Research Article
Breast Imaging
October 2004

Sensitivity of MRI Versus Mammography for Detecting Foci of Multifocal, Multicentric Breast Cancer in Fatty and Dense Breasts Using the Whole-Breast Pathologic Examination as a Gold Standard

Abstract

OBJECTIVE. Our aim was to compare the effectiveness of mammography and MRI in the detection of multifocal, multicentric breast cancer.
SUBJECTS AND METHODS. Ninety patients with planned mastectomies (nine bilateral) underwent mammography and dynamic gadolinium-enhanced MRI. Off-site reviewers aware of the entry criterion (planned mastectomy) evaluated both examinations for the presence of malignant foci, recording the density pattern on mammography. The gold standard was pathologic examination of the whole excised breast (slice thickness, 5 mm).
RESULTS. Of 99 breasts, pathologic findings revealed 52 unifocal, 29 multifocal, and 18 multicentric cancers for a total of 188 malignant foci (158 invasive and 30 in situ). Overall sensitivity was 66% (124/188) for mammography and 81% (152/188) for MRI (p < 0.001); 72% (113/158) and 89% (140/158) for invasive foci (p < 0.001); and 37% (11/30) and 40% (12/30) for in situ foci (p > 0.05, not significant), respectively. Mammography and MRI missed 64 and 36 malignant foci, respectively, with median diameters of 8 and 5 mm (p = 0.033) and an invasive–noninvasive ratio of 2.4:1 (45:19) and 1.0:1 (18:18) (p = 0.043), respectively. The overall positive predictive value (PPV) was 76% (124/164) for mammography and 68% (152/222) for MRI (not significant). In breasts with an almost entirely fatty pattern, sensitivity was 75% for mammography and 80% for MRI (not significant), and the PPV was 73% and 65% (not significant), respectively. In breasts with fibroglandular or dense pattern, the sensitivity was 60% and 81% (p < 0.001), and the PPV was 78% and 71% (not significant), respectively.
CONCLUSION. MRI was more sensitive than mammography for the detection of multiple malignant foci in fibroglandular or dense breasts. Mammography missed larger and more invasive cancer foci than MRI. A relatively low PPV was a problem for both techniques.

Introduction

Early diagnosis of breast cancer has increasingly resulted in a more conservative surgical approach to the disease. However, this approach should be based on the exclusion of undetected malignant foci in the breast [1, 2]. In pathology-controlled studies, multifocal (one quadrant involved) or multicentric (two or more quadrants involved) [3, 4] cancer occurs with a frequency ranging between 14% and 47%. In particular, Holland et al. [4] reported that a second invasive or in situ lesion was found at a distance of 1 cm from the margins of the first lesion in 59% of cases, at a distance of 2 cm in 42% of cases, at a distance of 3 cm in 17% of cases, and at a distance of 4 cm in 10% of cases. The distance between malignant foci can be also used to define multifocality (≤ 5 cm) or multicentricity (> 5 cm). Relapses after conservative surgery are frequently due to undetected malignant foci [1, 5].
The sensitivity of mammography for detecting multiple malignant foci frequently depends on the setting but is often less than 50% [511]. With gadolinium-enhanced dynamic MRI, sensitivities of 94–99% for invasive cancers and 50–80% for in situ cancers have been reported [7, 1217], whereas specificities of approximately 65–79% have been obtained when integrating morphologic and dynamic criteria [16, 18]. Several reports have unequivocally shown the higher sensitivity of MRI compared with that of mammography for the detection of multifocal, multicentric breast cancer [2, 5, 79, 11, 16, 1921]. In these reports, patient treatment was changed in 11–18% of the cases [7, 16, 22, 23]. Typically, MRI and mammographic findings in the previously mentioned studies were compared with a pathologic gold standard involving the suspected areas in each breast. Unfortunately, the whole breast was rarely considered, resulting in an underestimation of the false-negative findings with both techniques.
Generally, only a few reports are available on the importance of the whole-breast pattern for the diagnosis of multifocal, multicentric breast cancer [11, 16, 24]. The aim of our study was therefore to compare mammography and MRI for the detection and diagnosis of multiple breast cancer foci in potential candidates for mastectomy using the whole-breast pathologic examination as a gold standard.

Subjects and Methods

Study Design and Requirements

A multicenter prospective nonrandomized trial was performed between 1998 and 2000 at 18 European centers with proven breast imaging experience based on at least 5,000 mammographic examinations per year and at least 10,000 mammographic and 300 breast MRI examinations reviewed by the radiologist in chief [25]. The minimum MRI equipment requirements were a magnet of 1 T equipped with a gradient of 15 mT/m or greater and a bilateral synchronous breast coil.

Patients

Patients who were 18 years old or older of any race, with proven breast cancer and a planned mastectomy (proposed by the surgeon or oncologist or both or based on the patient's preference) were eligible for recruitment. Exclusion criteria were absolute contraindications to MRI, pregnancy or breast feeding, severe renal failure, known hypersensitivity to gadolinium chelates, inclusion in other clinical trials during the month before enrollment, and clinical status that would limit data reliability. The study was approved by the ethics committee of each center. Patients gave informed written consent to the protocol and to the use of the data obtained from statistical analysis for publication. A total of 153 patients were enrolled, with a mean ± SD of 9 ± 7.8 patients at each center (range, 1–32; median, 7).

