Contrast-Enhanced Mammography Versus MRI in the Evaluation of Neoadjuvant Therapy Response in Patients With Breast Cancer: A Prospective Study
Abstract
Please see the Editorial Comment by Stefanie Woodard discussing this article.
To listen to the podcast associated with this article, please select one of the following: iTunes, Google Play, or direct download.
BACKGROUND. Contrast-enhanced mammography (CEM) is rapidly expanding as a credible alternative to MRI in various clinical settings.
OBJECTIVE. The purpose of this study was to compare CEM and MRI for neoadjuvant therapy (NAT) response assessment in patients with breast cancer.
METHODS. This prospective study included 51 patients (mean age, 46 ± 11 [SD] years) with biopsy-proven breast cancer who were candidates for NAT from May 2015 to April 2018. Patients underwent both CEM and MRI before, during, and after NAT (pre-NAT, mid-NAT, and post-NAT, respectively). Post-NAT CEM included a 6-minute delayed acquisition. One breast radiologist with experience in CEM reviewed CEM examinations; one breast radiologist with experience in MRI reviewed MRI examinations. The radiologists assessed for the presence of an enhancing lesion; if an enhancing lesion was detected, its size was measured. RECIST version 1.1 response assessment categories were derived. Pathologic complete response (pCR) was defined as absence of both invasive cancer and ductal carcinoma in situ (DCIS).
RESULTS. Of 51 patients, 16 achieved pCR. CEM yielded systematically lower size measurements compared with MRI (mean difference, –0.2 mm for pre-NAT, –0.7 mm for mid-NAT, and –0.3 mm for post-NAT). All post-NAT imaging tests yielded systematically larger size measurements compared with pathology (mean difference, 0.8 mm for CEM, 1.2 mm for MRI, and 1.9 mm for delayed CEM). Of 12 patients with residual DCIS, an enhancing lesion was detected in seven on post-NAT CEM, eight on post-NAT MRI, and nine on post-NAT delayed CEM. Agreement of RECIST response categories between CEM and MRI, expressed as kappa coefficient, was 0.791 at mid-NAT and 0.871 at post-NAT. For detecting pCR by post-NAT imaging, sensitivity and specificity were 81% and 83% for CEM, 100% and 86% for MRI, and 81% and 89% for delayed CEM. Sensitivity was significantly higher for MRI than CEM (p = .001) and delayed CEM (p = .002); remaining comparisons were not significant (p > .05).
CONCLUSION. After NAT for breast cancer, CEM and MRI yielded comparable assessments of lesion size (both slightly overestimated vs pathology) and RECIST categories and showed no significant difference in specificity for pCR. MRI had higher sensitivity for pCR. Delayed CEM acquisition may help detect residual DCIS.
CLINICAL IMPACT. Although MRI remains the preferred test for NAT response monitoring, the findings support CEM as a useful alternative when MRI is contraindicated or not tolerated.
HIGHLIGHTS
Key Finding
▪ Compared with pathology, post-NAT CEM, MRI, and delayed CEM systematically overestimated residual tumor size by 0.8 mm, 1.2 mm, and 1.9 mm, respectively. For detecting pCR by post-NAT imaging, sensitivity and specificity were 81% and 83% for CEM, 100% and 86% for MRI, and 81% and 89% for delayed CEM.
Importance
▪ CEM may be a useful alternative test for NAT response assessment in patients with breast cancer who are unable to undergo MRI.
Neoadjuvant therapy (NAT) is increasingly used in breast cancer management [1–3]. NAT aims to reduce the volume of inoperable breast lesions, such as locally advanced or inflammatory carcinomas, to facilitate surgery. More recently, NAT has been extended to patients with operable disease, reflecting data showing improved disease-free survival and overall survival in such patients if a pathologic complete response (pCR) is achieved [4, 5]. Accumulating evidence supports dynamic contrast-enhanced MRI (DCE-MRI) for monitoring early response to NAT [6] and assessing for pCR [7, 8]; professional society guidelines currently recommend performing MRI both before and after NAT for response evaluation, recognizing the test's high diagnostic performance [7, 9, 10]. However, the use of MRI continues to be limited by contraindications, patient preferences, high cost, incomplete access, and potential treatment delays.
Contrast-enhanced mammography (CEM) is an emerging breast imaging technique that combines digital dual-energy mammography and IV administration of an iodinated contrast agent [11]. CEM entails acquisition of a two-view low-energy image, equivalent to a standard digital mammogram, and a high-energy image using an x-ray spectrum above the 33.2-keV K-edge of iodine. The high-energy images are not suitable for diagnostic purposes and are postprocessed to obtain recombined images showing areas of contrast enhancement [12]. Numerous studies have shown substantially higher diagnostic performance in breast cancer detection for CEM than for conventional morphologic techniques, such as digital mammography and breast tomosynthesis [12–14]. Studies have also reported comparable performance of CEM and MRI in breast cancer detection and staging [15, 16], indicating that CEM could serve as a credible alternative to MRI for such purposes [17–23]. Supporting the emergence of CEM, the American College of Radiology BI-RADS recently released a supplement with a CEM lexicon [24].
Findings on both CEM and DCE-MRI reflect tumor neoangiogenesis, which is essential for breast cancer growth, progression, and dissemination. Hypoxia, along with other physiologic and pathologic stimuli, leads to the formation of mural defects within the vasculature that allow contrast medium to extrava-sate and temporarily accumulate in the tumor interstitium. This extravasated contrast medium may be visualized on gadolinium-enhanced T1-weighted MR images and on recombined CEM images. Systemic therapy alters breast vascularity, diminishing enhancement of both fibroglandular tissue and tumors on MRI [25]. After systemic therapy, residual viable breast cancer may exhibit slow initial and persistent late enhancement that is typically associated with benign tumors. A similar pattern after NAT may be observed on CEM, such that any residual enhancement at the site of a previously known tumor, even in a delayed phase of acquisition, would warrant suspicion for residual disease.
The aim of this study was to compare CEM and MRI in the evaluation of NAT response in patients with breast cancer, with additional attention to the utility of a delayed CEM acquisition.
Methods
Patients
The institutional review board approved this prospective single-center study. All patients provided signed written informed consent.
