Breast magnetic resonance imaging: are those who need it getting it?

Background Indications for breast magnetic resonance imaging ( mri ), a very sensitive but less-specific tool for breast investigation, remain controversial, and accessibility is limited. The purposes of our study were to determine the proportion of breast mri exams performed for various clinical indications, to assess the wait times for breast mri , and to create a list of evidence-based indications for breast mri . The indications for breast mri exams performed in September 2013 at our academic centre were audited. A multidisciplinary meeting held in May 2014 established a list of evidence-based indications for breast mri , after which, in September 2014 and 2015, breast mri exams were re-audited for clinical indications, and pending requests were calculated. In September 2013, surveillance of women with a prior diagnosis of breast cancer represented 21% of breast mri exams (24 of 113), with preoperative staging representing 18% of exams (20 of 113) and high-risk screening representing 12% (13 of 113). Of pending mri requests, 230 were within the recommended delay period, and 457 exceeded the recommended delay. After elaboration of evidence-based guidelines, repeat audits in September 2014 and September 2015 showed that mri performed for women with a prior breast cancer diagnosis represented 23% (33 of 141) and 7% (10 of 143) of exams respectively, with preoperative staging having declined to 9% (13 of 141) and 11% (16 of 143) of exams, and high-risk screening having increased to 36% (51 of 141) and 46% (66 of 143) of exams. Overall, wait times were improved for all breast mri indications. improved wait times.


BACKGROUND
The Canada Health Act governs every health system in Canada and ensures universal access to health care for all Canadian residents. However, accessibility, another principle of the Canada Health Act, can become compromised. Lately, the federal government and the Wait Time Alliance have been particularly concerned about accessibility to cancer care and diagnostic imaging 1,2 . In the 2015 Wait Time Alliance report 3 , wait times for magnetic resonance imaging (mri) were on the rise in Ontario and Manitoba and exceeded recommended benchmarks for more than 50% of the population in Alberta and Prince Edward Island. (Data for the other 6 provinces are unknown because sufficient data for the assessment were not provided.) Despite the limited availability of mri, indications for mri exams are increasing exponentially. In breast imaging, mri is indicated for the detection and management of breast cancer, offering a sensitivity for cancer detection superior to that with other imaging modalities-ranging between 71% and 100% [4][5][6][7][8][9][10][11] . Between 2000 and 2009, demand for breast mri soared, increasing by a factor of more than 20 12 . However, breast mri must be used with discernment given its limited specificity: very low in the earliest reports, but now generally accepted to be in the 60%-80% range [4][5][6][7][8][9][10][11] . Moreover, evidence is increasing that preoperative staging mri might be detrimental to patients, unnecessarily lengthening their investigation time because of the detection of additional nonspecific findings 13,14 and increasing the mastectomy rate with little reduction in re-excision e206 Current Oncology, Vol. 24 and negative margins rates 15 . In that context, there has been a call for moderation with respect to routine orders for staging breast mri because no randomized controlled trial has yet demonstrated improved survival for women who undergo breast mri 16 .
Although several publications have analyzed breast mri and its use, application of the resulting knowledge does not necessarily translate in day-to-day practice. In 2014, the Breast Cancer Surveillance Consortium reported that only 25% of screening mri exams were performed for appropriate indications as defined by the American Cancer Society guidelines 17 . Conversely, fewer than 5% of patients at high risk of breast cancer received breast mri 17 . In this era in which wait lists for mri exams and health costs are constantly growing, improved patient selection is urgently needed 18 .
In the fall of 2013, given the increasing controversy about the use of breast mri, particularly as a preoperative staging tool, and given the lengthy wait lists for breast mri at our institution, we audited our breast mri use. Using a problem-solving model to bridge the gap between academic knowledge and clinical practice, we strove to improve quality of care by a method that could be applied at other institutions as well. The objectives of our project were to n determine how breast mri was being used at our academic centre, and particularly the proportions of breast mri exams performed for various clinical indications; n assess the wait times for breast mri; n review the literature about breast mri indications with our multidisciplinary panel of breast specialists; and n create a list of evidence-based indications for breast mri.

METHODS
The study used a clinical audit strategy model.