Imaging Methods

Mammography was performed with standard high-quality methods involving two- or three-view examinations, using additional projections or targeted compression and magnification when required.
Dynamic MRI was performed on days 7–14 of the menstrual cycle in premenopausal women but without scheduling limitations in postmenopausal women. A long-line cubital or hand venous access was obtained before starting the examination. Precontrast images were acquired using a 3D spoiled gradient-echo coronal sequence. The sequence parameters were as follows: TR range/TE range, 12–18.5/1.7–7 (chosen to avoid phase opposition between fat and water, depending on the field strength); flip angle, 25–45°; 50% rectangular field of view, from 16 × 32 cm to 20 × 40 cm; and matrix, from 93 × 256 to 230 × 256 (in-plane spatial resolution from 0.7 × 1.3 mm to 2.2 × 1.6 mm). A slice thickness of 3 mm or less with no interslice gap covered both breasts. The overall acquisition time ranged between 84 and 90 sec.
An IV bolus injection of gadoteridol (Pro-Hance, Bracco Imaging SpA) was given at a standard single dose of 0.1 mmol/kg and at a rate of 2 mL/sec. This was followed by a flush of 20 mL of saline solution at the same rate. Injection of contrast agent and saline flush was performed using an automatic injector.
The postcontrast MRI sequence was identical to the precontrast sequence. The former was acquired beginning 12 sec after initiation of the contrast injection and was repeated with a temporal resolution of 90 sec or less, up to 8 min after injection. The 12-sec delay ensured that the contrast agent reached the breast when the central k-space lines (with image-contrast information) were acquired.
Postprocessing involved temporal subtraction (postcontrast minus precontrast), the acquisition of maximum intensity projections and multiplanar reconstructions, and the construction of dynamic signal intensity–time or percentage enhancement–time curves or both for regions of interest (3 × 3 pixels) positioned within the suspected lesion on subjectively determined areas of maximal enhancement.

Pathologic Examination

Pathologic examinations were performed onsite and covered the entire excised breast. Slices of 5 mm were acquired after a consensus meeting of pathologists from each center. Breast lesions were located using a nine-region map (eight segments plus the nipple region) and were diagnosed according to the 1981 World Health Organization breast cancer classification [26].

Off-Site Image Evaluation and Radiologic–Pathologic Correlation

All mammographic, MRI, and pathologic data were collected and processed by a central unit. The off-site reviewing of mammograms and MR images was performed by two experienced radiologists in consensus. The reviewers were aware of the entry criterion (planned mastectomy) but were blinded to all patient data and to the results of the pathologic examination. A diagnostic score from 1 to 5 (1, negative; 2, benign; 3, probably benign; 4, suspicious; and 5, highly suggestive of malignancy) was assigned to each mammographic or MRI finding. For MRI, morphologic and dynamic data (initial signal-intensity increase, post-initial-signal behavior, shape, border, and contrast agent distribution within enhancing tumors) were combined according to the criteria defined by Fischer et al. [16] and reported by Baum et al. [27]. Scores of 1–3 according to Baum's classification were considered negative, whereas scores of 4–8 were considered positive. The classification method is explained in detail in Appendix 1. The density pattern on mammography was recorded as almost entirely fatty, scattered fibroglandular or heterogeneously dense, or extremely dense.
APPENDIX 1. Criteria for Evaluating Breast Lesions on Contrast-Enhanced MRI
CriteriaScore
Shape
 Round, oval, lobular0
 Linear, ductal, dendritic, spiculate1
Margins
 Regular0
 Irregular1
Kinetic pattern: initial enhancement a 
 < 50%0
 50–100%1
 > 100%2
Kinetic pattern: postinitial enhancement 
 Continuousb0
 Plateauc1
 Washoutd2
Enhancement pattern
 Homogeneous0
 Inhomogeneous1

Rim
2
Total score interpretation
0–2 = Benign
3 = Probably benign
4–5 = Suspicious abnormality
6–8 = Highly suggestive of malignancy
a
Initial peak signal intensity within the first 3 min after contrast administration relative to the precontrast signal intensity
b
Continuous signal intensity increase = signal-intensity increase of more than 10% relative to the initial peak signal intensity at 3–6 min
c
Plateau = deviation of the signal intensity at 3–6 min of ± 10% relative to the initial peak signal intensity
d
Washout = signal intensity decrease of more than 10% relative to the initial peak signal intensity at 3–6 min
Comparison of the location of the breast foci at pathologic examination, on mammography, and on MRI was performed off-site using the same nine-region map. A third radiologist matched mammographic and MRI findings and pathologic reports. Care was taken during the lesion-matching procedure to account for vertical displacement in the oblique projection on mammograms and to consider all available MRI data (subtracted scans, maximum intensity projections, and multiplanar reconstructions). The database obtained from the lesion-matching procedure permitted analysis on both a lesion-by-lesion basis and a breast-by-breast basis.
On the basis of the pathologic report, unifocal cancer was indicated when only one malignant focus was found in the breast, multifocal cancer was indicated when more than one malignant focus was shown in the same or in contiguous segments of the nine-region map, and multicentric cancer was indicated when noncontiguous regions were involved.