From May 2015 to April 2018, consecutive women with biopsy-proven breast cancer who were candidates for NAT were screened for potential study eligibility. On the basis of this screening process, women were considered ineligible if they were under age 18 because a history of adverse reaction to iodinated and/or gadolinium-based contrast agents or severe allergiclike reaction to drugs and/or foods, had renal function impairment, were unable to undergo MRI, were pregnant, or were breastfeeding. The number of women deemed ineligible on the basis of these screening criteria was not recorded. Eligible patients were approached for study enrollment. Enrolled patients underwent evaluation by both MRI and CEM at three time points: pre-NAT (before treatment), mid-NAT (within 1–3 months after treatment start), and post-NAT (2 weeks after the last treatment cycle). Enrolled patients who did not complete imaging at all time points were excluded from the final analysis.
The MRI and CEM examinations were performed on separate days at each time point. In premenopausal women, DCE-MRI was obtained during the follicular phase between the 7th and 14th days of the menstrual cycle to minimize background parenchymal enhancement [26]. No attempts were made to perform CEM examinations at a certain menstrual phase. Adverse reactions to the contrast material administrations for CEM and MRI were recorded.
A medical oncologist selected the NAT regimen for each patient according to the cancer's biologic characteristics. The NAT selection and administration were performed as part of patients' clinical care, unrelated to this investigation. No patient received neoadjuvant endocrine therapy. Before the start of NAT, each lesion was marked with a charcoal suspension [27] to facilitate later identification of the original tumor location, particularly in patients with pCR. Charcoal is routinely used for this purpose because it does not yield MRI artifact, has low cost, does not entail radioisotope injection, and has an excellent safety profile with very rare complications.
Contrast-Enhanced Mammography Protocol
CEM examinations were performed using a Selenia Dimensions mammography system (Hologic) equipped with software for dual-energy imaging (I-View, Hologic). Before image acquisition, 1.5 mL/kg body weight of iodinated contrast agent (350 mg I/mL ioversol, Optiray, Mallinckrodt Pharmaceuticals) was administered IV using a power injector at a flow rate of 3 mL/s; 2 minutes after contrast material injection, the breast was compressed, and a set of low-energy and high-energy images were acquired.
For pre-NAT CEM examinations, two standard craniocaudal and mediolateral oblique views were acquired in each breast. For mid-NAT and post-NAT examinations, CEM was performed only in the affected breast to reduce radiation exposure for study participants. In addition, for post-NAT examinations, a delayed CEM acquisition of the affected breast was obtained at 6 minutes after contrast material injection. A 6-minute interval was used for the delayed CEM acquisition to mirror the timing of the final contrast-enhanced MRI acquisition (as described in the next section) and to obtain images before a time point at which most benign and malignant lesions would be expected to show physiologic washout.
MRI Protocol
MRI examinations were performed with the patient in the prone position using a 1.5-T unit (Signa, GE Healthcare) equipped with an 8-channel phased-array breast coil. In each patient, the three examinations used the same protocol, which included an axial T2-weighted STIR sequence (TR/TE, 5362/50; inversion time, 150 ms; flip angle, 160°; slice thickness, 2 mm; matrix, 356 × 356), an axial DWI echo-planar sequence (TR/TE, 8883/70; slice thickness, 3 mm; matrix, 356 × 356; b values, 50 and 800 s/mm2), and an axial T1-weighted gradient-echo 3D Vibrant-Flex (GE Health-care) sequence (TR/TE, 6.6/4.5; slice thickness, 1.6 mm; matrix, 356 × 356) that was acquired once before and five times after IV administration of 0.1 mmol/kg of gadobenate dimeglumine (Multi-Hance, Bracco Imaging) using a power injector at a flow rate of 2 mL/s. Postprocessed subtraction, multiplanar reconstruction, and maximum-intensity-projection images were generated.
Image Analysis
One radiologist (D.B., with 20 years of posttraining experience in breast imaging including 3 years in CEM) reviewed the CEM examinations, and a different radiologist (C.F., with 15 years of post-training experience in breast imaging including 15 years in breast MRI) reviewed the MRI examinations. The readers were blinded to the results of the other imaging modality and to other clinical details. For each examination, the readers recorded whether an enhancing lesion was identified; the presence of an enhancing lesion was considered to indicate residual disease. If an enhancing lesion was identified, then the readers recorded the lesion's largest dimension. For CEM, this measurement was performed on the recombined images; for MRI, this measurement was performed on the subtracted postcontrast images. For multifocal lesions, the largest lesion was measured. For post-NAT CEM examinations, during the same interpretation session, the reader also recorded whether an enhancing lesion was identified on delayed CEM and, if so, measured the lesion's size on delayed CEM. For all modalities, if no enhancing lesion was identified, then a lesion size of 0 mm was recorded. The readers reviewed the examinations from all three time points for each patient in a single session. Before the readings, to help standardize tumor size measurements, the readers jointly reviewed the imaging in 10 patients with breast cancer who underwent both CEM and MRI but who were not included in the study analysis.
Treatment response was determined between pre-NAT and mid-NAT imaging and between pre-NAT and post-NAT imaging using the lesion size measurements and response assessment categories based on RECIST version 1.1 (v1.1) [28]. Baseline and follow-up examinations at the given time points were compared separately for CEM and MRI. Response was classified as a complete response if the lesion was no longer visualized, partial response if the lesion showed at least a 30% diameter decrease, progressive disease if the lesion showed at least a 20% diameter increase, and stable disease if the lesion showed neither a diameter decrease of at least 30% or a diameter increase of at least 20% [28].
Histopathologic Reference Standard
All patients underwent surgical resection of the tumor after completion of NAT. One of two breast pathologists (nonauthors, both with 20 years of posttraining experience in breast pathology) performed the histologic assessment of the surgical specimen in each patient to determine whether pCR was achieved and to measure the size of residual tumor. For the purpose of the study, pCR was defined as the absence of both invasive cancer and ductal carcinoma in situ (DCIS). The presence of DCIS as the only residual tumor (i.e., T category of ypTis) was recorded. The tumor size was recorded as 0 mm in patients with pCR and in patients with only DCIS.
Post Hoc Analysis
Patients with a size discrepancy of more than 10 mm between imaging modalities at any time point or between any post-NAT imaging modality and final pathology were considered outliers in terms of size measurement. For these patients, the two study readers performed a joint post hoc assessment to identify reasons for the discrepancy. This assessment included evaluation of all imaging tests, with attention to additional lesion characteristics and imaging findings that may have contributed to the discrepancy, and review of the pathologic findings. The two study readers also performed a joint post analysis of all patients with a false-positive or false-negative interpretation for pCR on any post-NAT imaging test.
