RESUMO
BACKGROUND: Many cT3 breast cancer patients are treated with mastectomy, regardless of response to neoadjuvant systemic therapy (NST). We evaluated local control of cT3 patients undergoing breast-conserving therapy (BCT) based on magnetic resonance imaging (MRI) evaluation post-NST. In addition, we analyzed predictive characteristics for positive margins after breast-conserving surgery (BCS). METHODS: All cT3 breast cancer patients who underwent BCS after NST between 2002 and 2015 at the Netherlands Cancer Institute were included. Local recurrence-free interval (LRFI) was estimated using the Kaplan-Meier method, and predictors for positive margins were analyzed using univariable analysis and multivariable logistic regression. RESULTS: Of 114 patients undergoing BCS post-NST, 75 had negative margins, 16 had focally positive margins, and 23 had positive margins. Of those with (focally) positive margins, 12 underwent radiotherapy, 6 underwent re-excision, and 21 underwent mastectomy. Finally, 93/114 patients were treated with BCT (82%), with an LRFI of 95.9% (95% confidence interval [CI] 91.5-100%) after a median follow-up of 7 years. Predictors for positive margins in univariable analysis were hormone receptor-positive/human epidermal growth factor receptor 2-negative (HR+/HER2-) subtype, lobular carcinoma, and non-mass enhancement (NME) on pre-NST MRI. MRI response was not correlated to positive margins. In multivariable regression, the odds of positive margins were decreased in patients with HER2-positive (HER2+; odds ratio [OR] 0.27, 95% CI 0.10-0.73; p = 0.01) and TN tumors (OR 0.17, 95% CI 0.03-0.82; p = 0.028). A trend toward positive margins was observed in patients with NME (OR 2.38, 95% CI 0.98-5.77; p = 0.055). CONCLUSION: BCT could be performed in 82% of cT3 patients in whom BCT appeared feasible on post-NST MRI. Local control in these patients was excellent. In those patients with HR+/HER2- tumors, NME on MRI, or invasive lobular carcinoma, the risk of positive margins should be considered preoperatively.