| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Breast Imaging |
1 From the Departments of Radiology (C.K.K., C.C.L., H.H.S.), Gynecology and Gynecologic Oncology (R.K.S., A.K., A.H., D.K.), Pathology (E.W., U.P.), and Human Genetics (M.M.), University of Bonn Medical Center, Sigmund-Freud-Strasse 25, D-53105 Bonn, Germany. Received November 2, 1998; revision requested December 17; final revision received August 30, 1999; accepted September 14. C.K.K., R.K.S. supported in part by a grant from the Deutsche Krebshilfe, and C.K.K. supported in part by a grant from the Förderverein Radiologie an der Universität Bonn. Address reprint requests to C.K.K.
 |
Abstract |
|---|
 TOP Abstract IntroductionMATERIALS AND METHODSRESULTSDISCUSSIONReferences |
|---|
MATERIALS AND METHODS: This prospective trial included 192 asymptomatic and six symptomatic women who, on the basis of personal or family history or genetic analysis, were suspected or proved to carry a breast cancer susceptibility gene.
RESULTS: Fifteen breast cancers were identified: nine in the 192 asymptomatic women (six in the first and three in the second screening round) and six in the symptomatic patients. Concerning the asymptomatic women, four of the nine breast cancers were detected and correctly classified with mammography and ultrasonography (US) combined; another two cancers were visible as well-circumscribed masses and were diagnosed as fibroadenomas. MR imaging allowed the correct classification and local staging of all nine cancers. In 105 asymptomatic women with validation of the 1st-year screening results, the sensitivities of mammography, US, and MR imaging were 33%, 33% (mammography and US combined, 44%), and 100%, respectively; the positive predictive values were 30%, 12%, and 64%, respectively.
CONCLUSION: The accuracy of MR imaging is significantly higher than that of conventional imaging in screening high-risk women. Difficulties can be caused by an atypical manifestation of hereditary breast cancers at both conventional and MR imaging and by contrast material enhancement associated with hormonal stimulation.
Index terms: Breast, MR, 00.121412, 00.12143 • Breast neoplasms, diagnosis, 00.30, 00.311, 00.32 • Breast neoplasms, MR, 00.121412, 00.12143 • Breast neoplasms, radiography, 00.11 • Breast neoplasms, US, 00.1298 • Cancer screening • Genes and genetics
|
Introduction |
|---|
|
TOPAbstractIntroductionMATERIALS AND METHODSRESULTSDISCUSSIONReferences |
|---|
An important feature of familial breast cancer is the patient's age at diagnosis. As opposed to women with sporadic breast cancer, women with a BRCA mutation develop breast cancer at a significantly younger age and, accordingly, more often in the premenopausal period. By the age of 50 years, more than 50% of the BRCA1 or BRCA2 mutation carriers have already developed the disease (10,11). Accordingly, the current recommendations for breast cancer screening may not be sufficient for gene carriers. Because of the high risk of developing breast cancer and the early onset of the disease, close screening examinations of proved or suspected gene carriers should start at a substantially earlier age than is recommended for the general population, that is, at the age of 25–30 years.
There is a high degree of uncertainty concerning the choice of adequate screening modalities to be offered to these women. Mammography is limited in the diagnosis of breast cancer in very young patients because of the very dense breast tissue in this age group (12–18). Moreover, on the basis of findings of in vitro studies, there is evidence that breast tissue harboring a BRCA mutation might be more vulnerable to ionizing radiation than the genetically intact parenchyma (19–25).
Magnetic resonance (MR) imaging of the breast has been shown to be sensitive for invasive breast cancer (26–33); its sensitivity is not impaired by dense parenchyma. In addition, MR imaging is not associated with ionizing radiation. It is, however, a cost-intensive modality. By the time this article was written, breast MR imaging had not been used in a screening setting, where because of the low prevalence of subjects with disease not only maximum sensitivity but also high predictive values are required. Particularly in young premenopausal subjects, breast MR imaging studies have been shown to be difficult to interpret; because spontaneous, hormone-induced enhancement may arise, false-positive findings may result, and, as a result, unnecessary biopsies may be performed (34–36).
The German Cancer Aid Society (Deutsche Krebshilfe) has initiated a conjoined multiinstitutional trial on familial breast cancer diagnosis and treatment. It serves to establish the identification, genetic testing, registration, and oncologic care of women who are suspected to carry a breast cancer suceptibility gene. Within the scope of this trial, we performed a prospective comparative study on the value of breast MR imaging screening in these women; as a pilot study, it was restricted to a single study site.
