American Cancer Society Guidelines for Breast Screening with MRI as an Adjunct to Mammography
←
→
Page content transcription
If your browser does not render page correctly, please read the page content below
CA Cancer J Clin 2007;57:75–89
Dr. Saslow is Director, Breast and
American Cancer Society Guidelines Gynecologic Cancer, Cancer Control
Science Department, American Cancer
for Breast Screening with MRI as an Society, Atlanta, GA.
Dr. Boetes is Professor, Department
Adjunct to Mammography of Radiology, University Medical Center
Nijmegen, Nijmegen, The Netherlands.
Dr. Burke is Chair; and Professor of
Medical History and Ethics, University
Debbie Saslow, PhD; Carla Boetes, MD, PhD; Wylie Burke, MD, PhD; Steven Harms, MD; of Washington, Seattle, WA.
Martin O. Leach, PhD; Constance D. Lehman, MD, PhD; Elizabeth Morris, MD; Etta Pisano,
Dr. Harms is Radiologist, The Breast
MD; Mitchell Schnall, MD, PhD; Stephen Sener, MD; Robert A. Smith, PhD; Ellen Warner, Center of Northwest Arkansas, Fay-
MD; Martin Yaffe, PhD; Kimberly S. Andrews; Christy A. Russell, MD (for the American Cancer etteville, AR; and Clinical Professor of
Radiology, University of Arkansas for
Society Breast Cancer Advisory Group) Medical Sciences, Little Rock, AR.
Dr. Leach is Professor of Physics as
ABSTRACT New evidence on breast Magnetic Resonance Imaging (MRI) screening has Applied to Medicine; Co-Chairman,
Section of Magnetic Resonance; and
become available since the American Cancer Society (ACS) last issued guidelines for the early Co-Director, Cancer Research; UK
detection of breast cancer in 2003. A guideline panel has reviewed this evidence and developed Clinical Magnetic Resonance Research
Group, The Institute of Cancer Research
new recommendations for women at different defined levels of risk. Screening MRI is recom- and The Royal Marsden NHS Founda-
mended for women with an approximately 20–25% or greater lifetime risk of breast cancer, tion Trust, Surrey, UK.
including women with a strong family history of breast or ovarian cancer and women who were Dr. Lehman is Professor of Radiol-
ogy; and Section Head of Breast Imag-
treated for Hodgkin disease. There are several risk subgroups for which the available data are ing, University of Washington Medical
insufficient to recommend for or against screening, including women with a personal history of Center and the Seattle Cancer Care
Alliance, Seattle, WA.
breast cancer, carcinoma in situ, atypical hyperplasia, and extremely dense breasts on mam-
Dr. Morris is Director, Breast MRI,
mography. Diagnostic uses of MRI were not considered to be within the scope of this review. Department of Radiology, Memorial
(CA Cancer J Clin 2007;57:75–89.) © American Cancer Society, Inc., 2007. Sloan-Kettering Cancer Center, New
York, NY.
To earn free CME credit for successfully completing the online quiz based on this article, go to Dr. Pisano is Vice Dean for Academic
http://CME.AmCancerSoc.org. Affairs; Kenan Professor of Radiology
and Biomedical Engineering; and Dir-
ector, UNC Biomedical Research Imag-
ing Center, UNC School of Medicine,
INTRODUCTION
University of North Carolina, Chapel
Hill, NC.
Mammography has been proven to detect breast cancer at an early stage and,
Dr. Schnall is Matthew J. Wilson
when followed up with appropriate diagnosis and treatment, to reduce mortality
Professor of Radiology; and Associate
from breast cancer. For women at increased risk of breast cancer, other screening Chair for Research, Department of
technologies also may contribute to the earlier detection of breast cancer, particu- Radiology, University of Pennsylvania,
larly in women under the age of 40 years for whom mammography is less sensitive. Philadelphia, PA.
The American Cancer Society (ACS) guideline for the early detection of breast can- Dr. Sener is Vice Chairman, Depart-
ment of Surgery, Evanston North-
cer, last updated in 2003, stated that women at increased risk of breast cancer might
western Health Care, Evanston, IL; and
benefit from additional screening strategies beyond those offered to women at aver- Professor of Surgery, Northwestern
age risk, such as earlier initiation of screening, shorter screening intervals, or the University Feinberg School of Medi-
addition of screening modalities (such as breast ultrasound or magnetic resonance cine, Chicago, IL.
imaging [MRI]) other than mammography and physical examination. However, the Dr. Smith is Director, Cancer Screen-
ing, Cancer Control Science Depart-
evidence available at the time was insufficient to justify recommendations for any
ment, American Cancer Society, Atlanta,
of these screening approaches. The ACS recommended that decisions about screen- GA.
ing options for women at significantly increased risk of breast cancer be based on Dr. Warner is Associate Professor,
shared decision making after a review of potential benefits, limitations, and harms of Department of Medicine, University
different screening strategies and the degree of uncertainty about each.1 of Toronto and Toronto-Sunnybrook
Regional Cancer Centre, Toronto,
Although there still are limitations in the available evidence, additional pub-
Canada.
lished studies have become available since the last update, particularly regarding
Volume 57 • Number 2 • March/April 2007 75American Cancer Society Guidelines for Breast Screening with MRI as an Adjunct to Mammography
Dr. Yaffe is Senior Scientist, Imaging use of breast MRI. The ACS guide- SUMMARY OF RECOMMENDATIONS
Research, Sunnybrook Health Sciences line panel has sought to provide addi-
Centre; and Professor, Departments
tional guidance to women and their Table 1 summarizes the ACS recommendations
of Medical Imaging and Medical for breast MRI screening.
Biophysics, University of Toronto, health care providers based on these
Toronto, Canada. new data.
Ms. Andrews is Research Associate, BACKGROUND
Cancer Control Science Department,
American Cancer Society, Atlanta, GA. GUIDELINE DEVELOPMENT MRI
Dr. Russell is Associate Professor
of Medicine, Keck School of Medicine, The ACS convened an expert MRI utilizes magnetic fields to produce detailed
University of Southern California; and
panel to review the existing early cross-sectional images of tissue structures, pro-
Co-Director of USC/Norris Breast
Center, Los Angeles, CA. detection guideline for women at viding very good soft tissue contrast. Contrast
This article is available online at
increased risk and for MRI screen- between tissues in the breast (fat, glandular tis-
http://CAonline.AmCancerSoc.org ing based on evidence that has accu- sue, lesions, etc.) depends on the mobility and
mulated since the last revision in magnetic environment of the hydrogen atoms in
2002 to 2003. Literature related to water and fat that contribute to the measured
breast MRI screening published between signal that determines the brightness of tissues
September 2002 and July 2006 was identified in the image. In the breast, this results in images
using MEDLINE (National Library of Medi- showing predominantly parenchyma and fat, and
cine), bibliographies of identified articles, and lesions, if they are present. A paramagnetic small
unpublished manuscripts. Expert panel mem- molecular gadolinium-based contrast agent is
bers reviewed and discussed data during a series injected intravenously to provide reliable detec-
of conference calls and a working meeting in tion of cancers and other lesions. Thus, contrast
August, 2006. When evidence was insufficient enhanced MRI has been shown to have a high
or lacking, the final recommendations incor- sensitivity for detecting breast cancer in high-
porated the expert opinions of the panel mem- risk asymptomatic and symptomatic women,
bers. The ACS Breast Cancer Advisory Group although reports of specificity have been more
members and the National Board of Directors variable.2–8 This high signal from enhancing
discussed and voted to approve the recommen- lesions can be difficult to separate from fat, lead-
dations. ing to the use of subtraction images or fat
TABLE 1 Recommendations for Breast MRI Screening as an Adjunct to Mammography
Recommend Annual MRI Screening (Based on Evidence*)
BRCA mutation
First-degree relative of BRCA carrier, but untested
Lifetime risk ~20–25% or greater, as defined by BRCAPRO or other models that are largely dependent on family history
Recommend Annual MRI Screening (Based on Expert Consensus Opinion†)
Radiation to chest between age 10 and 30 years
Li-Fraumeni syndrome and first-degree relatives
Cowden and Bannayan-Riley-Ruvalcaba syndromes and first-degree relatives
Insufficient Evidence to Recommend for or Against MRI Screening‡
Lifetime risk 15–20%, as defined by BRCAPRO or other models that are largely dependent on family history
Lobular carcinoma in situ (LCIS) or atypical lobular hyperplasia (ALH)
Atypical ductal hyperplasia (ADH)
Heterogeneously or extremely dense breast on mammography
Women with a personal history of breast cancer, including ductal carcinoma in situ (DCIS)
Recommend Against MRI Screening (Based on Expert Consensus Opinion )
Women at ⬍15% lifetime risk
*Evidence from nonrandomized screening trials and observational studies.
