High Risk Screening Breast MRI

by Elizabeth A. Morris, M.D.
Associate Professor of Radiology
Director of Breast Imaging research and Breast MRI
Department of Radiology MSKCC

Breast cancer is the most frequently diagnosed malignancy in women in the United States (US) accounting for 30% and it is the second leading cause of death as approximately 30% of all women who develop breast cancer will die of the disease. Having the ability to screen for pre-clinical breast cancer (cancer in the in-situ or early invasive state) with mammography has resulted in an approximately 30% decrease in mortality in the US. It is therefore agreed that early detection of breast cancer (before it has had a chance to metastasize) by mammography is partially responsible for the reduction in mortality. Currently, breast cancer screening is performed by mammography and recommended annually for all women over the age of 40 years in the US. Many randomized controlled trials have demonstrated the efficacy and importance of mammography in saving lives as well as improving treatment options for patients with this disease.

However, mammography is not a perfect test and is limited in certain patient populations and in certain breast densities, particularly the heterogeneously dense and extremely dense parenchymal pattern. It has been well recognized that mammography has an overall false negative rate of up to 20% and this number is higher in younger pre-menopausal average-risk patients with dense breasts, approaching 50%. These limitations have spurned interest in alternate modalities, above and beyond mammography, to detect early breast cancer, including breast magnetic resonance imaging (MRI).

Why consider MRI?

Although there is fairly clear consensus that mammography saves lives in the general population, there is also frustration and anxiety about the possibility of a false negative mammogram, particularly in the young, high-risk population where breast density may obscure a mass. Investigation into improvements in mammographic technique has and is currently being explored to improve sensitivity of detection. Digital mammography has demonstrated a slight improvement in cancer detection in young women with dense breasts and possibly should be used in this patient population.

Interest in MRI for breast cancer screening developed as breast MRI has a very high sensitivity for the detection of invasive carcinoma; higher than mammography. As MRI relies on the enhancement of cancers after contrast injection, the issue of breast density is not as problematic as in mammography. In numerous screening studies in different patient populations and with different techniques, MRI can detect most but not all invasive cancers, with a higher sensitivity than mammography. One issue for screening however is the ability to detect DCIS and the published MRI screening studies are not as uniform in the ability to detect DCIS.
Defining the high risk population

Although a woman’s average lifetime risk for developing breast cancer in the Unites States is one in eight, some women are at high risk. Risk factors for breast cancer include genetic mutations of BRCA1 or BRCA2, family history of breast cancer, previous history of breast cancer, previous biopsy of atypia (ductal or lobular), lobular carcinoma in situ (LCIS), or prior mantle radiation for Hodgkin’s disease.

There are three methods to identify high risk patients: genetic testing, family history assessment and review of clinical history. It is well known that women in general over-estimate their individual risk of breast cancer.

Risk stratification for breast cancer is very complicated process and likely beyond the expertise and knowledge of the average radiologist. There are a plethora of risk assessment models that have attempted to easily assess risk – many are designed for specific patient groups and all have their strengths and weaknesses. Computer programs based on these models exist that can calculate life time risk and can be helpful to assess risk but these may under- or over-estimate the actual risk. A genetic counselor who has been trained in the vagaries of the assessment models is the best person to assess the level of risk. Counseling is an important part of the entire process so that test results can be interpreted in the context of the individual patient’s situation. Also, counseling can be helpful for intensified surveillance, such as adding MRI, as the strengths and limitations of the additional screening tests can be discussed. Also the counselor can discuss the results of the genetic testing for either BRCA gene, if performed, with the patient.

Family History

Screening issues for high-risk young women who are documented gene carriers or have a family pedigree very suspicious for hereditary breast cancer are limited. For carriers, the average age at first diagnosis is younger than the general population who develops sporadic breast cancer. Recent reports suggest that 50% of carriers succumb to breast cancer before the age of 50 years. Also apparent is the fact that many patients who are successfully treated initially will develop a second breast cancer (up to 60% of patients). There may even be a younger incidence peak (late 40’s) for BRCA1 patients, much earlier than sporadic breast cancer. Breast cancers developing in mutation carriers tend to be of higher nuclear grade, exhibit more medullary differentiation and exhibit more receptor negativity than sporadic breast cancers, indicating their more aggressive nature.