The objectives addressed by this pilot study were the following:
1. Does breast MR imaging screening allow subclinical breast cancer diagnosis in women with familial breast cancer?
2. How does breast MR imaging compare with mammography and high-frequency breast ultrasonography (US) in screening in these high-risk women? What are the diagnostic indexes (sensitivity, specificity, positive and negative predictive values, diagnostic accuracy) of breast MR imaging screening compared with those of mammographic and breast US screening? Specifically, does breast MR imaging screening increase the rate of biopsies performed because of false-positive findings in these patients?
3. Is the breast MR imaging protocol that is in use for the diagnosis of sporadic breast cancer suitable for patients with familial breast cancer?
4. What are the MR imaging features of breast cancers in patients with familial breast cancer? Are there differences when compared with features in patients with regular, sporadic breast cancers?
5. Is it possible to derive recommendations concerning the selection of women who should be enrolled in breast MR imaging screening?
6. Is it possible to derive recommendations concerning adequate screening intervals?
In the following, we report on the preliminary results of a first breast MR screening round in 192 women and of follow-up screening rounds in 101 of the 192 women who on the basis of personal and/or family history or genetic testing results were suspected or proved to carry a breast cancer susceptibility gene. For comparison, we report data on a group of six patients with familial breast cancer who were symptomatic at the date of presentation. It should be noted that the cost-effectiveness of screening in this high-risk cohort was not addressed in this study.
|
MATERIALS AND METHODS |
|---|
|
TOPAbstractIntroductionMATERIALS AND METHODSRESULTSDISCUSSIONReferences |
|---|
The study was scheduled for a period of 5 years. The recruitment of women at high risk for breast cancer was initiated in March 1996; the data presented here stem from a preliminary analysis performed after the first 2.5 years (30 months). By October 1998, 256 women had been recruited who met the inclusion criteria listed in the next section. Of these, 192 underwent screening examinations in the first round, and 101 underwent examinations in the second and, in part, further rounds; this yielded a total of 363 breast MR imaging studies.
For comparison, we also provide data from six patients who showed signs or symptoms of breast cancer and who were referred because of a specific family history.
Patient Selection and Inclusion and Exclusion Criteria
Asymptomatic women were enrolled by the Gynecologic Oncology Section, Familial Breast Cancer Study Group. Women were eligible if, on the basis of personal and/or family history data, the high-risk status for familial breast cancer was confirmed by the gynecologic oncologist (R.K.S., A.K.) in cooperation with the geneticist (M.M.). Positive mutational analysis findings are not required for this diagnosis, because at present only the sequences of the BRCA1 and BRCA2 genes are known and are amenable to genetic testing. However, only about 50% of the familial breast cancer cases are due to mutations at the BRCA1 or BRCA2 site. A negative mutational analysis finding can therefore by no means prove the absence of familial (hereditary) breast cancer genes. Instead, historical data are used to assess the individual woman's risk of carrying a susceptibility gene for familial breast cancer (9,10).
In accordance with the guidelines established by the interdisciplinary research project on familial breast cancer by the Deutsche Krebshilfe, the following entrance criteria were used: a personal history or a history of a relative with breast cancer diagnosed at or before the age of 35 years; a personal history or a history of a relative with ovarian cancer diagnosed at or before the age of 40 years; a personal history or a history of a relative with bilateral breast cancer; a personal history or a history of a relative with both breast and ovarian cancers; a history of at least two relatives with breast and/or ovarian cancer, one of whom received a diagnosis at or before the age of 50 years; or a man with a personal history of breast cancer or a history of a male relative with breast cancer.
In addition to the women who underwent screening, six patients with symptoms were recruited. Here, a young patient age (n = 2) or a suggestive family history (n = 4) prompted the referral. Symptoms included a carcinoma from an unknown primary cancer with bone metastases in one patient and palpable breast lumps in five patients. In one of these patients, the palpable lump proved in fact unrelated to the breast cancer that was present in another quadrant of the same breast.
Genetic Testing
Asymptomatic women underwent free gynecologic, psychologic, and genetic counseling; only thereafter and only if the woman herself explicitly consented to be tested, the genetic mutational analysis was performed. Genetic testing for the BRCA1 and BRCA2 mutations was performed by the departments of gynecologic oncology and of human genetics. Testing included protein truncation testing of the large exons (exon 11 of BRCA1 and exons 10 and 11 of BRCA2) and direct sequencing of the remaining exons. The sensitivity of this approach is estimated to be 95% (37).
