Abstract
Rationale and Objectives
We implemented a rapid abridged multiphase (RAMP) breast magnetic resonance imaging (MRI) protocol to reduce scan time and increase workflow efficiency. In this study, we compared delayed-phase kinetic analyses of benign and malignant lesions on the RAMP protocol versus a full dynamic contrast-enhanced (DCE) MRI protocol.
Materials and Methods
Consecutive breast MRI examinations obtained from October 2015 to August 2016 with tissue diagnoses of suspicious MRI lesions were identified. RAMP MRI included one pre-contrast and two post-contrast phases. Full DCE MRI included one pre-contrast and at least three post-contrast phases. Lesion kinetic analyses including mean delayed-phase volume percentage of washout, predominant curve type, and worst curve type were assessed. Kinetic analyses assessed on RAMP and DCE MRI protocols were compared using Wilcoxon rank-sum test and Chi-Square test. Receiver operating characteristic (ROC) analysis was performed to discriminate benign and malignant lesions based on delayed-phase parameters.
Results
The study included 177 consecutive breast lesions (50 benign, 127 malignant) in 162 women. RAMP MRI (23 benign, 61 malignant) and DCE MRI examinations (27 benign, 66 malignant) demonstrated 8.4% vs. 9.3% washout (p=0.36) for benign lesions and 18.5% vs. 17% washout (p=0.66) for malignancies, respectively. There was no difference in the predominant and worst curve types for malignant and benign lesions or in area under the ROC curves for delayed-phased parameters between the two protocols (p>0.05).
Conclusion
Lesion kinetic analyses from the RAMP MRI protocol can achieve the same discriminatory ability as the full DCE protocol. By reducing scan time, the RAMP MRI protocol improves patient comfort and enhances workflow efficiency and can be easily implemented in any clinical setting.
Keywords: Breast MRI, kinetic analysis, abbreviated breast MRI protocol, breast cancer
INTRODUCTION
Enthusiasm for abbreviated breast magnetic resonance imaging (MRI) protocols continues to grow, given their potential time- and cost-saving benefits by improving workflow efficiency and patient tolerance. Research has shown that dynamic contrast-enhanced (DCE) breast MRI offers high sensitivity for breast cancer detection (1–7). As a result, organizations including the American Cancer Society, National Comprehensive Cancer Network, and Society of Breast Imaging/American College of Radiology (ACR) recommend breast MRI as a supplemental screening tool for women at elevated risk for breast cancer (7–9). Despite these recommendations, breast MRI remains widely underutilized (10). Reasons for underutilization are partly attributed to the lengthy examination time of a standard full DCE MRI protocol, which increases patient anxiety and discomfort, decreases scanner throughput, and increases cost. Abbreviated breast MRI protocols shorten examination time by reducing the number of sequences performed in each examination (11–21), most commonly acquiring one pre-contrast sequence and one early-phase (first) post-contrast sequence for interpretation of the first post-contrast subtracted (FAST) images and the reconstructed maximum intensity projection (MIP) images (11).
However, the ACR Breast MRI Accreditation Program requires four sequences in each examination besides the localizer or scout sequence (22). These four sequences are: (a) one T2-weighted/bright fluid sequence, (b) one pre-contrast T1 sequence, (c) one early-phase (first) post-contrast T1 sequence completed within four minutes of completion of contrast injection, and (d) one delayed-phase (last) post-contrast T1 sequence. The mandated inclusion of one pre- and at least two post-contrast sequences allows for capture of information on both early and late phases of lesion enhancement. The ACR Breast Imaging Reporting and Data System (BI-RADS) manual further defines the initial-phase enhancement pattern as “enhancement within the first two minutes after injection or until peak enhancement is reached,” and the delayed-phase enhancement pattern as enhancement “after two minutes or after the peak enhancement is reached” (23). Although morphology of masses and internal enhancement patterns of non-mass enhancement (NME) represent important features when assessing lesions on breast MRI, kinetic analysis of both the early and delayed-phase enhancement patterns is recommended by the ACR BI-RADS for classifying lesions seen on breast MRI (24, 25). Lesion washout is a strong predictor of breast cancer, and some studies have identified malignancy in 76% to 87% of lesions with any washout (24, 25).
Importantly, in the absence of a delayed-phase post-contrast sequence, most of the previously described abbreviated breast MRI protocols do not fulfill the ACR Breast MRI Accreditation requirements (11–21). Not only does this deficiency present a challenge to billing and reimbursement, kinetic analysis cannot be performed on these abbreviated examinations. Patients may therefore need to return for a full diagnostic DCE MRI protocol with repeat gadolinium-based contrast injection for complete characterization, including kinetic analysis, of findings seen on the abbreviated MRI.
Our institution first implemented a novel ACR-accredited rapid abridged multiphase (RAMP) breast MRI protocol in December 2015 and transitioned completely to RAMP MRI by April 2016. The RAMP MRI protocol contains all four sequences required by the ACR Breast MRI Accreditation Program, including one T2-weighted/bright fluid sequence, one pre-contrast sequence, one early-phase post-contrast sequence, and one delayed-phase post-contrast sequence. Scan time of RAMP MRI is approximately 10 minutes. Prior to December 2015, all breast MRI examinations were performed using the full DCE MRI protocol, which contained one pre-contrast sequence and at least three post-contrast sequences and took approximately 45 minutesto complete. Details of our RAMP versus full DCE MRI protocols are presented in Appendix A. Despite fulfilling the ACR accreditation requirements and allowing for kinetic analyses of both early-phase and delayed-phase enhancement patterns, effects of our RAMP MRI protocol on delayed-phase lesion kinetic analyses had not been investigated. The purpose of our study wasto validate the use of kinetic information from RAMP MRI in discriminating benign from malignant lesions by comparing the performances of delayed-phase kinetic analyses between our RAMP MRI protocol and our full DCE MRI protocol.
