Clinical Utility of Neoadjuvant MRI in Early-Stage Breast Cancer Patients

With the increasing use of neoadjuvant treatment (NAT) for patients with early-stage breast cancer (ESBC), adequate clinical staging is essential to inform treatment. While the use of MRI with NAT has been proposed to help with accuracy of pre-treatment clinical staging, its impact in clinical practice remains controversial. A prospective institutional database of patients with ESBC treated with NAT between May 2012 and December 2020 was analyzed in order to compare the management of patients who received an MRI prior to NAT to those who did not. The indications for MRI and correlation of MRI ndings to conventional breast imaging were evaluated. The impact of MRI on management was compared between the MRI and non-MRI groups.


Results
A total of 530 patients met inclusion criteria. Of these, 186 (35.1%) had an MRI and 344 (64.9%) did not.
The most frequent indication for MRI was the determination of disease extent (54.5%). Patients who had an MRI prior to neoadjuvant treatment were signi cantly more likely to be younger (47 years versus 57 years; p<0.001) and have multifocal disease (32.3% versus 22.1%; p<0.05). When compared to conventional imaging, MRI reported a greater extent of disease in the breast (37.6%), more nodal involvement (18.8%) and multifocal disease (15.1%). Additional diagnostic interventions were advised in 52.2% of patients. who underwent MRI. Rates of mastectomies were greater in the MRI group (80.0% versus 58.9%; p<0.05) in addition to more axillary dissections (28.0% versus 17.4%; p<0.01). Rates of locoregional recurrences were low in both groups, with similar disease-free survival outcomes at 5 years.
Conclusions MRI identi ed signi cantly more disease in contrast to conventional imaging and lead to more aggressive surgical management. Prospective studies evaluating the role of neoadjuvant MRI and its impact on long term outcomes are needed.

Background
Neoadjuvant chemotherapy is being increasingly utilized in the management of patients with early stage breast cancer [6]. While the primary objective of neoadjuvant treatment (NAT) is to achieve tumor downstaging and ideally a pathologic complete response (pCR) [3], in many cases offering neoadjuvant treatment can improve or increase the surgical options for a patient, especially if mastectomy is the only approach initially recommended. Adequate baseline clinical staging is of utmost importance to identify patients suitable for conversion to breast conserving surgery (BCS), to increase the chances of achieving clear margins and to inform locoregional and systemic treatment [4].
One of the modalities proposed to improve baseline clinical staging includes breast MRI, which has been extensively studied and demonstrates superior sensitivity for invasive and in situ cancer detection when compared with physical examination, ultrasonography, and mammography [19], as well as a more accurate depiction of disease extent in breast cancer patients [5,13,19]. MRI often detects additional areas suspicious for cancer foci [9], which, if not included in the surgical plan, increases the risk of involved margins at the time of surgery [7]. It is unclear, however, if the detection of additional disease leads to improved survival outcomes following neoadjuvant treatment [13,19]. While MRI is highly sensitive, it is not as speci c, and its high rate of false-positive ndings can lead to higher rates of additional biopsies and mastectomies [17].
Many studies have evaluated the role of preoperative MRI, with the hypothesis that its improved sensitivity may lead to more precise surgery. While some studies did in fact report lower reoperation rates in patients who received a pre-operative MRI [20], in a meta-analysis of 12 studies by Houssami et al., the impact of pre-operative MRI on surgical planning showed that approximately 11.1% of surgical alterations from the initial plan were based on true positive ndings and approximately 5.5% were due to false-positive ndings [9]. The use of pre-operative MRI therefore remains controversial, and there are currently no universally accepted guidelines for MRI use in the neoadjuvant setting [8]. Additionally, there are varying opinions regarding the potential use of neoadjuvant MRI, even amongst multidisciplinary expert panels (Simmons et al., 2015). In order to better understand the clinical utility of neoadjuvant MRI, we sought to understand the current use of MRI in the neoadjuvant setting in clinical practice, its correlation to conventional imaging ndings and its impact on patient management.

Study design
A prospective institutional database at the British Columbia Cancer Centre in Vancouver, Canada, was interrogated to identify patients treated with neoadjuvant therapy and subsequent surgical resection between January 1, 2012 and September 30, 2020. Patients with unresectable or advanced disease at the time of presentation were excluded, as were patients who did not go on to have surgical resection. No additional exclusion criteria were applied. Database accuracy was ensured with quarterly quality assessments with in-depth audits. This project was conducted in accordance with the University of British Columbia -BC Cancer Research Ethics Board.

