Effect of preoperative MRI on surgery program of patients with breast invasive lobular carcinoma by propensity score matching

Abstract

Objective To evaluate the effect of magnetic resonance imaging(MRI)examination on the treatment of patients with breast invasive lobular carcinoma(ILC)by using propensity score matching, and to study the relationship between preoperative breast MRI and the surgical outcomes of ILC patients. Methods The clinical and imaging data of 210 patients with ILC who underwent surgery from January 2006 to December 2016 were retrospectively analyzed who had been admitted to Beijing Hospital of Integrated Traditional Chinese and Western Medicine and Beijing Shangdi Hospital.61.9% of patients(130/210)underwent MRI examination.The MR detection rate of undetected lesions in mammography and ultrasonography was calculated and the patients′ treatment processes were analyzed.After the propensity scores matching, 34 patients with 1∶1 matched ratio, the differences among 14 basic characteristics and tumor characteristics in patients undergoing preoperative MRI(MRI group)and patients without preoperative MRI(control group)became consistent.Then the surgical results were compared. Results Of the 130 patients in the MRI group, 34.6%(45/130)had occult lesions; 62.2%(28/45)had malignant lesions.In 24.6% of patients(32/130), the surgical plan changed due to preoperative MRI examination, and 91.1%(41/45)of the results were consistent with pathology.In the propensity score matching analysis, breast MRI was associated with a lower incidence of re-operation, with an odds ratio(OR)of 0.135, P<0.05.The first mastectomy(OR=0.736, P=0.663)was similar to the final mastectomy(OR=0.691, P=0.141)and there was no statistical correlation with preoperative breast MRI. Conclusion Preoperative MRI can be used to detect occult malignancies and significantly reduce the incidence of recurrent surgery without increasing the rate of mastectomy in patients with ILC. Key words: Invasive lobular carcinoma; Magnetic resonance imaging; Propensity score matching