Quality Check

A number of patients among the 153 enrolled patients were not eligible for evaluation. A total of 42 patients were excluded because mastectomy was ultimately not performed (some patients opted for presurgical adjuvant chemotherapy), resulting in the absence of a pathologic gold standard for the entire breast or because the mammography, MRI, or pathologic examinations were either incomplete or subject to imaging artifacts (e.g., patient movement between the pre- and postcontrast MRI). An additional 21 patients were excluded from analysis at the central off-site unit because of protocol violations during MR image acquisition (e.g., a slice thickness of 4 mm rather than 3 mm in patients with large breasts). No exclusion from data analysis was due to the results of mammography or MRI.
Ninety patients (58.6 ± 16.1 years old) with complete mammographic, MRI, and pathologic correlation were therefore available for evaluation. This population included nine bilateral synchronous breast cancers to give a total of 99 breasts for subsequent statistical analysis.

Focus-by-Focus Analysis

The focus-by-focus analysis comprised determinations of the sensitivity and positive predictive value (PPV) of MRI compared with those of mammography for the overall population and for the fatty, scattered fibroglandular or heterogeneously dense, and extremely dense breast subgroups. Specificity was not calculated because neither mammography nor MRI was aimed at detecting benign lesions. The malignant foci missed with both techniques were pathologically classified as either invasive or noninvasive for calculation of an invasive–noninvasive ratio. A comparison of the dimensions of foci missed on mammography with those missed on MRI was performed. Finally, sensitivities for the detection of invasive and in situ foci were calculated.

Breast-by-Breast Analysis

For the breast-by-breast evaluation, understaging was defined as occurring when a malignant focus of 1 or greater was missed on mammography or MRI, correct staging was defined as occurring when mammography or MRI detected the exact number of foci according to the nine-region location of malignant foci revealed at pathology, and overstaging was defined as occurring when one false-positive focus of 1 or greater was present on mammography or MRI.

Statistical Analysis

Nonparametric McNemar, Mann-Whitney U, and chi-square tests were used.

Results

Of 99 breasts examined, pathology revealed 52 unifocal (53%), 29 multifocal (29%), and 18 multicentric (18%) cancers, for a total of 188 malignant foci. These foci comprised 158 invasive (median diameter, 18 mm) and 30 in situ (median diameter, 5 mm) cancers; details on the pathologic type and dimension of these lesions are reported in Table 1. The results of focus-by-focus matching in terms of sensitivity (overall and for invasive and in situ foci separately) and PPV are given in Table 2 for both mammography and MRI, along with the total numbers of true-positive, false-negative, and false-positive findings. Also reported in Table 2 are the ratio between missed invasive and missed noninvasive foci and the dimension of the foci missed with both techniques.
TABLE 1 Pathologic Type of the 188 Malignant Lesions in 99 Excised Breasts
Pathologic FindingsDiameter Dimension (mm) of Lesion
TypeNo.%MedianMean ± SD< 55–1010–20> 20Not Assessed
No.%No.%No.%No.%No.%
Invasive15884.018.021.8 ± 17.896221447307044106
   IDC9148.417.522.8 ± 20.66713142831414533
   ILC2814.912.015.7 ± 10.527518932829414
   IDC+DCIS189.622.523.3 ± 14.816211528105600
   IDC+ILC94.830.024.3 ± 9.90011122266700
   Othera126.417.521.1 ± 15.40018325542325
In situ3016.05.08.8 ± 9.2517723310271343
   DCIS2613.85.59.3 ± 9.3415727312281038
   LCIS31.6133000000267
   DCIS+LCIS
1
0.5