Informed Consent and Review Board Approval
The study design and protocol were reviewed and approved by the authors' institutional review board; all patients gave informed consent to be examined after the nature of the procedure had been fully explained.
Study Protocol
The study protocol consisted of clinical examinations, high-frequency breast US, mammography, and breast MR imaging. Screening was begun at the age of 25 years or at an age 5 years prior to the earliest onset of the disease in the family (whichever occurred earlier).
The imaging follow-up protocol differed slightly with patient age: The 75 study patients 40 years of age or older underwent a yearly triple assessment with US, mammography, and breast MR imaging plus additional US every 6 months. The 117 patients 39 years of age or younger underwent the same follow-up protocol, with one exception: On the basis of the parenchymal density pattern on the baseline mammogram, the radiologist decided whether the mammographic density was such that a reasonable mammographic sensitivity could be expected and whether further screening mammograms were indicated before the age of 40 years (38).
Patients in both age groups underwent an extra complete triple assessment (US, mammography, and breast MR imaging) in case any abnormality was noticed at any time by the patient or during the half-yearly clinical visits or the half-yearly US screening studies. Imaging in premenopausal patients was performed during the 2nd week of the menstrual cycle to reduce cyclical phase–induced breast densities or enhancing areas (34–36).
Bilateral two-view mammography (Mammomat 3000; Siemens Medical Systems, Erlangen, Germany) was performed in craniocaudal and mediolateral oblique projections, with spot compression and additional views obtained where appropriate. Mammographic findings were reported by using the American College of Radiology Breast Imaging Reporting and Data System (BI-RADS) categories (39).
High-frequency breast US (40) was performed by using a 7.5-MHz or a 10–12-MHz probe (Elegra, Siemens Medical Systems; and Esaote Biomedica, Munich, Germany). Both breasts were systematically inspected for suspicious lesions, that is, lesions with posterior acoustic shadowing, irregular morphology or unsharp borders, a vertical growth pattern (perpendicular to the surface of the pectoral muscle), or echogenic material within dilated ducts. The reports were coded in a pattern equivalent to the BI-RADS categories (39).
Details of MR image analysis are given in the dedicated section on this issue. To allow assessment of the individual contribution of each of the imaging modalities, all imaging studies (breast MR imaging studies, mammograms, and breast US scans) were read prospectively and independently of each other. Breast MR studies and mammograms were each read in consensus by two radiologists experienced in the respective modality. The respective readers (C.K.K., C.C.L., A.H., H.H.S.) were aware of the high-risk status of the patients, but they were blinded to possible clinical findings or the results of other imaging modalities.
Breast MR Imaging Protocol
Breast MR imaging was performed with a standard 1.5-T system (Gyroscan ACS II; Philips Medical Systems, Best, the Netherlands) by using a regular double-breast surface coil. The imaging protocol corresponded to the standard bilateral, dynamic, subtraction technique, as previously published (34). It consisted of two-dimensional, gradient-echo, contrast material–enhanced, dynamic imaging (250/4.6 [repetition time msec/echo time msec], 90° flip angle) before and nine times after bolus injection of 0.1 mmol of gadopentetate dimeglumine (Magnevist; Schering, Berlin, Germany) per kilogram of body weight, with a section thickness of 3–4 mm, a full 256 x 256 imaging matrix, and a field of view adjusted to cover both breasts (260–320 mm).
Each of the 10 dynamic image acquisitions consisted of a stack of 21–31 sections that were carefully positioned to cover the entire parenchymal volume. If contrast-enhancing lesions were identified on the subtraction images, region-of-interest–based time–signal intensity curves were plotted to show the enhancement behavior during the dynamic study. The time course of the time–signal intensity curve was assigned to three different types, as described in reference 41, as a steady, a plateau, or a washout type. By using the early postcontrast subtraction images, maximum intensity projection images were calculated.
MR Imaging Study Analysis
Breast MR imaging studies were read by C.K.K. and C.C.L. in consensus. Diagnostic coding was equivalent to the BI-RADS classification. Criteria to distinguish between benign and malignant contrast-enhancing lesions were based on both morphology and enhancement kinetics (41). The following criteria were considered: overall lesion configuration, lesion margins, internal architecture (in particular, the presence of internal septations), wash-in rates, and the time course of signal intensity changes in the intermediate and late postcontrast periods.
