br The th th and
The 12th, 13th and 14th (2011, 2013 and 2015) St Gallen International Breast Cancer Conference Expert Panel adopted a new approach to the classification of patients for therapeutic purposes based on the recognition of intrinsic biological subtypes within the breast cancer spectrum (Luminal A, Luminal B HER2 -, Luminal B HER 2 +, HER 2 + not Luminal, Triple negative) [36–38]. The aim of our study was to retrospectively evaluate the distribution of MRI BPE among different breast cancer subtypes searching for any sig-nificant difference in terms of immunohistochemical and receptorial panels. European Journal of Radiology 113 (2019) 148–152
We found that mild BPE pattern was significantly more prevalent than other BPE types only in the luminal B (HER-) tumors; no other significant distribution difference was found among other breast cancer subtypes. The only exception was represented by triple negative group, in which an 208538-73-2 of this trend, where moderate and marked BPE prevailed over minimal and mild types. Among all patients with mild BPE, luminal B (HER2-) tumors were significantly higher while among all patients with marked BPE, triple negative subtypes were sig-nificantly more prevalent. None of the five tumor subtypes had sig-nificant higher rates in minimal and moderate BPE groups in our series.
Our results differ from what Kim already observed, with no statis-tical evidence of correlation between BPE grade and the receptorial cancer subtype, and are confident with J. Wang’s, who stated that the heterogeneity of BPE characterized by texture on DCE-MRI was asso-ciated with triple negative cancers [18,20].
In fact, this study shows a significant distribution of mild BPE pat-tern within Luminal B (HER2-) tumor subtypes, with higher rates of mild BPE among all Luminal B (HER2-) tumors and Luminal B (HER2-) subtypes among all mild BPE patients. These results are slightly dif-ferent from the ones obtained by Ha et al.  and Wu et al.  who found that patients with high BPE may be at increased risk for breast cancer but not necessarily for molecular subtypes with poor prognosis with a higher risk for Luminal B cancers.
Another important data emerging from our study reveals an in-creasing trend for BPE in triple negative tumors, respectively with no patient with minimal BPE, 2 patients with mild BPE, 6 patients with moderate BPE, and finally 10 patients with marked BPE. This dis-tribution does not show a statistical significance, probably because of the small number of triple negative patients in our sample, that re-presents one of the study limitations.
On the other side, among all patients with marked BPE, triple ne-gative tumors showed significant higher rates than other subtypes.
These results strongly suggest a potential role of BPE as a predictive tool for Luminal B (HER2-) and triple negative tumor diagnosis; how-ever, more studies in this field are required to confirm this hypothesis on larger series.
Our study only indicates a potential role of breast MRI for a prog-nostic stratification; in fact, minimal, mild and moderate BPE patterns are poorly distributed among triple negative tumors and could indicate a more favorable prognosis.
On the other side, an additional risk stratification for triple negative tumors could be proposed with dedicated screening programs and shortened intervals for MRI surveillance in case of marked BPE. With regard to the BPE qualitative evaluation in our study and the possible effect of any confounding factor, the multivariate regression ana-lysis and the dichotomized approach on BPE with logistic regression found no statistical significant effect of age, menopausal status and lesion diameter on BPE distribution; these results could be due to the small number of the enrolled patients and to the fact that only patients affected by breast cancer, and not at high risk of breast cancer, were enrolled.
Our study has some important limitations. First of all, the small number of the enrolled patients with an unequal distribution of mole-cular subtypes, with few triple negative cancer affected patients, al-though they usually represent the 10–20% of all breast cancers, as in our series. The second limitation concerns the qualitative and potentially subjective BPE evaluation, but this bias is reduced by the almost perfect inter-rater agreement observed in our study. A quantitative or auto-mated BPE measurement could be useful to standardize and compare data among different experiences.
Furthermore, we did not evaluate the symmetrical or asymmetrical distribution of BPE on the contralateral breast.
Of course, the retrospective design of the study is also an important limitation, especially regarding the possibility of demonstrating a cause-effect association between BPE grade and tumor subtype devel-opment.