Identify the Problem
The study was performed at the Centre hospitalier de l'Université de Montréal, an academic tertiary-care hospital where the medical team includes 5 fellowship-trained breast imagers and 5 onco-surgeons specializing in breast surgery. Our breast clinic is an investigation site (Centre de référence pour investigation désigné) for the Quebec Breast Cancer Screening Program, and it receives referrals from other institutions in the province of Quebec, mostly in the wider Montreal area. Annually, about 1600 breast mri exams are performed.
To evaluate the situation at our institution and to determine the improvements required, all breast mri exams performed in September 2013 were retrospectively audited. All breast mri exams completed during that 1-month period were retrieved and collected from our picture archiving and communication system, and the indication for each exam was recorded. We then classified indications into these categories: preoperative staging; screening of high-risk women; follow-up of mri-identified anomalies; investigations for women with a personal history of breast cancer; surveillance of women with a history of lobular carcinoma in situ, atypical ductal hyperplasia, or atypical lobular hyperplasia; troubleshooting (after any one or a combination of inconclusive mammography, ultrasonography, or biopsy); investigation of nipple discharge; and "other," including evaluation of breast implants, search for a primary cancer in the presence of malignant adenopathy, and positive margins after breast surgery.

Evaluate Outcomes
In September 2013, 16 hours were dedicated every week for breast mri exams, allowing for the evaluation of approximately 26 patients having a standard 30-minute diagnostic exam and 2 patients hav ing mri-guided breast biopsies (90 minutes allotted per procedure). At the end of September 2013, we analyzed pending breast mri requests as tracked by our administrative system. We classified every request as being within wait times recommended by the Canadian Association of Radiologists 19 and set out in Quebec governmental guidelines (provincial recommendations are for imaging within 90 days after reception of a first request and within 30 days for follow-up examinations); exceeding wait times by less than 6 months; exceeding wait times by 6-12 months; and exceeding wait times by more than 12 months. We also projected the number of days before the next non-urgent breast mri would be available.

Adapt Knowledge to Local Context
To verify that our use of breast mri complied with the latest published recommendations and to improve our service to patients, we organized a multidisciplinary halfday scientific session for the breast team, which was held in May 2014. Of 28 possible participants, 21 attended the meeting, representing all invited fields: 5 radiologists, 4 surgical oncologists, 2 medical oncologists, 2 radiation oncologists, 2 geneticists, 3 breast physicians, and 3 radiology residents. An informal review of the literature about mri use for breast cancer staging, high-risk screening, and other indications was presented. There was then a discussion until multidisciplinary consensus was reached for a scientifically acceptable compromise between published evidence and mri availability at our institution.
The ordering radiology form for breast mri was subsequently modified to reflect the actualized accepted indications for mri, and steps were taken to make the document easily accessible to all clinicians in their daily clinical work.

Monitor Knowledge Use
In September 2014 and September 2015, we re-audited all breast mri exams performed during the 1-month period and collected data for each exam's indication. We calculated the number of pending breast mri requests at the end of both months and categorized them in the same manner as before the consensus.

Pre-consensus Evaluation
Our initial institutional review from September 2013 showed that 27% of breast mri exams (30 of 113) were performed for follow-up of an mri-identified lesion. Women with a personal history of breast cancer or needing preoperative staging constituted the 2nd and 3rd most common indications for women to undergo breast mri (21%, 24 of 113, and 18%, 20 of 113, respectively). Screening of high-risk women came in 4th place and represented only 12% of all breast mri studies (13 of 113, Table i).
Review of 687 pending breast mri requests at the end of September 2013 revealed that 67% exceeded recom mended wait times, with 294 (43%) exceeding recommended wait times by less than 6 months, 93 (14%) exceeding them by 6-12 months, and 70 (10%) exceeding them by more than 12 months (Table i). Actuarial calculations estimated that the next elective breast mri slot would be available in 320 days.

