A total of 1149 participants were eligible. Background parenchymal uptake was assessed by one of two readers (one reader 680 patients and the other, 469). Substantial agreement was observed between and within readers, according to criteria established by Landis and Koch [28]. The interreader kappa value was 0.84 (95% CI, 0.71–0.97). The intrareader values were 0.87 (95% CI, 0.70–1) for one reader and 0.94 (95% CI, 0.81–1) for the other.

Background parenchymal uptake was photopenic in 252 (22%) patients, minimal-mild in 728 (63%), moderate in 130 (11%), and marked in 39 (3%). The distribution of each background parenchymal uptake category by patient characteristics is shown in Table 1. Background parenchymal uptake was associated with both age and BMI (p < 0.0001, for both). Women with moderate or marked background parenchymal uptake had the lowest mean age, and the women with photopenic background parenchymal uptake had the lowest mean BMI.

The frequency of BI-RADS density categories differed across background parenchymal uptake categories (p < 0.0001) (Table 1). These differences are shown in a mosaic plot (Fig. 2A) that illustrates the distribution of background parenchymal uptake by BI-RADS density and the relative proportion of women in each category. Although the proportion of women with minimal-mild background parenchymal uptake decreased across BI-RADS density categories, the proportion of women with photopenic background parenchymal uptake increased with increasing BI-RADS density, and the proportion of women with moderate or marked background parenchymal uptake increased with increasing BI-RADS density. Among the 164 women with extremely dense breasts, 72 (44%) had photopenic, 55 (34%) had minimal-mild, and 37 (22%) had moderate or marked background parenchymal uptake.

Background parenchymal uptake was associated with menopausal status. Approximately three fourths (70–71%) of women with photopenic or minimal-mild background parenchymal uptake were post-menopausal, and among women with moderate or marked background parenchymal uptake, approximately three fourths (76%) were premenopausal or perimenopausal (p < 0.0001) (Fig. 2B). Among postmenopausal women, background parenchymal uptake was associated with use of systemic estrogen or progesterone: women with moderate or marked uptake were more likely to be currently using hormones (p < 0.0001) (Fig. 2C). Among premenopausal and perimenopausal women, background parenchymal uptake was not statistically associated with hormonal contraceptive use (p = 0.64) or day of menstrual cycle (p = 0.65). Background parenchymal uptake also was not statistically associated with history of benign biopsy (p = 0.40) or first-degree family history of breast cancer (p = 0.60).

We performed adjusted analyses to account for age and BMI and found that percentage density was lowest in the women with minimal-mild background parenchymal uptake (p < 0.001). The estimated percentage densities for the photopenic background parenchymal uptake versus moderate to marked background parenchymal uptake groups were not statistically different (p = 0.24) (Table 1). The distribution of percentage density by age and BMI groups is shown in Figure S1. (Figures S1 and S2 can be seen in the AJR electronic supplement to this article, available at www.ajronline.org.) The general tendency of women with minimal-mild background parenchymal uptake to have lower percentage density was appreciable in all age and BMI combinations with the possible exception of women with BMI greater than 30 and age older than 50 years. We also examined the association between 5-year risk according to the Gail model and background parenchymal uptake category but found no significant differences. The estimated 5-year Gail risk ranged from 1.6 (95% CI, 1.4–1.8) for women with moderate or marked background parenchymal uptake to 1.7 (95% CI, 1.4–2.1) for those with photopenic uptake after adjustment for age and BMI (p = 0.79) (Table 1 and Fig. S2).

In this evaluation of background parenchymal uptake, we found that in women with primarily dense breasts, a range of background parenchymal uptake was observed at MBI, indicating functional differences in the behavior of fibroglandular tissue among women with similar-appearing fibroglandular tissue on mammograms. In general, women with moderate or marked background parenchymal uptake were younger, more likely to be premenopausal or perimenopausal, and more likely to be using postmenopausal hormone therapy than were women with photopenic or minimal-mild background parenchymal uptake. Background parenchymal uptake was also associated with BMI, BI-RADS density category, and percentage density (minimal-mild background parenchymal uptake being the least dense) but was not statistically associated with day of menstrual cycle, history of benign biopsy, family history of breast cancer, or 5-year risk according to the Gail model. Whether background parenchymal uptake is predictive of subsequent breast cancer risk is yet to be determined but should be the focus of future research.

Most of the women undergoing adjunct MBI screening in this study had minimal-mild background parenchymal uptake, in which uptake in fibroglandular tissue is the same as in fat. Because background parenchymal uptake depicts the component of fibroglandular tissue that is functionally active (through uptake of ^99mTc-sestamibi), it was expected that moderate or marked background parenchymal uptake would be more frequently observed in denser breasts, which are associated with the highest risk of breast cancer [4]. We observed, however, that a substantial proportion of women with the densest breasts had photopenic background parenchymal uptake, in which there is a lack of uptake in fibroglandular tissue, showing less uptake compared with fat. The question becomes: What is different between mammographically dense tissue with moderate or marked background parenchymal uptake and that with photopenic background parenchymal uptake? Interestingly, photopenic background parenchymal uptake was more frequent in postmenopausal women and those with lower BMI, a subgroup in whom breast cancer risk is known to be lower [29].