MATERIALS AND METHODS
Study Inclusion
Our Institutional Review Board approved this Health Insurance Portability and Accountability Act-compliant study. Informed consent was waived. We retrospectively reviewed consecutive breast MRI examinations performed between October 1, 2015 and August 31, 2016 at our institution, which consisted of examinations performed under the full DCE MRI protocol between October 1, 2015 and March 31, 2016 (DCE group, before complete transition to RAMP protocol), and examinations performed under the RAMP MRI protocol between December 1, 2015 and August 31, 2016 (RAMP group, after initial implementation of RAMP protocol). We searched our breast imaging information system (MagView, Burtonsville, MD) for MRI examinations with BI-RADS assessment categories 4, 5, or 6, and confirmed histopathologic diagnoses based on core biopsies and/or surgical excisions. A breast imaging fellow (S.C.) retrospectively reviewed the images of all MRI examinations and excluded examinations assessed as BI-RADS category 6 without visible lesion enhancement status post core biopsy.
Imaging Technique
RAMP MRI examinations were performed on GE HDx 1.5T scanners with an eight-channel HD breast coil (General Electric Healthcare, Milwaukee, WI), or on a GE Discovery MR750 3T scanner with a 16-channel breast coil (Sentinelle coil, Invivo, Gainesville, FL). Full DCE MRI examinations were performed on GE HDx 1.5T scanners with an eight-channel HD breast coil. Schematic comparison of our RAMP and full DCE MRI protocols, along with the most common abbreviated breast MRI protocol reported in prior literature (11–21), is shown in Figure 1. Both RAMP and full DCE MRI protocols included a three-plane localizing sequence, an axial pre-contrast T1-weighted non-fat suppressed sequence, and an axial fat-suppressed T2-weighted or short tau inversion recovery sequence.
Figure 1.
Comparison of sequences comprising our full dynamic contrast-enhanced (DCE) MRI protocol, rapid abridged multiphase (RAMP) MRI protocol, and previously published abbreviated MRI (AB-MRI) protocol (a). Schematic depiction of enhancement kinetics in the early phase (Phase 1) and delayed phase (last phase) with respect to the timing of image acquisition after contrast injection for the full DCE protocol, RAMP protocol, and previously published AB-MRI protocol (b). Loc = localizer images; FS = fat-suppressed images; STIR = short tau inversion recovery; Pre = pre-contrast images; Phase = post-contrast phase images; Post = post-contrast images.
In the RAMP MRI protocol, the axial 3D T1-weighted fat-suppressed Volume Imaging for Breast Assessment (VIBRANT) series (General Electric Healthcare, Milwaukee, WI) consisted of one pre-contrast sequence and two post-contrast sequences; specifically, one early-phase (first) post-contrast sequence centered at 60–75 seconds after contrast injection, followed by one delayed-phase (last) post-contrast sequence centered at 180–205 seconds after contrast injection. In the full DCE MRI protocol, the VIBRANT series consisted of one pre-contrast sequence and at least three post-contrast sequences; these post-contrast sequences include one early-phase (first) post-contrast sequence centered at 60–75 seconds after contrast injection, followed by intervening post-contrast sequence(s) at 180–300 seconds until the delayed-phase (last) post-contrast sequence centered at 420–450 seconds. The full DCE MRI protocol also included a delayed sagittal post-contrast sequence at 480–600 seconds after contrast injection. Detailed parameters of the sequences in each protocol are presented in Appendix A.
Both protocols utilized image post-processing to generate sagittal reformations of T1-weighted fat-suppressed pre- and post-contrast images, axial and sagittal post-contrast subtraction images, and 3D summation post-contrast subtraction MIP images. The contrast agent used was gadopentetate dimeglumine [Gd-DTPA] (Magnevist®, Bayer Healthcare, Whippany, NJ) before December 2015 and gadoterate meglumine [Gd-DOTA] (Dotarem®, Guerbet LLC, Bloomington, IN) after December 2015, administered at a dose of 0.1 mmol/kg body weight and power-injected intravenously at a rate of 2.0 mL/s for all breast MRI examinations.
MRI Interpretation and Data Collection
All examinations were prospectively interpreted by one of nine breast-imaging specialized attending radiologists (1–10 years breast MRI experience). Interpreting radiologists reported descriptors and assessments per BI-RADS MRI lexicon, including the amount of fibroglandular breast tissue, degree of background parenchymal enhancement (BPE), and enhancing lesion type – with a focus defined as a <5 mm enhancing dot that was unique from background and did not clearly represent a space-occupying lesion, a mass defined as a 3D convex space-occupying lesion, and NME defined as a discrete area of enhancement from background that was neither a mass nor a focus (23). Computer-aided evaluation using commercially available software, CADstream (version 5.2.9, Confirma, Bellevue, WA), was provided at the time of initial interpretation.
Electronic medical records were retrospectively reviewed by a breast imaging fellow (S.C.) to extract the indication, amount of fibroglandular breast tissue, BPE, and MRI lesion type reported at the time of initial interpretation for each MRI examination. Patient demographics and histopathological results were also collected.
Retrospective Kinetic Analyses using Computer-aided Evaluation
Images were reviewed and processed using CADstream at an independent workstation. For both MRI protocols, CADstream utilized the VIBRANT series to compare pixel signal intensities on the pre-contrast and post-contrast sequences to generate time-intensity curves based on early-phase and delayed-phase kinetic analyses as previously described (26–28). The user specified an early-phase enhancement threshold of 50% or 100%, representing an increase in pixel intensity by 50% (1.5-fold) or 100% (2-fold) on the first post-contrast phase relative to the pre-contrast phase. CADstream registered pixels with a color-overlay when they reached the threshold. Initial kinetics was deemed “medium” enhancement if pixel intensity increased by 50–100% and “fast” enhancement if pixel intensity increased by >100% per BI-RADS terminology (23). For delayed-phase kinetics per BI-RADS (23), CADstream assessed changes from the early-phase (first) post-contrast sequence to the delayed-phase (last) post-contrast sequence. Increased pixel intensity on the delayed phase relative to the early phase by ≥10% was considered “persistent” enhancement, decreased pixel intensity by >10% was considered “washout” enhancement, and pixel intensity change by <10% in either direction was considered “plateau” enhancement.
