Return To Abstract Listing

Scientific Session 03 — Breast - Digital Breast Tomosythesis

Monday, May 1, 2017

Abstracts 1108-2863

2863. Pathologic Upgrade Rates of High-Risk Lesions With Digital 2D Versus Tomosynthesis Mammography

Lamb L*,  Bahl M,  Hughes K,  Lehman C. Massachusetts General Hospital, Boston, MA

Address correspondence to M. Bahl (

Objective: Research in the pretomosynthesis era demonstrated that most high-risk histologic breast lesions diagnosed by core needle biopsy are not associated with malignancy, yet the majority of patients undergo surgical excision because of reported upgrade rates to malignancy of 5% to 30%. The pathologic upgrade rates of high-risk lesions (HRLs) detected on digital breast tomosynthesis (DBT) have not been reported to our knowledge. The purpose of this study is to compare the pathologic upgrade rates of HRLs in the breast detected on digital mammography (DM) versus DBT.

Materials and Methods: The study cohort was comprised of patients with HRLs diagnosed by image-guided core needle biopsy from December 2007 to February 2011 (DM group, before DBT integration) and from January 2013 to March 2016 (DBT group, after DBT integration). HRLs include atypical ductal hyperplasia (ADH), atypical lobular hyperplasia (ALH), lobular carcinoma in situ (LCIS), radial scars, papillary lesions, flat epithelial atypia (FEA), nonspecific atypia, and fibroepithelial lesions. Medical records were reviewed for surgical outcomes and follow-up imaging. Z tests were used to calculate differences in upgrade rates between the DM and DBT groups.

Results: The analysis set was comprised of 1025 lesions in 996 patients with HRLs diagnosed by image-guided biopsy who subsequently underwent surgical excision (951 lesions, 92.8%) or had at least 2 years of imaging follow-up (74 lesions, 7.2%). The most common HRL was ADH in both the DM and DBT groups (327/1025 [31.9%]). ADH represented a higher proportion of HRL diagnoses in the DM group than in the DBT group (35.4% vs 28.7%, p = 0.02), and papillary lesions and ALH represented a lower proportion (6.7% vs 12.4% and 6.9% vs 10.7%, p < 0.04 for both). The overall pathologic upgrade rate of HRLs to ductal carcinoma in situ (DCIS) or invasive malignancy was 12.8% (131/1025). There were no statistically significant differences in overall upgrades rates of HRLs detected on DM versus DBT (61/491 [12.4%] vs 70/534 [13.1%], p = 0.74) or in upgrade rates of HRL subtypes detected on DM versus DBT. However, the upgrade rate to invasive carcinoma (rather than DCIS) was higher for HRLs detected on DBT than DM (31/70 [44.3%] vs 15/61 [24.6%], p = 0.02). Overall, the most common HRL to upgrade was ADH (65/327 [19.9%]) and least common was fibroepithelial lesions (1/48 [2.1%]). None of the 23 fibroepithelial lesions detected on DBT upgraded to malignancy.

Conclusion: Although the risk of upgrade of HRLs to invasive carcinoma is higher with DBT than is observed with DM, the upgrade rate to in situ disease is higher with DM. Surveillance rather than surgical excision of fibroepithelial lesions may be a reasonable option for patients evaluated with DM or DBT.