0
0
0
0
0
0
0
0
1
100
Total
188
100.0
16.0
20.4 ± 17.5
14
8
29
15
50
27
72
38
23
12
Note.—Dash (—) indicates data unavailable. IDC = invasive ductal carcinoma, ILC = invasive lobular carcinoma, DCIS = ductal carcinoma in situ, LCIS = lobular carcinoma in situ.
a
Ductal–lobular (no other specifications, n = 3); mixed (no other specifications, n = 2); endolymphatic carcinosis (n = 2); metaplastic carcinoma (n = 2); ductal (no other specifications, n = 1); ILC + DCIS (n = 1); ILC + LCIS (n = 1)
TABLE 2 Focus-by-Focus Analysis of Diagnostic Performance of Mammography and Dynamic MRI in Pathology-Controlled Studies in 99 Breasts
FeaturesMammographyMRIp
True-positive124152
False-negative6436
Overall sensitivity66% (124/188)81% (152/188)< 0.001a
Sensitivity for invasive foci72% (113/158)89% (140/158)< 0.001a
Sensitivity for in situ foci37% (11/30)40% (12/30)NSa
Invasive–noninvasive ratio of false-negative2.4 (45/19)1.0 (18/18)0.043b
Diameter of false-negative (mm)   
   Mean ± SD10.9 ± 18.25.6 ± 4.50.033c
   Median8.05.0
   Range0.5–130.00.5–15.0
False-positives4070
Positive predictive value
76% (124/164)
68% (152/222)
NSb
Note.—Dash (—) indicates data not available. For 22% (14/64) of the malignant missed foci on mammography and for 31% (11/36) of the malignant missed foci on MRI, the pathology report did not indicate the diameter of the focus. NS = not significant.
a
McNemar test
b
Chi-square test
c
Mann Whitney U test
The pathologic type of foci missed on mammography and MRI is reported in Table 3, whereas the numbers of concordant and discordant cases for mammography versus MRI for the 188 malignant foci are reported in Table 4.
TABLE 3 Pathologic Type of Malignant Foci Missed on Mammography and Dynamic MRI in Pathology-Controlled Studies in 99 Breasts
Pathologic TypeMammographyMRI
Invasive4518
   IDC208
   ILC177
   IDC+ILC51
   Other3a2b
In situ1918
   DCIS1716
   LCIS
2
2
Total
64
36
Note.— IDC = invasive ductal carcinoma, ILC = invasive lobular carcinoma, DCIS = ductal carcinoma in situ, LCIS = lobular carcinoma in situ.
a
Endolymphatic carcinosis (n = 2), metaplastic carcinoma (n = 1)
b
Endolymphatic carcinosis (n = 2)
TABLE 4 Concordant and Discordant Results Between Mammography and Dynamic MRI in Detecting 188 Malignant Foci in 99 Breasts
CasesNo.%
True-positives on both mammography and MRI12164
True-positive on mammography and false-negative on MRI32
False-negative on mammography and true-positive on MRI3116
False-negative on both mammography and MRI
33
18
Total
188
100
Overall, the global sensitivity for detection was 66% (124/188) for mammography and 81% (152/188) for MRI (p < 0.001, McNemar test). The sensitivity was 72% (113/158) and 89% (140/158), respectively (p < 0.001, McNemar test), for the detection of invasive foci and 37% (11/30) and 40% (12/30), respectively (p > 0.05 [not significant], McNemar test), for the detection of in situ foci. Mammography missed a total of 64 malignant foci compared with a total of 36 malignant foci missed on MRI. The median diameters of the missed malignant foci were 8 and 5 mm, respectively (p = 0.033, Mann-Whitney U test). The invasive–noninvasive ratio of missed lesions was of 2.4:1 (45:19) on mammography compared with 1.0:1 (18:18) on MRI (p = 0.043, chi-square test). The global PPV was 76% (124/164) for mammography and 68% (152/222) for MRI (p > 0.05 [not significant], chi-square test).
The sensitivity and PPV of both techniques for the subgroups of breast pattern together with statistical comparisons are reported in Table 5. In breasts with a fatty pattern, the sensitivity for detection was 75% for mammography and 80% for MRI (p > 0.05 [not significant], McNemar test). Likewise, the PPV was 73% on mammography and 65% on MRI (p > 0.05 [not significant], chi-square test). For the remaining breast pattern subgroups considered together, the sensitivity for detection was 60% for mammography and 81% for MRI (p < 0.001, McNemar test), whereas the PPV was 78% and 71% (p > 0.05 [not significant], chi-square test), respectively. Examples comparing MRI with mammography are given in Figures 1A, 1B, 1C, 1D, 2A, 2B, 2C, 2D, and 2E.
TABLE 5 Sensitivity and Positive Predictive Value of Mammography and MRI in Detecting 188 Malignant Foci in 99 Breasts for Different Patterns on Mammography
StatisticsFatty BreastsScattered Fibroglandular, Heterogeneously Dense, and Extremely Dense PatternsTotal
MammographyMRIpMammographyMRIpMammographyMRIp
Sensitivity75% (56/75)80% (60/75)NS60% (68/113)81% (92/113)< 0.00166% (124/188)81% (152/188)< 0.001
Positive predictive value
73% (56/77)
65% (60/92)
NS
78% (68/87)
71% (92/130)
NS
76% (124/164)
68% (152/222)
NS
Note.—McNemar test was used for comparisons of sensitivity and chi-square test for comparisons of positive predictive values. NS = not significant.
Fig. 1A. 47-year-old woman with multicentric breast cancer in dense breast. Craniocaudal mammogram shows round hyperdense opacity (arrow) at external quadrants.
Fig. 1B. 47-year-old woman with multicentric breast cancer in dense breast. Mediolateral oblique mammogram does not show clear depiction of tumor.
Fig. 1C. 47-year-old woman with multicentric breast cancer in dense breast. Contrast-enhanced subtracted coronal MR image (first dynamic phase) shows not only major tumor as large round area of contrast uptake with irregular borders at external quadrants but also three small satellite foci (arrows) showing multifocal cancer.
Fig. 1D. 47-year-old woman with multicentric breast cancer in dense breast. Subtracted coronal image (first dynamic phase), obtained in plane anterior to that shown in C, confirms multifocality at external quadrants (arrows); however, three other small foci (arrowheads) of contrast uptake are detected at internal quadrants, showing multicentric cancer.
Fig. 2A. 47-year-old woman with multicentric breast cancer with scattered fibroglandular pattern. Lateromedial 90° (A) and craniocaudal (B) mammograms show at least four roundish nodules clustered at internal quadrants of right breast (circle). Small posterior opacity (arrow, A) was recognized only retrospectively.
Fig. 2B. 47-year-old woman with multicentric breast cancer with scattered fibroglandular pattern. Lateromedial 90° (A) and craniocaudal (B) mammograms show at least four roundish nodules clustered at internal quadrants of right breast (circle). Small posterior opacity (arrow, A) was recognized only retrospectively.
Fig. 2C. 47-year-old woman with multicentric breast cancer with scattered fibroglandular pattern. Contrast-enhanced breast MR images from subtracted first dynamic phase, coronal partial maximum intensity projection (C) and two axial multiplanar reconstructions (D and E), confirm group of nodules at internal quadrants (circle) and reveal other three malignant nodules, one at external quadrant (arrow, C), another in retroareolar region (arrow, D), and last at internal quadrant in prepectoral region (arrow, E). Only last one could be retrospectively recognized in lateral mammogram (arrow, A).
Fig. 2D. 47-year-old woman with multicentric breast cancer with scattered fibroglandular pattern. Contrast-enhanced breast MR images from subtracted first dynamic phase, coronal partial maximum intensity projection (C) and two axial multiplanar reconstructions (D and E), confirm group of nodules at internal quadrants (circle) and reveal other three malignant nodules, one at external quadrant (arrow, C), another in retroareolar region (arrow, D), and last at internal quadrant in prepectoral region (arrow, E). Only last one could be retrospectively recognized in lateral mammogram (arrow, A).
Fig. 2E. 47-year-old woman with multicentric breast cancer with scattered fibroglandular pattern. Contrast-enhanced breast MR images from subtracted first dynamic phase, coronal partial maximum intensity projection (C) and two axial multiplanar reconstructions (D and E), confirm group of nodules at internal quadrants (circle) and reveal other three malignant nodules, one at external quadrant (arrow, C), another in retroareolar region (arrow, D), and last at internal quadrant in prepectoral region (arrow, E). Only last one could be retrospectively recognized in lateral mammogram (arrow, A).
The diagnostic performance in terms of understaging, correct staging, and overstaging on a breast-by-breast basis are reported in Table 6. Both mammography and MRI detected all nine cases of synchronous bilateral cancer.
TABLE 6 Staging Obtained with Mammography and MRI of Breast Cancers in 99 Breasts
TypeTotalMammographyMRI
UnderCorrectOverUnderCorrectOver
Unifocal521 (2)40 (77)11 (21)0 (0)35 (67)17 (33)
Multifocal2914 (48)9 (31)6 (21)12 (41)9 (31)8 (28)
Multicentric
18
15 (83)
1 (6)
2 (11)
7 (39)
6 (33)
5 (28)
Total
99
30 (30)
50 (51)
19 (19)
19 (19)
50 (51)
30 (30)
Note.—Numbers in parentheses are percentages. Understaging was defined as occurring when one or more malignant foci were missed on mammography or MRI; correct staging, when mammography or MRI detected the exact number and location of malignant foci revealed at pathology; and overstaging, when one or more false-positive findings were present on mammography or MRI.

