What’s new in Breast Imaging – July 2020

In this blog, we provide a synopsis of the latest in Breast Imaging published during the month of June 2020.

Are Breast Imagers Making Appropriate Follow Up Recommendations for Incidental Liver Lesions on Breast MRI?

Dipiro, P. J., Alper, D. P., Giess, C. S., Glazer, D. I., Lee, L. K., Lacson, R., & Khorasani, R. (2020). Comparing Breast and Abdominal Subspecialists’ Follow-Up Recommendations for Incidental Liver Lesions on Breast MRI. Journal of the American College of Radiology, 17(6), 773–778. doi: 10.1016/j.jacr.2019.12.024

https://pubmed.ncbi.nlm.nih.gov/32004482/

In this retrospective study, breast imagers’ follow up recommendations for incidental liver lesions (ILLs) on breast MRI were compared to abdominal subspecialists’ opinions. A total of 2,181 breast MRIs were divided into three categories: (1) ILLs with follow up recommendations, (2) ILLs without follow up recommendations, and (3) studies without ILLs. Two experienced abdominal radiologists reviewed all MRI liver findings and made follow up recommendations. Approximately 11% of all breast MRIs contained ILLs, and breast imagers recommended follow up for 12% of ILLs. Abdominal radiologists agreed with 97% cases in which breast imagers identified ILLs that did not require follow up. However, abdominal radiologists disagreed with 93% of cases in which breast imagers identified ILLs that required follow up. Abdominal subspecialists agreed on 93% of cases in which breast imagers did not identity ILLs on breast MRI. The authors concluded that there is a high disagreement rate between breast imagers and abdominal radiologists when breast imagers recommended follow up for ILLs on breast MRI.  The authors suggested that abdominal subspecialty consultation or integrated liver imaging decision support might reduce unnecessary follow up for ILLs on breast MRI.

Does Breast Tumor Kinetic Heterogeneity Predict Development of Distant Metastasis?

Kim, J. Y., Kim, J. J., Hwangbo, L., Suh, H. B., Kim, S., Choo, K. S., … Kang, T. (2020). Kinetic Heterogeneity of Breast Cancer Determined Using Computer-aided Diagnosis of Preoperative MRI Scans: Relationship to Distant Metastasis-Free Survival. Radiology, 295(3), 517–526. doi: 10.1148/radiol.2020192039

https://pubs.rsna.org/doi/10.1148/radiol.2020192039

In this retrospective study, the authors investigated the relationship between breast cancer kinetic kinetic heterogeneity and mean time to distant metastasis development. Preoperative breast MRIs from 276 consecutive patients were evaluated with computer-aided diagnosis (CAD) software, which quantified tumor kinetic heterogeneity and peak enhancement. Compared to women without distant metastasis, women with distant metastasis exhibited a higher degree of tumor kinetic heterogeneity (p<0.001), higher peak enhancement (p=.045), and higher presence of lymphovascular invasion (p=.044). The authors concluded that higher values of tumor kinetic heterogeneity and peak enhancement are associated with a shorter time to development of distant metastasis. This information may assist radiologists in predicting disease course with preoperative breast MRI.

How Does AI-based Decision Support Influence the Assessment of Sonographic Breast Lesions?

Mango, V. L., Sun, M., Wynn, R. T., & Ha, R. (2020). Should We Ignore, Follow, or Biopsy? Impact of Artificial Intelligence Decision Support on Breast Ultrasound Lesion Assessment. American Journal of Roentgenology, 214(6), 1445–1452. doi: 10.2214/ajr.19.21872

https://www.ajronline.org/doi/full/10.2214/AJR.19.21872

This multicenter retrospective review study investigated the utility of artificial intelligence (AI) decision support (DS) in the interpretation of breast lesions identified on ultrasound (US). Fifteen physicians (11 radiologists, 2 obstetric/gynecologists, and 2 surgeons) reviewed 900 sonographic breast lesions in two sittings. One session included review with US only, and the second session included US plus DS. Data points included DS and reader sensitivity, specificity, and positive likelihood ratio (PLR). Area under the curve (AUC) was also calculated for readers with and without the use of DS. Mean reader AUC with US plus DS was 0.87 compared to 0.83 for US only. PLR was higher for all readers with US plus DS except for one reader. In addition, US plus DS reduced inter- and intrareader variability. The study concluded that US plus DS improved accurate assessment of sonographic breast lesions for most physicians in addition to reducing inter- and intraobserver variability.

2D Synthetic Mammography versus Full-Field Digital Mammography: Which is more Accurate in Microcalcifcation Detection?

