Impact of preoperative magnetic resonance imaging (MRI) on conservative surgery (CSur) among curable breast cancer (BC) patients treated in a specialized center in Latin America.

Abstract

e12595 Background: Preoperative breast MRI can provide important clinical information on the surgical management of BC patients (pts). However, there is no evidence that it can produce a statistical impact on surgical outcomes in the early BC setting. This retrospective study sought to evaluate whether preoperative MRI was significantly correlated with CSur among early BC pts. Methods: Digital and physical records from 532 pts treated in a specialized BC center in Brazil from 2005 to 2018 were analyzed. The institutional database registered pathological, radiological and clinical information. The categorical variables CSur and MRI were analyzed by Chi-squared coefficient of correlation, whereas significant correlations were included in a multivariate logistic regression model. All statistical analysis were performed in STATA version 15.1C. Results: The present analysis included 532 pts. The diagnosis was suspected through suspicious self-exam in 380 pts (71%), while 129 pts (24%) had altered mammary screening tests. Clinical T and N stages were: T0-1=154pts (29%); T2=179 pts (33%); T3=135 pts (25%); T4=49 pts (9%); N0=267 pts (50%); N1=172 pts (32%); N2=80 pts (15%). Neoadjuvant treatment was performed in 223 pts (42%). MRI was indicated in 248 pts (46%). CSur was performed in 149 pts (28%). Margins were positive in 15 cases. This study found a statistically significant correlation between CSur and MRI ( X2 = 8.07; p=0.018). Although, when controlled for neoadjuvant treatment, T, N, and age, the independent variable MRI was not a statistical predictor of CSur (R=0.008; t=0.25; p=0.8; 95%CI -0.57 – 0.74). Advanced T and neoadjuvant treatment were inversely correlated with CSur in the multivariate analysis (R= -0.11; t= -7.8; p<0.001 and R= -0.11; t= -2.88; p=0.004, respectively). One multivariate analysis to evaluate predictors of T stages showed that suspicious self-exam and N stage had a significant linear relationship with T stages when controlled for age (R=0.70; t=7.9; p<0.001; and R=0.76; t=15; p<0.001, respectively). Conclusions: Preoperative MRI can have impact on the conservative surgical management of curable BC patients and the present analysis showed a positive correlation. In the logistic regression model, MRI was not identified as a predictor of CSur, however the proportion of locally advanced BC may have influenced the results. Further studies must be strengthened for identifying in which subgroup of patients MRI may be a predictor of surgical outcomes.