Management options for high-risk women with genetic mutations have ranged from close mammographic surveillance to prophylactic mastectomy. Bilateral prophylactic mastectomy is a treatment that is usually recommended to the documented patient who is gene positive. The surgery is usually well tolerated by the high-risk patient though has potential drawbacks, including incomplete removal of all breast tissue, loss of nipples and therefore sensation, and psychological issues. An added concern about recommending this treatment option stems from the fact that not all high risk patients are documented carriers thus making it difficult to recommend this potentially deforming surgery to a patient who is considered high risk but does not have documentation. Bilateral salpingo-oophorectomy is also an option and has been shown to decrease risk of breast cancer by 30%. Obviously this procedure reduces the chance of ovarian cancer as well and it best used after the age of child-bearing. Chemo-preventive agents such as Tamoxifen have become available that may reduce the likelihood of breast cancer in women at high-risk. However, these interventions do not completely eliminate the risk of breast cancer death.

Approximately 5% to 10% of all breast cancer is a result of hereditary susceptibility. The prevalence of BRCA mutations is estimated to be 1/500 and 1/1000 in the general population, however, in women of Jewish ancestry, the prevalence may be 1/50. Women who inherit a mutated form of the breast cancer susceptibility gene such as BRCA1 or BRCA2 have a high risk of breast cancer (20% by age 40, 50% by age 50, and 87% by age 70) or ovarian cancer. BRCA1 and BRCA2 account for approximately 50% of the hereditary cancer cases; the remainder probably attributable to yet unidentified genes. The BRCA1 gene is thought to be responsible for the breast/ovarian syndrome. BRCA1 accounts for half of familial breast cancer and 5% to 8% of all breast cancer. BRCA1-associated cancers tend to be highly proliferative, with a high grade, and tend to be receptor-negative. The BRCA1 gene is located on chromosome 17 and is a tumor suppressor gene, possibly important in the growth and regulation of epithelial cells in the breast.

BRCA2 is located on chromosome 13 and accounts for approximately 35% of familial breast cancer, less than BRCA1. It is thought that BRCA2contibutes fewer cases of early-onset breast cancer than does BRCA1. BRCA2 carriers are at risk of developing other cancers: prostate, bladder, pancreatic, and Hodgkin’s disease.

Only 1 – 2% of women have a family history that is suggestive for an autosomal dominant gene despite the fact that the majority of women in the general population have al least one relative with breast cancer. In most of these the family history does not confer an increase risk (ie, the cancer was sporadic) or a low increase risk due to a low penetrance gene. Features that suggest that the breast cancer may be due to a high penetrance gene include two or more close relatives with breast or ovarian cancer, breast cancer occurring before age 50 in a close relative, a family history of both breast and ovarian cancer, one or more relatives with two cancers (breast and ovarian cancer or two independent breast cancers) and male relatives with breast cancer.

Several models can help clinicians assess the likelihood of a BRCA mutation including Gail, Claus, Tyrer-Cusick. These models attempt to assess breast cancer risk based on family history, sometimes in combination with other factors such as prior breast biopsy or reproductive history, however an individual women’s risk may vary with these models. Two additional decision models have been developed to estimate the likelihood of a BRCA mutation called the BRCAPRO and BOADICEA and these models may fare better at predicting an individual’s risk.