Each MRI lesion that was assessed as BI-RADS categories 4, 5 or 6 based on the finalized radiology report was segmented as a region of interest (ROI) semi-automatically with the aid of the CADstream color overlay (S.C.). Detailed kinetic analyses were automatically generated by CADstream for the ROI (Fig. 2) and recorded. Specifically, we recorded the proportion of the volume of the lesion (i.e. lesion volume percentages) that demonstrated persistent, plateau, or washout kinetics. The pattern of delayed-phase enhancement kinetic (persistent, plateau, or washout) comprising the largest percentage volume of the ROI determined the predominant curve type classification (29). The most suspicious delayed-phase enhancement kinetic component (any washout > any plateau > any persistent) determined the worst curve type classification (29).
Figure 2.
Computer-aided evaluation color overlay map with kinetics assessment of two known malignancies assessed with the rapid abridged multiphase (RAMP) breast MRI protocol (left)and the full dynamic contrast-enhanced (DCE) MRI protocol (right). Lesion size, peak enhancement, lesion volume percentages of medium and fast kinetics on early phase andpersistent, plateau, and washout kinetics on delayed phase are displayed.
Statistical Analysis
We performed descriptive analyses comparing lesion and patient characteristics between our RAMP MRI and full DCE MRI protocols. We compared the distribution of age, examination indication, amount of fibroglandular tissue, degree of BPE, categorized lesion size, BI-RADS classification of lesion type, BI-RADS assessment categories, and lesion pathology by MRI protocol (RAMP vs. DCE) using Wilcoxon signed rank test for continuous variables and the Pearson’s chi-squared test for categorical variables. Lesion volume percentages of persistent, plateau, and washout delayed-phase kinetics were compared between the two protocols using Wilcoxon rank-sum test. Lesion delayed-phase predominant and worst curve type classifications were compared between the two protocols using Chi-square and Fisher’s exact tests. Receiver operating characteristic (ROC) curve analyses were performed to assess the diagnostic accuracy of each protocol in discriminating benign from malignant lesions based on delayed-phase lesion volume percentage washout, predominant curve type, and worst curve type. Data was analyzed using MATLAB (version 9.1, The MathWorks, Inc., Natick, MA).
RESULTS
Our study included a total of 177 consecutive lesions documented on 162 MRI examinations in 162 women (mean age, 51.6 years ± 11.1 [standard deviation]). Among the total 162 breast MRI examinations, 77 (47.5%) examinations were performed using the RAMP protocol and 85 (52.5%) examinations were performed using the full DCE protocol (Table 1). Slightly over two-thirds of examinations in each group (68.8% of RAMP group and 69.4% of DCE group) were performed for diagnostic purposes, with extent of disease evaluation for known malignancy being the most common indication. Personal history of breast cancer was the most common indication for screening. We found no significant difference in the indications of examinations (screening vs. diagnostic) between the two study groups (p=0.93). Majority of patients in each group had heterogeneous or extreme fibroglandular tissue (62.4% of RAMP group and 64.7% of DCE group) and majority of MRI examinations in each group demonstrated minimal or mild BPE (72.8% of RAMP group and 78.8% of DCE group).
Table 1.
Characteristics of patients, MRI examinations, and lesions evaluated by rapid abridged multiphase (RAMP) and full dynamic contrast-enhanced (DCE) MRI protocols
| Characteristics | RAMP | DCE | p-value |
|---|---|---|---|
| Patients and Examinations | N=77 | N=85 | |
| Age (y) | 51.4 + 9.9 | 51.8 + 12.0 | 0.78 |
| Indications | 0.93 | ||
| Screening | 24/77 (31.2) | 26/85 (30.6) | |
| Diagnostic | 53/77 (68.8) | 59/85 (69.4) | |
| Amount of fibroglandular tissue | 0.68 | ||
| Almost entirely fat | 1/77 (1.3) | 0 | |
| Scattered | 28/77 (36.4) | 30/85 (35.3) | |
| Heterogeneous | 38/77 (49.4) | 46/85 (54.1) | |
| Extreme | 10/77 (13.0) | 9/85 (10.6) | |
| Background parenchymal enhancement | 0.51 | ||
| Minimal | 20/77 (26.0) | 20/85 (23.5) | |
| Mild | 36/77 (46.8) | 47/85 (55.3) | |
| Moderate | 17/77 (22.1) | 12/85 (14.1) | |
| Marked | 4/77 (5.2) | 6/85 (7.1) | |
| Lesions | N=84 | N=93 | |
| Size | 0.58 | ||
| ≤ 1 cm | 25/84 (29.8) | 23/93 (24.7) | |
| > 1 cm to ≤ 2 cm | 26/84 (31.0) | 34/93 (36.6) | |
| > 2 cm to ≤ 5 cm | 25/84 (29.8) | 23/93 (24.7) | |
| > 5 cm | 8/84 (9.5) | 13/93 (14.0) | |
| Type | 0.32 | ||
| Focus | 3/84 (3.6) | 1/93 (1.1) | |
| Mass | 61/84 (72.6) | 63/93 (67.7) | |
| Non-mass enhancement | 20/84 (23.8) | 29/93 (31.2) | |
| BI-RADS assessment category | 0.43 | ||
| 4 | 40/84 (47.6) | 50/93 (53.8) | |
| 5 | 0/84 (0.0) | 1/93 (1.1) | |
| 6 | 44/84 (52.4) | 42/93 (45.2) | |
| Pathology | 0.62 | ||
| Benign | 23/84 (27.4) | 27/93 (29.0) | |
| Malignant | 61/84 (72.6) | 66/93 (71.0) | |
| Ductal carcinoma in situ | 9/61 (14.8) | 6/66 (9.1) | |
| Invasive cancer | 52/61 (85.2) | 60/66 (90.9) | |
| Invasive ductal carcinoma | 47/61 (77.0) | 51/66 (77.3) | |
| Invasive lobular carcinoma | 5/61 (8.2) | 9/66 (13.6) |
Note – Unless otherwise indicated, data are numbers of patients, examinations, or lesions with percentages in parentheses, or means ± standard deviation.
