Incidence round screening performance among women with dense breasts undergoing abbreviated breast MRI and digital breast tomosynthesis (ECOG-ACRIN EA1141).

Abstract

10502 Background: The sensitivity of mammography, including digital breast tomosynthesis (DBT), is limited by breast density. Abbreviated breast MRI (AB-MR) significantly increases breast cancer detection in women with dense breasts. In the prevalence (baseline) screening round of the EA1141 trial, AB-MR offered a 2.45-fold higher cancer detection rate (CDR) than DBT. Here we report the CDR and diagnostic accuracies of AB-MR and DBT during the EA1141 incidence screening round and the overall interval cancer rate. Methods: Informed consent was obtained from all participants. Asymptomatic average risk women aged 40-75 years with mammographically dense breasts scheduled for routine screening with DBT were enrolled. All women underwent DBT and AB-MR at baseline (prevalence) and year 1 (incidence) screens which were interpreted independently by 2 different radiologists blinded to the other modality. Women were contacted 11-13 months after the year 1 screen to assess for a subsequent breast cancer diagnosis. Pathology of biopsy was the reference standard for CDR and positive predictive value of biopsy (PPV3). Interval cancers reported during 11-13 months of follow-up until the next annual screening were included in the reference standard for sensitivity and specificity. A post hoc adjustment was made for multiple testing using the Bonferroni method; p-values are reported for comparisons which met the adjusted significance threshold (0.05/5 = 0.01). Reported 95% confidence intervals were not adjusted for multiplicity. Results: Of 1516 enrolled subjects, 1444 completed the baseline and 1291 also the incidence screen. During the latter, 9 women were diagnosed with breast cancer. DBT detected cancer in 5 women (3 DCIS, 2 invasive) for a CDR of 3.9/1000 (5/1291, CI 1.7, 9.0). AB-MR detected cancer in 6 women (0 DCIS, 6 invasive), 4 of which were not detected on DBT, for a CDR of 4.6/1000 (6/1291, CI 2.1, 10.1). The additional imaging (subject-level) rate for DBT was 7.9% (102/1291, CI 6.6%, 9.5%) vs 3.7% (48/1291, CI 2.8%, 4.9%) for AB-MR (p < 0.001). The PPV3 (lesion-level) for DBT was 29.4% (5/17, CI 13.5%, 52.6%) and 15.0% (6/40, CI 6.9%, 29.6%) for AB-MR. During the 2 years of study screening (baseline and year 1), there was 1 interval cancer (palpable lump 348 days after year 1 screen), for an overall interval cancer rate of 0.37/1000 person-years (1/2677 person-years, CI 0.05, 2.65). Sensitivity and specificity were 50.0% (5/10, CI 23.7%, 76.3%) and 98.5% (1261/1280, CI 97.7%, 99.0%) for DBT vs 60% (6/10, CI 31.3%, 83.2%) and 93.3% (1193/1278, CI 91.8%, 94.6%, p < 0.001) for AB-MR. Conclusions: On the incidence screening round, AB-MR detected additional invasive breast cancers not seen on DBT. In addition, a low overall interval cancer rate was observed with combined screening. Although the need for additional imaging was lower for AB-MR, specificity was higher for DBT. Clinical trial information: NCT02933489 .