Dodelzon, K., Simon, K., Dou, E., Levy, A. D., Michaels, A. Y., Askin, G., & Katzen, J. T. (2020). Performance of 2D Synthetic Mammography Versus Digital Mammography in the Detection of Microcalcifications at Screening. American Journal of Roentgenology, 214(6), 1436–1444. doi: 10.2214/ajr.19.21598

https://www.ajronline.org/doi/abs/10.2214/AJR.19.21598

This retrospective performance study compared the accuracy of microcalcification detection between 2D synthetic mammography (SM) and full-field digital mammography (FFDM). The authors also sought to determine radiologists’ preferences between these two modalities for evaluating calcifications. The reviewed studies included 70 recalled screening mammograms with microcalfications and 90 mammograms with negative findings.  In individual sessions, radiologists interpreted either SD or FFDM mammograms by assigning a BI-RADS assessment. In a final session, breast imagers assigned their level of conspicuity and diagnostic confidence after review of the 70 mammograms with microcalcifications.  There was no statistical difference in diagnostic accuracy between SM (AUC 88%) and FFDM (AUC 91%) with p=0.392. In addition, SM diagnostic accuracy was noninferior to FFDM with p=0.011. Sensitivities of SM and FFDM were 77% and 73%, respectively, and not significantly different (p=0.366). All radiologists found microcalcifications more conspicuous on SM, but subjective diagnostic confidence was not statistically different. The authors concluded that detection of microcalcifications on SM is noninferior to FFDM. In addition, radiologists felt a similar degree of diagnostic confidence in detecting microcalfications with both modalities.

Do intramammary edema and intratumoral necrosis on T2- weighted breast MRI determine treatment response for triple-negative breast cancer?

Harada, T. L., Uematsu, T., Nakashima, K., Sugino, T., Nishimura, S., Takahashi, K., … Ito, T. (2020). Is the presence of edema and necrosis on T2WI pretreatment breast MRI the key to predict pCR of triple negative breast cancer? European Radiology, 30(6), 3363–3370. doi: 10.1007/s00330-020-06662-7

https://pubmed.ncbi.nlm.nih.gov/32062698/

In this retrospective study, the authors sought to determine if intramammary edema (intra-E) and intratumoral necrosis (intra-N) on T2-weighted MRI are useful predictors of pathological complete response (pCR) in breast cancer patients. The study group included patients with biopsy-proven triple negative breast cancer with invasive carcinoma of no specific type histological subtype (TNBC-NST). The authors reviewed fifty-seven breast MRIs prior to patients receiving neoadjuvant therapy, and degrees of intra-E and intra-N were evaluated on T2-weighted images. Intra-E and intra-N did not correlate with pCR in this specific population of breast cancer patients. However, the authors concluded that tumor size positively correlated with intra-E.

Can quantitative total choline (tCho) values derived from MR spectroscopy characterize suspicious breast lesions on multiparametric MRI?

Sodano, C., Clauser, P., Dietzel, M., Kapetas, P., Pinker, K., Helbich, T. H., … Baltzer, P. A. (2020). Clinical relevance of total choline (tCho) quantification in suspicious lesions on multiparametric breast MRI. European Radiology, 30(6), 3371–3382. doi: 10.1007/s00330-020-06678-z

https://pubmed.ncbi.nlm.nih.gov/32065286/

In this prospective study, the authors investigated the added value of quantified total choline (tCho) in diagnosing malignancy and metastatic lymph nodes in suspicious breast lesions on multiparametric breast MRI. One hundred and twenty-one patients with suspicious lesions on multiparametric breast MRI were enrolled and underwent single-voxel proton MR spectroscopy with and without water suppression (103 patients had full imaging data at time of analysis). Two experienced radiologists measured and quantified tCho as mm/L. ROC-analysis was used to assess the relationship between malignancy and lymph node status with tCho values. Areas under the ROC curve for malignancy prediction with tCho values for the two radiologists were 0.816 and 0.809. It was determined that tCHo values >0.8 mm/L were diagnostic of malignancy with >95% sensitivity. Metastatic lymph nodes were not seen in patients with tCHo values <2.4 mm/L. The authors concluded that quantitative tCho not only differentiates benign and malignant suspicious breast lesions but also forecasts metastatic lymph node status.

Does Contrast-Enhanced Ultrasound Outperform Breast MRI in Breast Cancer Detection?