Among the total 177 lesions, four (2.3%) were classified as foci, 124 (70.1%) as masses, and 49 (27.7%) as areas of NME. More than two-thirds of lesions were malignant (127/177, 71.8%; mean size 1.4 ± 1.3 cm) and the remaining 50 lesions were benign (28.2%; mean size 3.0 ± 2.5 cm). There were no statistically significant differences in lesion size (p=0.58), type (p=0.32), or pathology (p=0.62) between the RAMP and DCE groups. Among the 84 lesions in the RAMP MRI group, 23 (27.4%) were benign and 61 were (72.6%) malignant, 52 of which were invasive cancers. Among the 93 lesions in the full DCE MRI group, 27 (29.0%) were benign and 66 (71.0%) were malignant, 60 of which were invasive cancers. Invasive ductal carcinoma was the most common malignancy in both groups (47/61, 77.0% in RAMP group, and 51/66, 77.3% in DCE group). Forty of 84 (47.6%) lesions were assessed as BI-RADS category 4 and 44 (52.4%) lesions were assessed as BI-RADS category 6 in the RAMP group. Fifty of 93 (53.8%) lesions assessed as BI-RADS category 4, one (1.1%) lesion as BI-RADS category 5, and 42 (45.2%) as BI-RADS category 6 in the DCE group. Of the total 177 lesions, 43 (24.3%) lesions were imaged on a 3T scanner and 134 (75.7%) lesions were imaged on a 1.5T scanner.
Mean delayed-phase lesion volume percentages of washout kinetics were not significantly different between the two protocols, with 8.4% [range 0–45%] washout in the RAMP group vs. 9.3% [range 0–62%] washout in the DCE group (p=0.36) for benign lesions, and 18.5% [range 0–92%] washout in the RAMP group vs. 17.0% [range 0–71%] washout in the DCE group (p=0.66) for malignant lesions (Table 2). Mean delayed-phase lesion volume percentage of persistent kinetics for malignant lesions was lower with the RAMP protocol (49.8% [range 3–97%]) vs. DCE protocol (59.4% [range 7–100%], p=0.04). On the other hand, there was no significant difference in percentages of persistent kinetics for benign lesions between the RAMP protocol (68.0% [range 5–100%]) and DCE protocol (76.8% [range 4–100%], p=0.21). Mean delayed-phase lesion volume percentage of plateau kinetics for malignant lesions was higher with the RAMP protocol (31.6% [range 3–79%]) vs. DCE protocol (23.6% [range 0–59%], p=0.005).
Table 2.
Comparison of delayed-phase lesion volume percentages of persistent, plateau and washout curve types between MRI protocols
| Curve Types | RAMP Mean % [Range] |
DCE Mean % [Range] |
p-value |
|---|---|---|---|
| Benign lesions, N=50 | |||
| Persistent | 68.0 [5–100] | 76.8 [4–100] | 0.21 |
| Plateau | 23.6 [0–70] | 13.9 [0–51] | 0.07 |
| Washout | 8.4 [0–45] | 9.3 [0–62] | 0.36 |
| Malignant lesions, N=127 | |||
| Persistent | 49.8 [3–97] | 59.4 [7–100] | 0.04 |
| Plateau | 31.6 [3–79] | 23.6 [0–59] | 0.005 |
| Washout | 18.5 [0–92] | 17.0 [0–71] | 0.66 |
| Invasive cancer, N=112 | |||
| Persistent | 48.0 [3–95] | 59.6 [7–100] | 0.02 |
| Plateau | 31.3 [4–63] | 23.5 [0–59] | 0.005 |
| Washout | 20.6 [0–92] | 16.8 [0–70] | 0.30 |
RAMP = rapid abridged multiphase MRI, DCE = full dynamic contrast-enhanced MRI
In both RAMP and DCE groups, malignant lesions demonstrated delayed-phase washout kinetics as the predominant and/or worst curve type more often than benign lesions (Table 3). Specifically, in the RAMP MRI group, 14.8% (9 of 61) of malignant lesions vs. 0% (0 of 23) of benign lesions demonstrated washout kinetics as the predominant curve type. In the DCE MRI group, 18.2% (12 of 66) of malignant lesions vs. 11.1% (3 of 27) of benign lesions demonstrated washout kinetics as the predominant curve type. For worst curve type classification, 73.8% (45 of 61) of malignant lesions vs. 69.6% (16 of 23) of benign lesions demonstrated washout kinetics in the RAMP group, whereas 65.2% (43 of 66) of malignant lesions vs. 48.1% (13 of 27) of benign lesions demonstrated washout kinetics in the DCE group. Overall, there were no significant differences in the predominant curve type classification for benign (p=0.27) and malignant (p=0.14) lesions between the two protocols. Likewise, no significant differences were seen in the worst curve type classification for benign (p=0.44) and malignant (p=0.42) lesions between the two protocols.
Table 3.