Genetic Testing

Genetic testing for BRCA mutation is generally offered to adult members of families with a known BRCA mutation or to women with at least a 10% chance of carrying a mutation based on clinical history or one of the risk assessment models. If a woman from a family with a documented BRCA mutation tests negative then it can be safely assumed that she has no elevated risk for breast cancer. However in a family without a known mutation, failure to find a mutation does not decrease the risk estimate.
Clinical history

Mammographic density has been shown to increase the risk of breast cancer. Most cancers develop in the breast in the dense areas of the mammogram and women with dense breasts have a 4 – 6 times increase risk for developing breast cancer than those with fattier breasts. Also patients with a prior history of breast cancer are at increase risk over the general population for developing a new breast cancer (0.5 – 1% per year).

Lobular carcinoma in situ (LCIS) and atypical lobular hyperplasia (ALH) are associated with an increased risk of breast cancer developing, ranging from 10 – 20% (approximately 0.5 – 1% per year). This translates to a 6 – 10 times increase risk of developing breast cancer compared to the general population. The invasive cancers are usually invasive lobular cancers though can be of ductal histology as well.

Atypical ductal hyperplasia (ADH) is also a high risk lesion leading to a 4 – 5 times increase risk of breast cancer. If the patient has a family history and a history of ADH the risk doubles.

Treatment for Hodgkin’s disease (HD) between the years 1955 and 1995 confers an increased risk due to mantle radiation. Risk increases significantly 15 – 30 years following treatment. More recent treatment uses lower doses of radiation and limits the radiation field therefore patients with HD treated more recently are not at elevated risk for developing breast cancer.

Overview of High Risk MRI Screening Studies

Several studies performed in the United States and in Europe have shown that MRI benefits high risk patients by detecting breast cancer when it is mammographically occult. Most of these studies are comprised of small patients groups; nevertheless, nearly all the studies have found that there is a 2-4% prevalence of cancer detected on MRI that is occult on mammography and occult to physical examination. The majority of these cancers detected with MRI are small (under 2 cm) and are associated with negative axillary nodes. Negative nodal status and small cancer size are associated with the reduction of mortality seen with mammographic screening. Therefore, it is interesting to reason that if MRI is able to detect small node negative cancers, including DCIS, that there may be an overall mortality benefit for these patients. These end-points however at this time are surrogate end-points. No randomized controlled trial has been mounted to date. At this time, a trial is likely not feasible and may arguably be unethical given the encouraging results published so far in these high risk patient groups.

Patient populations that are considered high risk in the published MRI screening trials have differed. Some trials have performed screening on patients with high suspicion for or proven BRCA 1 & 2 mutations. Others have looked at those with a broader range of risk factors including those with a personal history of breast cancer, those with previous biopsy demonstrating LCIS or ADH, or strong family history. Others have used risk assessment models such as the Gail model to determine high risk and have set entry criteria at varying levels. It appears from the trials that the higher the risk, the more likely an occult cancer will be detected on MRI. Moreover, some trials demonstrated that risk factors can be additive. However, despite these variations in patient selection for MRI screening, there is surprising consistency of the data from multiple sites in different countries with different imaging techniques. It appears that screening MRI is able to detect occult cancer in 2 – 4 % of patients.

While MRI can detect additional cancers the down-side of MRI screening is the added risk (in addition to mammographic screening) of false positive findings as well as call-backs for findings that are not entirely normal but that don’t warrant a biopsy. In general, when screening MRI is performed in high risk populations, 15-25% of patients will require biopsy based on MRI findings and 25 - 65% of patients who undergo biopsy will have breast cancer. Most of the time these biopsies can be performed with percutaneous needle biopsy under ultrasound or MR guidance though in some cases which are technically unfeasible, surgical biopsy may be required with pre-operative MR needle localization. Specificity, while an issue with MRI, may be improved in the future with MR Spectroscopy (MRS). Initial results have in a small trial have demonstrated that interrogation of suspicious lesions undergoing biopsy with MRS resulted in one false positive but no false negatives. Importantly, by using MRS, the biopsy rate could have been reduced more that 50%. Though breast MRS is not currently available clinically, its potential to improve specificity is exciting.