A BI-RADS category 1 equivalent, or "nothing special to report," was attributed in cases without any contrast material enhancement.
A BI-RADS category 2 equivalent, or benign enhancement, was assigned in cases where enhancement was detectable but where it was rated as benign (focal masses with well-circumscribed morphology, internal septations but otherwise homogeneous enhancement, steady time course of the time–signal intensity curve, and centrifugal progression of enhancement; or non–mass-related gradual regional or patchy enhancement). Typical fibroadenomas and simple fibrocystic changes were grouped into this category.
A BI-RADS category 3 equivalent, or probably benign enhancement with follow-up required, was assigned in cases that were compatible with "unidentified bright breast objects," or UBOs (spontaneous, hormone-induced enhancement; see references 34 and 35 for a description), and in cases with presumably benign masses that lacked some of the BI-RADS category 2 features.
A BI-RADS category 4 equivalent, or possibly malignant with biopsy recommended, was assigned in lesions with a washout time course, irrespective of morphologic features, and in cases with suspicious morphology, irrespective of enhancement kinetics. Suspicious morphology was present in cases with stellate or irregular lesion configuration, heterogeneous internal architecture (particularly rim enhancement), and asymmetric segmental or linear enhancement compatible with ductal carcinoma in situ (DCIS).
Finally, a BI-RADS category 5 equivalent, or probably malignant with appropriate action to be taken, was attributed in cases where both morphologic and kinetic features were suggestive of a malignant lesion.
Establishing the Final Diagnosis
The final diagnosis was established by means of core-needle and/or excisional biopsy in 27 patients. If a lesion was detected with MR imaging alone, a hook wire was placed preoperatively with MR imaging guidance to guide the subsequent excisional biopsy (42).
In 101 of the remaining 165 patients, the MR diagnoses were confirmed with follow-up examination findings. At the time of data analysis, the follow-up interval was 1 year in 25 patients, 2 years in another 56 patients, and more than 2 years in 20 patients. In addition, all patients were receiving clinical, US, or mammographic follow-up, as described.
For the calculation of the diagnostic indexes of the different imaging modalities under investigation (breast MR, mammography, breast US), only those patients were considered in whom a verification of the breast MR imaging diagnosis was obtained with excisional biopsy (n = 27) and/or breast MR imaging follow-up (n = 101). Twenty-three patients underwent both biopsy and breast MR follow-up studies. So, at the time this article was written, the total number of patients in whom a validation of the first breast MR screening study was available equaled 105.
Statistical Analysis
Data were analyzed by using SPSS software package 6.13 (SPSS, Chicago, Ill). On the basis of the data for the 105 patients with validated MR diagnoses, the true-positive, true-negative, false-positive, and false-negative rates of the different imaging modalities (mammography, US, and MR imaging) were determined. The Bayessian parameters of sensitivity, specificity, positive and negative predictive values, and diagnostic accuracy were calculated.
The 
2 and McNemar tests were used to evaluate the accuracy of breast MR imaging as opposed to the accuracy of the conventional imaging modalities (ie, mammography and breast US combined). A P value of less than .05 was regarded as indicating a statistically significant difference.
|
RESULTS |
|---|
|
TOPAbstractIntroductionMATERIALS AND METHODSRESULTSDISCUSSIONReferences |
|---|
Diagnoses
Surveys of clinical, mammographic, US, and breast MR imaging findings in patients with malignant lesions are given in Tables 1 (asymptomatic women) and 2 (symptomatic women).
|
|
Symptomatic women.—Breast cancer was identified in all six patients. Four had palpable cancer. In one patient, the palpable mass that had prompted referral to our center turned out to be a fibroadenoma, but a clinically occult invasive lobular cancer was present in another quadrant of the same breast. One patient presented with bone pain, which was caused by bone metastases from an unknown primary cancer. In this patient, MR imaging revealed a multicentric invasive breast cancer that turned out to be the searched-for primary cancer.