Comparison of delayed-phase predominant curve type and worst curve type classifications between MRI protocols
| Curve Types | RAMP N (%) |
DCE N (%) |
p-value |
|---|---|---|---|
| Predominant Curve Type Classification | |||
| Benign lesion, N | 23 | 27 | |
| Persistent | 20 (87.0) | 23 (85.2) | |
| Plateau | 3 (13.0) | 1 (3.7) | 0.27 |
| Washout | 0 (0) | 3 (11.1) | |
| Malignant lesion, N | 61 | 66 | |
| Persistent | 37 (60.7) | 48 (72.7) | |
| Plateau | 15 (24.6) | 6 (9.1) | 0.14 |
| Washout | 9 (14.8) | 12 (18.2) | |
| Invasive cancer, N | 52 | 60 | |
| Persistent | 31 (59.6) | 44 (73.3) | |
| Plateau | 13 (25.0) | 5 (8.3) | 0.12 |
| Washout | 8 (15.4) | 11 (18.3) | |
| Worst Curve Type Classification | |||
| Benign lesion, N | 23 | 27 | |
| Persistent | 1 (4.3) | 2 (7.4) | |
| Plateau | 6 (26.1) | 12 (44.4) | 0.44 |
| Washout | 16 (69.6) | 13 (48.1) | |
| Malignant lesion, N | 61 | 66 | |
| Persistent | 9 (14.8) | 8 (12.1) | |
| Plateau | 7 (11.5) | 15 (22.7) | 0.42 |
| Washout | 45 (73.8) | 43 (65.2) | |
| Invasive cancer, N | 52 | 60 | |
| Persistent | 8 (15.4) | 8 (13.3) | |
| Plateau | 7 (13.5) | 14 (23.3) | 0.62 |
| Washout | 37 (71.2) | 38 (63.3) | |
RAMP = rapid abridged multiphase MRI, DCE = full dynamic contrast-enhanced MRI
There were no significant differences between the RAMP and DCE MRI protocols in the areas under the ROC curves (AUC) for discriminating benign from malignant lesions based on delayed-phase lesion volume percentage washout (AUC 0.66 vs. 0.69, p=0.72), predominant curve type (AUC 0.61 vs. 0.55, p=0.52), or worst curve type (AUC 0.49 vs. 0.54, p=0.60) (Table 4).
Table 4.
Receiver operating characteristic (ROC) curve analysis for discriminating benign versus malignant lesions between MRI protocols
| ROC Curve Comparison | RAMP (N=84) AUC (95% CI) |
DCE (N=93) AUC (95% CI) |
p-value |
|---|---|---|---|
| Percentage Washout | 0.66 (0.53–0.78) | 0.69 (0.54–0.80) | 0.72 |
| Predominant Curve Type | 0.61 (0.47–0.74) | 0.55 (0.40–0.68) | 0.52 |
| Worst Curve Type | 0.49 (0.33–0.65) | 0.54 (0.39–0.68) | 0.60 |
RAMP = rapid abridged multiphase MRI, DCE = full dynamic contrast-enhanced MRI
DISCUSSION
We found that the delayed-phase enhancement kinetic analyses are similar for breast lesions seen on our RAMP and full DCE MRI examinations. Specifically, we found no significant difference between the RAMP and full DCE MRI protocols in discriminating benign from malignant breast lesions based on delayed-phase volume percentage of washout, predominant curve type, or worst curve type classifications. These findings indicate that by abridging our multiphase series to one pre-contrast sequence, one early-phase post-contrast sequence within two minutes of contrast injection and one delayed-phase post-contrast sequence after two minutes of contrast injection, we are able to preserve the utility of enhancement kinetic analyses in our RAMP breast MRI examinations.
Our study is distinct from prior studies on abbreviated breast MRI (11–21) in several ways. The most commonly described abbreviated breast MRI protocol consists of a single early-phase post-contrast sequence, which therefore does not allow for delayed-phase enhancement kinetic analysis. Our RAMP MRI protocol differs from the typical abbreviated breast MRI protocol by acquiring both early-phase and delayed-phase post-contrast sequences. Furthermore, although Grimm et al. previously described a similar abbreviated protocol (13), we are the first to demonstrate comparable performances of delayed-phase kinetic analyses between such an abridged MRI protocol and the full DCE MRI protocol in discriminating benign from malignant breast lesions. Our RAMP MRI protocol also includes one T2-weighted/bright fluid sequence and one T1-weighted non-fat suppressed sequence. As such, it serves as a complete breast MRI examination that can be acquired in approximately 10 minutes and is fully adequate for both screening and diagnostic indications without the need to bring patients back for further MRI evaluation of possible findings.
Delayed-phase enhancement patterns represent key kinetic parameters for differentiating benign and malignant breast lesions on MRI (25, 28, 30). Although malignancies can display any delayed-phase kinetic features, prior studies have identified malignancy in 76% to 87% of lesions with any washout (24, 25), indicating washout as a strong independent predictor of malignancy. Our study shows similar results, where 100% (9 of 9) of lesions with predominantly washout enhancement and 74% (45 of 61) of lesions with any washout enhancement proven to be malignant in the RAMP group; 80% (12 of 15) lesions with predominantly washout enhancement and 77% (43 of 56) of lesions with any washout enhancement proven to be malignant in the DCE group.
The overall distributions of the three delayed-phase enhancement patterns for benign and malignant lesions in our study are in keeping with previously reported results (27–29), considering varying MRI protocols across the different studies. Malignant lesions in our RAMP protocol (delayed phase at 180–205 seconds after contrast injection) demonstrated 46.0%, 33.0% and 11.0% of median delayed-phase lesion volume of persistent, plateau and washout curve types, respectively. These results are in agreement with the 45.5%, 32.5% and 12.0% of median delayed-phase volume of persistent, plateau and washout curve types, respectively, for malignant lesions in a prior study with delayed phase at 270 seconds after contrast injection (29). Delayed-phase washout was observed as the predominant curve type in 11.9–33.0% of malignant lesions in prior studies (27–29). Our results showing 14.8% (RAMP group) and 18.2% (DCE group) of malignant lesions with delayed-phase washout as the predominant curve type are within the range of previously reported data (27–29).
Notably, the ROC curve analysis showed comparable discriminating ability for benign and malignant lesions based on delayed-phase percentage washout and predominant curve type between our protocols and protocols from a prior study (29). Lesion classification based on delayed-phase percentage washout curve type produced an AUC of 0.66 (95% confidence interval [CI]: 0.53–0.78, RAMP group) and an AUC of 0.69 (95% CI: 0.54–0.80, DCE group); the AUC based on percentage washout curve type in the aforementioned study with delayed phase at 270 seconds was 0.70 (95% CI: 0.64–0.71). Lesion classification based on delayed-phase predominant curve type produced an AUC of 0.61 (95% CI: 0.47–0.74, RAMP group) and an AUC of 0.55 (95% CI: 0.40–0.68, DCE group); the AUC based on the predominant curve type in the prior study was 0.60 (95% CI: 0.54–0.65).