†Based on evidence of lifetime risk for breast cancer.
‡Payment should not be a barrier. Screening decisions should be made on a case-by-case basis, as there may be
particular factors to support MRI. More data on these groups is expected to be published soon.
76 CA A Cancer Journal for CliniciansCA Cancer J Clin 2007;57:75–89
suppression, or both, to assess disease. Because cancer. Features of the family history which
parenchymal tissue also enhances, but generally suggest the cancers may be due to such a high-
more slowly than malignant lesions, and also penetrance gene include 2 or more close (gen-
because contrast can wash out rapidly from some erally first- or second-degree) relatives with breast
tumors, it is important to look at images at an or ovarian cancer; breast cancer occurring before
early time point after contrast injection (typically age 50 years (premenopausal) in a close relative;
1 to 3 minutes). MRI examinations may involve a family history of both breast and ovarian can-
examining images at one time point or, more cer; one or more relatives with 2 cancers (breast
often, will collect a preinjection image with and ovarian cancer or 2 independent breast can-
sequential sets of images after contrast injection cers); and male relatives with breast cancer.15–18
(dynamic contrast-enhanced [DCE]-MRI). Both Two breast/ovarian cancer susceptibility genes,
the appearance of lesions and, where available, BRCA1 and BRCA2, have been identified.19,20
the uptake and washout pattern can be used to Inherited mutations in these genes can be found
identify malignant disease and discriminate it in approximately 50% of families in which an
from benign conditions. inherited risk is strongly suspected based on the
These techniques, which have been widely frequency and age of onset of breast cancer cases,
employed for assessing symptomatic disease, have and in most families in which there is a much
recently been shown to provide good sensitiv- higher than expected incidence of both breast
ity as a screening tool for breast cancer in women and ovarian cancer.
at increased risk based on family history.9–14 The Several models can assist clinicians to estimate
approach requires appropriate techniques and breast cancer risk or the likelihood that a BRCA
equipment, together with experienced staff. mutation is present (Online Supplemental
Higher quality images are produced by dedi- Material). The Gail, Claus, and Tyrer-Cusick
cated breast MRI coils, rather than body, chest, models estimate breast cancer risk based on fam-
or abdominal coils. ily history, sometimes in combination with other
risk factors, such as reproductive history or prior
breast biopsies.16,21–23 Although risk prediction
IDENTIFICATION OF WOMEN WITH A
HIGH RISK OF BREAST CANCER is generally similar for the different models, an
individual woman’s risk estimate may vary with
Three approaches are available for identify- different models.21,24,25
ing women with a high risk of breast cancer: Two decision models have been developed
family history assessment, genetic testing, and to estimate the likelihood that a BRCA mutation
review of clinical history. All contribute to iden- is present, BRCAPRO18,26 and the Breast and
tifying women who are candidates for breast Ovarian Analysis of Disease Incidence and Carrier
MRI screening. Estimation Algorithm (BOADICEA) 27 ; the
BOADICEA model also provides estimates of
Family History
breast cancer risk (Online Supplemental Material).
Although a high proportion of women in the
Genetic Testing
general population have at least one relative with
breast cancer, for the majority of these women, The prevalence of BRCA mutations is esti-
this “family history” either does not increase risk mated to be between 1/500 and 1/1,000 in the
at all (ie, the cancer was sporadic) or is associ- general population28; however, in women of Jewish
ated with, at most, a doubling of lifetime risk ethnicity, the prevalence is 1/50.29,30 Women with
(due to either shared environmental risk factors cancer-predisposing mutations in either BRCA1
or an inherited gene of low penetrance). Only or BRCA2 have an increased risk of both breast
1% to 2% of women have a family history sug- and ovarian cancer. From population-based
gestive of the inheritance of an autosomal dom- studies, women with BRCA1 mutations are esti-
inant, high-penetrance gene conferring up to mated to have a 65% risk by age 70 years for
an 80% lifetime risk of breast cancer. In some developing breast cancer (95% confidence inter-
families, there is also a high risk of ovarian val [CI], 44% to 78%); the corresponding risk
Volume 57 • Number 2 • March/April 2007 77American Cancer Society Guidelines for Breast Screening with MRI as an Adjunct to Mammography
for BRCA2 mutations is 45% (95% CI, 31% to studies, 39,42 which may reflect the effect of
56%).31 Risks estimated from cancer-prone fam- chemotherapy on earlier onset of menopause; risk
ilies seen in referral centers are higher, with limit was equivalent in a third study.43 Risk of breast
of risk in the 85% to 90% range.31 These muta- cancer significantly increased 15 to 30 years after
tions follow an autosomal dominant pattern of radiation therapy.41 More recently, treatment
transmission, which means that the sister, mother, approaches have used lower doses of radiation and
or daughter of a woman with a BRCA muta- limited-field radiotherapy. In one study, which
tion has a 50% chance of having the same muta- compared patients who received radiation ther-
tion. apy in 1966 to 1974 and 1975 to 1985, treat-
The benefits and risks of genetic testing are ment in the later timeframe was not related to
beyond the scope of this article, but are reviewed increased risk of breast cancer after a median
in the American Society of Clinical Oncology follow up of 13 years, whereas patients treated
policy statement update on genetic testing for between 1966 and 1974 were at increased risk,
cancer susceptibility. 32 Genetic testing for a suggesting that Hodgkin disease survivors treated
BRCA1 or BRCA2 mutation is generally offered with current approaches will not face substan-
to adult members of families with a known tially increased breast cancer risk.44
BRCA mutation, or to women with at least a Lobular carcinoma in situ (LCIS) and atypi-
10% likelihood of carrying such a mutation, cal lobular hyperplasia (ALH), together described
based on either validated family history criteria as lobular neoplasia, are associated with substan-
or one of the above-mentioned models. If a tially increased risk of subsequent breast cancer,
woman from a family in which a BRCA muta- with lifetime risk estimates ranging from 10%
tion has been previously identified does not have to 20%.45 This equates to a continuous risk of
that mutation, one can generally safely conclude about 0.5% to 1.0% per year. The invasive can-
that her breast cancer risk is no higher than it cers may be ipsilateral or contralateral, are usu-
would have been if she did not have a family ally invasive lobular cancers, and more than 50%
history of breast cancer. However, in a high-risk of these diagnoses occur more than 15 years after
family without a known mutation, failure to the original diagnosis of LCIS. Similar findings
find a mutation in a particular member does not have been reported by Fisher et al,46 describing
reduce her risk estimate. a 12-year update of 180 women with LCIS who
A high risk of breast cancer also occurs with were treated with local excision alone and fol-
mutations in the TP53 gene (Li-Fraumeni syn- lowed by the National Surgical Adjuvant Breast
drome) and the PTEN gene (Cowden and Project (NSABP), as well as Li et al, who de-
Bannayan-Riley-Ruvalcaba syndromes). 33 scribed the risk of invasive breast cancer among
Accurate prevalence figures are not available, but 4,490 LCIS patients using Surveillance, Epidemi-
these conditions appear to be very rare.34,35 ology, and End Results (SEER) data between
1988 to 2001.47
Clinical Indicators of Risk
Atypical ductal hyperplasia (ADH) is part of
Some clinical factors are associated with sub- the continuum of ductal proliferative breast dis-
stantial breast cancer risk. Among women with eases ranging from usual ductal hyperplasia to
Hodgkin disease, increased breast cancer risk has ductal carcinoma in situ (DCIS). The literature
been consistently and significantly associated review by Arpino et al45 suggests a 4- to 5-fold
with mantle field radiation treatment. In several increased risk of invasive breast cancer (com-
studies of women treated between 1955 and pared with a 6- to 10-fold risk with LCIS) at a
1995, risk was inversely related to age at treat- median follow up of 17 years, which is doubled
ment in patients diagnosed between the ages of if the woman has an associated family history of
10 to 30 years, with only slight or no increased breast cancer. It is unclear, however, what per-
risk when diagnosis was before age 10 years or centage of the women with this family history
after age 30 years.36–41 Risk following treatment and ADH are at this significantly increased risk
with radiation and chemotherapy was half that because they are carriers of a BRCA1 or 2 gene
of treatment with radiation alone in two mutation.
