Introduction
A positive estrogen receptor (ER) in breast cancer determines if patients should receive endocrine treatment. However, not all patients with ER+ breast cancer benefit from endocrine treatment: 40–50% relapse after adjuvant endocrine therapy [
1] and 50–70% show a clinical response after neoadjuvant endocrine therapy (NET) [
1‐
3]. A more accurate prediction whether endocrine treatment will be effective would benefit these patients, and allow for better selection and personalization of endocrine treatment.
Early prediction of NET efficacy could be used to personalize the course of treatment, i.e., expedite surgery or switch to neoadjuvant chemotherapy (NAC) in poor responders.
Typically, response monitoring during neoadjuvant therapy is performed with imaging. Magnetic resonance imaging (MRI) of the breast is the most accurate and recommended modality [
4,
5]. Several MRI features have been identified as predictors of tumor response during NAC [
6‐
11]. However, research regarding response monitoring in NET is limited [
12,
13].
A potential predictor of endocrine treatment efficacy is contralateral parenchymal enhancement (CPE). CPE is a quantitative measure of the relative late parenchymal enhancement of the healthy breast on MRI [
14,
15] and differs from background parenchymal enhancement (BPE), which is a qualitative measure of early parenchymal enhancement. CPE is calculated as the mean of the top-10% relatively most enhancing voxels. A high CPE was shown to be associated with improved survival in unilateral ER+ human epidermal growth factor 2 receptor–negative (HER2−) breast cancer patients after adjuvant endocrine therapy [
14,
15]. If CPE is also associated with NET efficacy, it could be used to personalize the course of NET in breast cancer patients.
It is hypothesized that the contralateral breast represents the diseased breast before tumorigenesis [
14], or may represent systemic (inflammatory) effects induced by the tumor [
16]. CPE represents the highest delayed enhancement in healthy fibroglandular tissue. CPE might be affected by hormonal activity, as parenchymal enhancement varies during the menstrual cycle [
17]. The underlying biological reason for the observed association between CPE and survival after endocrine treatment is unknown, but was demonstrated in two independent studies [
14,
15]. Investigating the behavior of CPE during NET might not only provide a tool for the personalization of NET but could also provide insights into the underlying biological mechanisms.
Pathologic complete response (pCR) after neoadjuvant treatment is a controversial surrogate endpoint of prognosis in ER+/HER2− breast cancer [
18,
19]. pCR is poorly associated with prognosis in ER+/HER2−, and rate of pCR is low in both NAC and NET (about 7.5% and < 10% respectively) [
18‐
20]. To understand how tumor response after NET is related to prognosis, the preoperative endocrine prognostic index (PEPI) was developed [
21]. PEPI is derived from the surgical excision specimen after NET and is based on pT- and pN-stage, Ki67 index, and ER-status. PEPI stratifies patients in three groups with distinct prognoses: PEPI-1 has the most favorable prognosis, whereas PEPI-3 has the poorest prognosis. PEPI can be used to personalize treatment after NET: patients with PEPI-1 have such a favorable prognosis that adjuvant endocrine monotherapy could suffice, whereas appropriate adjuvant treatment should be considered for PEPI-2 and PEPI-3 patients [
21,
22]. PEPI was validated in the IMPACT trial [
21] and the ACOSOG Z1031 trial [
22].
In this study, we present a retrospective observational cohort study of patients with invasive unilateral ER+/HER2− breast cancer treated with NET. The aim was to determine whether pretreatment CPE or changes in CPE during treatment are associated with prognosis (on the basis of PEPI) after NET.
Discussion
In this retrospective single-center observational cohort study, we showed that pretreatment CPE, a quantitative measure of relative late parenchymal enhancement on MRI, and change in CPE during NET were associated with PEPI-group in the post-treatment surgical specimen: a high pretreatment CPE and a decrease in CPE during NET were associated with a higher PEPI-group (poor prognosis).
Research regarding response imaging during NET is limited. Our results are in agreement with the findings of Hilal et al, who found that high pretreatment BPE, classified according to the BI-RADS lexicon, was associated with non-responders after NET [
13]. In the NAC setting, BPE has been linked to several treatment outcomes [
6]: a high BPE before start of NAC was associated with worse recurrence-free survival (RFS) [
30], while a decrease in BPE during NAC was associated with pCR [
31‐
33].
