Residual microcalcifications after neoadjuvant chemotherapy for locally advanced breast cancer: comparison of the accuracies of mammography and MRI in predicting pathological residual tumor

Neoadjuvant chemotherapy (NAC) has been established as the standard treatment for inoperable or locally advanced breast cancer (LABC). Moreover, NAC has been increasingly used as one of the emerging treatment options for operable breast cancer. The advantages of NAC are that it facilitates breast-conserving surgery by reducing primary tumor burden and that it improves survival by treating micrometastasis [1‐5]. The achievement of pathological complete response (pCR) after NAC is a known surrogate marker of disease-free survival as well as overall survival [6]. Therefore, an accurate assessment of response to NAC and residual tumor is important for planning the extent of surgery and predicting prognosis.

To date, breast magnetic resonance imaging (MRI) is the most accurate imaging method for assessing the extent of residual tumor after NAC than other imaging modalities [7‐14]. However, MRI has limitations in evaluating the accurate extent of breast cancer associated with suspicious malignant microcalcifications on mammography (MG). Controversy remains as to whether all residual microcalcifications after NAC reflect residual tumor or whether changes in the number and pattern of microcalcifications represent a therapeutic response. Previous studies have shown that residual microcalcifications after NAC are not always correlated with residual tumor burden [15‐20]. Residual microcalcifications can represent not only remnant malignant tumors but also necrotic tumor cell products in patients after treatment [21‐25].

Accordingly, the aim of this study was to correlate residual mammographic microcalcifications after NAC with pathological results. We also compared the accuracy of MG and MRI in predicting the size of pathological residual tumors.

In our study, microcalcifications within the tumor bed after NAC were unchanged in 82.8% of patients; residual microcalcifications were malignant in 55.2% of patients and benign in 44.8% of patients, and these findings were similar those obtained in previous studies [15‐18]. Additionally, we found that fine pleomorphic microcalcifications were significantly correlated with residual malignancy, but amorphous microcalcifications were correlated with benign microcalcifications according to the final pathology. However, there was no significant difference between malignant and benign microcalcifications regarding distribution or changes after NAC.

We found that the accuracies of both MG and MRI in predicting the size of the pathological residual tumor were insufficient in cases of residual microcalcifications after NAC. In the study, the size of the residual tumor with remaining microcalcifications after NAC tended to be overestimated by MG and underestimated by MRI. The size of residual enhancement on post-NAC MRI was correlated to a greater degree to the size of the pathological residual tumor than to the extent of residual microcalcifications on post-NAC MG (CCC = 0.196 vs. 0.566). These findings are similar to those of Weiss et al. [18], who reported that the CCCs of MG and MRI with pathology were − 0.12 and 0.55, respectively.

We also found that the accuracies of both MG and MRI in predicting pathological tumor size were affected by tumor receptor status (Table 4). The agreement between the residual microcalcifications and the size of the pathological residual tumors was highest for ER(+) and HER2(−) tumors (CCC = 0.5629 and 0.5472, respectively) and lowest for ER(−) tumors (CCC = − 0.0162). The reliability of MRI in predicting the size of the pathological residual tumor was highest (CCC = 0.8584) for ER(−) tumors and substantial in PR(−) and HER2(+) tumors (CCC = 0.7920 and 0.6606, respectively). These results indicate that patients with ER(−) tumors, even if extensive microcalcifications remain on post-NAC MG, can be considered candidates for breast-conservation surgery if there is no residual enhancement on post-NAC MRI. In contrast, in patients with ER(+) or HER2(−) tumors, the residual microcalcifications should be completely resected because MRI might underestimate the residual lesions. Recently, Kim et al. [19] demonstrated that the reliabilities of both MG and MRI for the prediction of residual tumor differed among breast cancer subtypes. In all subtypes, the agreement between the extent of residual microcalcifications and the size of the pathological residual tumor was fair and was lower than that obtained using MRI (ICC = 0.368 and 0.723, p < 0.0001). The reliability of MG for the prediction of residual tumor was highest for the HR+/HER2+ subtype (ICC = 0.417) and lowest for the triple negative (TN) subtype (ICC = 0.205). The extent of residual tumor as measured by MG was more frequently underestimated in HR+/HER2(−) and TN tumors and was overestimated in HR+/HER2+ and HR−/HER2+ tumors. Although the statistical analysis methods that they used and their results differed from ours, both studies showed that the accuracies of both MG and MRI in predicting the pathological tumor size can be influenced by molecular subtype and receptor status. This means that the concern expressed about the incomplete resection of microcalcifications may depend on the subtype or receptor status of the breast cancer. Further investigations with a large, prospective cohort are needed to validate these results.

The presence of residual microcalcifications on post-NAC may confuse surgeons when they decide whether to perform breast-conserving surgery or mastectomy. Until now, there has been no consensus or guideline on how to handle residual microcalcifications after NAC. Feliciano et al. suggested that the complete resection of residual microcalcifications should remain the standard because the likelihood of residual malignancy remains high for patients who showed a decrease in microcalcifications on post-NAC MG and for those who showed a loss of enhancement on post-NAC MRI [17]. Neither the changes of microcalcifications on MG nor the combined assessment of MG and MRI after NAC predicted the presence of pCR. Although the absence of residual enhancement on MRI was strongly correlated with pCR, the discordant rate between absent residual enhancement on post-NAC MRI and pCR was 26%. In our study, the absence of residual enhancement on post-NAC MRI did not sufficiently predict the absence of residual tumor in 33% of cases. Therefore, we agree with the suggestion of Feliciano et al. that all microcalcifications within the tumor bed should be completely excised, although not all residual microcalcifications on post-NAC MG reflect residual malignancy.

The main limitation of our study was the small sample size. Additionally, the intra- and inter-observer variabilities between the measurement of residual lesions using MG and MRI were not evaluated. Third, our study population mostly included luminal cancers or HER2-enriched cancers. Therefore, the generalization of our results to all subtypes of breast cancer may be difficult. Fourth, patients who did not show microcalcifications within their tumor were excluded from this study. Because triple-negative cancer usually presents as a mass with a relatively well-circumscribed margin rather than as suspicious malignant microcalcifications, most triple-negative cancers were excluded from this study. Furthermore, these considerations explained the low pCR rate of 13.8% found in our study because previous studies have shown a higher pCR rate after NAC in TN cancer [27, 28].

Most microcalcifications within the tumor bed were unchanged after the NAC, and residual microcalcifications did not correlate with residual malignancy in 44.8% of cases. Among mammographic features, the shape of microcalcifications within the tumor bed significantly differed between benign and malignant microcalcifications according to final pathology. The amorphous microcalcifications more frequently became benign, while fine pleomorphic microcalcifications were more frequently correlated with residual malignant microcalcifications. The extent of residual microcalcifications as measured by MG had poor agreement with the size of the pathological residual tumor, which was lower than that by MRI. The accuracies of both MG and MRI in predicting residual tumor size after NAC depended on tumor receptor status, which might impact surgical planning or concerns about the incomplete excision of residual microcalcifications in subsets of patients with breast cancer. However, further investigations with a large, prospective cohort are needed to validate these results. Although MRI provided more reliable information about residual tumor than MG, all residual microcalcifications should be completely excised because the absence of residual enhancement on post-NAC MRI did not sufficiently predict the absence of residual tumor in 33% of cases.