Description of High Risk Screening MRI Studies

Several prospective non-randomized studies have demonstrated the utility of breast MRI. These have been performed in several countries (Netherlands, UK, Canada, Germany and US) with different patient populations (risk status, age etc) and different MRI techniques (low spatial resolution/high temporal resolution versus high spatial resolution/low temporal resolution). All added MRI to mammography and some studies also included ultrasound to detect early breast cancer. All of these studies reported significantly higher detection rates with MRI compared to mammography (or ultrasound). Overall sensitivities range from 71 – 100% for MRI and 16 – 40% for mammography in these high risk populations. Table 1 demonstrates sensitivities and specificities of these trials.

The largest study to date was performed by Kriege et al (2004) in The Netherlands where 1909 unaffected women were screened at 6 centers. The patients ranged from 25 – 70 years and needed an estimated risk of 15% to be included in the study (19% were BRCA proven carriers). After a median follow up of 3 years 50 breast cancers were diagnosed (mostly invasive 88%). 43% of the invasive cancers were 1 cm or less and one third had positive axillary lymph nodes. In this study mammography detected more DCIS than MRI, possibly related to the low spatial resolution techniques used.

Another multi-center trial (22 centers) in the UK called the MARIBS trial by Leach et al. screened 649 patients age 35 – 49 who had a lifetime risk of at least 25% (19% proven BRCA carriers). After a median of 3 rounds, 35 cancers were diagnosed (again mostly invasive 83%). 45% of the cancers were 1 cm or less and 14% had spread to the axilla. 2 interval cancers were detected.

In Canada a single center trial by Warner et al. studied 236 documented BRCA carriers aged 25 – 65 years for up to 3 years. 22 cancers (72% invasive) were detected with 50% of the cancers being 1 cm or less and 13% were node positive. One interval cancer was detected.
Kuhl et al in Germany screened 529 women with a lifetime risk of 20% at a single center for a mean of 5 years. 43 cancers were detected (79% invasive) with positive nodes in 16%. One interval cancer was detected.

Morris et al in the US reported on a single center trial of 367 high risk women (5% proven BRCA carriers) with a lifetime risk of greater than 15% for a single round of screening. 14 cancers were detected (43% invasive) and 14% had positive nodes.

Lehman et al in the US reported on a multi-center trial (13 centers predominantly in the US) of 390 women with a life-time risk of more than 25% for one round of screening. Four cancers were found on MRI an only one on mammography.

Podo et al in Italy studied 105 documented BRCA women at 9 sites for a single round of screening. 8 cancers were documented – all identified on MRI and one identified on mammography.
National Guidelines

In 2006, the National Institute for Health and Clinical Excellence (NICE), the independent organization responsible for providing national guidance on the promotion of good health and the National Collaborating Centre for Primary Care in the United Kingdom published an update of its familial breast cancer screening guideline, recommending yearly MRI screening for some women between the ages of 20 and 49 if they have a high risk of breast cancer, including women who have one of the high-risk genes. This recommendation was based on the results from the MARIBS study that was mounted in the UK as well as the other multi-center trials performed in Europe.

In the United States, the ACS (American Cancer Society) guideline for the early detection of breast cancer was updated in 2007 and stated that women at increased risk should undergo additional screening with annual breast MRI. The use of breast MRI in the high-risk population is limited to those women with documented BRCA 1 or 2 gene or those women with a family member who is a documented carrier but they themselves are untested; any woman with a greater than 20% lifetime risk (as defined by the BRCAPRO or other models dependent on family history); women with a history of mantle radiation; women with a breast cancer syndrome such as Li-Fraumeni, Cowden and Bannayan-Riley-Ruvalcaba. There is very little published clinical information that exists for screening patients who are at increased risk based on a prior benign biopsy yielding lobular carcinoma in situ (LCIS), atypical ductal hyperplasia (ADH) or atypical lobular hyperplasia (ALH). The ACS did not issue a recommendation for or against screening in these clinical scenarios as there is not yet enough compelling clinical data. The decision whether or not to screen patients with these high risk lesions should be made on an individual basis by the referring clinician. In all of these cases, annual mammography is still recommended and should be performed either at the same time as MRI screening or staggered 6 months later. It is important to state that the decision to screen additionally should be carefully evaluated by the physician and patient with clear understanding of the limitations of the additional tests as well as the benefits.