Data collection and outcomes
The charts of eligible patients were examined, and data was obtained from the database. The patient's demographic, clinical and pathological characteristics, imaging results, surgical outcomes and survival data were retrieved. Patients were then divided into two groups: those who received an MRI in the neoadjuvant setting, labeled the "MRI" group and those who did not, labeled the "non-MRI" group. For the MRI group, indications for MRI use, MRI ndings and additional interventions following MRI were collected. MRI ndings were compared to routine imaging modalities (mammography, tomosynthesis, ultrasonography) to determine if there was a discrepancy of disease extent in the breast, de ned as a greater or equal to 20% difference in size based on the largest measurement in one dimension or a change in T stage based on the American Joint Committee on Cancer (AJCC) 8th Edition staging system TNM classi cation; a discrepancy of disease extent in the lymph nodes, de ned as greater or equal to a 20% difference in size of nodal involvement or changes to nodal status (N1, N2, N3); differences in number of foci in the breast (multifocality), or multicentricity; discovery of contralateral ndings; and/or additional unexpected systemic ndings. Baseline surgical plans and nal surgical outcomes were noted and changes were classi ed as "less surgery" if patients were downgraded from mastectomy to BCS or axillary dissection to sentinel node biopsy or "more surgery" if the surgical plan was changed to a mastectomy instead of BCS or axillary dissection instead of sentinel node biopsy. Intervals between time to consultation and treatment were also of interest for the MRI and non-MRI groups, as well as instances of local, regional and distant recurrences.

Patient characteristics
A total of 530 patients who received neoadjuvant therapy between January 2012 and September 2020 were eligible for this study. Among them, 186 patients had a breast MRI prior to NAT and 344 patients did not ( Table 1). The median age at diagnosis was 47 years in the MRI group and 57 in the non-MRI group (p < 0.001). Patients with BI-RADS breast density C and D density was 32.3% in the MRI group and 22.6% in the non-MRI group (p=0.017) and was not reported in more than half of included patients. A total of 32.3% versus 22.1% had multifocal disease in the MRI and non-MRI groups (p=0.012). No signi cant differences were noted between the grade, biomarker status, histology, pre-treatment stage or lymphovascular invasion (LVI). Rates of neoadjuvant chemotherapy use were 90.9% in the MRI group and