Statistical Analysis
Continuous variables were summarized as mean ± SD or as median and interquartile range depending on the given variable's distribution as assessed by the Shapiro-Wilk test. Bland-Altman analysis was used to assess agreement in measurements of lesion size among combinations of CEM, MRI, delayed CEM, and final pathology; results of the Bland-Altman analyses were reported as the bias and limits of agreement. Pearson correlation coefficients were used to assess agreement of size measurements among imaging tests and pathology and were classified as follows: 0.00–0.09, negligible correlation; 0.10–0.39, weak correlation; 0.40–0.69, moderate correlation; 0.70–0.89, strong correlation; 0.90–1.00, very strong correlation [29]. Unweighted Cohen kappa coefficients were used to assess agreement of RECIST response categories between mid-NAT CEM and MRI and between post-NAT CEM and MRI. Overall results were classified as follows using the scale created by Landis and Koch [30]: 0.00–0.20, slight agreement; 0.21–0.40, fair agreement; 0.41–0.60, moderate agreement; 0.61–0.80, substantial agreement; and 0.81–1.00, almost perfect agreement. The diagnostic performance of post-NAT CEM, MRI, and delayed CEM for the detection of pCR was calculated using presence of pCR on final pathology as the reference standard. Findings were considered true-positive when imaging showed no enhancement and pathologic assessment showed pCR, true-negative when imaging showed enhancement and pathology showed residual disease, false-positive when imaging showed no enhancement and pathology showed residual disease, and false-negative when imaging showed enhancement and pathology showed pCR. Diagnostic performance was compared between imaging tests using the McNemar test. A p value less than .05 was considered statistically significant. MedCalc for Windows (version 19.2, MedCalc Software) was used for all statistical analyses.
Results
Patients
During the study period, a total of 63 patients met the initial eligibility criteria and were approached for study participation. No patient declined participation, resulting in 63 initially enrolled patients. Of the enrolled patients, one patient declined NAT and instead opted for up-front surgery; two patients were unable to complete the pre-NAT MRI examination because of anxiety; three patients did not complete all mid-NAT imaging (no CEM in one, no CEM and no MRI in two); and six patients did not complete all post-NAT imaging (incomplete MRI in one, no CEM in one, no CEM and no MRI in four). These exclusions resulted in a final analyzed sample of 51 patients (mean age, 46 ± 11 years; range, 28–74 years). Figure 1 describes the flow of study enrollment, and Table 1 shows the baseline characteristics of the 51 patients. All 51 patients had invasive ductal carcinomas (unifocal in 29/51 [57%], multifocal in 22/51 [43%]).
| Characteristic | Value (n = 51) |
|---|---|
| Age (y) | |
| Mean ± SD | 46 ± 11 |
| Median | 45 |
| Range | 28–74 |
| Menopausal status | |
| Premenopausal | 36 (71) |
| Postmenopausal | 15 (29) |
| Histology | |
| IDC | 51 (100) |
| Focality | |
| Unifocal | 29 (57) |
| Multifocal | 22 (43) |
| Grade | |
| 1 | 0 (0) |
| 2 | 12 (24) |
| 3 | 39 (76) |
| ER | |
| Positive | 35 (69) |
| Negative | 16 (31) |
| PR | |
| Positive | 27 (53) |
| Negative | 24 (47) |
| Molecular subtype | |
| Luminal | 19 (37) |
| HER2 positive | 22 (43) |
| Triple negative | 10 (20) |
| Pre-NAT T categorya | |
| T1 | 14 (27) |
| T2 | 27 (53) |
| T3 | 10 (20) |
| Pre-NAT N categorya | |
| N0 | 27 (53) |
| N1 | 21 (41) |
| N2 | 2 (4) |
| N3 | 1 (2) |
| Post-NAT T categorya | |
| ypT0b | 16 (31) |
| ypTisc | 12 (24) |
| ypT1 | 14 (27) |
| ypT2 | 9 (18) |
| Post-NAT N categorya | |
| ypN0 | 38 (75) |
| ypN1 | 12 (24) |
| ypN2 | 1 (2) |
Note—Unless otherwise indicated, data are expressed as number of patients with percentage in parentheses. IDC = invasive ductal carcinoma, ER = estrogen receptor, PR = progesterone receptor, HER2 = human epidermal growth factor receptor 2, NAT = neoadjuvant therapy.
a
Based on AJCC breast cancer staging manual, 8th edition [49].
b
Complete pathologic response.
c
Only residual ductal carcinoma in situ.
The mean interval between CEM and MRI was 5.9 ± 2.0 days for pre-NAT imaging, 6.4 ± 2.3 days for mid-NAT imaging, and 6.0 ± 2.0 days for post-NAT imaging. No patient experienced an adverse reaction to contrast material administration for any of the MRI or CEM examinations. Histologic assessment of the surgical specimens after completion of NAT showed pCR in 16 of 51 (31%) patients (6/10 [60%] in those with triple-negative breast cancer, 8/22 [36%] in those with human epidermal growth factor receptor 2 positive–positive breast cancer, 2/19 [11%] in those with luminal breast cancer). A total of 12 of 51 (24%) patients had residual DCIS (i.e., T category ypTis). Mean pathologic tumor size after NAT was 8.1 ± 7.7 mm. No patient had multifocal residual disease at final pathologic evaluation.