Discussion

The increasing worldwide tendency for conservative breast surgery places great emphasis on the need to assess precisely the full extent of cancer in the breast. Indeed, inadequate surgical management of breast cancer is common, as shown by the number of excision biopsies with positive resection margins [2830]. Although radiation therapy and chemotherapy play important roles in managing undetected malignant foci, the fact that these foci are not surgically removed may lead to an increased local recurrence [9]. This increase is of particular interest for multicentric cancers, when the distant focus is probably outside the field of boost radiotherapy [31]. Similarly, it is of dramatic importance for synchronous bilateral cancers.
Physical examination, mammography, and fine-needle aspiration cytology are known to be inadequate solutions to this problem, as shown recently by Drew et al. [9]. The contribution of sonography may also be important [32]. However, MRI has recently been shown to be more sensitive than a combination of mammography and sonography for the detection of multiple breast cancer foci [11].
To our knowledge, this is the first large study in which a whole-breast pathologic examination was used as a gold standard for defining the diagnostic performance of mammography and MRI in looking for multifocal, multicentric disease. The number of malignant foci was approximately 1.9:1 (188:99) per breast overall and 2.9:1 per breast if only the 29 multifocal and 18 multicentric cancers are considered in the absence of the 52 unifocal cancers.
In our setting, mammography achieved an overall sensitivity of 66% for the detection of malignant foci. This value is higher than that previously reported [511]. On the other hand, there was a trade-off in terms of false-positive lesions because a total of 40 false-positive lesions were determined on mammography for a PPV of 76%. Although this number was higher than the 68% obtained for MRI, the results were not significantly different. In terms of multifocal, multicentric cancers, false-positive results were a clear problem for both techniques. A PPV of 76% for mammography may be considered a striking performance, but this can be attributed in large part to the selection of potential candidates for mastectomy with almost two (1.9) malignant foci per breast, of whom 47% had multifocal, multicentric cancer.
MRI was certainly better than mammography for the detection of multiple malignant foci in our series. However, 19% of the malignant foci remained undetected on MRI. This may have been a result of insufficient spatial resolution, as suggested by the dimensions of the missed foci (median diameter of missed foci on MRI, 5 mm; median diameter of detected foci on MRI, 8 mm). Alternatively, the missed cancers may have had a low level of angiogenesis.
MRI achieved a sensitivity of 89% for the detection of invasive foci, a 17% gain over mammography (72%). However, the sensitivity for the detection of in situ foci was only 40%, which was comparable to the 37% achieved with mammography. The difference between our MRI sensitivities for both invasive and noninvasive cancers and those generally reported in the literature (94–99% and 50–80%, respectively) is probably due to the more accurate gold standard in our study (whole-breast pathologic examination). This advantage is found in particular for in situ cancers on both mammography and MRI because these cancers can often be recognized only serendipitously at pathology. The ability to recognize microcalcifications on mammography does not offer a real gain in sensitivity in comparison with MRI.
Our results were expected. Hlawatsch et al. [11] reported a sensitivity on MRI of 81% for the correct diagnosis of multifocal, multicentric cancer compared with that of mammography alone (48%) and with a combination of mammography and sonography (63%). Similarly, Malur et al. [17] reported a sensitivity of only 26% for the detection of multifocal cancers on combined mammography and sonography compared with 67% on MRI. The sensitivities for the detection of multicentric cancers were 56% and 89%, respectively [17]. Fischer et al. [16] reported a sensitivity of only 58% on MRI for the detection of in situ cancers. The higher sensitivity of mammography (86%) and MRI (93%) reported by Fischer et al. in their large series of 405 malignant lesions in 336 patients was probably due to the lower overall prevalence of multifocal, multicentric cancers (1.2 malignant lesions per patient compared with 1.9 in our series).
The invasive–noninvasive ratio of our missed lesions with mammography was 2.4:1 versus 1.0:1 for MRI. These data enhance dramatically the difference in sensitivity between the two techniques. Interestingly, combining mammography and MRI achieved a sensitivity of only 83% (156/188 detected lesions) because of the presence of only three lesions that were false-negative on MRI and true-positive on mammography.
Despite the fact that MRI is the best technique available for the detection of multifocal, multicentric breast cancers, improved performance may still be achievable. In terms of sensitivity, a gain could be achieved using a higher in-plane and through-plane spatial resolution. In terms of specificity, improved performance may be achieved by integrating the best morphologic and dynamic data [26], by means of proton MR spectroscopy [33], and by the use of short-term antiestrogen (tamoxifen) medication before MRI [34].
As indicated previously, the PPV of 68% for MRI was not significantly different from that of mammography. This well-known problem makes the availability of MRI-guided biopsy mandatory for centers that intend to use MRI as a diagnostic staging tool.
The analyses of mammographic patterns to give indications for clinical practice. The difference in sensitivity between mammography and MRI was not significant in breasts with a fatty pattern but was largely significant in nonfatty breasts. MRI sensitivity, on the other hand, was not influenced by the density of the breast, and the PPV does not depend on breast pattern. In patients with a nonfatty pattern and at least one suspicious lesion on mammography, sonography could be performed after and not before MRI. Thus, the first and the second sonographic examinations could be combined to exploit the good sonographic sensitivity for multifocal, multicentric breast cancer [11, 32] and to recognize (with immediate biopsy) a large number of the foci detected on MRI [35]. Our results confirm those of Hlawatsch et al. [11], who used a score density of 1–5 and showed treatment-relevant additional information on MRI in patients with a mean score of 4.0.
The results of the breast-by-breast analysis can also be considered to have been influenced by the whole-breast gold standard. Both mammography and MRI correctly staged only 51% of the breasts, whereas a mirrorlike pattern of understaging and overstaging was noted for the remaining breasts: 30% and 19%, respectively, for mammography; and 19% and 30%, respectively, for MRI. These results indicate that understaging of 30% of breast tumors would have occurred on mammography alone, whereas in 19% of the cases, biopsy under stereotactic guidance would have been indicated. Conversely, on MRI, 19% of breast tumors would have been understaged, whereas 30% would have undergone biopsy under MRI guidance. Practically, these results suggest that MRI may give more information if a second conventional evaluation is performed (e.g., a second mammography to acquire new tailored views or magnifications or a targeted sonography after MRI). Moreover, MRI-guided biopsy should be performed only when MRI-detected foci remain undefined.
Synchronous bilateral breast cancer affects approximately 10% of our potential candidates for mastectomy. All the bilateral cancers in our series were found with both techniques. Unlike reviewers in other studies, the mammography reviewers in our study were aware that both breasts were candidates for mastectomy. In the study by Fischer et al. [16], additional contralateral carcinomas were depicted on MRI alone in just 79% (15/19) of the cases. Without MRI and using a clinical reviewing setting for mammography, a substantial number of bilateral breast cancers would have been missed with the result that the undetected contralateral focus would have been untreated.
In conclusion, starting from the viewpoint that conservative therapy for breast cancer requires the ability to screen the entire breast for additional sites of malignancy with a high sensitivity (a difficult task in dense breasts), we compared MRI to mammography in breasts of various densities. Our results indicate that MRI is significantly more sensitive than mammography for the detection of multiple malignant foci in scattered fibroglandular or heterogeneously and extremely dense breasts, but is not significantly more sensitive in fatty breasts. Mammography misses more invasive and larger cancer foci than MRI. However, false-positives are a problem for both techniques. The relatively low specificity of MRI makes the availability of MRI-guided breast biopsy mandatory. Despite the extra cost involved in introducing MRI into the diagnostic workup of breast cancer, the conclusion of Essermann et al. [2] that MRI “would be valuable as a staging tool in the preoperative setting even if the cost is in the range of $1,300 to $2,000” should be considered.
We are aware that the real clinical significance of malignant foci detected solely on MRI is still a matter for debate [36]. If combined with radiation therapy, breast-conserving surgery gives survival rates that are not significantly different from those obtained with mastectomy [31, 37]. Randomized studies comparing the outcome of patients undergoing pretreatment MRI with a control group not undergoing pretreatment MRI are needed to define the effects of a more precise evaluation of the extent of disease on relapse rate, quality of life, and survival rate. One such trial has just begun in the United Kingdom [36]. However, until the results of this or other similar studies become available, we suggest a dynamic MRI examination before treatment planning in patients with a nonfatty breast pattern.