Consensus
Review of the September 2013 breakdown of breast mri examinations according to indications and compilation of the overall wait times for such exams confirmed that improvement was in order.
Discussion by the 21 members of the breast team who participated in the May 2014 multidisciplinary meeting achieved a consensus regarding the use of breast mri. Evidence-based accepted indications for breast mri fell into two major categories: preoperative staging in specific clinical situations and high-risk screening (Table ii). For the latter category, given that several requests for women with a strong family history of breast cancer were awaiting formal genetic evaluation, screening mri was also accepted for women fulfilling the high-risk criteria for familial breast cancer based on the (Australian) National Breast and Ovarian Cancer Centre classification 62 . Discussions also covered when to start and stop mri screening. Table ii presents the full list of evidence-based indications for breast mri applied at the Centre hospitalier de l'Université de Montréal.
Although evidence from the literature formed the base of our institutional guidelines, the reality of our particular practice was taken into consideration during the multidisciplinary meeting. Our genetics experts suggested referring women for a genetic evaluation when they fulfilled any of these criteria: less than 50 years of age with breast cancer; having a first-degree relative diagnosed with breast cancer before the age of 50; having multicentric cancer; and being diagnosed with triple-negative cancer at a young age. However, because the wait time for a genetic consultation can be relatively long, potentially delaying breast mri screening for patients who might be at high risk, the consensus of the breast team was to allow patients considered high risk according to clinical criteria who were awaiting genetics evaluation to immediately benefit from annual breast mri evaluation. The high-risk clinical criteria were based on the Australian classification because that classification is easily applicable 62 .
The decision to use mri to evaluate patients with triplenegative breast cancer was another point of discussion during the meeting. Although results from the alliance A011104/acrin 6694 study on the subject are still pending, radiology experts determined they had the capacity to honour requests for this indication given that patients with triple-negative cancer represent only a small proportion of our patient population. It was determined that all requests for breast mri that deviated from the consensus would no longer be routinely accepted at our institution, including specific situations in which, although the literature might show some evidence of benefit, the benefit is not sufficient to warrant routine use of mri (Table iii). For example, women whose family history suggests a moderate risk for breast cancer are not candidates for mri screening before formal genetic evaluation, nor are women with no risk factor other than a personal history of breast cancer. Similarly, our literature review did not support continued mri screening for high-risk women who had undergone bilateral mastectomy whether prophylactic or therapeutic, nor did it support continued mri surveillance of high-risk women after 69 years of age.
The conclusions from our exchanges were summarized in a written document that was circulated to all 28  15,[20][21][22][23][24][25][26][27][28][29][30][31][32][33][34][35][36][37][38][39] Recommend or consider MRI in these situations: members of the breast team for final approval (the 21 who had attended the meeting and the few who had not been able to attend). The final document was then sent by e-mail to all physicians involved in breast care at our institution, and a printout of the consensus was made easily accessible to clinicians and radiologists at their clinics and workstations. It was decided that all breast mri requests received before the consensus meeting would be honoured, without questioning the indication. However, for every new mri request received after May 2014 that did not meet the new consensus, a letter informing the referring physician of the rejected request was sent together with the summarized recommendations from the consensus meeting. It was clearly stated that the consensus guidelines aimed to standardize our institution's use of breast mri, but did not replace the physician's clinical judgment. They were therefore invited to contact the radiologist directly to discuss the case in further detail, if desired.
The ordering radiolog y form for breast mri was modified to reflect the consensus guidelines (Table iv). Although our initial prescription form included subcategories of preoperative imaging (re-excision for positive margins and search for primary breast cancer), all subcategories were merged under the single category of preoperative staging. To improve access to screening mri for women with a genetic susceptibility, the ordering form was also modified to subdivide all screening mri requests into two groups: women with a proven genetic susceptibility to breast cancer, such as carriers of a BRCA mutation and their first-degree relatives; and all other women with sufficiently strong risk factors to justify mri screening, but with no formal genetic evaluation. Specific reference to our consensus document was added to the new ordering form to reaffirm our determination to respect the institutional consensus.
In addition to detailing the indications for breast mri, our new ordering form also provided guidelines for exam priority: preoperative staging and evaluation of chemotherapy response are first-priority evaluations (P1); follow-up for probably benign lesions (Breast Imaging Reporting and Data System class 3), problem-solving after mammography and ultrasonography, and investigation of nipple discharge are moderate-priority evaluations (P2); and screening of high-risk women with a genetic susceptibility receive the next priority rating (P3). All other screening requests and implant evaluations are last-priority examinations (P4).

Post-consensus Evaluation
After implementation of the consensus guidelines, the proportion of breast mri exams performed for each indication changed markedly (Table i, Figure 1). One year after applying the new guidelines, at the first re-audit, screening mri rose from the 4th most common indication to the most common indication. That trend continued in the subsequent year, with the proportion of screening mri exams reaching 46% of all breast mri exams in September 2015, compared with 36% 1 year earlier and only 12% in 2013.
Conversely, the proportion of breast mri exams performed for preoperative staging was cut in half 1 year after the consensus, dropping to 9% (2014) from 18% (2013). In September 2015, the proportion of preoperative breast mri exams remained relatively stable at 11%.