Tumor Size on Contrast-Enhanced Mammography and MRI
Figure S1 (available in the online supplement) shows Bland-Altman plots comparing CEM and MRI at the three study time points. At all time points, CEM resulted in a systematically lower size in comparison with MRI (mean difference, –0.2 mm for pre-NAT, –0.7 mm for mid-NAT, –0.3 mm for post-NAT). Figure S2 (available in the online supplement) shows Bland-Altman plots comparing delayed CEM with the other imaging tests at the post-NAT time point. Delayed CEM resulted in a systematically larger size in comparison with the other tests (mean difference, 1.1 mm vs CEM, 0.8 mm vs MRI). The mean size after NAT was 10.3 ± 9.5 mm for CEM, 10.6 ± 9.9 mm for MRI, and 11.5 ± 10.0 mm for delayed CEM. Figure 2 shows Bland-Altman plots comparing the various post-NAT imaging tests with pathology. All imaging tests resulted in a systematically larger size in comparison with pathology (mean difference, 0.8 mm for CEM, 1.2 mm for MRI, 1.9 mm for delayed CEM). Size on post-NAT CEM, MRI, and delayed CEM was within 10 mm of the pathologic tumor size in 44 (86%), 45 (88%), and 43 (84%) patients and within 5 mm in 36 (71%), 35 (69%), and 34 (67%) patients, respectively. Correlation of size on post-NAT imaging with size on pathology was strong for CEM (r = 0.795), MRI (r = 0.755), and delayed CEM (r = 0.742). Of the 12 patients with residual DCIS on final pathology, an enhancing lesion was detected on post-NAT CEM in 7 (58%), on post-NAT MRI in 8 (67%), and on post-NAT delayed CEM in 9 (75%).
Fig. 2A —Bland-Altman plots comparing post–neoadjuvant therapy (post-NAT) imaging tests and pathology.
A, Plots show agreement of tumor measurements on final pathology and post-NAT contrast-enhanced mammography (CEM) (A), MRI (B), and delayed CEM (C). Dots represent imaging and pathology measurements for each patient. Top and bottom lines denote 95% limits of agreement, and middle line denotes bias.
Fig. 2B —Bland-Altman plots comparing post–neoadjuvant therapy (post-NAT) imaging tests and pathology.
B, Plots show agreement of tumor measurements on final pathology and post-NAT contrast-enhanced mammography (CEM) (A), MRI (B), and delayed CEM (C). Dots represent imaging and pathology measurements for each patient. Top and bottom lines denote 95% limits of agreement, and middle line denotes bias.
Fig. 2C —Bland-Altman plots comparing post–neoadjuvant therapy (post-NAT) imaging tests and pathology.
C, Plots show agreement of tumor measurements on final pathology and post-NAT contrast-enhanced mammography (CEM) (A), MRI (B), and delayed CEM (C). Dots represent imaging and pathology measurements for each patient. Top and bottom lines denote 95% limits of agreement, and middle line denotes bias.
A total of 7 of 51 (14%) patients were deemed outliers owing to a difference in size of more than 10 mm between imaging modalities at any time point or between any post-NAT imaging modality and pathology. Post hoc analysis of the imaging in these patients was performed. In one patient, who had luminal B invasive ductal carcinoma, lesion size was 90 mm on pre-NAT CEM and 70 mm on pre-NAT MRI; this overestimation was attributed to marked background parenchymal enhancement on CEM that made lesion measurement challenging (Fig. 3). In another patient, who had triple-negative invasive ductal carcinoma, lesion size was 15 mm on pre-NAT CEM and 30 mm on pre-NAT MRI; this discrepancy was attributed to low conspicuity of enhancement on CEM, on which the lesion appeared as a nonmass enhancement; in comparison, on pre-NAT MRI, the lesion appeared as a mass enhancement. In another patient, who had pCR (pathologic size of 0 mm), a lymph node visible on the mediolateral oblique view at the axillary tail was interpreted as an enhancing lesion on post-NAT CEM (measuring 13 mm) and delayed CEM (measuring 15 mm); the node showed benign morphologic features on MRI, which was interpreted as showing no enhancing lesion. In another patient, who had luminal B invasive ductal carcinoma, lesion size was 30 mm on pathology but 8 mm on post-NAT CEM, 13 mm on post-NAT MRI, and 10 mm on post-NAT delayed CEM; no specific cause for the underestimation of lesion size on imaging was identified.
Fig. 3A —41-year-old woman with invasive ductal carcinoma (grade 2, luminal B) of right breast.
A, Craniocaudal (CC, left) and mediolateral oblique (MLO, right) low-energy images from pre–neoadjuvant therapy (pre-NAT) contrast-enhanced mammography (CEM) performed on 25th day of luteal phase show 60-mm architectural distortion (line) in central portion of right breast.
Fig. 3B —41-year-old woman with invasive ductal carcinoma (grade 2, luminal B) of right breast.
B, CC (left) and MLO (right) recombined images from pre-NAT CEM show marked background parenchymal enhancement, hindering lesion evaluation; 90-mm nonmass regional enhancement (line) was described.
Fig. 3C —41-year-old woman with invasive ductal carcinoma (grade 2, luminal B) of right breast.
C, Axial fat-saturated T1-weighted unenhanced MR image shows 70-mm hypointense irregular mass (line) with noncircumscribed margins in central portion of right breast.
Fig. 3D —41-year-old woman with invasive ductal carcinoma (grade 2, luminal B) of right breast.
D, Axial fat-saturated T1-weighted postcontrast MR image shows corresponding fast initial enhancement (line) with delayed plateau.
The three other outliers were attributed to the presence of DCIS on final pathology (T category of ypTis). In all three of these patients, the tumor size on pathology was recorded as 0 mm, contributing to the size discrepancy on imaging. In two of these patients, the lesion was not visible on post-NAT CEM but was visualized on post-NAT delayed CEM and MRI as a nonmass enhancement (measuring 15 mm on post-NAT delayed CEM and 15 mm on post-NAT MRI in one patient, and measuring 27 mm on post-NAT delayed CEM and 30 mm on post-NAT MRI in the other patient) (Fig. 4). In the remaining patient, all three post-NAT imaging tests showed nonmass enhancement measuring from 27 to 30 mm.
Fig. 4A —36-year-old woman with invasive ductal carcinoma (grade 2, human epidermal growth factor receptor 2 positive) of right breast.
A, Low-energy craniocaudal (CC) image from pre–neoadjuvant therapy (pre-NAT) contrast-enhanced mammography (CEM) shows 60-mm asymmetric area of dense tissue with associated fine pleomorphic microcalcifications (line) in lateral portion of right breast.
Fig. 4B —36-year-old woman with invasive ductal carcinoma (grade 2, human epidermal growth factor receptor 2 positive) of right breast.
B, Recombined CC image from pre-NAT CEM shows corresponding 75-mm regional heterogeneous nonmass enhancement (line).
Fig. 4C —36-year-old woman with invasive ductal carcinoma (grade 2, human epidermal growth factor receptor 2 positive) of right breast.
