RPS 602a - Contrast-enhanced breast MRI and beyond

Purpose:

To investigate whether the additional use of ultrafast MRI can improve the diagnostic performance of conventional dynamic contrast-enhanced MRI (DCE-MRI) in evaluating MRI-detected lesions in breast cancer patients.

Methods and materials:

This retrospective study enrolled 103 consecutive breast cancer patients with 204 breast lesions (62 benign and 142 malignant) who underwent preoperative DCE-MRI with ultrafast imaging (9 image sets with 6.5-second temporal resolution). Two reviewers assessed the BI-RADS categories of breast lesions using conventional DCE-MRI and assessed the following parameters on the ultrafast MRI: initial enhancement phase, maximum relative enhancement, slope, and maximum slope (slope_max) on the kinetic curve. Interobserver agreement was analysed between the two reviewers. The ultrafast MRI parameters were compared between benign and malignant tumours, and cut-off values were determined. For 97 additional MRI-detected lesions, the BI-RADS category was re-assessed using the cut-off values and the diagnostic performance was compared between the conventional DCE-MRI and the combined conventional and ultrafast DCE-MRI.

Results:

All ultrafast MRI parameters differed significantly between malignant and benign tumours (all, p<0.001). Initial enhancement phase by reviewer and slope_max were the top two parameters showing significant differences between benign and malignant tumours with high reliability. With the use of cut-off values for initial enhancement phase (≤phase 2) and slope_max (>9.8%/sec), the specificity of conventional DCE-MRI was significantly increased (29.4% vs 64.7%, p<0.001) without loss of sensitivity (100% vs 88.2%, p=0.157) in evaluating masses.

Conclusion:

The additional use of ultrafast MRI can improve the specificity of conventional DCE-MRI when evaluating MRI-detected masses in breast cancer patients.

Limitations:

A large proportion of malignant lesions might have influenced the study results. The kinetic curve analysis for conventional and ultrafast DCE-MRI was performed by manually selecting the most enhanced voxel.

Ethics committee approval

n/a

Funding:

No funding was received for this work.

Purpose:

Breast diffusion-weighted imaging (DWI) appears to have strong potential for prediction of therapy outcome in patients treated with neoadjuvant systemic therapy (NST). While summary statistics show decent results in recent meta-analyses, absolute data are variable between studies. This systematic review aims to identify methodological differences between studies and their effect on reported results in order to improve the value of DWI as a biomarker for the monitoring of NST.

Methods and materials:

Pubmed was searched for papers reporting on DWI for response prediction/monitoring/evaluation of NST in breast cancer, published before February 2019. The selection was first performed on title/abstract by two independent researchers and, subsequently, full texts were screened for eligibility. Quality of selected studies was assessed using QUADAS2. Population and cancer characteristics (e.g. age, tumour types, type of NST), MRI parameters (e.g. field strength, b-values, voxel size), and post-processing approaches (e.g. ADC mapping, region of interest (ROI) selection) were extracted.

Results:

The search yielded 31 papers. After assessment, 15 papers were included in this analysis. Risk of bias of included studies was low. Studies varied with regard to breast cancer types, NST types and pCR-definition (in-/exclusion of residual DCIS). Technical variation was observed in MRI-settings including b-values, ROI-selection, and timing of the MRI scans. These differences all contribute to the observed overlapping ranges of ADC values for pCR and non-pCR.

Conclusion:

This systematic review reveals strong heterogeneity in DWI-studies on prediction of NST response. Consequently, DWI metrics for response monitoring should be interpreted with caution. Standardisation of the technique is important.

Limitations:

The risk of bias of included studies was low.

Ethics committee approval

n/a

Funding:

No funding was received for this work.

Purpose:

To determine BI-RADS categories of patients based on diffusion-weighted imaging (DWI) by a novel evaluation model and compare DWI-derived categories to the previous multiparametric magnetic resonance imaging (MP-MRI) interpretations of the same group and the histopathology data.

Methods and materials:

The anonymised DWIs of consecutive 400 patients were evaluated retrospectively according to a classification system based on lesion morphology and pattern of diffusion changes. To investigate the performance of DWI, the highest BI-RADS category for each patient was determined and the results were compared with BI-RADS categories found in MP-MRI reports, and, in possible cases, were also compared to pathology results. nADC values were measured and the diagnostic performance of nADC in benign and malignant discrimination and the threshold values were analysed.

Results:

Evaluation system based on DWI findings provided 64% sensitivity, 98% specificity, 90% positive predictive value (PPV), 89% negative predictive value (NPV), and 89.5% accuracy when the MP-MRI was accepted as the reference standard. 87% sensitivity, 68% specificity, 70% PPV, 86% NPV, and 77% accuracy were determined according to pathology results. A threshold value of 0.784 for nADC provided 68% sensitivity, 73% specificity, 72% PPV, 69% NPV, and 71% accuracy for benign-malignant discrimination. nADC values were significantly lower in malignant lesions.