78 CA A Cancer Journal for CliniciansCA Cancer J Clin 2007;57:75–89
Mammographic density has been shown to higher lifetime risk of breast cancer (19% proven
be a strong independent risk factor for the devel- to have a BRCA mutation) at 6 centers across
opment of breast cancer.48–51 In several studies, The Netherlands.9 After a median of 3 rounds of
women with the most breast density were found screening, 50 breast cancers (44 invasive) were
to have a 4- to 6-fold increased risk of breast can- diagnosed. Eighty percent of the invasive can-
cer, compared with women with the least dense cers were detected by MRI, compared with 33%
breasts.52–56 For example, women with 75% or by mammography. However, mammography
higher mammographic density had a more than outperformed MRI for detecting DCIS. Of the
five-fold increased risk of breast cancer, com- invasive cancers, 43% were 1 cm or smaller in
pared with women with less than 1% density.57 diameter, and 33% had spread to axillary lymph
In addition, it has been shown that malignant nodes. The specificity of MRI was 90%, com-
tumors of the breast are more likely to arise in pared with 95% for mammography.
the areas of greatest mammographic density, com- Leach et al screened 649 unaffected women
pared with the more fatty areas of the breast.58 aged 35 to 49 years who had at least a 25% life-
The absolute risk of contralateral breast can- time risk of breast cancer (19% proven to have a
cer in women with a personal history of breast BRCA mutation) at 22 centers in the UK.11 After
cancer is estimated to be 0.5% to 1% per year, or a median of 3 rounds of screening, 35 cancers
5% to 10% during the 10 years following diag- (29 invasive) were diagnosed. Sensitivity of MRI
nosis, significantly higher than that of the gen- was 77%, compared with 40% for mammography,
eral population.59 Hormone therapy and/or with specificities of 81% and 93%, respectively.
chemotherapy for the primary cancer is likely MRI was most sensitive and mammography least
to subsequently lower the risk of contralat- sensitive for women with BRCA1 mutations.
eral breast cancer. Forty-five percent of the cancers were 1 cm or less
in size, and 14% had spread to axillary lymph
nodes. There were two interval cancers.
EVIDENCE AND RATIONALE Warner et al screened 236 women aged 25
Evidence of Efficacy from MRI Screening Studies to 65 years with a BRCA mutation at a single
center in Toronto for up to 3 years and detected
In the mid to late 1990s, at least 6 prospective, 22 cancers (16 invasive).14 Sensitivity of MRI
nonrandomized studies were initiated in The was 77%, compared with 36% for mammogra-
Netherlands, the United Kingdom (UK), Canada, phy, with 50% of the cancers 1 cm or smaller,
Germany, the United States (US), and Italy to and 13% were node positive. There was one
determine the benefit of adding annual MRI to interval cancer. Specificity was 95% for MRI
(film) mammography for women at increased and 99.8% for mammography.
risk of breast cancer. Some of the studies included Kuhl et al screened 529 women aged 30 years
ultrasound and/or clinical breast examination, and older with a lifetime breast cancer risk of at
as well. Despite substantial differences in patient least 20% at a single center in Bonn for a mean of
population (age, risk, etc.) and MRI technique, 5 years.10 They detected 43 cancers (34 invasive),
all reported significantly higher sensitivity for with 1 interval cancer. The sensitivity of MRI
MRI compared with mammography (or any of was 91%, compared with 33% for mammogra-
the other modalities). All studies that included phy. The node positive rate was 16%. Specificity
more than one round of screening reported inter- of both MRI and mammography was 97%.
val cancer rates below 10%. Participants in each The International Breast MRI Consortium
of these 6 studies had either a documented screened 390 women aged 25 years and older
BRCA1 or BRCA2 mutation or a very strong with more than a 25% lifetime risk of breast can-
family history of breast cancer. Some of the stud- cer at 13 centers (predominantly in the US) on
ies included women with a prior personal his- a single occasion.12 Four cancers were found by
tory of breast cancer. MRI, and only one of these by mammography.
Kriege et al screened 1,909 unaffected women However, because the patients were not followed
aged 25 to 70 years with an estimated 15% or after screening, the false-negative rate could not
Volume 57 • Number 2 • March/April 2007 79American Cancer Society Guidelines for Breast Screening with MRI as an Adjunct to Mammography
TABLE 2 Published Breast MRI Screening Study Results
The Netherlands Canada United Kingdom Germany United States Italy
No. of centers 6 1 22 1 13 9
No. of women 1,909 236 649 529 390 105
Age range 25–70 25–65 35–49 ⱖ30 ⱖ25 ⱖ25
No. of cancers 50 22 35 43 4 8
Sensitivity (%)
MRI 80 77 77 91 100 100
Mammogram 33 36 40 33 25 16
Ultrasound n/a 33 n/a 40 n/a 16
Specificity (%)
MRI 90 95 81 97 95 99
Mammogram 95 ⬎99 93 97 98 0
Ultrasound n/a 96 n/a 91 n/a 0
n/a = not applicable.
be determined. MRI specificity was 95%, com- and/or genetic mutations. More recently, smaller
pared with 98% for mammography. studies have provided information on the poten-
In a study in Italy with 9 participating centers, tial benefit of MRI screening for women with
Sardanelli et al screened 278 women aged 25 years clinical factors that put them at increased risk.
and older; 27% carried a BRCA mutation or Preliminary data were obtained from one retro-
had a first-degree relative with a BRCA muta- spective study, in which Port et al61 reviewed
tion.13 After a median of 1.4 rounds of screen- the screening results of 252 women with biopsy-
ing, 18 cancers (14 invasive) were found. MRI confirmed LCIS and 126 women with atypical
sensitivity was 94%, compared with 59% for hyperplasia (either ductal or lobular), of whom
mammography, 65% for ultrasound, and 50% half were screened with annual mammography
for clinical breast examination. MRI specificity and biennial clinical exams and half were also
was 99%. screened with MRI. The women who were
Overall, studies have found high sensitivity screened with MRI were younger and more
for MRI, ranging from 71% to 100% versus 16% likely to have a strong family history. MRI screen-
to 40% for mammography in these high-risk ing offered a small advantage to patients with
populations. Three studies included ultrasound, LCIS, but not atypical hyperplasia, and also
which had sensitivity similar to mammography. resulted in increased biopsies: 6 cancers were
The Canadian, Dutch, and UK studies 9,11,14 detected by MRI in 5 women with LCIS (4% of
reported similar sensitivity (71% to 77%) within patients undergoing MRI), and none were
CIs for MRI, although the single-center study detected in women with atypical hyperplasia.
