MRI对乳腺癌肿瘤大小测量准确性及影响因素分析

Analysis on accuracy and influencing factors of MRI in predicting breast tumor size

目的肿瘤大小是判断乳腺癌患者预后的重要依据之一,同时对于治疗方案的选择及新辅助化疗疗效的评估至关重要。本研究探讨磁共振成像(magnetic resonance imaging,MRI)对于乳腺癌肿瘤大小测量的准确性,并对其影响因素进行分析。方法回顾性分析2015-01-01-2016-08-30河北医科大学第四医院乳腺中心行MRI检查的195例(198个病灶)乳腺癌患者的临床病理及MRI检查资料。根据免疫组化ER、PR、HER-2及Ki-67值,将分子分型分为Luminal A型、Luminal B型、HER-2过表达型及三阴型。采用多因素Logistic回归分析MRI测量乳腺癌肿瘤大小不准确的预测因素。结果病理学及MRI测量肿瘤大小的中位数(四分位间距)分别为1.6(1.0)cm、1.8(0.9)cm。198个病灶中,MRI检查准确评估病灶大小为78.3%(155/198),高估病灶14.6%(29/198),低估病灶7.1%(14/198)。Spearman相关分析显示,MRI与病理所测乳腺癌肿瘤大小呈高度相关,r=0.79,P<0.001。Bland-Altman分析显示,与病理结果比较,MRI略高估病灶大小,平均差值为0.19cm,95%CI为1.42～1.80。多因素Logistic回归分析提示,非肿块型病变及HER-2过表达型乳腺癌是MRI测量肿瘤大小不准确的独立预测因素(OR=5.45,95%CI为1.91～15.54,P=0.002;OR=5.03,95%CI为1.30～19.51,P=0.02)。结论 MRI对于乳腺癌肿瘤大小的测量准确性较高,但仍存在高估或低估肿瘤大小的情况。病变强化形态和分子分型与MRI测量肿瘤大小准确性相关,非肿块型病变及HER-2过表达型乳腺癌更易出现测量肿瘤大小不准确。

OBJECTIVE The precise measurement of tumor size in breast cancer is important for the prediction of prognosis,the planning of appropriate treatment and monitoring the therapeutic response.The goal of this study was to evaluate whether preoperative MRI imaging of the breast was an accurate measure of tumor size by comparing it to the histopathological size,and evaluate the impact of different clinicopathologie and radiological tumor characteristics on the accuracy of these measurement.METHODS The clinicopathologic and radiological data of 198 breast cancer lesions of 195 patients were retrospectively analyzed(from January 1,2015 to August 30,2016 at the Fourth Hospital of Hebei Medical University).Molecular subtypes were divided into Luminal A,Luminal B,HER-2 positive and triple negative types according to immunohistochemistry and Ki-67.Multivariate logistic regression analysis was performed to identify independent variables that were associated with the accuracy of the assessments of MRI sizes and pathology sizes.RESULTS The median tumor size was 1.8 cm on MRI,compared to 1.6 cm on pathology.MRI-pathology concordance within 5 mm was found in 78.3%cases(155/198).MRI overestimated the size of 14.6%tumors(29/198)and MRI underestimated the size of 7.1%tumors(14/198).MRI had a significant correlation with final pathology,r=0.79,P<0.001.Bland-Airman analysis showed that the mean difference between MRI and histology was 0.19 cm.Multivariate Logistic regression analysis showed that non-mass type breast cancer and HER-2 positive breast cancer were independent predictor of tumor regression models(OR=5.45,P=0.002;OR=5.03,P=0.02).CONCLUSIONS Quality of tumor size measurement using MRI is good,but there are still overestimated and underestimated the size of tumors.The lesion type on MRI and molecular subtypes are related to the inaccurate estimation of tumor size based on MRI.Non-mass type breast cancer and HER-2 positive breast cancer are strongly associated with discordance between MRI and pathology sizing of breast cancer.