TNM stages of breast cancers.—All nine patients with breast cancers detected purely with screening had tumors that corresponded to a pTis through pT1 tumor stage; the average size of the invasive screening-detected breast cancers was 10.5 mm. All nine patients had negative lymph node findings (pN0) at extensive axillary dissection (Figs 1, 2). In the six symptomatic patients, the breast cancers corresponded to a pT1c tumor stage in three patients, a pT2 tumor stage in two, and a pT3 tumor stage in one. The average size of breast cancers in symptomatic patients was 21.6 mm. Metastatic disease was present in three of the six patients (positive lymph nodes in two and positive lymph nodes plus distant [bone] metastases in one patient).
|
|
|
|
|
|
|
|
|
|
In four of the six symptomatic patients, mammography did show breast cancer and helped correctly classify the palpable lump in one of the patients as probably benign (fibroadenoma). It did not show, however, the small breast cancer in another quadrant of the same breast and the multicentric breast cancer that was the primary cancer in the patient who had bone metastases and an unknown primary carcinoma at presentation.
US examination.—The 192 asymptomatic women had 19 false-positive breast US results. With consideration of the subgroup of 105 patients who underwent screening and had validated diagnoses, US did depict and help correctly classify breast cancer in two patients. However, it did not show the multicentric foci in one of the two patients. In one patient, US did depict a lesion, and the prospective rating was that of either a complicated cyst or a papilloma within a cyst (BI-RADS category 0); because breast MR imaging was recommended for clarification, the case can be regarded as true-positive. In one patient with breast cancer, US did depict a well-circumscribed mass, but it was rated as probably benign (fibroadenoma). In five patients with biopsy-confirmed breast cancer, the lesion was not identified at US. Accordingly, US findings were true-positive in three, true-negative in 77, false-positive in 19, and false-negative in six patients.
In the six symptomatic patients, the three palpable breast cancers were correctly classified with US. A palpable lump was correctly classified as a fibroadenoma in one patient, but US did not show the breast cancer in the same breast in this patient. In addition, the primary cancer was not detected with US in the patient with bone metastases.
Mammography and US combined.—With mammography and breast US combined, four of the nine asymptomatic women and four of the six symptomatic patients had breast cancer detected and correctly classified. If we disregard the underestimation of disease extent in the asymptomatic patients, this adds up to sensitivities of 44% (four of nine patients) for the asymptomatic women and 67% (four of six patients) for the symptomatic women.
Breast MR imaging.—In five of the 192 asymptomatic women, biopsy was performed because of a false-positive MR imaging finding (three with fibroadenoma, one with sclerosing adenosis, and one with atypical ductal hyperplasia). With consideration of the subgroup of 105 patients who underwent screening and in whom a validation of the first screening round findings exists, all nine patients with breast cancer had detection and correct classification with breast MR imaging. Moreover, the multicentric foci in four of the nine patients were correctly identified and changes were made in the treatment plan, in that primary mastectomy was performed. One of the patients with DCIS did not exhibit calcifications at mammography but had a prospective breast MR imaging diagnosis of intraductal cancer owing to its segmental enhancement pattern. Thus, the following results emerged: Breast MR imaging was true-negative in 91 patients, true-positive in nine patients, false-positive in five patients, and false-negative in none of the patients.
In all six symptomatic patients, breast cancer was depicted and correctly classified with breast MR imaging. The searched-for primary cancer (including the multicentric foci) was correctly identified in the patient with bone metastases. The palpable lump in one patient was correctly classified as a fibroadenoma and the unexpected, incidental, invasive breast cancer in another quadrant of the same breast was identified. Two of the three symptomatic patients with small-stage disease (pT1) had a diagnosis with breast MR imaging alone.
Statistical analysis.—By using the 2 test to compare the diagnoses with conventional imaging studies versus the diagnoses with breast MR imaging studies, a 2 of 5.08 was obtained, which suggests that breast MR imaging was significantly more accurate in this cohort, with a P value of .012. The McNemar test results revealed a 2 of 21.55, which indicated that breast MR imaging diagnoses were correct significantly more often than were conventional methods (P < .003).
Prevalence of Contrast-enhancing Areas in the First Screening Round
With consideration of the subgroup of 105 patients with validated diagnoses, 91 were determined to be cancer-free on the basis of the first screening breast MR imaging study findings. Of these 91 patients, 47 (52%) showed no enhancement or only gradual diffuse enhancement. In the remaining 44 patients (48%), the breast MR imaging study was categorized as benign or probably benign in spite of the presence of contrast-enhancing areas (Figs 3, 4). These contrast-enhancing areas were classified as probably representing hormone-induced or cyclical phase–induced enhancement, solitary or multiple fibroadenomas, benign fibrocystic changes, or intramammary lymph nodes. The prevalence of the different enhancing areas was as follows: solitary focal enhancement, 19 (21%) of 91 patients; multifocal spotty or patchy enhancement, 22 (24%) of 91 patients; and asymmetric regional (but not segmental) enhancement, three (3%) of 91 patients.
|
|
|
|
|
In the second screening round, short-term follow-up studies were requested in seven patients. In one of these patients, a follow-up MR imaging study performed after 4 months showed an increasing lesion size. Biopsy results revealed an invasive ductal cancer, pT1c, N0. At this stage, the tumor also was evident at mammography and breast US.