Conclusion:

In this study, DWI-derived new conceptualise provided acceptable accuracy rates. Although DWI is a functional examination, its morphological data may also aid in the characterisation of suspicion of lesions.

Limitations:

The evaluation was made by two radiologists in consensus, intraobserver and interobserver comparisons could not be made. For technical reasons, some patients were excluded. No reflection of the general population due to MRI indications. Retrospective design.

Ethics committee approval

DYBEAH 29.04.2019-62/18.

Funding:

No funding was received for this work.

Purpose:

To assess the performance of diffusion-weighted imaging (DWI) for breast cancer detection and characterisation by combining acquired b-values (800 s/mm2) and different synthetic b-values (1000, 1200, 1500, and 1800 s/mm2).

Methods and materials:

In this IRB-approved retrospective study, 80 women with 103 benign and malignant breast lesions underwent multiparametric breast MRI at 3T with dynamic contrast-enhanced images (DCE-MRI) and DWI were included. Three readers (r1, r2, r3) independently evaluated CE-MRI and DWI combined with ADC maps for lesion visibility, malignancy likelihood, and preferred b-value. Histopathology was the standard of reference for all the lesions.

Results:

Synthetic intermediate b-values (1200–1500 s/mm2) provided the best lesion conspicuity and image quality across all readers. Only 9 (8.7%) enhancing lesions (5 benign, 4 malignant) were missed by all readers, comprising diffuse tumours and masses <1cm. Overall lesion detectability for DWI ranged between 79.6%-83.4% and was 100% for DCE-MRI.

Sensitivity and specificity for lesion characterisation with DWI was r1 61.7% and 94.1%, r2 61.7% and 82.4%, and r3 65% and 82.4%, respectively. Sensitivity and specificity for DCE-MRI was r1 88.3% and 76.5%, 95% and 50%, and r3 86.7% and 70.6%. DWI was significantly less sensitive across all readers (p<0.001) but significantly more specific for two readers (p=0.007) than DCE-MRI.

Conclusion:

Synthetic intermediate b-values (1200–1500 s/mm2) provided the best lesion conspicuity and image quality. DWI with synthetic high b-values increases specificity as a valuable adjunct to DCE-MRI.

Limitations:

No normal cases were included in this study and therefore specificity values were unable to be obtained for lesion detectability.

Ethics committee approval

IRB-approved.

Funding:

Spanish Foundation Alfonso Martin Escudero, the NIH/NCI Cancer Center Support Grant (P30 CA008748) and the Breast Cancer Research Foundation.

Purpose:

To evaluate the diagnostic performance of multiparametric magnetic resonance imaging (mpMRI) with dynamic contrast-enhanced (DCE) and diffusion-weighted imaging (DWI) in non-mass enhancing (NME) breast tumours.

Methods and materials:

A prospectively populated database was searched for patients who underwent MRI of the breast with T2wi, DWI, and DCE MRI for a BIRADS 45 imaging abnormality, and demonstrated a suspicious NME tumour on DCE-MRI NME according to the 5th edition of the BI-RADS lexicon. Two readers independently assessed DWI and DCE, as well as the two combined as mpMRI. Appropriate statistical tests were used to compare diagnostic values with respect to sensitivity, specificity, diagnostic accuracy, and AUC.

Results:

66 patients with 66 NME breast lesions were included: 37/66 (56%) malignant and 29/66 (44%) non-malignant. DCE-MRI had a sensitivity and specificity of 94.9%, 67% (r1) and 100%, 77.8% (r2). DWI 84.1%,55.6% (r1) and 76.9%,81.5% (r2), respectively. Diagnostic accuracy was 78.8% (r1), 72.7% (r2) for DWI, and 83.3% (r1), 90.1% (r2) for DCE-MRI, respectively. When both parameters were used together as mpMRI, sensitivity was 87.2% (r1), 90% (r2), and specificity 85.2% (r1 and r2), resulting in the best diagnostic accuracy of 90.6% (r1) and 86.2% (r2). MpMRI allowed a reduction of false positives of 33.3% (6 vs 4) (r1) and 55.5% (9 vs 4) (r2). There was substantial inter-reader agreement for both DCE-MRI and MpMRI readings (k=0.610 vs k=0.719), while scarce for DWI (k=0.380)

Conclusion:

Multiparametric MRI using DWI and DCE-MRI increases diagnostic accuracy for breast cancer detection for lesions presenting as NME on DCE MRI.

Limitations:

Retrospective study.

Ethics committee approval

Institutional review board-approved study

Funding:

Breast Cancer Research Foundation.

Purpose:

We aimed to evaluate the diagnostic performance of multiparametric approach for breast lesions using maximum slope (MS) from ultrafast dynamic contrast-enhanced (UF-DCE) sequence, apparent diffusion coefficient (ADC), lesion size, and patient’s age.