Background parenchymal uptake appears to be influenced by hormonal activity in the breast: moderate or marked uptake was more common in postmenopausal women using hormone therapy than in nonusers and in women with menstrual cycles (premenopausal or perimenopausal) than in postmenopausal women. However, among postmenopausal women, moderate or marked background parenchymal uptake was still observed in a small proportion of nonusers (18 of 587 [3%]), and photopenic background parenchymal uptake was observed in a substantial proportion of postmenopausal women using hormone therapy (25 of 145 [17%]), which could suggest variability in response to hormone therapy, as has been seen with percentage density on mammograms [30]. A number of studies have shown similar influence of menopausal status and use of postmenopausal therapy on MRI-depicted BPE [11–13].

A previous report showed that background uptake of ^99mTc-sestamibi during MBI is influenced by menstrual cycle in premenopausal women, moderate or marked uptake being more likely to occur in the luteal phase than in the follicular phase [17, 31]. Likewise, menstrual cycle has been found to affect background enhancement on MR images, enhancement being lowest on days 7–14 of the menstrual cycle among women undergoing serial imaging throughout the cycle [14]. We did not find an association between background parenchymal uptake and day of menstrual cycle. This discrepancy may be explained by the fact that the participants in our study were at various phases in their cycles. Previous studies, however, assessed background parenchymal uptake during MBI examinations performed specifically during peak follicular and peak luteal phases in the same woman.

Background uptake can occasionally obscure tumors during MBI (Fig. 3), and therefore, efforts to minimize background uptake may be considered. It has been recommended that MBI be scheduled on days 7–10 of the menstrual cycle [17], although our current data do not support this recommendation. According to our results coupled with those in the literature, cessation of hormonal therapy for a few months before MBI may also be considered. An example of a patient in whom background parenchymal uptake decreased when hormone therapy was discontinued is shown in Figure 4. Likewise, BPE has been observed to obscure tumors at MRI [32].

The mechanism of ^99mTc-sestamibi uptake in the breast is poorly understood. This radionuclide is a perfusion agent, most common to myocardial imaging. In cancer, up-take of ^99mTc-sestamibi is associated with blood flow, tissue viability, mitochondrial activity, and mitotic activity of tissue [33–35]. Similarly, angiogenesis and vascular permeability are the basis of contrast enhancement of breast tumors for the gadolinium chelate agents used in MRI [36]. Our experience to date has shown that MBI uptake of ^99mTc-sestamibi in breast cancers correlates with gadolinium MRI contrast enhancement. Further study is needed to verify that these contrast agents are correlated in benign background breast tissue. A recent report [37] supports a correlation between ^99mTc-sestamibi uptake and gadolinium contrast enhancement in background parenchyma.

MRI screening is currently indicated only for women at high risk, such as those with BRCA mutations or greater than 20% lifetime risk based on family history risk prediction models. It is not recommended on the basis of breast density alone, primarily owing to concerns about variable specificity, high cost, and patients’ refusal to undergo screening MRI [38–40]. With the recent finding that MBI interpretation can be rapidly learned [18] and performed at radiation doses low enough to be acceptable for screening [16, 41], it may be feasible to implement MBI on a widespread basis for women with mammographically dense breasts who do not meet the risk criteria to qualify for screening MRI. Thus, understanding of background parenchymal uptake will be central to future risk stratification in this population. Further work is necessary to fully characterize the underlying cause of background parenchymal uptake and determine its utility in prediction of subsequent development of breast cancer.

The underlying biologic link between breast density and risk is not yet clear. Fibroglandular tissue in the breast is not homogeneous but comprises both epithelium and supporting stroma in varying proportions among women. Although there are different hypothesized mechanisms by which these tissues influence breast cancer development, study results have indicated that cellular proliferation of epithelium, stroma, or both is associated with mammographic density [42, 43]. Lobular involution (the age-related atrophy of epithelium associated with reduction in breast cancer risk) has been inversely associated with mammographic density. Examination of biopsy specimens has shown that nondense breasts on mammography are more likely to have undergone complete lobular involution [44]. However, this relation is not exact; a substantial proportion of the densest breasts were found to also have undergone complete lobular involution. Because epithelium cannot be easily differentiated from stroma on a mammogram and lobular involution status cannot be evaluated, mammography is limited in the use of underlying tissue composition for predicting risk at the individual level. These issues underscore the need for functional imaging tools, such as MRI and perhaps MBI, which may help further delineate which mammographically dense fibroglandular tissue is likely to increase risk of breast cancer and which is not. Such functional imaging tools may allow appropriate individualization of supplemental screening and risk reduction strategies for women with dense breasts.

We thank the study coordinators, Tammy Hudson and Beth Connelly; nuclear medicine technologists, Carley Pletta, Karlie Homann, Thuy Tran, and Tiffinee Swanson; and Fang Fang Wu, who performed percentage density analysis.