While a decrease in parenchymal enhancement on MRI during NAC is reported to be associated with pCR, in our study, a decrease in CPE was associated with an unfavorable prognosis after NET. Perhaps one would expect parenchymal enhancement to decrease in patients with effective endocrine treatment due to depressed hormonal activity, as BPE is increased during physiological hormonal activity [
34] or during hormone replacement therapy [
35,
36]. BPE was associated with increased microvessel density [
37]: persistent or increased parenchymal enhancement during NET might reflect increased perfusion and better drug delivery. CPE was not associated with percent staining of ER or progesterone receptor on immunohistochemistry, nor with genomic ER-pathway activity in the tumor [
15,
38]. A different explanation for these opposing effects between the different neoadjuvant therapies might be due to different immunohistochemical subtypes of breast cancer. It is known that breast cancer is a heterogeneous disease with different prognoses, treatment, and imaging characteristics, especially in ER+/HER2− breast cancer [
39]. Differences in tumor biology and treatment mechanisms (cytotoxic chemotherapy vs antiproliferative endocrine therapy) could have had different systemic effects on the fibroglandular tissue, which could lead to differences in the behavior of parenchymal enhancement. Without a clear understanding of the biological basis of parenchymal enhancement and treatment efficacy, and the (dis)similarity between BPE and CPE, it is difficult to provide an explanation for these opposing findings between NAC and NET.
Although the changes in parenchymal enhancement are counterintuitive in the context of chemotherapy, a high CPE was previously associated with a favorable prognosis after adjuvant endocrine therapy [
14,
15]. In our study, an increase of CPE is associated with a favorable prognosis after NET. In that sense, a high CPE after NET was also associated with a favorable prognosis (PEPI-1).
Remarkably, high pretreatment CPE was related to a poor prognosis (PEPI-3) at final pathology, whereas high CPE was previously shown to be related with improved overall and invasive disease-free survival after adjuvant endocrine therapy [
14,
15]. The exact reason for this finding is unknown, although the difference might simply be due to different endpoints. Additionally, pretreatment CPE alone was not useful in distinguishing between the different PEPI-groups at final pathology.
PEPI was used as a surrogate endpoint of prognosis because pCR and change in tumor size are poorly associated with prognosis in ER+/HER2− breast cancer [
18,
19]. Specifically for ER+/HER2− breast cancer, change in tumor size during NAC is a poor predictor of response and a poorly reproducible surrogate endpoint of survival [
40,
41]. Change in tumor size during NAC yielded a non-significant AUC for the prediction of pCR in one study [
42] and was not associated with survival after NAC in another study [
39]. Additionally, clinical response during NET was not associated with survival [
21]. In our study, change in CPE during NET was associated with prognosis (on the basis of PEPI) and performed similarly to other mid-treatment predictors of tumor response in ER+/HER2− breast cancer after NAC: change in CPE discriminated PEPI with an AUC of 0.77, and change in apparent diffusion coefficient discriminated pCR with an AUC of 0.76 [
11]. To our knowledge, CPE is the first quantitative imaging feature that was observed to be associated with prognosis at final pathology after NET.
Our results support the hypothesis that the healthy breast contains information about endocrine treatment success for patients with unilateral ER+/HER2− breast cancer. CPE was reported to stratify patients within high-risk groups based on genomic assays (70-gene signature and 21-gene recurrence score) [
43]. These results suggest that CPE contains prognostic information independent of these genomic assays and could potentially be used to further personalize treatment.
The main limitation of this study is its relatively small size, which is reflected in the wide CIs of the estimates, and limits the power to detect small effects. To account for the small population size, we took full advantage of the statistical efficiency of a linear mixed model for the repeated measurements analysis, and the association between CPE and prognosis after NET was strong enough to reach the a priori defined significance threshold of < .05. The association between survival and CPE was previously shown to reproduce between different MRI vendors and small differences in imaging parameters [
15]. For nine patients, the pretreatment MRI was performed in the referring hospital on a different MRI vendor which could have led to variability in the CPE measurements. However, the flip angle and repetition time, being the imaging parameters with the most influence on intensity [
44], were similar over the entire cohort. Despite the differences in parameters, CPE was observed to be significantly associated with PEPI. Additionally, exclusion of the nine referred patients did not influence the results. Although there is currently no consensus on the optimal duration of NET, recent clinical studies treat patients for up to 24 weeks (about 6 months) [
20], as there is evidence that maximum tumor response may be reached after 6 to 7 months of NET [
45]. In this study, patients received NET for a median duration of 7.2 months. The findings should be validated in a larger cohort to assess the discriminatory ability of CPE during NET. Lastly, an important step for the implementation of quantitative measurements of parenchymal enhancement is the development of software for use in clinical practice.
In conclusion, pretreatment and changes in contralateral parenchymal enhancement during neoadjuvant endocrine treatment were associated with PEPI-group in unilateral ER+/HER2− breast cancer patients: a high pretreatment CPE and a decrease in CPE during NET were associated with a poor prognosis after NET on the basis of PEPI. Future research will focus on the potential of CPE to assess endocrine treatment effectiveness.
Publisher’s note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.