Li, C., Yao, M., Shao, S., Li, X., Li, G., & Wu, R. (2020). Diagnostic efficacy of contrast-enhanced ultrasound for breast lesions of different sizes: a comparative study with magnetic resonance imaging. The British Journal of Radiology, 93(1110), 20190932. doi: 10.1259/bjr.20190932

https://pubmed.ncbi.nlm.nih.gov/32216631/

In this prospective study, the authors compared the diagnostic accuracy of contrast-enhanced ultrasound (CEUS), MRI, and a combination of the two modalities in evaluating breast lesions of different sizes. Two size groups included lesions less than or equal to 20 mm and lesions greater than 20 mm. Four hundred and six breast lesions seen on conventional ultrasound underwent MRI and CEUS (194 lesions were benign and 212 were malignant according to histological analysis). MRI and CEUS alone demonstrated similar sensitivities for detecting breast cancer in all size categories (p=1.000). Combined MRI and CEUS modalities outperformed CEUS alone for all sized lesions (p<0.05). Combined modalities superseded the diagnostic accuracy of MRI alone only in lesions greater than 20mm.  The authors concluded that CEUS alone has comparable sensitivity to MRI for detecting breast cancer, and combined modalities improved specificity and accuracy of breast cancer detection.

Mammographic Architectural Distortion: Can Breast MRI Reliably Exclude Malignancy?

Amitai, Y., Scaranelo, A., Menes, T. S., Fleming, R., Kulkarni, S., Ghai, S., & Freitas, V. (2020). Can breast MRI accurately exclude malignancy in mammographic architectural distortion? European Radiology, 30(5), 2751–2760. doi: 10.1007/s00330-019-06586-x

https://pubmed.ncbi.nlm.nih.gov/32002641/

In this retrospective review study, the authors sought to determine the diagnostic accuracy of breast MRI to exclude malignancy in patients with mammographic architectural distortion (MAD). From January 2008 to September 2018, one hundred and seventy-five patients with confirmed MAD underwent problem-solving breast MRI. Patient outcomes were defined by pathology (from biopsy or surgical excision) or at least one year of radiological follow up. Out of the 106 patients with negative breast MRIs, zero cancers were diagnosed. The negative predictive value for excluding malignancy in MAD cases with breast MRI was 100% (Confidence Interval 95 to 100%). The authors concluded that malignancy is accurately excluded in patients with MAD and negative breast MRIs, which may prevent unnecessary tissue sampling for diagnosis.

How Should the Breast Imager Approach Suspected Breast TB?

Mathew, D., Rubin, G., Mahomed, N., & Rayne, S. (2020). Imaging and  in clinical features of breast tuberculosis: a review series of 62 cases. Clinical Radiology, 75(7). doi: 10.1016/j.crad.2020.03.017

https://www.sciencedirect.com/science/article/pii/S0009926020301161

In this retrospective study, the authors assessed the disease burden of breast tuberculosis (TB) in three South African tertiary hospitals. Additionally, the authors correlated demographic, clinical, and imaging findings for breast TB cases. Over an 18-month period 2,516 patients underwent laboratory workup for breast disease.  Of these patients, 2.5% (62 patients) were diagnosed with breast TB with a median age at presentation of 38.5 years. Of note, 80% of these patients were HIV positive. The most common presentation of breast TB cases was breast abscess. In 92% of cases, histological analysis showed necrotizing granulomatous inflammation. Interestingly, mycobacterium tuberculosis cultures were only positive in 27% of cases. The authors concluded that suspected breast TB should be evaluated with clinical assessment, ultrasound, mammogram, and FNA/ultrasound (US)-guided core biopsy. The authors also recommended use of US-guided core biopsy over FNA as this sampling method provided better tissue samples and positive results.

Do Silicon Skin Markers Compromise Image Interpretability in Automated Breast Ultrasound?

Jong, L. D., Welleweerd, M. K., Zelst, J. C. V., Siepel, F. J., Stramigioli, S., MNine ann, R. M., … Fütterer, J. J. (2020). Production and clinical evaluation of breast lesion skin markers for automated three-dimensional ultrasonography of the breast: a pilot study. European Radiology, 30(6), 3356–3362. doi: 10.1007/s00330-020-06695-y

https://pubmed.ncbi.nlm.nih.gov/32060713/

In this prospective pilot study, the authors sought to determine if a palpable skin marker utilized during automated breast ultrasound compromised image interpretability. Sixteen patients were prospectively enrolled in this IRB-approved study. Patients underwent automated breast ultrasound twice, including with and without a silicone disk placed on the skin. Nine radiologists reviewed the images and rated image quality, image interpretability, and diagnostic confidence. Although observer scores for image quality and image noise decreased from “excellent” to “good” in the presence of a palpable marker (p<0.01), 93% of radiologists agreed that image noise did not influence image interpretability. The authors concluded that easily producible silicone skin markers do not decrease image interpretability during automated breast ultrasound.