C, Axial postcontrast fat-saturated T1-weighted image from pre-NAT MRI shows 70-mm regional heterogeneous nonmass enhancement (line).
Fig. 4D —36-year-old woman with invasive ductal carcinoma (grade 2, human epidermal growth factor receptor 2 positive) of right breast.
D, Recombined CC image from post-NAT CEM shows no abnormal enhancement in area of lesion.
Fig. 4E —36-year-old woman with invasive ductal carcinoma (grade 2, human epidermal growth factor receptor 2 positive) of right breast.
E, Recombined CC image from post-NAT delayed CEM shows area of 27-mm segmental nonmass enhancement (line).
Fig. 4F —36-year-old woman with invasive ductal carcinoma (grade 2, human epidermal growth factor receptor 2 positive) of right breast.
F, Axial postcontrast fat-saturated T1-weighted subtracted image from post-NAT MRI shows 30-mm segmental nonmass enhancement (line). Pathologic assessment after NAT showed residual ductal carcinoma in situ (DCIS, T category of ypTis). Overestimation of tumor size on post-NAT delayed CEM and MRI was attributed to assignment of pathologic size of 0 mm for residual DCIS.
Table 2 summarizes the RECIST v1.1 response categories at the mid-NAT and post-NAT time points for CEM and MRI. The agreement of response categories between CEM and MRI was substantial at mid-NAT (κ = 0.791; 95% CI, 0.635–0.948) and almost perfect at post-NAT (κ = 0.871; 95% CI, 0.749–0.993).
| Time Point | MRI | CEM | Agreement | ||||||
|---|---|---|---|---|---|---|---|---|---|
| CR | PR | SD | PD | CR | PR | SD | PD | ||
| Mid-NAT | 14 (27) | 31 (61) | 5 (10) | 1 (2) | 13 (25) | 29 (57) | 8 (16) | 1 (2) | 0.791 (0.635–0.948) |
| Post-NAT | 21 (41) | 23 (45) | 6 (12) | 1 (2) | 19 (37) | 26 (51) | 5 (10) | 1 (2) | 0.871 (0.749–0.993) |
Note—Data are expressed as number of patients with percentage in parentheses or as kappa coefficient with 95% CI in parentheses. CEM = contrast-enhanced mammography, CR = complete response, PR = partial response, SD = stable disease, PD = progressive disease.
Diagnostic Performance for Pathologic Complete Response
Table 3 presents the diagnostic performance of post-NAT imaging for pCR. CEM yielded sensitivity of 81%, specificity of 83%, and accuracy of 82%; MRI yielded sensitivity of 100%, specificity of 86%, and accuracy of 90%; and delayed CEM yielded sensitivity of 81%, specificity of 89%, and accuracy of 86%. The sensitivity of MRI was significantly higher than that of CEM (p = .001) and delayed CEM (p = .002); sensitivity was not significantly different between CEM and delayed CEM (p > .99). Specificity and accuracy were not significantly different between any pairwise combinations of tests (all p > .05).
| Performance Index | CEM (A) | MRI (B) | Delayed CEM (C) | p (A vs B) | p (A vs C) | p (B vs C) |
|---|---|---|---|---|---|---|
| Sensitivity | 81 (54–96) [13/16] | 100 (79–100) [16/16] | 81 (54–96) [13/16] | .001 | > .99 | .002 |
| Specificity | 83 (66–93) [29/35] | 86 (70–95) [30/35] | 89 (73–97) [31/35] | .70 | .41 | .66 |
| PPV | 68 (43–87) [13/19] | 76 (53–92) [16/21] | 76 (71–100) [13/17] | .25 | .16 | .95 |
| NPV | 91 (75–98) [29/32] | 100 (88–100) [30/30] | 91 (76–98) [31/34] | .08 | .16 | .09 |
| Accuracy | 82 (69–97) [42/51] | 90 (79–97) [46/51] | 86 (74–94) [44/51] | .25 | .59 | .54 |
Note—Data are expressed as percentage with 95% CI in parentheses and numerator and denominator in brackets. Imaging tests were considered positive for pCR if showing an absence of an enhancing lesion; pathology was considered positive for pCR if showing an absence of invasive cancer or ductal carcinoma in situ.
At post hoc analysis, among six patients with false-positive findings for pCR (i.e., no enhancing lesion but residual tumor on pathology) on post-NAT CEM, five had DCIS (grade 3 in two patients, grade 2 in three); five patients with false-positive findings for pCR on post-NAT MRI and four with false-positive findings for pCR on delayed CEM were among those six patients. One patient showed false-positive findings on all post-NAT imaging modalities (an 8-mm invasive grade 3 ypT1b tumor on final pathology initially diagnosed as a 30-mm estrogen receptor–positive progesterone receptor–negative invasive ductal carcinoma). Three patients showed false-negative findings for pCR (i.e., presence of an enhancing lesion but no residual tumor on pathology) on post-NAT CEM and post-NAT delayed CEM; these three patients showed mass enhancement measuring 13 mm, 7 mm, and 5 mm, respectively, on both modalities. No false-negative finding for pCR was observed for post-NAT MRI.
Discussion
This prospective single-center study compared CEM, MRI, and delayed CEM in the evaluation of tumor response to NAT in patients with breast cancer. Lesion size was consistently lower on CEM than on MRI across the study time points, with a maximal systematic bias of –0.7 mm. On post-NAT imaging, both CEM and MRI systematically overestimated lesion size with respect to final pathology (by 1.2 mm for MRI and 0.8 mm for CEM). Size on post-NAT imaging was within 10 mm of the size on final pathology in 86–88% of patients for both modalities. RECIST v1.1 response assessment categories on CEM and MRI showed substantial agreement on mid-NAT imaging and almost perfect agreement on post-NAT imaging. Finally, on post-NAT imaging, sensitivity for pCR was significantly higher for MRI than for CEM, although specificity and accuracy were not significantly different. Though MRI remains the preferred imaging test for post-NAT response assessment, the findings indicate potential utility of CEM in evaluating NAT response when MRI cannot be performed (e.g., because of unavailability, contraindication, or patient preference).