Current Issues with Using MRI for Screening

There are multiple hurdles still to overcome to disseminate this important technology to achieve high level screening for the high-risk population. On a practical level, MRI is expensive, and that simple fact may restrict this test to only those who can afford it. Until insurers uniformly pay for MRI screening this may be a barrier to obtaining this exam for some women. The technique and the criteria for interpretation have been standardized in the United States in the form of the BI-RADS® (Breast Imaging Reporting and Data System) lexicon, supported by the American College of Radiology (ACR). The ACR has also supported formation of a voluntary accreditation process where facilities can obtain accreditation for performing breast MRI. It is expected that the criteria for accreditation will be released in 2009. As part of the accreditation process it will likely insist that centers perform intervention under MRI so that patients are treated in a timely and appropriate fashion. It is only reasonable for women to expect the same quality assurance programs incorporated into mammography practices to be part of other image-based screening. Quality assurance includes ability to localize or biopsy, correlation of biopsy results with imaging patterns and adequate surgical referral.

MRI may not be feasible in some women, such as those with pacemakers, aneurysm clips or severe claustrophobia. Moreover, breast MRI may result in a benign biopsy: among high risk women who undergo breast MRI screening, 3% to 15% have a subsequent biopsy yielding benign results. Also, whether detection of cancer on MRI positively affects patient survival (ie, affects mortality) is not yet known.

Whether MRI would benefit the average risk woman is completely unknown as no data exist. It is speculated that MR screening performed on women with average risk would have a much lower detection rate so that false positives would become overwhelming and the positive biopsy rate would be too low to justify its use. At this writing there is no indication to perform screening breast MRI in an average risk patient.

It should be emphasized that the data obtained in the studies is performed at tertiary referral centers that have a high risk population. Additionally, radiologists at these centers undoubtedly have more experience at breast MRI interpretation, biopsy and follow up. Rolling this technology out into the general radiology community will be challenging as there is a learning curve involved in the performance and the interpretation of breast MRI. Furthermore, radiologists who offer this service will need to be well versed in all breast imaging techniques, be able to biopsy and correlate biopsy results and offer appropriate follow up to their patients, in effect creating a whole new level of service.

Increased Call Backs and Biopsies

When considering a screening test, many important variables need to be satisfied, particularly a high sensitivity. As MRI has probably the highest sensitivity to detection of invasive cancer of all tests at this time, interest in screening for breast cancer with MRI has evolved. Initial reports that breast MRI could detect early breast cancer in the screening setting were positive however there was appropriate skepticism. The disadvantage of MRI screening is the possibility of a false positive diagnosis. The morbidity, anxiety, and cost associated with false positive findings cannot be underestimated. Because specificity of breast MRI is not extremely high, in fact it is lower than mammography in all published screening studies, the chance of finding a false positive finding far outweighs the possibility of cancer detection. Nevertheless, when the positive biopsy rate of MRI is examined in the literature, it is approximately 30%, meaning that one in three women recommended for a biopsy for an MRI finding will have cancer. For many, this is an acceptable rate and offers women with high risk a chance to undergo a screening test in lieu of more radical options such as tamoxifen or bilateral prophylactic mastectomies.

Additionally, the cost and anxiety associated with short-term follow-up examinations need to be considered for all those incidental foci that are seen on MRI and are not clearly benign. There are no guidelines on how soon to follow these patients, and no clear-cut way to manage these sometimes difficult findings. Recall rates in a screening population of high risk women of 12% have been described; however in our experience with a high risk population, the recall rate was 25%; a significant percentage, and certainly more than the recommended mammographic follow-up rate.