Our study underestimates the performances of delayed-phase kinetic parameters of both the RAMP and DCE MRI protocols in discriminating benign from malignant lesions. We included only lesions assessed as BI-RADS categories 4 and 5 that underwent tissue diagnosis and BIRADS category 6—known biopsy-proven malignancies. As such, the benign lesions included in this study were more likely to have displayed suspicious enhancement kinetics such as delayed-phase washout to prompt a suspicious BI-RADS category assessment. Both MRI protocols would show better discriminatory ability of benign vs. malignant lesions with greater AUCs based on kinetic analyses by including MRI lesions assessed as definitely benign (BI-RADS category 2) and probably benign (BI-RADS category 3), which more likely have non-concerning enhancement kinetics (e.g. persistent delayed-phase enhancement). Additionally, our analyses of lesion classification relied solely on delayed-phase kinetic parameters, without inputting additional useful information from the early-phase enhancement pattern. Lesion classification would improve when taking into consideration both early-phase and delayed-phase kinetic parameters; for example, 16.0% (8 out of 50) of benign lesions did not demonstrate any fast early-phase enhancement kinetic compared to 3.1% (4 out of 127) of malignant lesions included in this study.
Previous studies investigating breast MRI enhancement kinetics included lesions displaying early-phase enhancement above a pre-specified minimum threshold, BI-RADS category 3 lesions, and benign lesions that did not undergo histopathologic confirmation. These differences in study methods and inclusion criteria would account for the apparent underperformance, but in reality an underestimation, of our malignancy classification and ROC curve analysis results based on the delayed-phase worst curve type compared to prior studies (27–29). The percentages of malignant lesions in our study demonstrating delayed-phase washout as the worst curve type were 73.8% (RAMP group) and 65.2% (DCE group), compared to prior results of 76.2–92.7% (27–29). The AUC based on the delayed-phase worst curve type in our study were 0.49 (95% CI: 0.33–0.65, RAMP group) and 0.54 (95% CI: 0.39–0.68, DCE group), compared to the AUC of 0.67 (95% CI: 0.61–0.72) from the prior study with 270-second delayed phase (29). Most importantly, given that these factors equally impacted both RAMP and DCE groups, we found no significant difference in the results between our RAMP and DCE MRI protocols.
Our study also suggests improved delayed-phase kinetics characterization of malignant lesions observed with the RAMP MRI protocol compared to the full MRI protocol. Statistically significant differences were present in lesion volume percentages of persistent and plateau delayed-phase kinetics for malignant lesions between the two protocols. Malignant lesions demonstrated significantly lower percentage of persistent kinetics and greater percentage of plateau kinetics with the RAMP protocol than with the full DCE protocol. We theorize that due to earlier acquisition of the delayed phase in our RAMP protocol (180–205 seconds vs. 420–450 seconds for DCE MRI), malignant lesions that would eventually demonstrate persistent kinetics at a later delayed-phase time point with DCE MRI displayed plateau kinetics on RAMP MRI as they did not reach the ≥10% threshold on the delayed phase of RAMP MRI to qualify for persistent enhancement. The underlying pathophysiology accounting for this significant difference is unclear. One consideration is the possibility of presence of intermixing pixels representing intervening progressively enhancing background parenchyma within the malignant lesions/ROIs, which predominated at later delayed-phase time points on DCE MRI. Because plateau kinetics are considered more suspicious for malignancy whereas persistent kinetics are considered more frequently benign in etiology (25, 31), this significant difference with greater volume percentages of plateau than persistent kinetics among malignant lesions in the RAMP protocol suggests better performance of the RAMP protocol in detecting indeterminate or suspicious (plateau) kinetics among malignant lesions than the full protocol, which may provide diagnostic value. In short, visualization of more plateau kinetics and less persistent kinetics in malignancies, as seen in our RAMP MRI protocol, can more accurately guide radiologists toward higher suspicion for malignancies.
A retrospective study by Partridge et al. assessed the kinetic characteristics of lesions on delayed-phase sequences acquired at different time points during a single MRI examination, which demonstrated comparable diagnostic accuracy of delayed-phase kinetic analyses between sequences performed at an earlier and later time points, providing proof of concept for our RAMP MRI protocol (29). Our findings give support that beyond two minutes after contrast injection, the timing of our delayed-phase sequence may not significantly impact the discriminatory performance of delayed-phase kinetic analyses.
Our study has limitations. It was a retrospective, single-institution study. All DCE MRI scans were performed on a 1.5T scanner, whereas RAMP MRI scans were performed on both 1.5T and 3T scanners using different breast coils for the 1.5T versus 3T scanners. Two contrast agents were used during the study period due to a hospital-wide transition. Also, our study was not powered to perform subgroup analyses of performance based on clinical indication (screen vs. diagnostic), specific lesion type, or histologic subtypes.
In conclusion, our study demonstrates comparable delayed-phase enhancement kinetic analyses derived from our clinically implemented ACR-accredited RAMP MRI protocol and our full DCE MRI protocol. There was no significant difference between the two protocols in assessing washout kinetic characteristics of lesions and in discriminating benign from malignant lesions based on delayed-phase percentage washout, predominant curve type, or worst curve type. Our RAMP MRI protocol preserves the complete diagnostic information in lesion characterization and vital kinetic assessment of a full DCE MRI protocol. By reducing scan time, the RAMP MRI protocol improves patient comfort and workflow efficiency with similar advantages offered by abbreviated MRI, meets the ACR Breast MRI Accreditation requirements, and can be readily implemented across the full spectrum of clinical practices.
Disclosures:
Dr. Chou is a co-investigator on a General Electric research agreement; Dr. Kalpathy-Cramer is a consultant for Infotech Software Solution; Dr. Lehman is on the medical advisory board of General Electric Company and has a research grant with General Electric.