Acknowledgments

The Italian Trial for Breast MR in Multifocal/Multicentric Cancer was promoted by the Italian Association for Medical Radiology (Sections of Senology and Magnetic Resonance) and supported by Bracco Imaging Spa. Enrolling centers were Ferrara, Ospedale S. Anna (S. Corcione); Ancona, University, Ospedale Torrette (G. M. Giuseppetti); S. Giovanni Rotondo, Scientific Institute (M. Cammisa) and Chieti, University (L. Bonomo); Aviano, Centro di Riferimento Oncologico (S. Morassut); Udine, University, Policlinico (M. Bazzocchi); Trieste, University, Ospedale Gattinara (R. Pozzi Mucelli); Siena, University, Policlinico Le Scotte (P. Stefani); Rome, Ospedale Fatebenefratelli (A. Orlacchio); Milan, European Institute of Oncology (M. Bellomi); Milan, Vita-Salute University, Ospedale S. Raffaele (P. Panizza); Milan, Istituto Nazionale Tumori (R. Musumeci); Rome, Tor Vergata University (G. Simonetti); Firenze, University, Ospedale Careggi (N. Villari) and Center for Study and Prevention of Cancer (D. Morrone); Reggio Emilia, Ospedale (A. Troiso); Genoa, University, Ospedale S. Martino (F. Sardanelli, now at Istituto Policlinico San Donato, San Donato Milanese, Milan, Italy); L'Aquila, University, Ospedale S. Maria (G. Masciocchi); Como, Ospedale S. Anna (G. Gozzi); Milan, Ospedale S. Carlo (D. Vergnaghi); and Modena, University, Policlinico (R. Romagnoli).

Footnotes

The Italian Trial for Breast MR in Multifocal/Multicentric Cancer was promoted by the Italian Association for Medical Radiology (Sections of Senology and Magnetic Resonance) and supported by Bracco Imaging SpA.
Address correspondence to F. Sardanelli ([email protected]).

References

1.
Osteen RT. Selection of patients for breast-conserving surgery. Cancer 1994; 74:366-371
2.
Esserman L, Hyolton N, Yassa L, Barclay J, Frankel S, Sickles E. Utility of magnetic resonance imaging in the management of breast cancer: evidence for improved preoperative staging. J Clin Oncol 1999; 17:110-119
3.
Lagios MD, Westdahl PR, Rose MR. The concept and implications of multicentricity in breast carcinoma. Pathol Annu 1981; 16:83-102
4.
Holland R, Veling SMJ, Mravunac M, Hendriks JHCL. Histologic multifocality of Tis, T1, T1-2 breast carcinomas: implication for clinical trials of breast-conserving surgery. Cancer 1985; 56:979-990
5.
Harms SE, Flaming DP, Hesley KL, et al. MR imaging of the breast with rotating delivery of excitation off resonance: clinical experience with pathological correlation. Radiology 1993; 187:79-84
6.
Fischer U, Vosshenrich R, Probst A, Burchardt H, Grabbe E. Preoperative MR-mammography in diagnosed breast carcinoma: useful information or useless extravagance [in German]? Rofo Fortschr Geb Rontgenstr Neuen Bildgeb Verfahr 1994; 161:300-306
7.
Orel SG, Schnall MD, Powell CM, et al. Staging of suspected breast cancer: effect of MR imaging and MR-guided biopsy. Radiology 1995; 196:115-122
8.
Mumtaz H, Hall-Craggs MA, Davidson T, et al. Staging of symptomatic primary breast cancer with MR imaging. AJR 1997; 169:417-424
9.
Drew PJ, Chatterjee S, Turnbill LW, et al. Dynamic contrast-enhanced magnetic resonance imaging of the breast is superior to triple assessment for the preoperative detection of multifocal breast cancer. Ann Surg Oncol 1999; 6:559-603
10.
Orel SG, Schnall MD. MR imaging of the breast for the detection, diagnosis, and staging of breast cancer. Radiology 2001; 220:13-30
11.
Hlawatsch A, Teifke A, Schmidt M, Thelen M. Preoperative assessment of breast cancer: sonography versus MR imaging. AJR 2002; 179:1493-1501
12.
Boné B, Aspelin P, Bronge L, Isberg B, Perbeck L, Veress B. Sensitivity and specificity of MR mammography with histopathological correlation in 250 breasts. Acta Radiol 1996; 37:208-213
13.
Heywang-Kobrunner SH, Viehweg P, Heinig A, Küchler C. Contrast-enhanced MRI of the breast: accuracy, value, controversies, solutions. Eur J Radiol 1997; 24:94-108
14.
Del Maschio A, Panizza P. Breast MR: state of the art. Eur J Radiol 1998; 27:250-253
15.
Fischer U, Westerhof JP, Brinck U, Korabiowska M, Schauer A, Grabbe E. Ductal carcinoma in situ in dynamic MR-mammography at 1.5 [in German]. Rofo Fortschr Geb Rontgenstr Neuen Bildgeb Verfahr 1996; 164:290-294