from Germany10 reported a higher sensitivity, Biopsies were recommended for 25% of MRI
which may reflect the concentration of radio- screened patients; 13% of biopsies had a cancer
logical practice and higher patient volume per detected. All of the cancers in women screened
radiologist at a single center. There is evidence of with MRI were Stage 0 to I, whereas all of the
a learning curve for radiologists conducting MRI cancers in women who were not screened with
breast screening, with the number of lesions inves- MRI were Stage I to II. Cancer was detected
tigated falling with experience.60 The three mul- on the first MRI in 4 of 5 patients. The sensi-
ticenter studies reflect the likely initial effectiveness tivity of MRI was 75%, the specificity was 92%,
of this modality in a population context, and it and the positive predictive value was 13%.
is expected that, with training and advances in
technology, sensitivity will increase further. Technological Limitations and Potential Harms
Table 2 provides a summary of these six screen- Associated with MRI Screening
ing studies.
Most of the available data are based on screen- Although the efficacy of breast MRI has
ing women at high risk due to family history been demonstrated, it does not achieve perfect
80 CA A Cancer Journal for CliniciansCA Cancer J Clin 2007;57:75–89
TABLE 3 Rates of Detection and Follow-up Tests for Screening MRI Compared with Mammography
MRI Mammography
The Netherlands United Kingdom The Netherlands United Kingdom
Positives 13.7% 19.7% 6.0% 7.2%
Recalls 10.84% 10.7% 5.4% 3.9%
Biopsies 2.93% 3.08% 1.3% 1.33%
Cancers 1.04% 1.44% 0.46% 0.69%
False negatives 0.23% 0.43% 0.81% 1.52%
sensitivity or specificity in women undergo- more data are needed on factors associated with
ing screening, and as such, the issue of adverse lower specificity rates.
consequences for women who do, but espe- Table 3 compares the likelihood of detection
cially those who do not, have breast cancer is and follow-up tests for women who underwent
important to address. As with mammography screening MRI and mammography in two
and other screening tests, false negatives after screening studies (Dutch and UK). The study
MRI screening can be attributed to inherent populations differed, with the Dutch study hav-
technological limitations of MRI, patient char- ing a wider age group and lower risk category,
acteristics, quality assurance failures, and human compared with the UK study.9,11 This affected
error; false positives also can be attributed to both the prevalence of cancer and the pick-up
these factors, as well as heightened medical- rate by modality in the two studies. These results,
legal concerns over the consequence of missed drawn from two trials, demonstrate the rela-
cancers. A patient’s desire for definitive find- tively high recall rate in the high-risk popula-
ings in the presence of a low-suspicion lesion tion, as well as the fact that MRI is a relatively
may also contribute to a higher rate of benign new technique. Despite the high number of
biopsies. The consequences of all these fac- recalls, because of the high cancer rate, the rate
tors include missed cancers, with potentially of benign surgical biopsy in the UK study per
worse prognosis, as well as anxiety and poten- cancer detected was similar to that experienced
tial harms associated with interventions for in the population-based national breast screen-
benign lesions. ing service. Recalls will inevitably lead to addi-
The specificity of MRI is significantly lower tional investigations, many of which will not
than that of mammography in all studies to date, demonstrate that cancer is present.
resulting in more recalls and biopsies. Call-back Given the high rate of cancer combined with
rates for additional imaging ranged from 8% to the risk of false-positive scans in a high-risk pop-
17% in the MRI screening studies, and biopsy ulation undergoing MRI-based screening, the
rates ranged from 3% to 15%.9–14 However, sev- psychological health of these women merits
eral researchers have reported that recall rates study. In a subgroup of 611 women in the UK
decreased in subsequent rounds of screening: study, 89% reported that they definitely intended
prevalence screens had the highest false-positive to return for further screening, and only 1% def-
rates, which subsequently dropped to less than initely intended not to return. However, 4%
10%.9,62,63 Most call backs can be resolved with- found breast MRI “extremely distressing,” and
out biopsy. The call-back and biopsy rates of 47% reported still having intrusive thoughts about
MRI are higher than for mammography in high- the examination 6 weeks afterward.64
risk populations; while the increased sensitivity In a sample of 357 women from the Dutch
of MRI leads to a higher call-back rate, it also study, psychological distress remained within
leads to a higher number of cancers detected. normal limits throughout screening for the
The proportion of biopsies that are cancerous group as a whole. However, elevated breast
(positive predictive value) is 20% to 40%.9–14 cancer-specific distress related to screening was
Since false-positive results appear to be common, found in excessive (at least once per week) breast
Volume 57 • Number 2 • March/April 2007 81American Cancer Society Guidelines for Breast Screening with MRI as an Adjunct to Mammography
self-examiners, risk overestimators, and women mutation status, other risk factors, age, and
closely involved in the breast cancer case of a sis- mammographic breast density.
ter. At least 35% of the total sample belonged There are substantial concerns about costs of
to one of these subgroups. It was recommended and limited access to high-quality MRI breast
that patients in one of these vulnerable sub- screening services for women with familial risk.
groups be approached for additional psycho- In addition, MRI-guided biopsies are not widely
logical support.65 available. With many communities not provid-
In a small sample of women from the Toronto ing MRI screening and with MRI-guided biop-
study followed over a course of 2 years, there sies not widely available, it is recognized that
was no evidence of any effect on global anxiety, these recommendations may generate concerns
depression, or breast cancer-related anxiety.66 In in high-risk women who may have limited access
another sample of 57 women, almost 50% had to this technology.
elevated baseline general and/or breast cancer- The ability of MRI to detect breast cancer
specific anxiety, but in 77% of cases this was (both invasive and in situ disease) is directly related
attributed by the patients to life events, includ- to high-quality imaging, particularly the signal-
ing relatives with cancer. A nonsignificant increase to-noise ratio, as well as spatial resolution of the
in general anxiety and breast cancer-related anx- MR image. In order to detect early breast can-
iety, compared with baseline, was found in the cer (ie, small invasive cancers, as well as DCIS),
subset of women recalled for further imaging or simultaneous imaging of both breasts with high
biopsies.67 Follow-up time is still insufficient to spatial resolution is favored. High spatial resolu-
determine whether anxiety scores return to base- tion imaging should be performed with a breast
line once the work up has been completed. coil on a high field magnet with thin slices and
There is a special responsibility to alert high matrix (approximately 1 mm in-plane res-
patients to this technology, with its potential olution). These technical parameters are consid-
strengths and harms, and to be encouraging, ered to be the minimal requirements to perform
while allowing for shared decision making. The an adequate breast MRI study. The ability to per-
interplay between risks, benefits, limitations, form MRI-guided biopsy is absolutely essential
and harms is complicated by the fact that indi- to offering screening MRI, as many cancers (par-
vidual women likely will weigh these differ- ticularly early cancers) will be identified only on
ently depending on their age, values, perception MRI. The American College of Radiology
of risk, and their understanding of the issues. (ACR) is currently developing an accreditation
Steps should be taken to reduce anxiety asso- process for performing breast MRI, and, in addi-
ciated with screening and the waiting time to tion to the performance of high spatial resolu-
diagnosis, and conscientious efforts should be tion images, the ability to perfor m MRI
made to inform women about the likelihood intervention (ie, needle localization and/or biopsy)
of both false-negative and false-positive find- will be essential in order to obtain accreditation
ings. How information is conveyed to the patient by this group. Accreditation will be voluntary
greatly influences the patient’s response: it is and not mandatory. This guideline will likely be
important that providers not convey an undue available in 2007.