Genetic Status in Patients with a Diagnosis of Breast Cancer
At the time of data analysis, complete testing for the BRCA1 and BRCA2 genes had been performed in 54 of the 192 study participants. A mutation was documented at the BRCA1 site in 22 patients and at the BRCA2 site in 13 patients.
Of the nine asymptomatic women with breast cancer detected, seven had mutations at the BRCA1 (n = 6) or BRCA2 site (n = 1). One patient with DCIS refused to undergo testing; in the last patient, test results were pending at the time this article was written.
Imaging Features of Breast Cancers and BRCA Status
Of the 13 patients with invasive and the two patients with in situ cancers, seven had cancers with an atypical manifestation on all imaging studies, including breast MR imaging studies.
At mammography, suspicious microcalcifications were present in only three of the 15 patients with breast cancer (one with DCIS, two with invasive cancers). A completely round or ovoid, well-circumscribed mass was present in five patients; in two of these, the appearance corresponded to that of a typical fibroadenoma.
In breast US, the same five well-circumscribed lesions appeared as mobile, ovoid, hypo- or anechoic masses, with their long axes parallel to the surface of the pectoral muscle; in two cases, the lesions were compressible.
At breast MR imaging, the same five lesions corresponded to round or ovoid masses with smooth borders, and three of the five had a completely homogeneous internal architecture. Enhancement kinetics allowed the correct diagnosis in all five patients with well-circumscribed breast cancers because of a pronounced washout phenomenon in three patients and a markedly heterogeneous internal enhancement pattern in two patients. In four of 15 malignant lesions, gradual enhancement was present, but an irregular morphology (n = 2) or a heterogeneous internal architecture (n = 2) and/or washout time-course kinetics (n = 2) allowed the diagnosis. In the three patients with pure DCIS or with DCIS associated with invasive breast cancer, lesions manifested as typical segmental enhancement at breast MR imaging.
Of the five patients with breast cancers with a fibroadenoma-like morphology, four had a mutation at the BRCA1 site identified and one had contraindications for genetic testing (as revealed at pretest psychiatric exploration). A BRCA1 mutation was present in two of the patients who had cancer with gradual enhancement and irregular morphology. A BRCA2 mutation was present in only one of the patients with breast cancer so far; it corresponded to the DCIS that lacked microcalcifications and was diagnosed at breast MR imaging owing to its segmental enhancement. Results of genetic tests in the remaining patients with cancers were still pending at the time of data collection.
|
DISCUSSION |
|---|
|
TOPAbstractIntroductionMATERIALS AND METHODSRESULTSDISCUSSIONReferences |
|---|
The first data we know of that show that preventive bilateral mastectomy is effective became available in an article recently published in the New England Journal of Medicine (48); yet, it should be noted that in this study, prophylactic mastectomy was compared with "doing nothing" (particularly, no surveillance) in the reference cohort. According to current recommendations, preventive mastectomy is not recommended as a first-line option (49–53); it is chosen depending on the individual risk assessment and on sociopsychologic considerations. So, the most widely used way to care for these patients is lifelong surveillance ("secondary prevention"). This study was aimed at elucidating the value of breast MR imaging as a screening tool in these high-risk women.
The preliminary results of this pilot study suggest that breast MR imaging is a useful screening tool in women who are suspected or proved to be carriers of breast cancer susceptibility genes. Among the three imaging modalities under investigation—mammography, high-frequency breast US, and breast MR imaging—only breast MR imaging allowed the accurate detection, correct classification, and correct local staging of all breast cancer cases: It depicted breast cancer in three asymptomatic women in whom mammographic results had been completely negative or unrelated to the breast cancer that was actually present. In another three asymptomatic women with mammographically and ultrasonographically nonsuspicious or probably benign findings, breast MR alone correctly revealed the actual disease extent in all four patients with multicentric breast cancer.