In early studies after the advent of CEM, Fallenberg et al. [31] and Lobbes et al. [32] found no significant differences in breast cancer size measurements among CEM, MRI, and pathology, with good correlation among the techniques. However, in a prospective study of 33 patients who underwent NAT, Barra et al. [20] found that CEM and MRI exhibited mean differences in residual tumor size of 8 mm and 18 mm, respectively, in comparison with pathology. The size discrepancies with respect to pathology are substantially larger in the study by Barra et al. than in the current study. One possible explanation relates to the assessment of multifocal residual cancer, which was observed in the study by Barra et al. but not in the current study. For determination of histopathologic size in such cases, Barra et al. calculated the sum of the diameters for all residual lesions. The authors did not explicitly indicate whether the same approach was used for determining the size of multifocal lesions on imaging. In comparison, the largest dimension of the largest lesion was measured for multifocal lesions on imaging in the current study. Additionally, the prone position used for breast MRI results in a median lesion displacement of approximately 3–6 cm in all three orthogonal directions in comparison with the supine position used during surgery, possibly yielding greater distances among multiple enhancing foci [33]. Similar geometric issues apply, albeit to a lesser extent, as a result of breast compression during CEM. Finally, for both imaging and pathology, mean post-NAT sizes were larger for Barra et al. (1.6 cm for pathology, 2.4 cm for CEM, 3.6 cm for MRI) than in the current study (8.1 mm for pathology, 10.3 mm for CEM, 10.6 mm for MRI, 11.5 mm for delayed CEM), indicating the presence of additional underlying differences between studies in terms of patients and/or cancers.
Size discrepancies of more than 10 mm among various combinations of imaging modalities and pathology were observed in a small number of patients. One factor contributing to size discrepancy was background parenchymal enhancement on CEM. This finding supports scheduling CEM on day 7–14 of the menstrual cycle in trials evaluating NAT response. In a recent meta-analysis of the diagnostic performance of CEM that included over 10,000 patients [15], the timing of CEM examinations according to menstrual cycle phase was not significantly associated with the sensitivity or specificity of CEM, although caution was advised in interpreting these results given the fragmented nature in which the timing of the CEM examinations was reported across studies.
The definition of pCR itself is a central factor affecting the performance of imaging tests for pCR detection, though this definition varies across centers [7]. A key aspect of the variation in the definition of pCR is the classification of residual DCIS. In our study, pCR was defined as the absence of invasive cancer and DCIS. Residual DCIS was not considered pCR because accurate detection and size estimation of residual DCIS by imaging could guide tailored surgical plans, helping to achieve negative margins and thereby successful radical surgery [34]. Although DCIS was considered to represent residual disease, the pathologist did not record the exact size of residual DCIS, resulting in any residual enhancement correctly identified on imaging being deemed an overestimate of pathologic size. These considerations indicate ongoing challenges presented by residual DCIS for both pathology and imaging interpretation.
Past studies have also compared the performance of CEM and MRI in detecting pCR after NAT [18, 20–22]. Our study aligns with those by Iotti et al. [21] and Patel et al. [18] in terms of the definition of pCR and in terms of defining the tests' diagnostic performance with respect to detection of pCR (vs detection of residual disease). Iotti et al. evaluated 46 women with breast cancer who underwent NAT and reported sensitivities of 100% (95% CI, 63–100%) and 87% (95% CI, 47–100%) and specificities of 84% (95% CI, 69–94%) and 60% (95% CI, 43–76%) for CEM and MRI, respectively. Patel et al. evaluated 65 women with breast cancer who underwent NAT and reported a sensitivity of 95% (95% CI, 73–100%) for both CEM and MRI and a specificity of 66.7% (95% CI, 51–80%) for CEM and 68.9% (95% CI, 53–81%) for MRI. Contrary to those two studies, our study showed that MRI had significantly higher sensitivity than CEM, achieving a sensitivity for pCR of 100%. Variation across studies in the rates of pCR (31% for the current study vs 17% for Iotti et al. and 31% for Patel et al.) and DCIS (24% for the current study vs 7.8% for Patel et al.; rate not directly reported by Iotti et al.) may contribute to the variation in performance. The performance of MRI in the current study may also in part relate to the use of gadobenate dimeglumine, a high-relaxivity contrast agent that has shown superiority to other agents [35, 36]. The imaging criterion of an absence of enhancement for detecting pCR in the present study and the two earlier studies also impacts the overall observed levels of performance. A prior meta-analysis noted significant variation in the performance of MRI for detecting pCR after NAT on the basis of the selected threshold, with studies variably defining response as no enhancement or enhancement equal to or less than that of normal breast tissue [9]. However, enhancement equal to or less than that of normal breast tissue cannot be used for CEM because normal fibroglandular tissue is not recognizable on CEM images.
Several earlier studies explored the adjunct of delayed CEM acquired within 10 minutes after contrast material administration in clinical settings other than NAT monitoring, yielding variable results [37–39]. One study reported that the delayed acquisition had higher specificity for diagnosing mass lesions in dense breasts [37], whereas two other studies did not report significant performance improvements [38, 39]. To our knowledge, this is the first study to evaluate a delayed CEM acquisition for the assessment of response to NAT. At post-NAT imaging, delayed CEM yielded a systematically larger size with respect to CEM (by 1.1 mm) and detected additional cases of residual DCIS that were not detected by CEM or MRI. This resulted in delayed CEM having the fewest number of false-positive results and having a higher specificity for pCR in comparison with CEM, although this difference in specificity was not statistically significant. The detection of in situ disease may be the primary advantage of delayed CEM, facilitating individualized surgical plans [40]. On the basis of the findings, we suggest that a delayed acquisition be included if performing CEM rather than MRI after NAT.
The use of NAT has increased in the last decade, evolving from a therapy intended to enable breast conservation to an option for improving outcomes by achieving pCR [41]. In current practice, all patients undergoing NAT then undergo surgery as part of a standard treatment plan. However, omission of surgery in patients with imaging findings of pCR, supported by percutaneous needle biopsy, remains a future goal [42]. To adopt such an approach, highly reliable detection of residual disease by imaging is crucial. Though MRI remains the preferred test, the modality has a range of barriers, including high cost and heterogeneous availability. CEM may represent an accessible and affordable alternative. An advantage of CEM is its acceptance by patients. In a study by Hobbs et al. [43], 49 interviewed women overall preferred CEM to MRI because CEM was faster, more comfortable, and less noisy. A concern relating to use of CEM is the potential for adverse reactions to the iodinated contrast agent. In a systematic review of 84 studies and 14,012 patients, Zanardo et al. [44] reported a pooled rate of adverse reactions of 0.82%, with most reactions being mild.