Call-back rates and biopsy rates are higher than for mammography but the increased sensitivity of MRI leads to a higher call back rate and also higher number of cancers detected. Recall rates decrease on subsequent rounds of screening (in one study to less than 10%) reflecting improved diagnostic confidence when having a comparison study as well as increased experience of the reader. What is interesting is that there is no evidence to suggest, despite higher recall rates and biopsies, a negative psychological impact on women at high-risk for breast cancer undergoing MRI screening.

MRI Technique

The ability to detect breast cancer (both invasive and in-situ disease) is directly related to high quality imaging, particularly the signal to noise ratio as well as the spatial resolution of the MR image. In order to detect early breast cancer simultaneous imaging of both breasts with high spatial resolution is favored. High-spatial resolution should be performed with a breast coil on a high field magnet with thin slices and high matrix (approximately 1-mm in-plane resolution). These technical parameters are considered to be the minimal requirements to perform an adequate breast MRI study. The ability to biopsy under MRI guidance is crucial and should be a part of any MRI screening service. The American College of Radiology (ACR) is currently developing a voluntary accreditation process for performing breast MRI and this guideline will likely be available in 2009.

In the realm of screening, quality assurance will become important. Sites that are considering offering this service should partner with a high-risk clinic and/or clinician with experience in counseling women at high risk. Sites with significant experience in interpretation and capability of biopsy should only accept patients for screening breast MRI. Sites that perform this examination are strongly encouraged to audit their practice to track call-back rates, biopsies as one would in mammography.

The age at which women should start MRI screening is unknown and currently the decision is based on shared decision making taking into consideration individual situations. In general high risk screening is started between the ages of 25 and 35 given data from observational studies that women with BRCA1 mutations have a 3% incidence of breast cancer at age 30 and increase thereafter. Most investigators agree that high risk women should be screened annually starting at age 30. Conversely, the age at which to stop intensive surveillance is also unknown. There are no data beyond 69 years.

Screening intervals are also unknown. Most high risk MRI screening studies show few interval cancers however BRCA tumors are known to develop more rapidly than sporadic cancers. At this time it is reasonable to screen yearly with MRI with further research into this area much needed. Some investigators recommend staggering the screening examinations with mammography so that the patient can be screened with either mammography or MRI every 6 months. Other experts however recommend MRI and mammography at the same time. There is no evidence to support either approach. It would be a serious mistake for physicians and women to assume that a breast MRI would obviate the need for a mammogram. Peer-reviewed published studies have shown that there are cancers seen on mammography, particularly DCIS, that are not imaged on MRI. Any attempt to replace mammographic screening with any other modality will sacrifice the advantage to the screened population of the detection of these early, highly treatable breast cancers. Therefore, responsible medical advice to women can only state that if MRI has a role to play in breast cancer screening, it should only be used in conjunction with conventional mammography. It is possible that older women can be screened less than younger women based on the observation that the doubling times of inheritable cancers decrease with age.

Summary

While it is certain that the terrain of the medical field is changing with the advent of more consumer-controlled medical care and more patient choice in determining the use of procedures, it is our responsibility as physicians to honestly present all the data, so that patients can make a well informed decision. What we know so far about breast MRI screening is that it is very valuable in the high risk population with a life-time cancer risk of > 20% and should be recommended annually in that population.

Breast imaging radiologists are well aware of the limitation of mammographic screening and have been deeply involved in the search for techniques to further decrease breast cancer mortality. There are still many unanswered questions about breast MRI screening of high risk patients and as radiologists we must be cautious not to over sell this technology and expect it to be the panacea for breast cancer detection. Continued surveillance with mammography still appears to be important and it is entirely possible that in the future breast MRI may be supplanted by another modality. Currently in the new millennium we have a highly sensitive test to detect breast cancer and if applied to the appropriate patients will more than likely continue to improve breast cancer mortality.

Back to top