However, it should be kept in mind that these data are based on a still small number of breast cancers. In particular, multiinstitutional trials are needed to verify these results in a larger cohort of asymptomatic women and thus in a larger number of women with breast cancer. Moreover, it should be well understood that the validation of the diagnosis is based on a follow-up period of merely 1 year in 25 of 105 asymptomatic women, such that a definitive exclusion of false-negative diagnoses was not possible at the time of data analysis.
Although of course the sample size was still small, it is encouraging that the TNM stage of the screening-detected breast cancers was lower than that of their clinically evident counterparts: All nine patients with screening-detected breast cancers had tumor stages of pTis through pT1c, N0, M0. The average size of the screening-detected BRCA-induced breast cancers in this study (10.5 mm) was substantially smaller than the average size of BRCA-induced breast cancers as published in the literature (24 mm) (46) and the average size of breast cancer in our six symptomatic patients (21.6 mm). In contrast with the patients with screening-detected breast cancers, three of the six patients in the symptomatic group with breast cancers identified had stage pT2–3, with positive lymph nodes in three of the six patients. Of the three symptomatic patients with small-stage (pT1) breast cancer, two had cancers detected with breast MR imaging alone.
In addition, the low rate of five false-positive findings that resulted in biopsy is encouraging. In our view, this is the most important finding of this preliminary study: While the very high sensitivity of breast MR concerning the detection of breast cancer is well recognized (27–29,32), the modality's limited specificity and, notably, its allegedly high number of false-positive findings have led to the estimation that breast MR may not be suitable for screening.
In this study, the lowest number of biopsies performed because of false-positive findings was in breast MR imaging—in spite of the high prevalence of contrast material enhancement of benign lesions in our mostly premenopausal cohort (49% [47 of 96 patients]). On the basis of findings of previous studies on breast MR imaging in young women (34–36), the enhancing areas could be correctly classified as representing probably benign or physiologic (hormone-induced) enhancement. In 19 patients in whom a definite classification of enhancing areas was not possible right away, short-term follow-up breast MR studies were used for clarification (10% [19 of 192 patients]).
Probably at least in part because of the restrictive indication to perform biopsy in "MR-only" lesions (lesions seen only on MR imaging studies), the specificity for breast MR in this screening study (95% [91 of 96 patients]) was considerably higher than specificities published for the use of breast MR in symptomatic patients (these include our own experiences); the specificity compared favorably with that for mammography (93% [89 of 96 patients]) in the same high-risk population, and both specificities were equivalent to the published specificity for screening mammography in non-high-risk patients (13).
If we had recommended biopsy in all equivocal cases that were managed with follow-up, this would have halved the positive predictive value for MR from its actual 64% (nine of 14 patients) to 27% (nine of 33 patients) and reduced the specificity from 95% (91 of 96 patients) to 75% (72 of 96 patients). Accordingly, short-term follow-up studies can help reduce the number of decisions to perform biopsy because of false-positive breast MR imaging findings in premenopausal women (Figs 3, 4). Yet, in view of the rapid growth and the short lead time of hereditary breast cancers, this strategy is associated with the risk of inadvertently delaying cancer diagnoses, and thus one may question the success of the screening efforts as a whole. Therefore, the best way to deal with spontaneous, hormone-induced enhancement in breast MR imaging is yet to be found.
The breast MR technique used here (bilateral, dynamic, subtraction breast MR imaging) seems to work in the high-risk patient group. The bilateral approach is reasonable in a screening setting; imaging in a dynamic mode with high spatial resolution permits the use of diagnostic criteria related to morphology and to enhancement kinetics. This is critical in view of the imaging manifestation of familial breast cancer.
Of the 13 invasive cancers, as many as seven had an atypical manifestation on breast MR imaging studies and—if at all visible—on conventional images. In particular, BRCA1–induced breast cancers could manifest with morphologic and architectural features identical to those of fibroadenomas, that is, with benign morphology and absence of microcalcifications at mammography; with benign configuration or borders and mobility at breast US; and with a roundish configuration, smooth or lobulated borders, and high signal intensity on T2-weighted images or entirely homogeneous enhancement on breast MR imaging studies (Fig 2).