Limitations of our study must be considered. First, the study was conducted at a single center and had a small sample size. Second, results were not stratified by tumor subtypes. Third, only a single reader reviewed images for each modality. Fourth, the reader for CEM had substantially less experience than the reader for MRI, reflecting experience levels for the two modalities that are likely to be encountered in clinical practice. Indeed, breast MRI was described as early as 1986 [45], whereas CEM was initially described in trials from 2003 [13, 46, 47]. Finally, the additional glandular dose from the deferred CEM acquisitions was not recorded. According to a prior systematic review, the mean additional glandular dose derived from a delayed CEM acquisition ranges from 0.43 to 2.65 mGy per view [44]. In addition, Jeukens et al. [48] found that radiation exposure from a unilateral single CEM exposure posed a small health risk, considering not only the average glandular dose but also the age-dependent lifetime attributable risk for breast cancer incidence and mortality.
In conclusion, CEM and MRI yielded comparable lesion measurements and RECIST v1.1 response assessments after NAT for breast cancer and no significant difference in specificity or accuracy for pCR. Nonetheless, MRI had a sensitivity of 100% for detecting pCR, which was significantly higher than the sensitivity of CEM. Though MRI remains the preferred test when available, the findings support CEM as a useful alternative when MRI is unavailable. If CEM is performed after NAT, then inclusion of a delayed CEM acquisition may help detect residual DCIS.
Supplemental Content
File (22_27756_suppl.pdf)
- Download
- 972.80 KB
References
1.
Makhoul I, Kiwan E. Neoadjuvant systemic treatment of breast cancer. J Surg Oncol 2011; 103:348–357
2.
Kaufmann M, von Minckwitz G, Bear HD, et al. Recommendations from an international expert panel on the use of neoadjuvant (primary) systemic treatment of operable breast cancer: new perspectives 2006. Ann Oncol 2007; 18:1927–1934
3.
Cortazar P, Zhang L, Untch M, et al. Pathological complete response and long-term clinical benefit in breast cancer: the CTNeoBC pooled analysis. Lancet 2014; 384:164–172
4.
Trimboli RM, Giorgi Rossi P, Battisti NML, et al. Do we still need breast cancer screening in the era of targeted therapies and precision medicine? Insights Imaging 2020; 11:105
5.
Asaoka M, Gandhi S, Ishikawa T, Takabe K. Neoadjuvant chemotherapy for breast cancer: past, present, and future. Breast Cancer (Auckl) 2020; 14:1178223420980377
6.
Loo CE, Straver ME, Rodenhuis S, et al. Magnetic resonance imaging response monitoring of breast cancer during neoadjuvant chemotherapy: relevance of breast cancer subtype. J Clin Oncol 2011; 29:660–666
7.
Marinovich ML, Sardanelli F, Ciatto S, et al. Early prediction of pathologic response to neoadjuvant therapy in breast cancer: systematic review of the accuracy of MRI. Breast 2012; 21:669–677
8.
Lobbes MB, Prevos R, Smidt M, et al. The role of magnetic resonance imaging in assessing residual disease and pathologic complete response in breast cancer patients receiving neoadjuvant chemotherapy: a systematic review. Insights Imaging 2013; 4:163–175
9.
Marinovich ML, Houssami N, Macaskill P, et al. Meta-analysis of magnetic resonance imaging in detecting residual breast cancer after neoadjuvant therapy. J Natl Cancer Inst 2013; 105:321–333
10.
National Comprehensive Cancer Network website. Invasive breast cancer guidelines, version 2.2016. www.nccn.org/professionals/physician_gls/pdf/breast.pdf. Accessed May 12, 2022
11.
Cozzi A, Schiaffino S, Sardanelli F. The emerging role of contrast-enhanced mammography. Quant Imaging Med Surg 2019; 9:2012–2018
12.
Jochelson MS, Lobbes MBI. Contrast-enhanced mammography: state of art. Radiology 2021; 299:36–48
13.
Lewin JM, Isaacs PK, Vance V, Larke FJ. Dual-energy contrast-enhanced digital subtraction mammography: feasibility. Radiology 2003; 229:261–268
14.
Lobbes MB, Lalji U, Houwers J, et al. Contrast-enhanced spectral mammography in patients referred from the breast cancer screening programme. Eur Radiol 2014; 24:1668–1676
15.
Cozzi A, Magni V, Zanardo M, Schiaffino S, Sardanelli F. Contrast-enhanced mammography: a systematic review and a meta-analysis of diagnostic performance. Radiology 2022; 302:568–581
16.
Lobbes MBI, Heuts EM, Moossdorff M, van Nijnatten TJA. Contrast enhanced mammography (CEM) versus magnetic resonance imaging (MRI) for staging of breast cancer: the pro CEM perspective. Eur J Radiol 2021; 142:109883
17.
Savaridas S. Misinterpretation of raw data: fundamental flaws in ‘The diagnostic performance of CESM and CE-MRI in evaluating the pathological response to neoadjuvant therapy in breast cancer: a systematic review and meta-analysis.’ Br J Radiol 2021; 10:20210741
18.
Patel BK, Hilal T, Covington M, et al. Contrast-enhanced spectral mammography is comparable to MRI in the assessment of residual breast cancer following neoadjuvant systemic therapy. Ann Surg Oncol 2018; 25:1350–1356
19.
Barra FR, de Souza FF, Camelo REFA, Ribeiro ACO, Farage L. Accuracy of contrast-enhanced spectral mammography for estimating residual tumor size after neoadjuvant chemotherapy in patients with breast cancer: a feasibility study. Radiol Bras 2017; 50:224–230
20.
Barra FR, Sobrinho AB, Barra RR, et al. Contrast-enhanced mammography (CEM) for detecting residual disease after neoadjuvant chemotherapy: a comparison with breast magnetic resonance imaging (MRI). BioMed Res Int 2018; 2018:8531916
21.
Iotti V, Ravaioli S, Vacondio R, et al. Contrast-enhanced spectral mammography in neoadjuvant chemotherapy monitoring: a comparison with breast magnetic resonance imaging. Breast Cancer Res 2017; 19:106
22.
