We retrospectively collected data from 100 consecutive normal female Japanese patients who underwent screening mammography and supplemental screening breast ultrasonography on the same day between January 2021 and June 2021 at Shizuoka Cancer Center Hospital. The participants’ median age was 57 (range: 41–69) years, and 75% (n = 75/100) were postmenopausal. The institutional review board of our institution approved this retrospective study, and the need for informed consent was waived.

All ultrasonography examinations were performed in accordance with the ISO 15189 standard by one of 10 certified sonographers with 1–19 years of experience in breast ultrasonography using the ARIETTA 850, ARIETTA 70, or Prosound F75 system (Hitachi, Ltd., Tokyo, Japan); Aplio XG SSA-790A or 500 system (Toshiba Medical Systems, Tokyo, Japan); or LOGIQE-10 (GE HealthCare, Tokyo, Japan) with linear transducers (10–18 MHz). The screening breast ultrasonography technique was standardized. The sonographers were instructed to begin by scanning the left breast and then the right breast using the transverse orientations. In a routine screening examination, the sonographers were required to take representative images that included the thickest layer of the fibroglandular tissue (FGT) in each breast (Fig. 1). Thus, radiologists could assess the distribution and amount of FGT in each breast after ultrasonography examinations. In addition, the representative images clearly showed five main layers including the skin, subcutaneous fat, FGT, retromammary fat, and chest muscle in the normal breast sonographic anatomy (Fig. 1). We referred to total fat (subcutaneous fat and retromammary fat) as FAT in the representative images.

Routine breast ultrasonography examination image of the right breast. The numbers on the image correspond to the numbered anatomical structures: 1, skin; 2, subcutaneous fat; 3, fibroglandular tissue (FGT); 4, retromammary fat; and 5, chest muscle. This image does not show any site with flat anterior superficial fascia including the thickest FGT layer. The better representative point (arrowhead) is halfway between a site with a concave anterior superficial fascia and a site with a convex anterior superficial fascia due to the Cooper ligament. FGT thickness-to-FAT thickness ratio = FGT thickness (3)/FAT thickness (2 + 4). FAT means subcutaneous fat (2) and retromammary fat (4). The arrow indicates a site with a convex anterior superficial fascia due to the Cooper ligament that is not suitable for the representative point

In general, breast composition is defined according to FGT and FAT balance on breast imaging. Therefore, one breast radiologist evaluated ultrasonographic breast composition in each case, which was calculated as the FGT thickness-to-FAT thickness ratio, as follows:

FGT thickness-to-FAT thickness ratio = FGT thickness/FAT thickness (Fig. 1)

In order to reduce the measurement variability for calculating the FGT thickness-to-FAT thickness ratio, we determined that the best representative point would be a site with flat anterior superficial fascia, including the thickest layer of FGT. A site with convex anterior superficial fascia that is caused by a Cooper ligament would not be suitable as a representative point. In the absence of a site with flat anterior superficial fascia including the thickest layer of FGT, the better representative point would be a halfway site including the thickest layer of FGT between a site with concave anterior superficial fascia and a site with convex anterior superficial fascia.

In addition, the presence of a high glandular tissue component (GTC) in FGT was investigated because women with a higher mammographic breast composition have a greater GTC [8, 9] (Fig. 2). GTC can be dichotomized as low or high according to a GTC that represents 50% of the FGT. Fat lobules in the FGT, which are distinct from glandular tissues, were not counted in the GTC assessment [8]. Moreover, the presence of evident fat lobules in FGT, which are different from GTC, was examined because lobular involution usually occurs and FGT replaces fat with age (Fig. 3). Previous research has shown a positive association between nondense breasts, reflecting fat tissue, and complete lobular involution [9].

Mammographic dense breasts that are also ultrasonographic dense breasts with a high glandular tissue component in the fibroglandular tissue. a Mammograms in bilateral mediolateral oblique view showing extremely dense breasts. b This corresponding ultrasonography image shows a high glandular tissue component (GTC). Normal fibroglandular tissue (FGT) includes isoechoic dendritic structures that are equivalent to the GTC. 1, skin; 2, subcutaneous fat; 3, FGT; 4, retromammary fat; and 5, chest muscle

Mammographic nondense breasts that are also ultrasonographic nondense breasts with evident fat lobules in the fibroglandular tissue. a Mammograms in bilateral mediolateral oblique view showing scattered areas of fibroglandular density as nondense breasts. b This corresponding ultrasonography image shows evident fat lobules in the fibroglandular tissue (FGT). Normal FGT almost replaces fat lobules and has no GTC. 1, skin; 2, subcutaneous fat; 3, FGT; 4, retromammary fat; 5, chest muscle

Mammograms with two standard imaging planes (mediolateral oblique and craniocaudal views) were obtained using a commercially available full-field digital mammography unit (Selenia Dimensions; Hologic, Bedford, MA). Three radiologists retrospectively reviewed all mammograms for mammographic breast composition assessment according to the new Japanese breast density classification system [10], which is a modification of the Breast Imaging-Reporting and Data System (BI-RADS) with a two-grade scale (nondense and dense breasts).

A woman had a higher breast density if the density of one breast was different from that of the other. Thus, our analysis was performed at the examination level. All breasts with sonographic breast composition were classified as nondense or dense, and breast density was correlated with mammographic breast composition. Fisher’s exact test was used to compare the presence of a high GTC and evident fat lobules in FGT among mammographic breast composition groups. The receiver operating characteristic (ROC) curves of the FGT thickness-to-FAT thickness ratio was calculated, and the AUC was determined using the R statistical package version 4.3.2 (R Core Team, October 2023; www.r-project.org). The cutoff point between nondense and dense breasts was calculated using Youden’s index (highest value for [sensitivity + specificity − 1]).