16.
Fischer U, Kopka L, Grabbe E. Breast carcinoma: effect of the preoperative contrast-enhanced MR imaging on the therapeutic approach. Radiology 1999; 231:881-888
17.
Malur S, Wurdinger S, Moritz A, Michels W, Schneider A. Comparison of written reports of mammography, sonography and magnetic resonance mammography for preoperative evaluation of breast lesions, with special emphasis on magnetic resonance mammography. Breast Cancer Res 2001; 3:55-60
18.
Nunes LW, Schnall MD, Orel SG, et al. Breast MR imaging: interpretation model. Radiology 1997; 202:833-841
19.
Boetes C, Mus RD, Holland R, et al. Breast tumors: comparative accuracy of MR imaging relative to mammography and US for demonstrating extent. Radiology 1995; 197:743-747
20.
Rieber A, Merkle E, Bohm W, Brambs HJ, Tomczak R. MRI of histologically confirmed mammary carcinoma: clinical relevance of diagnostic procedures for detection of multifocal or contralateral secondary carcinoma. J Comput Assist Tomogr 1997; 21:773-779
21.
Gatzemeier W, Liersch T, Stylianou A, Buttler A, Becker H, Fischer U. Preoperative MR mammography in breast carcinoma: effect on operative treatment from the surgical viewpoint. Chirurg 1999; 70:1460-1468
22.
Tan JE, Orel SG, Schnall MD, Schultz DJ, Solin LJ. Role of magnetic resonance imaging and magnetic resonance imaging-guided surgery in the evaluation of patients with early-stage breast cancer for breast conservation treatment. Am J Clin Oncol 1999; 22:414-418
23.
Zhang Y, Fukatsu H, Naganawa S, et al. The role of contrast-enhanced MR mammography for determining candidates for breast conservation surgery. Breast Cancer 2002; 9:231-239
24.
Kawashima H, Matsui O, Suzuki M, et al. Breast cancer in dense breast: detection with contrast-enhanced dynamic MR imaging. J Magn Reson Imaging 2000; 11:233-243
25.
Del Maschio A, Bazzocchi M, Giuseppetti GM, et al. RM mammaria: rapporto sul trial nazionale multicentrico delle Sezioni di Studio di Risonanza Magnetica e Senologia. Radiol Med (Torino) 2002; 104:262-272
26.
[No authors listed] Cancer statistics in developing countries: report on a WHO meeting, Nagoya, Japan, 18–22 August, 1981. World Health Stat Q 1983; 36:213-217
27.
Baum F, Fischer U, Vosshenrich R, Grabbe E. Classification of hypervascularized lesions in CE MR imaging of the breast. Eur Radiol 2002; 12:1087-1092
28.
Frei KA, Kinkel K, Bonel HM, Lu Y, Esserman LJ, Hylton NM. MR imaging of the breast in patients with positive margins after lumpectomy: influence of the time interval between lumpectomy and MR imaging. AJR 2000; 175:1577-1584
29.
Orel SG, Reynolds C, Schnall MD. Breast carcinoma: MR imaging before re-excisional biopsy. Radiology 1997; 205:429-436
30.
Soderstrom CE, Harms SE, Farrell RS Jr, Pruneda JM, Flaming DP. Detection with MR imaging of residual tumor in the breast soon after surgery. AJR 1997; 168:485-488
31.
Harris JR, Recht A, Connolly J, et al. Conservative surgery and radiotherapy for early breast cancer. Cancer 1990; 66:1427-1438
32.
Moon WK, Noh DY, Im JG. Multifocal, multicentric, and contralateral breast cancers: bilateral whole-breast US in the preoperative evaluation of patients. Radiology 2002; 224:569-576
33.
Katz-Brull R, Lavin PT, Lenkinski RE. Clinical utility of proton magnetic resonance spectroscopy in characterizing breast lesions. J Natl Cancer Inst 2002; 94:1197-1203
34.
Heinig A, Lampe D, Kolbl H, Beck R, Heywang-Kobrunner SH. Suppression of unspecific enhancement on breast magnetic resonance imaging (MRI) by antiestrogen medication. Tumori 2002; 88:215-223
35.
Obdeijn IM, Brouwers-Kuyper EM, Tilanus-Linthorst MM, Wiggers T, Oudkerk M. MR imaging-guided sonography followed by fine-needle aspiration cytology in occult carcinoma of the breast. AJR 2000; 174:1079-1084
36.
Kneeshaw PJ, Turnbull LW, Drew PJ. Current applications and future direction of MR mammography. Br J Cancer 2003; 88:4-10
37.
Veronesi U, Salvadori B, Luini A, et al. Conservative treatment of early breast cancer: long-term results of 1232 cases treated with quadrantectomy, axillary dissection, and radiotherapy. Ann Surg 1990; 211:250-259