ElSaid NAE, Mahmoud HGM, Salama A, Nabil M, ElDesouky ED. Role of contrast enhanced spectral mammography in predicting pathological response of locally advanced breast cancer post neo-adjuvant chemotherapy. Egypt J Radiol Nucl Med 2017; 48:519–527
23.
Kamal RM, Saad SM, Moustafa AFI, et al. Predicting response to neo-adjuvant chemotherapy and assessment of residual disease in breast cancer using contrast-enhanced spectral mammography: a combined qualitative and quantitative approach. Egypt J Radiol Nucl Med 2020; 51:161
24.
Lee CH, Phillips J, Sung JS, Lewin JM, Newell MS. Contrast enhanced mammography (CEM): a supplement to ACR BI-RADS mammography 2013. American College of Radiology website. www.acr.org/-/media/ACR/Files/RADS/BI-RADS/BIRADS_CEM_2022.pdf. Published 2022. Accessed April 15, 2022
25.
Negrão EMS, Souza JA, Marques EF, Bitencourt AGV. Breast cancer pheno-type influences MRI response evaluation after neoadjuvant chemotherapy. Eur J Radiol 2019; 120:108701
26.
Sogani J, Morris EA, Kaplan JB, et al. Comparison of background parenchymal enhancement at contrast-enhanced spectral mammography and breast MR imaging. Radiology 2017; 282:63–73
27.
Arman A, Kilicoglu G, Guner HH, Celik L. Marking of nonpalpable breast lesions using a custom carbon suspension. Acta Radiol 2001; 42:599–601
28.
Eisenhauer EA, Therasse P, Bogaerts J, et al. New response evaluation criteria in solid tumours: revised RECIST guideline (version 1.1). Eur J Cancer 2009; 45:228–247
29.
Schober P, Boer C, Schwarte LA. Correlation coefficients: appropriate use and interpretation. Anesth Analg 2018; 126:1763–1768
30.
Landis JR, Koch GG. The measurement of observer agreement for categorical data. Biometrics 1977; 33:159–174
31.
Fallenberg EM, Dromain C, Diekmann F, et al. Contrast-enhanced spectral mammography versus MRI: initial results in the detection of breast cancer and assessment of tumour size. Eur Radiol 2014; 24:256–264
32.
Lobbes MBI, Lalji UC, Nelemans PJ, et al. The quality of tumor size assessment by contrast-enhanced spectral mammography and the benefit of additional breast MRI. J Cancer 2015; 6:144–150
33.
Carbonaro LA, Tannaphai P, Trimboli RM, Verardi N, Fedeli MP, Sardanelli F. Contrast enhanced breast MRI: spatial displacement from prone to supine patient's position—preliminary results. Eur J Radiol 2012; 81:e771–e774
34.
Javid SH, Anderson BO. Tailored strategies for DCIS management. Oncology (Williston Park) 2011; 25:861–863
35.
Martincich L, Faivre-Pierret M, Zechmann CM, et al. Multicenter, double-blind, randomized, intraindividual crossover comparison of gadobenate dimeglumine and gadopentetate dimeglumine for breast MR imaging (DETECT Trial). Radiology 2011; 258:396–408
36.
Carbonaro LA, Pediconi F, Verardi N, Trimboli RM, Calabrese M, Sardanelli F. Breast MRI using a high-relaxivity contrast agent: an overview. AJR 2011; 196:942–955
37.
Ainakulova AS, Zholdybay ZZ, Kaidarova DR, et al. Contrast-enhanced spectral mammography without and with a delayed image for diagnosing malignancy among mass lesions in dense breast. Contemp Oncol (Pozn) 2021; 25:17–22
38.
Huang JS, Pan HB, Yang TL, et al. Kinetic patterns of benign and malignant breast lesions on contrast enhanced digital mammogram. PLoS One 2020; 15:e0239271
39.
Xu W, Zheng B, Chen W, et al. Can the delayed phase of quantitative contrast-enhanced mammography improve the diagnostic performance on breast masses? Quant Imaging Med Surg 2021; 11:3684–3697
40.
Proulx F, Correa JA, Ferré R, et al. Value of pre-operative breast MRI for the size assessment of ductal carcinoma in situ. Br J Radiol 2016; 89:20150543
41.
Hennigs A, Riedel F, Marmé F, et al. Changes in chemotherapy usage and outcome of early breast cancer patients in the last decade. Breast Cancer Res Treat 2016; 160:491–499
42.
Apte A, Marsh S, Chandrasekharan S, Chakravorty A. Avoiding breast cancer surgery in a select cohort of complete responders to neoadjuvant chemotherapy: the long-term outcomes. Ann Med Surg (Lond) 2021; 66:102380
43.
Hobbs MM, Taylor DB, Buzynski S, Peake RE. Contrast-enhanced spectral mammography (CESM) and contrast enhanced MRI (CEMRI): patient preferences and tolerance. J Med Imaging Radiat Oncol 2015; 59:300–305
44.
Zanardo M, Cozzi A, Trimboli RM, et al. Technique, protocols and adverse reactions for contrast-enhanced spectral mammography (CESM): a systematic review. Insights Imaging 2019; 10:76
45.
Heywang SH, Hahn D, Schmidt H, et al. MR imaging of the breast using gadolinium-DTPA. J Comput Assist Tomogr 1986; 10:199–204
46.
Jong RA, Yaffe MJ, Skarpathiotakis M, et al. Contrast-enhanced digital mammography: initial clinical experience. Radiology 2003; 228:842–850
47.
Diekmann F, Diekmann S, Taupitz M, et al. Use of iodine-based contrast media in digital full-field mammography: initial experience. Rofo 2003; 175:342–345
48.
Jeukens CRLPN, Lalji UC, Meijer E, et al. Radiation exposure of contrast-enhanced spectral mammography compared with full-field digital mammography. Invest Radiol 2014; 49:659–665
49.
Zhu H, Doğan BE. American Joint Committee on Cancer's staging system for breast cancer, eighth edition: summary for clinicians. Eur J Breast Health 2021; 17:234–238
Information & Authors
Information
Published In
Copyright
© American Roentgen Ray Society.
History
Submitted: March 26, 2022
Revision requested: April 12, 2022
Revision received: May 13, 2022
Accepted: June 10, 2022
First published: June 22, 2022
Keywords
Authors
Metrics & Citations
Metrics
Citations
Export Citations
To download the citation to this article, select your reference manager software.