients with sentinel lymph node biopsy and lymph node clearance,and 235 patients with fulfilled data.Two models were established.Model 1 was analyzed according to multivariate analysis.Variables(P<0.05)were included in the model 1,and model 2 was established based on clinical and pathological data,then the two models were compared.RESULTS The multivariate analysis showed that the tumor diameter,vascular tumor thrombus,tumor type,the ratio of sentinel lymph node positive number and total sentinel lymph node ratio(RST),were included in model 1,tumor diameter,vascular tumor thrombus,tumor type,sentinel lymph node positive/sentinel lymph node total,age,estrogen receptor,progesterone receptor,HER2,Ki-67 were included in model 2.Model 1 ROC curve area was 0.739,model 2 ROC curve area was 0.747,and both models had good predictive power.By comparison,two models for non-sentinel lymph node metastasis was not statistically significant(P=0.428 5).CONCLUSION Model 1 created with clinical-pathological parameters,which is convenient and rapid for predicting the non-SLN metastasis of individual breast cancer patients,has certain guiding significance for the rational selection of breast cancer ALND.