Information & Authors

Information

Published In

American Journal of Roentgenology
Pages: 1149 - 1157
PubMed: 15385322

History

Submitted: August 14, 2003
Accepted: March 23, 2004

Authors

Affiliations

Francesco Sardanelli
Department of Diagnostic Imaging, Istituto Policlinico San Donato, Via Morandi 30, San Donato Milanese, Milan 20097, Italy.
Gian M. Giuseppetti
Institute of Radiology, University of Ancona, Via Conca, Ancona 60020, Italy.
Pietro Panizza
Department of Radiology, Vita-Salute University, San Raffaele Hospital, Via Olgettina 60, Milan 20132, Italy.
Massimo Bazzocchi
Institute of Radiology, Udine University, Via Colugna 50, Udine 33100, Italy.
Alfonso Fausto
Department of Diagnostic Imaging, Istituto Policlinico San Donato, Via Morandi 30, San Donato Milanese, Milan 20097, Italy.
Giovanni Simonetti
Department of Diagnostic Imaging and Interventional Radiology, University of Tor Vergata, Viale Oxford 81, Rome 00133, Italy.
Vincenzo Lattanzio
Department of Senology, Policlinico Universitario, Piazza G. Cesare 11, Bari 70124, Italy.
Alessandro Del Maschio
Department of Radiology, Vita-Salute University, San Raffaele Hospital, Via Olgettina 60, Milan 20132, Italy.
for the Italian Trial for Breast MR in Multifocal/Multicentric Cancer

Metrics & Citations

Metrics

Citations

Export Citations

To download the citation to this article, select your reference manager software.

Articles citing this article

View Options

View options

PDF

View PDF

PDF Download

Download PDF

Media

Figures

Other

Tables

Share

Share

Copy the content Link

Share on social media