Women with dense breasts
Screening MRI will not be offered to women at moderate risk based on family history, before a formal calculation of their lifetime risk n These criteria serve as reference to define moderate risk: n 1 or 2 first-degree relatives with breast cancer before 50 years of age (same side of the family) and without high-risk factors n 2 first-or second-degree relatives with breast or ovarian cancer (same side of the family) without high-risk factors  The 2nd most common indication for breast mri in 2013 was surveillance of women previously treated for breast cancer, but consensus guidelines established that breast mri would no longer be performed solely for that risk factor. Because of an important backlog of women awaiting mri for that indication, the effect of the recommendation was noticed only 2 years after implementation, when the proportion of breast mri exams for that indication dropped to 7% from the 21% observed in September 2013.
The number of scheduling slots available for breast mri increased slightly during the study period, to 19 hours from 16 hours weekly, thus allowing 143 breast mri exams to be performed in September 2015 compared with the 113 performed in September 2013. That measure, combined with an improved selection of women undergoing breast mri because of clear institutional consensus guidelines having been established, resulted in globally improved wait times for all breast mri exams (Table i): the 320-day delay for the next available mri slot in September 2013 was reduced to 176 days in September 2015. Breast mri requests awaiting scheduling and requests exceeding recommended wait times by less than 6 months also improved in the year after the consensus guidelines were established. Because of the accumulated backlog, the effect of the new guidelines on requests exceeding recommended wait times by more than 6 months was observed only during the 2nd year of follow-up.

DISCUSSION
With the introduction of mri into the detection and management of breast cancer, multiple associations, including the Canadian Association of Radiologists, the American Cancer Society, and the American College of Radiology, have published guidelines for the appropriate use of breast mri 38,63,65 . Taking into account more recently published data, those recommendations can be incomplete, contradictory, or outdated for some situations. Moreover, given that policies and distribution of medical resources can be quite different from country to country, recommendations by foreign medical associations should be carefully considered before they are adopted into local practice. The Choosing Wisely Canada campaign-based on the concept that good care for patients is achieved by prescribing the right tests to the right patients 64,65 -is crucial considering the high cost and limited availability of breast mri. With its excellent sensitivity but lower specificity, mri leads to additional tests and investigations, increased costs, and increased wait times for patients, which might lead to more harm than good.
Acknowledging the challenges associated with breast mri, the objectives of our clinical audit were successfully reached. We improved breast mri use at our institution by accepting only evidence-based indications, and as a result, availability for this imaging modality for all women needing it was improved. By better defining high-risk patients in need of screening, clearly listing situations in which preoperative imaging could be beneficial, and eliminating mri use where scientific evidence of benefit was not established, increased priority was given to high-risk screening studies, with improved scheduling of specific preoperative mri evaluations.
Our expert consensus on breast mri indications had a positive effect on wait times within the initial year of application. However, because of the high volume of pending mri requests, with delays for non-urgent mri exams of close to a year (320 days) at the project's onset, the effect of the new guidelines was not fully evident until the 2-year postimplementation cycle.
As soon as the consensus guidelines were introduced, compliance was high, reflecting the elevated participation rate in our multidisciplinary meeting, with at least 1 representative of each subspecialty being present. By significantly contributing to the project, all members of the breast team felt motivated to adhere to the group's consensus and to take pride in its success. Enough time was provided in the multidisciplinary consensus meeting to allow for all concerns about possible sequelae of the upcoming changes to be voiced, thus ensuring that, as soon as the consensus guidelines were introduced, they would be readily applicable for our breast centre and our patient population. Reminding the breast team and referring physicians of the consensus guidelines and making them easily available for quick reference during the daily workflow were also critical to the project's success.
Beyond our hospital, our institutional work was made available to all professionals interested in breast mri and striving to improve their use of this technology. It also formed the basis for a first draft of provincial Quebec guidelines, mandated by the Ministère de la Santé et des Services sociaux and proposed by the Direction générale de cancérologie, with publication in May 2016 66 .

CONCLUSIONS
Through collaboration by all members of the breast team, we actualized a list of breast mri indications at our institution. Application of the resulting guidelines improved the service to women awaiting breast mri, particularly those with a genetic predisposition, for whom annual screening is recommended. We expect to regularly update our expert consensus as new guidelines and studies are published.
Because access to technology is a constant challenge and a preoccupation in our health system and elsewhere in the world, the method we used to tackle the volume of breast mri requests in the presence of limited technology availability could be applicable in other settings. The process of a clinical audit is relatively simple, based on multidisciplinary work and a desire to improve access for all patients.

Addendum
Since the writing of this manuscript, the results of our project have been used as the foundation of a provincial evaluation of breast mri indications under the supervision of the Institut national d'excellence en santé et en services sociaux of Quebec.

CONFLICT OF INTEREST DISCLOSURES
We have read and understood Current Oncology's policy on disclosing conflicts of interest, and we declare that we have none.
AUTHOR AFFILIATIONS *Department of Radiology, † Department of Surgery, and ‡ Department of General Medicine, l'Université de Montréal, Montreal, QC.