Abbreviations and Acronyms
- ACR
American College of Radiology
- AUC
area under the receiver operating characteristic curve
- BI-RADS
Breast Imaging Reporting and Data System
- CI
confidence interval
- DCE
dynamic-contrast enhanced
- FAST
first post-contrast subtracted
- MIP
maximum intensity projection
- MRI
magnetic resonance imaging
- NME
non-mass enhancement
- RAMP
rapid abridged multiphase
- ROC
receiver operating characteristic
- VIBRANT
Volume Imaging for Breast Assessment
Appendix
| RAMP MRI | DCE MRI | ||
|---|---|---|---|
| Three-plane localizer | Three-plane localizer | ||
| Axial pre-contrast 3D T1-weighted non-fat suppressed fast spoiled gradient echo (FSPGR) (1.5T) | Axial pre-contrast 3D T1-weighted non-fat suppressed Volume Imaging for Breast Assessment (VIBRANT) (3T) | Axial pre-contrast 3D T1-weighted non-fat suppressed fast spoiled gradient echo (FSPGR) | |
| TE | 4.2 ms | 2.3 ms | 4.2 ms |
| TR | 7.8 ms | 5 ms | 7.8 ms |
| Matrix | 384 × 256 | 400 × 400 | 384 × 256 |
| Slice thickness | 3 mm | 0.8 mm | 3 mm |
| FOV | 32 cm | 33 cm | 32 cm |
| Flip angle | 10 degrees | 10 degrees | 10 degrees |
| Axial 2D fast short tau inversion recovery (STIR) T2 (1.5T) | Axial 3D CUBE T2-weighted fast spin echo (3T) | Axial 2D fast short tau inversion recovery (STIR) T2 (1.5T) | |
| TE | 60 ms | 90 ms | 60 ms |
| TR | 6600 ms | 2500 ms | 6600 |
| Matrix | 320 × 192 | 352 × 352 | 320 × 192 |
| Slice thickness | 4 mm | 0.8 mm | 4 mm |
| FOV | 35 cm | 33 cm | 35 cm |
| Axial 3D T1-weighted fat suppressed VIBRANT series (1.5T) | Axial 3D T1-weighted fat suppressed VIBRANT series (3T) | Axial 3D T1-weighted fat suppressed VIBRANT series | |
| Pre-contrast | Yes | Yes | Yes |
| Phase 1 | 60–75 sec | 60–75 sec | 60–75 sec |
| Phase 2 | 180–205 sec | 180–205 sec | 180–300 sec |
| Phase 3 | - | - | 420–450 sec |
| Sagittal post-contrast | - | - | 480–600 sec |
| TE | 2.4 ms | 2.4 ms | 2.3 ms |
| TR | 6.4 ms | 6.4 ms | 5 ms |
| Matrix | 400 × 400 | 400 × 400 | 384 × 384 |
| Slice thickness | 0.8 mm | 0.8 mm | 2 mm |
| FOV | 33 cm | 33 cm | 32 cm |
| Flip angle | 10 degrees | 10 degrees | 10 degrees |
RAMP MRI = rapid abridge multiphase MRI, DCE MRI = full dynamic contrast-enhanced MRI, TE= Echo time, TR= Repetition time, Matrix= Frequency by Phase Encoding Matrix, FOV= Field of View
Footnotes
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
Entire work originated from: Department of Radiology, Massachusetts General Hospital, 55 Fruit St, WAC 240, Boston, MA 02114
This work has been presented at the 2017 annual RSNA meeting in Chicago, IL as a scientific oral presentation. This project was supported by a training grant from the National Institute of Biomedical Imaging and Bioengineering (NIBIB) of the National Institutes of Health under award number 5T32EB1680 to K. Chang. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health.
Contributor Information
Sadia Choudhery, Department of Radiology, Massachusetts General Hospital, Boston, MA.
Shinn-Huey S. Chou, Department of Radiology, Massachusetts General Hospital, 55 Fruit St, WAC 240, Boston, MA 02114.
Ken Chang, Athinoula A. Martinos Center for Biomedical Imaging, Department of Radiology, Massachusetts General Hospital, Boston, MA.
Jayashree Kalpathy-Cramer, Athinoula A. Martinos Center for Biomedical Imaging, Department of Radiology, Massachusetts General Hospital, Boston, MA.
Constance D. Lehman, Department of Radiology, Massachusetts General Hospital, 55 Fruit St, WAC 240, Boston, MA.
