**3. Results**

The study included 550 mammographic lesions from 538 women, aged 40 to 94 years (mean age: 58.9, SD: ±8.94), recalled at breast cancer screening mammography between 2017 and 2019. Among the 550 lesions recalled, 60.4% were in dense breasts, and 39.6% were in nondense breasts. Breast density was found to influence recall decisions significantly. Mammographic abnormalities were more likely to be recalled when seen in dense breasts than in non-dense breasts (*p* < 0.001). The distribution of lesion types across dense and nondense breasts is shown in Table 2.


**Table 2.** Association between recall and breast density based on lesion features.

NSD: Nonspecific density; Nondense Breasts: almost entirely fatty & scattered areas of fibroglandular density; Dense Breasts: heterogeneously dense & extremely dense; Bold values indicate statistical significance at the *p*-value ≤ 0.05 level.

Table 3 shows that there is no difference in true negative proportions between DBT and ultrasound in nondense breasts (32.8% vs. 22%, respectively; *p* = 0.2). Conversely, in dense breasts, ultrasound showed a significantly higher proportion of true negatives than DBT (54.8% vs. 16.1%, respectively; *p* < 0.001).

**Table 3.** True negative proportions of DBT and ultrasound across dense and nondense breasts.


Nondense Breasts: almost entirely fatty & scattered areas of fibroglandular density; Dense Breasts: heterogeneously dense & extremely dense; DBT: digital breast tomosynthesis; US: ultrasound; Bold values indicate statistical significance at the *p*-value ≤ 0.05 level.

Table 4 shows the potential reduction of unnecessary biopsies for DBT and ultrasound stratified according to breast density. Differences between DBT and ultrasound in terms of preventing one unnecessary biopsy are also presented. Among all recalled mammographic abnormalities, ultrasound showed greater potential to decrease unnecessary biopsies than DBT: entirely fatty (21% vs. 5%, respectively; *p* = 0.04); scattered fibroglandular (23% vs. 10%, respectively; *p* = 0.003); heterogeneously dense (34% vs. 7%, respectively; *p* < 0.001); extremely dense (39% vs. 9%, respectively; *p* < 0.001) breasts. The number of cases needing assessment to prevent one unnecessary biopsy was significantly lower with ultrasound than with DBT in heterogeneously dense breasts (1.8 vs. 7, respectively; *p* < 0.001) and extremely dense breasts (1.9 vs. 5.1, respectively; *p* = 0.03), but there were no significant differences in entirely fatty breasts (3.2 vs. 4.3, respectively; *p* = 0.65) and scattered fibroglandular densities (2.6 vs. 4.6, respectively; *p* = 0.21).


**Table 4.** The potential reduction of DBT and ultrasound to decrease unnecessary biopsies following screening mammography and the number of DBT and ultrasound examinations required to be assessed to prevent one unnecessary biopsy.

BI-RADS A: almost entirely fatty; BI-RADS B: scattered areas of fibroglandular density; BI-RADS C: heterogeneously dense; BI-RADS D: extremely dense; DM: digital mammography; DBT: digital breast tomosynthesis; US: ultrasound; ARR: absolute risk reduction. Bold values indicate statistical significance at the *p*-value ≤ 0.05 level.
