Background Neoadjuvant chemotherapy (NAC) in breast cancer allows assessment of response to treatment and increases eligibility for breast conservation treatment (BCT). We previously reported that MRI tumor phenotype (based on loss of containment) predicted response to NAC and ability to achieve BCT. Since tumor subtype can affect MRI features, we sought to validate whether MRI phenotype and/or tumor subtype affected BCT eligibility in a prospective trial, I-SPY (CALGB 150007/ACRIN 6657). We also investigated the accuracy of post-NAC MRI in determining BCT eligibility, and the distribution of tumor subtypes within MRI phenotypes with the goal of identifying clinically useful factors associated with successful BCT. 

Methods We analyzed I-SPY 1 Trial data, a multi-center, prospective NAC trial with serial MRIs. Phenotypes 1 and 2 were well defined masses, while 3, 4, and 5 were more diffuse. Subjects were considered truly eligible for BCT if tumor diameter on surgical pathology was ≤ 4 cm (based on NSABP B-06 trial criteria). Post-NAC MRI and surgical pathology were considered discrepant if longest tumor diameter differed by > 2 cm. Categorical variables were compared using the chi-squared test. 

Results Of the 221 subjects, 198 had pre and post-NAC MRIs, and surgical pathology available. Of these, 174 had a tumor diameter > 4 cm prior to NAC, and were considered initially ineligible for BCT. After NAC, 141 (81%) patients became BCT eligible based on having ≤4 cm of residual tumor. In contrast, 46 (26%) had pathologic complete response (pCR, no invasive tumor). The rate of becoming BCT eligible differed by hormone receptor (HR)/Her2 status (p = 0.005) and MRI phenotype (p = 0.037), with higher rates of eligibility in Her2+ and triple negative tumors, and the well defined MRI phenotypes. Overall, 75 (38%) patients had a size discrepancy between tumor diameter on post-NAC MRI and size on surgical pathology. Subjects with diffuse MRI phenotypes had more size discrepancy than phenotypes 1 and 2 (p = 0.001). Subjects with HR+/Her2- tumors also had more size discrepancy than other subtypes (p = 0.004). The more diffuse MRI phenotypes were more commonly HR+/Her2- tumors, and less commonly HR−/Her2- (p = 0.006). There was no difference in actual receipt of BCT by HR/Her2 status. However, patients with a well-defined MRI phenotype were significantly more likely to actually receive BCT than those with a diffuse MRI phenotype (47% versus 27%, p = 0.023). 

Conclusions Patients with well defined patterns on MRI and with Her2+ or triple negative disease were more likely to become eligible for BCT after NAC. The same factors that predict pCR also predict BCT eligibility, but far more patients achieve BCT eligibility than pCR. MRI after NAC was less accurate in the diffuse phenotypes, which could reflect increased difficulty in measuring tumor size in this setting. Tumor phenotype and subtype can help predict likelihood of achieving eligibility for BCT, and can be useful in setting appropriate expectations for women at the start of NAC. Citation Information: Cancer Res 2011;71(24 Suppl):Abstract nr PD02-05.