sense of anxiety about a positive MRI finding. There is a learning curve with respect to inter-
While the high rate of biopsies and further pretation for radiologists. Published trial sites
investigations is acceptable in women with a that experience a high volume of cases are expe-
high risk of breast cancer, the number of such rienced, but community practice groups have
investigations in women at lower risk will be reported call-back rates over 50% in the major-
much higher than would be appropriate, lead- ity of the studies that are interpreted. Experience
ing to the need to counsel women in lower and familiarity with patterns of enhancement,
risk categories that MRI screening is not advis- normal and possibly abnormal, are thought to
able and that the harms are believed to out- decrease recall rates and increase positive biopsy
weigh the benefits. Such advice needs to be rates. The ACR accreditation process will stip-
based on considerations of family history, genetic ulate a minimum number of exams that must be
82 CA A Cancer Journal for CliniciansCA Cancer J Clin 2007;57:75–89
read for training purposes and a minimum num- 12% 10-year risk; and $96,379 for women with
ber for ongoing accreditation. Sites performing a 6% 10-year risk. For the 30- to 39-year-old
breast MRI are encouraged to audit their call- age range, the incremental costs per QALY are
back rates, biopsy rates, and positive biopsy rates. $24,275 for a BRCA1 carrier with an 11%
10-year risk and $70,054 for a women with a
Cost-effectiveness 5% 10-year risk. Based on these estimates, which
Only limited data are available on the cost- are based on costs within the UK National Health
effectiveness of breast MRI screening. One recent Service, MRI screening will be offered to women
study modeled cost-effectiveness for adding MRI at familial risk aged 30 to 39 years at a 10-year
to mammography screening for women of dif- risk greater than 8%, and to women at familial
ferent age groups who car ry a BRCA1 or risk aged 40 to 49 years at a 10-year risk greater
BRCA2 mutation.68 The authors concluded that than 20%, or greater than 12% when mammog-
the cost per quality-adjusted life year (QALY) raphy has shown a dense breast pattern.
saved for annual MRI plus film mammography,
compared with annual film mammography alone, Evidence Supporting Benefit of MRI Screening
Among Women in Different Risk Categories
varied by age and was more favorable in carri-
ers of a mutation in BRCA1 than BRCA2 The guideline recommendations were based
because BRCA1 mutations confer higher can- on consideration of (1) estimates of level of risk
cer risk, and higher risk of more aggressive can- for women in various categories and (2) the extent
cers, than BRCA2 mutations.31 Estimated cost to which risk groups have been included in MRI
per QALY for women aged 35 to 54 years was studies, or to which subgroup-specific evidence
$55,420 for women with a BRCA1 mutation is available. Because of the high false-positive rate
and $130,695 for women with a BRCA2 muta- of MRI screening, and because women at higher
tion. Cost-effectiveness was increased when the risk of breast cancer are much more likely to ben-
sensitivity of mammography was lower, such as efit than women at lower risk, screening should
in women with very dense breasts on mammog- be recommended only to women who have a
raphy: estimated costs per QALY were $41,183 high prior probability of breast cancer. There is
for women with a BRCA1 mutation and $98,454 growing evidence that breast cancer in women
for women with a BRCA2 mutation with dense with specific mutations may have biological and
breast tissue. The most important determinants histological features that differ from sporadic can-
of cost-effectiveness were breast cancer risk, cers. This may result in observed variations in
mammography sensitivity, MRI cost, and qual- the sensitivity of MRI relative to mammogra-
ity of life gains from MRI. phy in detecting cancer in women with a BRCA
An evaluation of the cost-effectiveness of the mutation and those at high familial risk, but with-
UK study69 has determined that the incremen- out mutations in these genes.11
tal cost per cancer detected for women at approx-
Women at Increased Risk Based on
imately 50% risk of carrying a BRCA gene
mutation was $50,911 for MRI combined with Family History
mammography over mammography alone. For The threshold for defining a woman as hav-
known mutation carriers, the incremental cost ing significantly elevated risk of breast cancer is
per cancer detected decreased to $27,544 for based on expert opinion. Any woman with a
MRI combined with mammography, compared BRCA1 or BRCA2 mutation should be con-
with mammography alone. Analysis supporting sidered at high risk. The panel has not restricted
the introduction of targeted MRI screening in its recommendations only to women with BRCA
the UK for high-risk women70 identified the mutations because BRCA testing is not always
incremental cost of combined screening per available or informative, and other risk indica-
QALY in 40- to 49-year-old women as $14,005 tors identify additional subsets of women with
for a BRCA1 carrier with a 31% 10-year risk— increased breast cancer risk. If mutation testing
the group in which MRI screening is seen to is not available, has been done and is noninfor-
be most effective; $53,320 for women with a mative, or if a woman chooses not to undergo
Volume 57 • Number 2 • March/April 2007 83American Cancer Society Guidelines for Breast Screening with MRI as an Adjunct to Mammography
TABLE 4 Breast Cancer Risks for Hypothetical Patients, Based on 3 Risk Models
Family History BRCAPRO*18 Claus†16 Tyrer-Cuzick‡23
35-year-old woman
Mother BC 33
Maternal aunt BC 42 19% 36% 28%
35-year-old woman
Paternal aunt BC 29, OC 49
Paternal grandmother BC 35 23% 24% 32%
35-year-old woman
Paternal aunt BC 29
Paternal grandmother BC 35 18% 24% 31%
35-year-old woman
Mother BC 51
Maternal aunt BC 60 13% 18% 23%
35-year-old woman of Jewish ancestry
Mother BC 51
Maternal aunt BC 60 18% 18% 28%
BC = breast cancer.
OC = ovarian cancer.
*BRCAPRO (1.4–2) Breast cancer risk calculated to age 85 years.
†Breast cancer risk calculated to age 79 years.
‡Breast cancer risk calculated for lifetime. Other personal characteristics included in the Tyrer-Cuzick risk model for each
case were age at menarche = 12; age at first birth = 28; height = 1.37 meters (5 feet, 4 inches); weight = 61 kg (134 lbs);
woman has never used hormone replacement therapy (HRT); no atypical hyperplasia or lobular carcinoma in situ (LCIS).
testing, pedigree characteristics suggesting high for MRI screening recommendations, as well as
risk may be considered. Very careful family his- areas of uncertainty. For some women, mammog-
tory analysis is required, using tools such as raphy may be as effective as for women at average
BRCAPRO.18,26 Risk assessment is likely to offer risk, and MRI screening may have little added
the greatest potential benefit for women under benefit. In contrast, mammography is less effec-
the age of 40 years. Table 4 provides examples of tive in women with very dense breasts, and MRI
women with a family history indicative of mod- screening may offer added benefit.
erate and high risk. The online supplemental Women who have received radiation treat-
material provides guidance for accessing and ment to the chest, such as for Hodgkin disease,
using risk assessment models. compose a well-defined group that is at high risk.