This is in good agreement with recently published results of specific histopathologic features of BRCA1-induced breast cancers (54–56): These cancers tend to appear as well-circumscribed lesions with pushing margins and lobulated borders and to have a high prevalence of the medullary type. Accordingly, BRCA1 cancers can look like fibroadenomas, and they arise in an age group where, also in high-risk women, fibroadenomas are much more prevalent than breast cancers. Mammography and breast US—both of which rely on morphologic criteria—failed to correctly help classify these cancers. Breast MR, however, allowed evaluation of lesion morphology and of enhancement kinetics. In cases with deceivingly benign morphology, kinetic features allowed the correct diagnosis of breast cancer.
Still, in view of the apparent difficult differential diagnosis of lesions in our first screening round and with the increasing availability of MR-compatible biopsy systems, we regularly perform large-core stereotactic biopsy with mammographic or MR guidance in lesions with questionable findings, particularly in well-circumscribed masses seen at mammography or breast MR imaging.
In conclusion, concerning the objectives of the study, the following statements can be made:
1. There is evidence that breast MR imaging allows early breast cancer diagnosis in women with familial breast cancer.
2. In this preliminary study (ie, with a still small number of breast cancers), breast MR imaging compared favorably with mammography and high-frequency US. MR imaging screening increased the sensitivity for breast cancer detection as compared with the sensitivities of mammography and breast US while maintaining specificity. It should be well understood that the high specificity is in part due to our tendency to recommend short-term follow-up instead of immediate biopsy of MR-only lesions. Moreover, owing to the relatively short follow-up interval that was used to validate the diagnoses in 25 of 105 patients, these data have to be considered as "under investigation" until more data are available.
3. The bilateral dynamic breast MR imaging protocol used here seems to work for high-risk screening. The follow-up protocol with adjusted screening intervals for US, mammography, and breast MR imaging seems to be an acceptable compromise of the diverging demands for maximum sensitivity, cost containment, and limited radiation exposure.
4. Breast MR screening in this young, high-risk cohort was difficult for two reasons: First, the imaging features of BRCA-induced breast cancers seemed atypical when compared with the average appearance of sporadic breast cancers. In particular, BRCA1-induced breast cancers could exhibit a fibroadenoma-like appearance, such that only kinetic features allowed the correct diagnosis. Second, diagnostic difficulties due to spontaneous contrast material enhancement (unidentified bright breast objects) occurred. We conclude that (a) morphologic and kinetic criteria have to be considered before a final diagnosis is made and that (b) even then, the timely diagnosis of BRCA-induced breast cancers (and the differentiation of malignant from hormone-induced enhancement) is difficult and requires specific expertise with MR imaging in premenopausal patients. Consequently, (c) imaging-guided core-needle biopsy should be considered generously until more data are available on imaging findings in BRCA-induced breast cancer. An important problem that remains is the management of the high-risk case with single or multiple, small (<5-mm), MR-only lesions. The positive predictive value of MR imaging for small enhancing lesions in premenopausal women is low, and they are hardly amenable to MR-guided core-needle biopsy; follow-up is associated with the risk of delaying breast cancer diagnosis. Accordingly, the best way to deal with this issue has yet to be found.
5. Our data do not yet justify recommendations as to the selection of high-risk patients who should undergo breast MR screening. What can be stated at this time is that seven of nine women with screening-detected breast cancers had a BRCA mutation.
6. In contrast with the initial data published on the nature of BRCA-induced breast cancers (57,58) and in accordance with more recent data (45,54–56), the tumors that we have seen were highly aggressive cancers with very rapid growth. If more of these aggressive tumors with short lead time are encountered, then a screening interval of even 1 year could be too long. Further comparative studies with different (possibly risk-adjusted) screening intervals will have to be performed to clarify this issue.
|
Acknowledgments |
|---|
|
Footnotes |
|---|
Author contributions: Guarantors of integrity of entire study, C.K.K., H.H.S.; study concepts and design, C.K.K.; definition of intellectual content, D.K., U.P., H.H.S.; literature research, C.K.K., R.K.S.; clinical studies, C.K.K., R.K.S., C.C.L., A.K., A.H., E.W., M.M.; data acquisition, C.K.K., R.K.S., C.C.L., A.K., A.H.; data analysis, C.K.K., H.H.S.; statistical analysis, C.K.K.; manuscript preparation and editing, C.K.K.; manuscript review, R.K.S., C.C.L., A.K., A.H., E.W., D.K., U.P., H.H.S., M.M.
|
References |
|---|
|
TOPAbstractIntroductionMATERIALS AND METHODSRESULTSDISCUSSIONReferences |
|---|