Methods and materials:

Two different groups with different UF-DCE sequences were analysed. Group 1: 204 lesions (152 malignant, 52 benign) with k-space-weighted image contrast (KWIC) sequence, and group 2: 114 lesions (78 malignant, 36 benign) with improved 3D-gradient-echo volumetric interpolated breath-hold examination (VIBE) sequence with compressed sensing (both were prototype sequences). Each UF-DCE sequence acquired the image up to 1 min after gadolinium injection. DWI sequences were the same between two groups with b value of 0 and 1,000 s/mm². MS and ADC values were measured by two radiologists and their inter-reader agreements were evaluated using intra-class correlation coefficients (ICC). MS derived from UF-DCE sequences, ADC value from DWI, lesion size, and patients’ age were used for a multiparametric model using ROC analysis. Univariate and multivariate logistic regression analysis were performed.

Results:

Inter-rater agreements were excellent (≥0.95). Univariate logistic regression analysis showed that all parameters differed significantly between malignant and benign lesion (p<0.05). AUC of MS and ADC were 0.74/ 0.92 in group 1 and 0.77/ 0.86 in group 2. The combination of all parameters using multiparametric model yielded an AUC of 0.95 for group 1 and 0.91 for group 2, respectively.

Conclusion:

A multiparametric approach based on UF-DCE MRI, DWI, lesion size, and patient’s age provides excellent diagnostic performance.

Limitations:

Retrospective design.

Ethics committee approval

This study protocol was approved by our institutional review board. The written informed consent was waived because of the retrospective design.

Funding:

JP15K09922, MEXT 25120002, 25120008.

Purpose:

Diffusion-weighted imaging (DWI) forms an integral part of breast MRI to detect cancer and assess disease response following neoadjuvant chemotherapy. Apparent diffusion coefficient (ADC) allows quantification of diffusion signal and facilitates differentiating benign from malignant lesions. Zoom DWI combines 2D spatially-selective RF excitation pulses for single-shot echo-planar-imaging with reduced field-of-view in the phase-encoding direction. It leads to a decreased number of k-space acquisition lines, significantly shortening EPI echo train and susceptibility artefacts, hence reducing image distortion and blurr. This study evaluates the feasibility of zoom DWI versus standard DWI in terms of ADC quantification.

Methods and materials:

25 patients with 30 lesions, 18 malignant, including 3 post NACT followup, and 12 benign (histopathologically proven) were scanned on 3T MRI, Ingenia, Philips, using a dedicated multichannel phased-array breast coil. The comparative protocol comprised of zoom DWI and routine DWI sequence (b0 & b1000). Both sets of images were used to generate ADC maps using in-house diffusion tool. ROI were drawn at corresponding points to obtain ADC1 (zoom DWI) and ADC2 (routine DWI) by the same radiologist with 14 years experience in reading breast MRI. The mean and standard deviation of ADC values were noted.

Results:

ADC1 values were lower than ADC2 in malignant lesions and were higher in most benign and post NACT lesions. Reference cut-off used for malignant lesions was < 1.2x10-2 mm2/s and for benign > 1.2. The explanation could be better image quality with less or no image distortion, blurring, and aliasing in zoom DWI, hence improved ADC map generation and accurate quantification.

Conclusion:

Zoom DWI enables accurate breast lesion characterisation, especially for those with borderline ADC values, into benign and malignant on 3T imaging due to better image quality.

Limitations:

n/a

Ethics committee approval

n/a

Funding:

No funding was received for this work.

Purpose:

To assess breast DTI (mean diffusivity (MD) and fractional anisotropy (FA)) in patients with invasive ductal carcinoma (IDC) and ductal carcinoma in situ (DCIS) in comparison with DWI (apparent diffusion coefficient (ADC)).

Methods and materials:

A retrospective analysis of breast MRI (N=30) included consecutive patients with unilateral lesion: IDC (n1=15) and DCIS (n2=15). All patients were examined with a 1.5T MRI system with DTI and DCE sequences. FA, MD, and ADC were computed for all lesions, and FA and MD were computed for parenchyma. Mann Whitney U test and Spearman correlations were performed.

Results:

MD is significantly different between IDC and DCIS: 0.991+/-0.006 vs. 1.331+/-0.053, p<0.0001. FA is not considered significantly different between IDC and DCIS: 0.300+/-0.021 vs. 0.318+/-0.027, p=0.085. MD and FA are considered significantly different between DCIS and parenchyma (1.331+/-0.053 vs. 1.627+/-0.026, p<0.0001; 0.318+/-0.027 vs. 0.351+/-0.021, p=0.002, respectively). Correlation between DTI parameters in DCIS (MD and FA) and ADC was considered significant (r=0.581, p=0.023; r=0.710, p=0.003, respectively).

Conclusion:

MD in DCIS is significantly lower in comparison with values for IDC and is significantly higher than parenchymal MD. FA is significantly different in DCIS compared to parenchyma. In correlation with ADC, DTI parameters may contribute to a better characterisation of non-mass enhancement.

Limitations:

A retrospective, single-centre study on a small sample of patients may lead to lower generalisability. Manually drawn ROI definition may be the source of bias. Technical limitations include EPI-based artefacts and the 1.5T field strength in DTI.

Ethics committee approval

Referent board approval obtained for retrospective analysis.

Funding:

No funding was received for this work.