Although evidence of the efficacy of MRI screen-
Women at Increased Risk Based
ing in this group is lacking, it is expected that
on Clinical Factors MRI screening might offer similar benefit as for
Additional factors that increase the risk of breast women with a strong family history, particularly
cancer, and thus may warrant earlier or more fre- at younger ages and within 30 years of treatment.
quent screening, include previous treatment with Because of the high risk of secondary breast can-
chest irradiation (eg, for Hodgkin disease), a per- cer in this group, MRI screening is recommended
sonal history of LCIS or ADH, mammographi- based on expert consensus opinion.
cally dense breasts, and a personal history of breast While lifetime risk of breast cancer for women
cancer, as discussed above. There are little data to diagnosed with LCIS may exceed 20%, the risk
assess the benefit of MRI screening in women of invasive breast cancer is continuous and only
with these risk factors. Women at increased risk moderate for risk in the 12 years following local
or who are concerned about their risk may find excision.46 Only one MRI screening study has
it helpful to have their provider clarify the bases included a select group of women with LCIS,61
84 CA A Cancer Journal for CliniciansCA Cancer J Clin 2007;57:75–89
which showed a small benefit over mammogra- and it will be necessary in the foreseeable future
phy alone in detecting cancer. This benefit was to rely on evidence of stage of disease and types
not seen in patients with atypical hyperplasia. of cancers. In the absence of randomized trials,
MRI use should be decided on a case-by-case recurrence and survival data will come from
basis, based on factors such as age, family his- observational study designs.
tory, characteristics of the biopsy sample, breast The age at which screening should be initi-
density, and patient preference. ated for women at high risk is not well estab-
Although there have been several tr ials lished. The argument for early screening is based
reported looking at the accuracy and positive on the cumulative risk of breast cancer in women
predictive value of MRI and mammography in with BRCA1 mutations and a strong family his-
women with high breast density, all of these tri- tory of early breast cancer, which is estimated
als have been conducted in women with known to be 3% by age 30 years and 19% by age
or highly-suspected malignancies within the 40 years.76 Population-based data also indicate
breast.71–74 To this point, there has been no Phase that risk for early breast cancer is increased by a
III randomized trial reported that has shown a family history of early breast cancer.16 Based on
reduction in either mortality or in the size of these observations, some experts have suggested
diagnosed breast cancer when comparing breast that breast cancer screening begin 5 to 10 years
MRI with mammography in women with high before the earliest previous breast cancer in the
mammographic density. family. In 1997, an expert panel suggested that
Scant data are available for MRI screening of screening be initiated at some time between the
women with a personal history of breast cancer. ages of 25 and 35 years for women with a
In one study, MRI detected more cancers in BRCA1 or BRCA2 mutation.77 Because these
women who had both a personal history and a recommendations were based on limited obser-
family history, compared with women at high vational data, the decision regarding when to
risk based on family history alone.75 While women initiate screening should be based on shared deci-
with a previous diagnosis of breast cancer are at sion making, taking into consideration individ-
increased risk of a second diagnosis, the ACS ual circumstances and preferences. No data are
panel concluded that the estimated absolute life- available related to the effectiveness of screening
time risk of 10% does not justify a recommenda- women beyond age 69 years with MRI and
tion for MRI screening at the present time. mammography versus mammography alone;
most of the current data are based on screening
Limitations of Evidence from MRI Studies
in younger women, and thus, similar investiga-
and Research Needs
tions are needed in older age cohorts. For most
Assiduous attempts were made to base rec- women at high risk, screening with MRI and
ommendations on solid evidence. However, out- mammography should begin at age 30 years and
come data from screening MRI studies are not continue for as long as a woman is in good health.1
sufficient to form a solid basis for many of the rec- Most of the available data are based on annual
ommendations. It was therefore necessary to rely MRI screening; there is a lack of evidence regard-
on available inferential evidence and expert opin- ing shorter or longer screening intervals. Further,
ion to provide the guidance needed for patients while good data are available for the first screen-
and their health care providers. ing exam (ie, the “prevalent screen”), consider-
Although the literature shows very good evi- ably less data are available from subsequent
dence for greater sensitivity of MRI than mam- screening exams (ie, “incidence screens”), and
mography and good evidence for a stage shift the available data include relatively short follow-
toward earlier, more favorable tumor stages by up times. Most studies of annual MRI have shown
MRI in defined groups of women at increased few interval cancers, certainly fewer than with
risk, there are still no data on recurrence or mammography. Given the probably shorter dura-
survival rates, and therefore, lead-time bias is tion of the detectable preclinical phase, or sojourn
still a concern. Further, a large randomized, mor- time, in women with BRCA mutations, MRI
tality endpoint study is unlikely to take place, has demonstrated superiority to mammography
Volume 57 • Number 2 • March/April 2007 85American Cancer Society Guidelines for Breast Screening with MRI as an Adjunct to Mammography
in this regard. Therefore, to the best of our knowl- economical to do prospective surveillance stud-
edge, MRI should be performed annually. ies since screening costs are covered by third par-
However, in view of data suggesting that tumor ties. A common surveillance protocol could
doubling time in women with an inherited risk permit pooling of data, much like presently is
decreases with age,78 it is conceivable that older done within the framework of the National
women can safely be screened less frequently than Cancer Institute’s Breast Cancer Surveillance
younger women. The available evidence is lim- Consortium, a collaborative network of seven
ited, and additional research regarding optimal mammography registries in the United States
screening interval by age and risk status is needed. with linkages to tumor and/or pathology reg-
Some experts recommend staggering MRI istries that was organized to study the delivery and
screening and mammography screening every quality of breast cancer screening and related
6 months. The potential advantage of this patient outcomes in the United States.79 We also
approach is that it may reduce the rate of inter- encourage seeking opportunities for broad inter-
val cancers. Other experts recommend MRI national research collaboration on study ques-
and mammography at the same time or within tions of common interest.
a short time period. This approach allows for Several further clinical trials of screening women
the results of both screening tests to be inter- at increased risk of breast cancer are underway,
preted together and reported to the patient at including an international study of MRI and
the same time. All of the clinical trials screened ultrasound in conjunction with the International
participants with both MRI and mammogra- Breast MRI Consortium and Cancer Genetics
phy at the same time. There is no evidence to Network, and the American College of Radiology
support one approach over the other. For the Imaging Network (ACRIN) 666 screening trial
majority of women at high risk, it is critical that of mammography compared with ultrasound.
MRI screening be provided in addition to, not An amendment to the ACRIN trial, 6666, will
instead of, mammography, as the sensitivity and screen patients with one round of MRI.
cancer yield of MRI and mammography com-
bined is greater than for MRI alone. However, CONCLUSION
where there is a concern about raised radiation
sensitivity, it may be advisable to employ MRI Often no available screening modality is
alone despite the overall lower sensitivity. uniquely ideal. For breast MRI, there is an increas-
In order to pursue answers to some of the ing body of observational data showing that screen-
unresolved questions related to the use of MRI ing can identify cancer in patients of specific risk
and mammography to screen women at increased groups, ie, high-risk patients facing a lifetime risk
risk, it is important to develop creative strategies of ~20–25% or greater related to family history as
related to data gathering and study design. estimated by one or more of the different risk
Multicenter studies can result in greater efficiency models. We have specified a range of risk because
in accumulating sufficiently large enough data estimates from the risk models vary and because
sets in this subgroup of women. Conventional each of the risk models is imperfect. Furthermore,
study designs with randomization may prove dif- these models likely will continue to be refined
ficult given the potential advantage of adding over time; therefore, these risk estimates for dif-
MRI to mammography in higher-risk groups, ferent family history profiles are likely to change.
and thus, design strategies that utilize surrogate Thus, when estimating patient risk it is impor-
markers and historic controls may prove both tant to always be certain that the most current
more practical and feasible. To move forward, model is being used. In addition to family his-
we encourage the development of a simple, com- tory, clinical factors as described earlier may be a
mon data collection protocol to capture infor- relevant factor in individualized decisions about
mation from the growing number of centers that MRI screening when family history alone does not
offer MRI and formal systems to collect out- predict a risk of approximately 20–25%.