Discussion
Our study reveals that the frequency of neoadjuvant MRI was greater in younger patients with dense breast tissue and multifocal disease. In the population of patients with known early-stage breast cancer, MRI was most frequently indicated to better delineate disease extent. Additionally, our study is the rst to report the clinical utility of neoadjuvant MRI, speci cally with regard to how MRI aided conventional imaging in further evaluation of tumor multifocality, multicentricity, nodal involvement and contralateral breast ndings. Ultimately, the identi cation of these ndings by MRI resulted in frequent additional diagnostic interventions and more frequent alterations to surgical management plans for the breast and axilla, with higher rates of mastectomies and axillary dissections.
The fact that MRI use was more frequently used in a younger patient population is consistent with previously published data. Traditionally, MRI has been proposed as an additional screening test for younger women at high risk of breast cancer or those with a germline mutation in the BRCA1 or BRCA2 genes in whom mammography alone has lower sensitivity [12]. Additionally, MRI is more frequently recommended for younger women due to increased breast density in this patient population [26]. Given that BI-RADS has only recently been mandated for mammographic reports in BC, this information was unavailable in more than half of the patients included in this study, limiting the generalizability of these ndings. Nevertheless, a higher density of breast tissue has been generally accepted as an indication for MRI [22], partially due to better fat suppression and a higher resolution [14]. This is particularly relevant prior to NAT where disease extent and accurate staging are utilized for assessment of treatment response and locoregional treatment planning [24]. Additionally, while there are no clear guidelines on who should receive an MRI in neoadjuvant settings, the most frequent reason for an MRI was to better delineate disease extent, which is consistent with earlier reports [4-5, 9, 17]. Although not adopted as a standardof-care, MRI prior to NAT is being increasingly used for the evaluation disease extent and the prediction of NAT response [24].
Our analyses demonstrated that discrepancies between clinical and radiological staging were frequent and noted in 73% of patients. To our knowledge, no prior study has directly correlated the ndings between other conventional imaging, clinical extent of disease and MRI in the pre-operative setting. However, in keeping with these ndings, DeMartini and colleagues summarized the results of multiple studies that collectively evaluated MRI in approximately 1,500 cases of newly diagnosed breast cancer. All found that MRI identi ed additional ipsilateral malignancy with a reported frequency that ranged from 10%-34% [5]. In another meta-analysis, MRI reported additional contralateral cancers in 4% of subjects [2]. One of the signi cant issues associated with breast MRI is the high rate of false-positive ndings which overestimates the extent of the disease and can result in additional investigations, unnecessary biopsies, and more extensive surgery [1,10]. Our study found that just over half of the patients who underwent MRI were subjected to additional investigations. Rates of false-positive MRI ndings were not evaluated as part of this study; however, the existing literature reports the false-positive rates ranging from 29-80% [10].
With regard to changes in surgical management, surgical plans for the breast and axilla were altered in nearly 27% and 28% of cases, respectively. In line with these ndings, it's been previously reported that 13-26% of women with invasive breast cancer have a change in surgical management based on preoperative MRI evaluation and 7-17% alter their surgical treatment from lumpectomy to mastectomy [10]. Chen and colleagues also conducted a systematic analysis to investigate how pre-and post-NAC MRI ndings affect the surgeon's recommendation [3]. They concluded that tumor size, multifocality, and disease extent on pre-treatment MRI indeed affected the initial surgical recommendation (mastectomy versus lumpectomy). However, it remains unclear that additional surgery results in lower rates of local, regional and distant relapses. While our study was not powered to detect these detect small differences, no signi cant differences were noted with the exploratory analyses.
Amongst the limitations, it should be noted that this was a non-randomized cohort study with a retrospective analysis of a prospective database. Additional information that was not accounted for could have impacted the frequency of changes in surgical management. Patient speci c factors, such as increased anxiety [11], more extensive MRI ndings than anticipated and patients seeking reconstruction for cosmesis or balancing surgery [20], may have contributed to the higher rates of change in surgical planning. We did not directly investigate if additional MRI ndings and patterns of MRI use added any bene t to long-term outcomes in the studied patient population given short follow-up data, but this will be examined when longer follow-up has been established. Furthermore, a number of questions remain unanswered, including the impact of neoadjuvant MRI on the frequency of positive margins, cancer recurrence and mortality, given that data to date have not clearly established the bene ts of MRI on rates of reoperation, disease recurrence, or survival [19]. To this effect, a prospectively designed trial is underway at BC Cancer in an attempt to shed light of some of these unanswered questions (NCT03790813). The results of the current study can be used to design future prospective trials to better capture the extent to which MRI ndings impact intended treatment, to de ne the subpopulation of patients with the highest bene t from MRI, and to establish appropriate guidelines for recommending MRI prior to NAT.

Conclusions
Overall, younger patients with known multifocal disease were more likely to get an MRI referral prior to NAT. The use of neoadjuvant MRI was associated with more frequent diagnostic interventions and more extensive surgery when compared to a group without MRI use neoadjuvantly. The impact of MRI on the management of neoadjuvant patients in a prospective setting are needed, in addition to long-term followup to evaluate the impact of MRI on rates of local, regional and distant relapses and patient reported outcomes Statements And Declarations Funding: No funding was received for this study.
Con ict of Interest: N. LeVasseur reports receiving honoraria for participation in advisory boards from Knight, Lilly, Novartis, P zer, Roche, Seagen, TerSera and research funds from Abbvie and Exact Sciences. CS has received research funding from P zer and Amgen, and has received fees as an advisory board member from P zer, Amgen, Novartis, Roche, Merck, Lilly, Sandoz, and Mylan.
Ethical approval: All procedures performed in studies involving human participants were in accordance with the ethical standards of the institutional and/or national research committee and with the 1964 Helsinki declaration and its later amendments or comparable ethical standards.
Informed consent: Informed consent was not required as data was anonymized and aggregated.

Declarations
Con ict of Interest: N. LeVasseur reports receiving honoraria for participation in advisory boards from Knight, Lilly, Novartis, P zer, Roche, Seagen, TerSera and research funds from Abbvie and Exact Sciences. CS has received research funding from P zer and Amgen, and has received fees as an advisory board member from P zer, Amgen, Novartis, Roche, Merck, Lilly, Sandoz, and Mylan.
Ethical approval: All procedures performed in studies involving human participants were in accordance with the ethical standards of the institutional and/or national research committee and with the 1964 Helsinki declaration and its later amendments or comparable ethical standards.
Informed consent: Informed consent was not required as data was anonymized and aggregated. Change to the surgical management plan in the MRI and non-MRI groups, divided by breast and nodal surgical management plan