We investigated conventional DCE-MRI and ultrafast MRI parameters to predict pCR and correlated MRI kinetic parameters with histologic MVD. Our study results suggest that smaller tumor size, the absence of axillary lymph node metastasis and HER2-enriched or TNBC subtypes could be significant factors for predicting pCR. However, ultrafast MRI parameters did not show significant predictive value in our study. In HER2-enriched cancer, lower initial enhancement and tumor angiovolume can be associated with pCR. MS on ultrafast MRI was only kinetic MRI parameter, associated with histologic MVD.
Conventional DCE-MRI exhibited a high AUC for predicting pCR (0.871–0.896), which is consistent with a previous meta-analysis by Gu et al. (0.88) [5]. Sensitivity was slightly higher on the first post-contrast phase and MIP images, but specificity and positive predictive value were higher on the last post-contrast phase. Recent studies have also emphasized that the absence of enhancement in the delayed phase was associated with pCR, while subtle enhancing lesions in the delayed phase could be indicative of pathologic residual tumor [24, 25]. Additionally, residual tumor size was more accurately evaluated on the last post-contrast phase and MIP images than the first post-contrast phase, which is consistent with previous studies [25–27]. It is speculated that the antiangiogenic effects of chemotherapy might slow down the rate of contrast enhancement in breast cancer [26, 28]. Furthermore, residual radiologic tumor size reflects the total histologic invasive and in situ tumor size rather than invasive tumor only, aligning with previous literature [27]. Therefore, when assessing tumor response after NAC and determining the extent of surgical intervention, reviewing delayed dynamic images in addition to early dynamic images could yield more accurate results.
In comparing pCR and non-pCR groups, we observed significant differences in tumor size, lymph node status, tumor angiovolume, delayed kinetic components, and hormone receptor and HER2 status. Advanced tumoral and nodal stages, indicating high tumor burden, are well-known prognostic factors in breast cancer patients receiving NAC [7, 29, 30]. Additionally, molecular subtypes (HER2-enriched or triple-negative vs. luminal cancer) were the most significant factors for predicting pCR, and this has been proven in many studies, likely due to advancements in chemotherapy agents and the addition of anti-HER2 target therapy [18, 24, 29, 30]. An increased delayed washout component was the significant kinetic parameter associated with pCR in univariate regression analysis, but not an independent factor for pCR in multiple analysis. Previous researchers have reported that an increased wash-out rate of tumor enhancement predicts pCR [6, 31]. We speculate that molecular subtype could be a confounding factor for the delayed washout component, because HER2-enriched and triple-negative cancers exhibited significantly higher delayed washout fractions compared to the luminal subtype (mean delayed washout components, 51.9 ± 19.0, 55.9 ± 14.4 vs. 38.6 ± 23.0, p = 0.048). In the comparison of MRI kinetic parameters according to molecular subtypes, the pCR group in HER2-enriched cancer showed a lower initial enhancement value and tumor angiovolume compared to the non-pCR group. Early strong enhancement and high perfusion in breast cancer imply increased tumor angiogenesis. Poorly organized and malformed vessels of tumors can reduce the efficacy of drug delivery [32]. Therefore, highly vascularized HER2-enriched cancer may respond poorly to NAC.
In this study, ultrafast MRI parameters were not predictive of pCR, regardless of molecular subtype. Previous studies have explored the relationship between ultrafast MRI kinetic parameters and tumor response in breast cancer patients receiving NAC, yielding inconsistent results [18, 19]. A relatively large-scale retrospective study by Kim et al. found no significant ultrafast MRI parameter predictive of pCR in all breast cancer subtypes, similar to our study. However, a larger ratio of enhancing tumor volume between the first and subsequent time points of tumor enhancement was associated with pCR only in TNBC [18]. On the other hand, in a prospective study with a smaller study population by Ramtohul et al., the pCR group exhibited increased ultrafast MRI parameters, compared to the non-pCR group. High wash-in slope (> 1.6%/s), HER2 positivity, and high tumor-infiltrating lymphocytes were important parameters for pCR in their study [19]. Ramtohul et al. explained that high tumor perfusion measured in the early enhancement phase could be associated with a greater blood supply and better oxygenation of the tumor, resulting in a favorable response to chemotherapy agents [19]. The differences between our findings and those of previous studies may be attributed to variations in study populations' size and characteristics, as well as differences in ultrafast MRI acquisition protocols and kinetic parameter analysis. Nevertheless, Kim et al.'s study that had used similar ultrafast MRI acquisition and analysis methods reported similar negative results regarding the association between ultrafast MRI parameters and NAC response in all subtypes of breast cancer [18]. Therefore, we cautiously conclude that the role of ultrafast MRI before NAC in predicting tumor response is unlikely to be significant, but future studies with a larger number of participants and stratification by molecular subtype are warranted.
Regarding the correlation between MRI kinetic parameters and histologic MVD, only MS on ultrafast MRI was associated with MVD. Several preceding studies have reported varying results in terms of the type of MRI parameters correlated and the degree of correlation [14, 17, 33–38]. Some studies have commonly reported that the initial signal enhancement ratio on conventional DCE-MRI was positively correlated with MVD [33, 35, 36], while others couldn't demonstrate any relationship between MRI kinetic parameters and MVD [34, 38]. A recent study by Mori et al. reported that the initial slope on ultrafast MRI was correlated with MVD (correlation coefficient 0.51), but conventional DCE-MRI parameters were not, similar to our study [17]. The diversity of these results may be due to variations in MVD counting methods, staining techniques, MRI acquisition protocols, and analysis methods, as well as differences in study participants' characteristics. Since our study included advanced breast cancer patients receiving NAC, most of the lesions would have been highly vascularized tumors, potentially explaining the lack of significant differences in vascularity between lesions. Furthermore, the MVD was assessed on biopsy specimens and therefore may not be representative of the MVD of the entire tumor. The field area of the microscope may also have had an impact. Some reports suggest that MVD measured at a higher magnification, 400X rather than 200X is more related to prognosis than at a lower magnification [39, 40]. Nevertheless, the correlation of MS with MVD implies that the initial fast and strong enhancement can predict the vascularity of the tumor to some extent. Future studies with more participants are needed to validate these results.
Our study has several limitations. First, the study included a relatively small number of participants, preventing a detailed analysis by molecular subtype and warranting caution when generalizing the study results. Second, we were unable to include mid-term MRI during NAC in our study because abbreviated MRI without ultrafast imaging was used for mid-term evaluation in routine clinical practice during the study period. Therefore, we couldn't evaluate the role of mid-term MRI or the significant changes in breast cancer during the early stages of NAC in predicting pCR. Finally, we used two different MRI vendors for this study, potentially introducing heterogeneity in study results.
In conclusion, larger tumor size, the absence of lymph node metastasis, and molecular subtypes such as HER2-enriched cancer or TNBC are significant factors for predicting pCR. While kinetic parameters of conventional DCE-MRI and ultrafast MRI may not be independently associated with pCR, our findings suggest that low initial enhancement and a small angiovolume of the whole tumor may be associated with pCR in HER2-enriched cancer. A prospective study with a larger study population should be performed to validate these results and to evaluate the role of ultrafast and DCE-MRI according to molecular subtypes in predicting tumor response in breast cancer patients undergoing NAC.