come data. Because many insurers presently cover Several studies have demonstrated the ability
MRI screening for high-risk women, it may be of MRI screening to detect cancer with early-
86 CA A Cancer Journal for CliniciansCA Cancer J Clin 2007;57:75–89
stage tumors that are associated with better out- Medicine, University of Southern California; and
comes. While survival or mortality data are not Co-Director of USC/Norris Breast Center, Los
available, MRI has higher sensitivity and finds Angeles, CA; Barbara Andreozzi; Prevention
smaller tumors, compared with mammography, Chair, Montana Comprehensive Cancer Control
and the types of cancers found with MRI are Coalition; and Community Development Specialist,
the types that contribute to reduced mortality. Montana State University Extension Service,
It is reasonable to extrapolate that detection of Anaconda, MT; Gena R. Carter, MD; Radi-
noninvasive (DCIS) and small invasive cancers ologist, InMed Diagnostic Women’s Center,
will lead to mortality benefit. Norwell, MA; Lynn Erdman, RN, MS; Senior
The guideline recommendations for MRI Vice President, North Carolina and South Carolina,
screening as an adjunct to mammography for American Cancer Society, South Atlantic Division,
women at increased risk of breast cancer take Charlotte, NC; W. P. Evans, III, MD; Professor
into account the available evidence on efficacy of Radiology; and Director of the University of
and effectiveness of MRI screening, estimates Texas Southwestern Center for Breast Care, Dallas,
of level of risk for women in various categories TX; Herschel W. Lawson, MD; Senior Medical
based on both family history and clinical fac- Advisor, Division of Cancer Prevention and Con-
tors, and expert consensus opinion where evi- trol, Centers for Disease Control and Prevention,
dence for certain risk groups is lacking. All of Atlanta, GA; Maggie Rinehart-Ayers, PhD,
these groups of women should be offered clin- PT; Associate Professor and Director of Clinical
ical trials of MRI screening, if available. Women Education, Thomas Jefferson University, Jefferson
should be informed about the benefits, limita- College of Health Professions, Philadelphia, PA;
tions, and potential harms of MRI screening, Carolyn D. Runowicz, MD; Professor of
including the likelihood of false-positive find- Obstetrics and Gynecology; NEU Chair in
ings. Recommendations are conditional on an Experimental Oncology; and Director of the Neag
acceptable level of quality of MRI screening, Comprehensive Cancer Center, University of
which should be performed by experienced Connecticut Health Center, Farmington, CT;
providers in facilities that provide MRI-guided Debbie Saslow, PhD; Director, Breast and
biopsy for the follow up of any suspicious results. Gynecologic Cancer, Cancer Control Science
Department, American Cancer Society, Atlanta,
ACKNOWLEDGEMENT
GA; Stephen Sener, MD; Vice Chairman,
Department of Surgery, Evanston Northwestern
We thank Julie Culver, MS, CGC, for her Health Care, Evanston, IL; and Professor of Surgery,
assistance in preparing materials related to breast Northwestern University Feinberg School of
cancer risk assessment. Medicine, Chicago, IL; Robert A. Smith, PhD;
Director, Cancer Screening, Cancer Control
ACS BREAST CANCER ADVISORY GROUP
Science Department, American Cancer Society,
Atlanta, GA; William C. Wood, MD; Professor;
Members: Christy A. Russell, MD (Chair); and Chairman, Department of Surgery, Emory
Associate Professor of Medicine, Keck School of University School of Medicine, Atlanta, GA
REFERENCES 3. Gilles R, Guinebretiere JM, Toussaint C, et al. 6. Boetes C, Barentsz JO, Mus RD, et al. MR
Locally advanced breast cancer: contrast-enhanced characterization of suspicious breast lesions with a
1. Smith RA, Saslow D, Sawyer KA, et al. subtraction MR imaging of response to preoper- gadolinium-enhanced TurboFLASH subtraction
American Cancer Society guidelines for breast can- ative chemotherapy. Radiology 1994;191:633–638. technique. Radiology 1994;193:777–781.
cer screening: update 2003. CA Cancer J Clin 4. Harms SE, Flamig DP, Hesley KL, et al. MR
2003;53:141–169. 7. Liu PF, Debatin JF, Caduff RF, et al. Improved
imaging of the breast with rotating delivery of exci-
diagnostic accuracy in dynamic contrast enhanced
2. Heywang-Kobrunner SH, Bick U, Bradley tation off resonance: clinical experience with patho-
logic correlation. Radiology 1993;187:493–501. MRI of the breast by combined quantitative and
WG Jr, et al. International investigation of breast
qualitative analysis. Br J Radiol 1998;71:501–509.
MRI: results of a multicentre study (11 sites) con- 5. Fischer U, von Heyden D, Vosshenrich R, et al.
cerning diagnostic parameters for contrast-enhanced Signal characteristics of malignant and benign 8. Orel SG, Schnall MD. MR imaging of the breast
MRI based on 519 histopathologically correlated lesions in dynamic 2D-MRT of the breast (in for the detection, diagnosis, and staging of breast
lesions. Eur Radiol 2001;11:531–546. German). Rofo 1993;158:287–292. cancer. Radiology 2001;220:13–30.
Volume 57 • Number 2 • March/April 2007 87American Cancer Society Guidelines for Breast Screening with MRI as an Adjunct to Mammography
9. Kriege M, Brekelmans CT, Boetes C, et al. 25. McTiernan A, Kuniyuki A, Yasui Y, et al. aphragmatic radiation for Hodgkin lymphoma:
Efficacy of MRI and mammography for breast- Comparisons of two breast cancer risk estimates in Mayo Clinic exper ience. Mayo Clin Proc
cancer screening in women with a familial or genetic women with a family history of breast cancer. Cancer 2003;78:708–715.
predisposition. N Engl J Med 2004;351:427–437. Epidemiol Biomarkers Prev 2001;10:333–338. 42. van Leeuwen FE, Klokman WJ, Stovall M,
10. Kuhl CK, Schrading S, Leutner CC, et al. 26. Berry DA, Parmigiani G, Sanchez J, et al. et al. Roles of radiation dose, chemotherapy, and
Mammography, breast ultrasound, and magnetic Probability of carrying a mutation of breast-ovar- hor monal factors in breast cancer following
resonance imaging for surveillance of women at ian cancer gene BRCA1 based on family history. Hodgkin’s disease. J Natl Cancer Inst 2003;
high familial risk for breast cancer. J Clin Oncol J Natl Cancer Inst 1997;89:227–238. 95:971–980.
2005;23:8469–8476. 27. Antoniou AC, Pharoah PP, Smith P, Easton 43. Bhatia S, Yasui Y, Robison LL, et al. High risk
11. Leach MO, Boggis CR, Dixon AK, et al. DF. The BOADICEA model of genetic suscepti- of subsequent neoplasms continues with extended
Screening with magnetic resonance imaging and bility to breast and ovarian cancer. Br J Cancer follow-up of childhood Hodgkin’s disease: report
mammography of a UK population at high famil- 2004;91:1580–1590. from the Late Effects Study Group. J Clin Oncol
ial risk of breast cancer: a prospective multicentre 28. Petrucelli N, Daly MB, Culver JOB, et al. 2003;21:4386–4394.
cohort study (MARIBS). Lancet 2005;365: BRCA1 and BRCA2 Hereditary Breast/Ovarian 44. Tinger A, Wasserman TH, Klein EE, et al.