REFERENCES
- 1.Berg WA, Zhang Z, Lehrer D, et al. Detection of breast cancer with addition of annual screening ultrasound or a single screening MRI to mammography in women with elevated breast cancer risk. JAMA 2012; 307:1394–1404. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Mahoney MC, Gatsonis C, Hanna L, et al. Positive predictive value of BI-RADS MR imaging. Radiology 2012; 264:51–58. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Morrow M, Waters J, Morris E. MRI for breast cancer screening, diagnosis, and treatment. Lancet 2011; 378:1804–1811. [DOI] [PubMed] [Google Scholar]
- 4.Kuhl CK, Schrading S, Bieling HB, et al. MRI for diagnosis of pure ductal carcinoma in situ: a prospective observational study. Lancet 2007; 370:485–492. [DOI] [PubMed] [Google Scholar]
- 5.Lehman CD, Gatsonis C, Kuhl CK, et al. MRI evaluation of the contralateral breast in women with recently diagnosed breast cancer. N Engl J Med 2007; 356:1295–1303. [DOI] [PubMed] [Google Scholar]
- 6.Bluemke DA, Gatsonis CA, Chen MH, et al. Magnetic resonance imaging of the breast prior to biopsy. JAMA 2004; 292:2735–2742. [DOI] [PubMed] [Google Scholar]
- 7.Saslow D, Boetes C, Burke W, et al. American Cancer Society guidelines for breast screening with MRI as an adjunct to mammography. CA Cancer J Clin 2007; 57:75–89. [DOI] [PubMed] [Google Scholar]
- 8.Daly MB, Axilbund JE, Buys S, et al. Genetic/familial high-risk assessment: breast and ovarian. J Natl Compr Canc Netw 2010; 8:562–594. [DOI] [PubMed] [Google Scholar]
- 9.Mainiero MB, Lourenco A, Mahoney MC, et al. ACR Appropriateness Criteria Breast Cancer Screening. J Am Coll Radiol 2016; 13:R45–R49. [DOI] [PubMed] [Google Scholar]
- 10.Miles R, Wan F, Onega TL, et al. Underutilization of Supplemental Magnetic Resonance Imaging Screening Among Patients at High Breast Cancer Risk. J Womens Health (Larchmt) 2018; [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Kuhl CK, Schrading S, Strobel K, et al. Abbreviated breast magnetic resonance imaging (MRI): first postcontrast subtracted images and maximum-intensity projection-a novel approach to breast cancer screening with MRI. J Clin Oncol 2014; 32:2304–2310. [DOI] [PubMed] [Google Scholar]
- 12.Mango VL, Morris EA, David Dershaw D, et al. Abbreviated protocol for breast MRI: are multiple sequences needed for cancer detection? Eur J Radiol 2015; 84:65–70. [DOI] [PubMed] [Google Scholar]
- 13.Grimm LJ, Soo MS, Yoon S, et al. Abbreviated screening protocol for breast MRI: a feasibility study. Acad Radiol 2015; 22:1157–1162. [DOI] [PubMed] [Google Scholar]
- 14.Heacock L, Melsaether AN, Heller SL, et al. Evaluation of a known breast cancer using an abbreviated breast MRI protocol: Correlation of imaging characteristics and pathology with lesion detection and conspicuity. Eur J Radiol 2016; 85:815–823. [DOI] [PubMed] [Google Scholar]
- 15.Moschetta M, Telegrafo M, Rella L, et al. Abbreviated Combined MR Protocol: A New Faster Strategy for Characterizing Breast Lesions. Clin Breast Cancer 2016; 16:207–211. [DOI] [PubMed] [Google Scholar]
- 16.Harvey SC, Di Carlo PA, Lee B, et al. An Abbreviated Protocol for High-Risk Screening Breast MRI Saves Time and Resources. J Am Coll Radiol 2016; 13:R74–R80. [DOI] [PubMed] [Google Scholar]
- 17.Chen SQ, Huang M, Shen YY, et al. Abbreviated MRI Protocols for Detecting Breast Cancer in Women with Dense Breasts. Korean J Radiol 2017; 18:470–475. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Machida Y, Shimauchi A, Kanemaki Y, et al. Feasibility and potential limitations of abbreviated breast MRI: an observer study using an enriched cohort. Breast Cancer 2017; 24:411–419. [DOI] [PubMed] [Google Scholar]
- 19.Jain M, Jain A, Hyzy MD, et al. FAST MRI breast screening revisited. J Med Imaging Radiat Oncol 2017; 61:24–28. [DOI] [PubMed] [Google Scholar]
- 20.Panigrahi B, Mullen L, Falomo E, et al. An Abbreviated Protocol for High-risk Screening Breast Magnetic Resonance Imaging: Impact on Performance Metrics and BI-RADS Assessment. Acad Radiol 2017; 24:1132–1138. [DOI] [PubMed] [Google Scholar]
- 21.Petrillo A, Fusco R, Sansone M, et al. Abbreviated breast dynamic contrast-enhanced MR imaging for lesion detection and characterization: the experience of an Italian oncologic center. Breast Cancer Res Treat 2017; 164:401–410. [DOI] [PubMed] [Google Scholar]
- 22.Breast RA Magnetic Resonance Imaging (MRI) Accreditation Program Requirements. Reston, VA: American College of Radiology, 2011 [Google Scholar]
- 23.Morris EA CC, Lee CH, et al. ACR BI-RADS Magnetic Resonance Imaging In: ACR BI-RADS Atlas, Breast Imaging Reporting and Data System. Reston, VA: Amerian College of Radiology, 2013 [Google Scholar]
- 24.Schnall MD, Blume J, Bluemke DA, et al. Diagnostic architectural and dynamic features at breast MR imaging: multicenter study. Radiology 2006; 238:42–53. [DOI] [PubMed] [Google Scholar]
- 25.Kuhl CK, Mielcareck P, Klaschik S, et al. Dynamic breast MR imaging: are signal intensity time course data useful for differential diagnosis of enhancing lesions? Radiology 1999; 211:101–110. [DOI] [PubMed] [Google Scholar]
- 26.Williams TC, DeMartini WB, Partridge SC, et al. Breast MR imaging: computer-aided evaluation program for discriminating benign from malignant lesions. Radiology 2007; 244:94–103. [DOI] [PubMed] [Google Scholar]
- 27.Gutierrez RL, Strigel RM, Partridge SC, et al. Dynamic breast MRI: does lower temporal resolution negatively affect clinical kinetic analysis? AJR Am J Roentgenol 2012; 199:703–708. [DOI] [PubMed] [Google Scholar]
- 28.Wang LC, DeMartini WB, Partridge SC, et al. MRI-detected suspicious breast lesions: predictive values of kinetic features measured by computer-aided evaluation. AJR Am J Roentgenol 2009; 193:826–831. [DOI] [PubMed] [Google Scholar]
- 29.Partridge SC, Stone KM, Strigel RM, et al. Breast DCE-MRI: influence of postcontrast timing on automated lesion kinetics assessments and discrimination of benign and malignant lesions. Acad Radiol 2014; 21:1195–1203. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30.Kuhl CK, Schild HH, Morakkabati N. Dynamic bilateral contrast-enhanced MR imaging of the breast: trade-off between spatial and temporal resolution. Radiology 2005; 236:789–800. [DOI] [PubMed] [Google Scholar]
- 31.El Khouli RH, Macura KJ, Jacobs MA, et al. Dynamic contrast-enhanced MRI of the breast: quantitative method for kinetic curve type assessment. AJR Am J Roentgenol 2009; 193:W295–300. [DOI] [PMC free article] [PubMed] [Google Scholar]