1769–1778. Cancer. Gene Reviews. Available at: http://www. The incidence of breast cancer following mantle
12. Lehman CD, Blume JD, Weatherall P, et al. genetests.org/query?dz⫽brca1. Accessed December field radiation therapy as a function of dose and
Screening women at high risk for breast cancer 28, 2006. technique. Int J Radiat Oncol Biol Phys 1997;
with mammography and magnetic resonance imag- 29. Roa BB, Boyd AA, Volcik K, Richards CS. 37:865–870.
ing. Cancer 2005;103:1898–1905. Ashkenazi Jewish population frequencies for com- 45. Arpino G, Laucirica R, Elledge RM. Pre-
13. Sardanelli F. Breast MR imaging in women at mon mutations in BRCA1 and BRCA2. Nat Genet malignant and in situ breast disease: biology and
high risk of breast cancer. Is something changing in 1996;14:185–187. clinical implications. Ann Intern Med 2005;
early breast cancer detection? Eur Radiol. In press. 30. Struewing JP, Hartge P, Wacholder S, et al. 143:446–457.
14. Warner E, Plewes DB, Hill KA, et al. Surveil- The risk of cancer associated with specific muta- 46. Fisher ER, Land SR, Fisher B, et al. Pathologic
lance of BRCA1 and BRCA2 mutation carriers tions of BRCA1 and BRCA2 among Ashkenazi findings from the National Surgical Adjuvant Breast
with magnetic resonance imaging, ultrasound, Jews. N Engl J Med 1997;336:1401–1408. and Bowel Project: twelve-year observations con-
mammography, and clinical breast examination. cerning lobular carcinoma in situ. Cancer 2004;
31. Antoniou A, Pharoah PD, Narod S, et al.
JAMA 2004;292:1317–1325. 100:238–244.
Average risks of breast and ovarian cancer associ-
15. U.S. Preventive Services Task Force. Genetic ated with BRCA1 or BRCA2 mutations detected 47. Li CI, Malone KE, Saltzman BS, Daling JR.
risk assessment and BRCA mutation testing for in case series unselected for family history: a com- Risk of invasive breast carcinoma among women
breast and ovarian cancer susceptibility: recom- bined analysis of 22 studies. Am J Hum Genet diagnosed with ductal carcinoma in situ and lob-
mendation statement. Ann Intern Med 2005; 2003;72:1117–1130. ular carcinoma in situ, 1988–2001. Cancer
143:355–361. 2006;106:2104–2112.
32. American Society of Clinical Oncology.
16. Claus EB, Risch N, Thompson WD. Autosomal American Society of Clinical Oncology policy 48. Boyd NF, Lockwood GA, Martin LJ, et al.
dominant inheritance of early-onset breast cancer. statement update: genetic testing for cancer sus- Mammographic densities and breast cancer risk.
Implications for risk prediction. Cancer 1994;73: ceptibility. J Clin Oncol 2003;21:2397–2406. Breast Dis 1998;10:113–126.
643–651. 49. Oza AM, Boyd NF. Mammographic paren-
33. Garber JE, Offit K. Hereditary cancer predis-
17. Palomaki GE, McClain MR, Steinort K, position syndromes. J Clin Oncol 2005;23:276–292. chymal patterns: a marker of breast cancer risk.
et al. Screen-positive rates and agreement among Epidemiol Rev 1993;15:196–208.
six family history screening protocols for breast/ 34. Schneider KA, DiGianni LM, Patenaude AF,
et al. Accuracy of cancer family histories: compar- 50. Saftlas AF, Szklo M. Mammographic paren-
ovarian cancer in a population-based cohort of chymal patterns and breast cancer risk. Epidemiol
21- to 55-year-old women. Genet Med 2006; ison of two breast cancer syndromes. Genet Test
2004;8:222–228. Rev 1987;9:146–174.
8:161–168.
35. Zbuk KM, Stein JL, Eng C. PTEN Hamartoma 51. Warner E, Lockwood G, Tritchler D, Boyd
18. Parmigiani G, Berry D, Aguilar O. Determining NF. The risk of breast cancer associated with mam-
carrier probabilities for breast cancer-susceptibil- Tumor Syndrome (PHTS). Gene Reviews. Available
at: http://www.genetests.org/query?dz⫽phts. mographic parenchymal patterns: a meta-analysis
ity genes BRCA1 and BRCA2. Am J Hum Genet of the published literature to examine the effect of
1998;62:145–158. Accessed October 17, 2006.
method of classification. Cancer Detect Prev
19. Miki Y, Swensen J, Shattuck-Eidens D, et al. A 36. Aisenberg AC, Finkelstein DM, Doppke KP, 1992;16:67–72.
strong candidate for the breast and ovarian cancer et al. High risk of breast carcinoma after irradia-
52. Barlow WE, White E, Ballard-Barbash R,
susceptibility gene BRCA1. Science 1994;266:66–71. tion of young women with Hodgkin’s disease.
et al. Prospective breast cancer risk prediction model
Cancer 1997;79:1203–1210.
20. Wooster R, Neuhausen SL, Mangion J, et al. for women undergoing screening mammography.
Localization of a breast cancer susceptibility gene, 37. Bhatia S, Robison LL, Oberlin O, et al. Breast J Natl Cancer Inst 2006;98:1204–1214.
BRCA2, to chromosome 13q12-13. Science cancer and other second neoplasms after childhood
53. Boyd NF, Byng JW, Jong RA, et al. Quantitative
1994;265:2088–2090. Hodgkin’s disease. N Engl J Med 1996;334:745–751.
classification of mammographic densities and breast
21. Amir E, Evans DG, Shenton A, et al. Evaluation 38. Hancock SL, Tucker MA, Hoppe RT. Breast cancer risk: results from the Canadian National
of breast cancer risk assessment packages in the cancer after treatment of Hodgkin’s disease. J Natl Breast Screening Study. J Natl Cancer Inst 1995;
family history evaluation and screening programme. Cancer Inst 1993;85:25–31. 87:670–675.
J Med Genet 2003;40:807–814. 39. Swerdlow AJ, Barber JA, Hudson GV, et al. 54. Brisson J, Merletti F, Sadowsky NL, et al.
22. Gail MH, Brinton LA, Byar DP, et al. Projecting Risk of second malignancy after Hodgkin’s disease Mammographic features of the breast and breast
individualized probabilities of developing breast in a collaborative British cohort: the relation to cancer risk. Am J Epidemiol 1982;115:428–437.
cancer for white females who are being examined age at treatment. J Clin Oncol 2000;18:498–509. 55. Byrne C, Schairer C, Wolfe J, et al. Mammo-
annually. J Natl Cancer Inst 1989;81:1879–1886. 40. Travis LB, Hill D, Dores GM, et al. Cumulative graphic features and breast cancer risk: effects with
23. Tyrer J, Duffy SW, Cuzick J. A breast cancer absolute breast cancer risk for young women treated time, age, and menopause status. J Natl Cancer
prediction model incorporating familial and per- for Hodgkin lymphoma. J Natl Cancer Inst Inst 1995;87:1622–1629.
sonal risk factors. Stat Med 2004;23:1111–1130. 2005;97:1428–1437. 56. Vacek PM, Geller BM. A prospective study of
24. Domchek SM, Eisen A, Calzone K, et al. 41. Wahner-Roedler DL, Nelson DF, Croghan breast cancer risk using routine mammographic
Application of breast cancer risk prediction models IT, et al. Risk of breast cancer and breast cancer breast density measurements. Cancer Epidemiol
in clinical practice. J Clin Oncol 2003;21:593–601. characteristics in women treated with supradi- Biomarkers Prev 2004;13:715–722.
88 CA A Cancer Journal for CliniciansYou can also read
