Because screening mammography has decreased sensitivity in women with dense breast parenchyma, several techniques are under consideration for supplemental screening of this population, including whole-breast screening ultrasound, MRI, digital tomosynthesis, and molecular breast imaging (MBI) [
1–
7]. Although supplemental screening can improve cancer detection rates by revealing mammographically occult cancers, the generation of additional recalls and benign biopsy results, and associated costs are major concerns [
8,
9]. In this era of health care cost containment, with recent reductions in Centers for Medicare and Medicaid Services reimbursement for breast imaging, consideration of the economic implications of additional breast screening tests is increasingly important [
10,
11].
Nuclear medicine–based breast imaging techniques, including both single photon and positron emission systems, depict mammographically occult breast cancer in dense breasts [
5,
12–
16]. The evaluation of these techniques in the screening environment has been limited, partly because of concerns about radiation dose [
17]. The use of low-radiation-dose MBI for supplemental screening of women with mammographically dense breasts has been reported [
6]. Adding MBI to mammography increased the cancer yield from 3.2 per 1000 with mammography alone to 12.0 per 1000 (
p < 0.001). However, the recall rate increased from 11% to 18%, and the biopsy rate increased from 1.3% to 4.2% (both
p < 0.001) [
6]. During that trial, we tracked the diagnostic imaging tests and biopsies generated by the screening approach of mammography with supplemental MBI versus that generated by screening mammography alone. The purpose of the current study was to examine the additional diagnostic workup and costs generated by the addition of a single MBI examination to screening mammography of women with dense breasts.
Discussion
The combination of screening mammography and adjunct MBI of dense breasts was associated with an almost fourfold increase in the number of cancers detected compared with the number detected with screening mammography alone while a favorable benign biopsy rate and PPV3 for additional findings were maintained. The additional cancers detected with MBI came at a price of a 3.2-fold increase in total costs in this study.
Adding a single MBI examination to mammography for women with dense breasts increased the benign biopsy rate 2.1% (33 of 1585) compared with the rate for mammography alone. A higher benign biopsy rate was reported for supplemental ultrasound in the American College of Radiologic Imaging Network (ACRIN) 6666 trial, which included women at elevated breast cancer risk who also had a least one dense breast quadrant. The addition of physician-performed screening ultrasound to mammography increased the benign biopsy rate 7.3% (193 of 2659) after a single screen and 4.7% (224 of 4814) in years 2 and 3 of annual screening [
1]. In two reports of supplemental screening ultrasound performed for women with dense breasts and normal mammographic findings, the benign biopsy rate was 0.9% (83 of 9157) in an Italian study using physician-performed ultrasound [
22] and 6.4% (60 of 935) in a Connecticut study using technologist-performed ultrasound [
2].
The PPV3 for additional biopsies prompted by supplemental MBI in this study was 30.0% (14 of 47), higher than that reported for supplemental ultrasound. In ACRIN 6666, the PPV3 for additional biopsies prompted by ultrasound was 5.8% (12 of 207) for a single screening ultrasound and 7.4% (18 of 242) for years 2 and 3 of screening ultrasound [
1]. In an overview of studies evaluating technologist-performed adjunct screening ultrasound of dense breasts [
23], the overall PPV3 for ultrasound findings was 5.8%. The reported average cancer detection rates for prevalent screening ultrasound are 2.5 per 1000 for technologist-performed examinations [
23] and 4.3 per 1000 for physician-performed examinations [
1,
22], considerably lower than the 12.0 per 1000 supplemental cancer detection rate found for prevalent MBI in our study.
The addition of a single MRI examination to mammography in ACRIN 6666 led to additional benign biopsies in 6.2% (38 of 612) of participants, higher than the rate of additional benign biopsies prompted by MBI in our study (2.1%). The PPV3 for additional biopsies prompted by MRI findings was 20.4% (10 of 49), somewhat lower than that observed with MBI (30.0%). Notably, the supplemental cancer detection rate of MRI in this population at elevated risk and with dense breasts was 18.0 per 1000, higher than that observed with MBI (12.0 per 1000) in our cohort recruited solely on the basis of having dense breasts.
Our results show the importance of comparison of adjunct MBI studies with a recent screening mammogram and review of the clinical history. In our study, comparison with screening mammograms explained the MBI findings as benign in 31.2% (54 of 173) of patients with initially positive MBI findings, which substantially lowered its false-positive rate. As has been seen with other breast imaging modalities [
24], we expect the specificity of MBI to improve in the incidence screening setting with the benefit of previous images to review, though we also expect the cancer detection rate to decrease.
Our results also show the importance of performing a thorough diagnostic workup of all MBI-detected lesions regardless of the assessment category assigned. The malignancy rate for probably benign (category 3) findings at adjunct MBI, after comparison with mammograms, was 5.1%, which is substantially higher than the less than 2% malignancy rate observed with probably benign assessments with mammography and ultrasound [
18,
25,
26]. An analysis of probably benign (BI-RADS 3) findings at supplemental ultrasound showed a low malignancy rate of 0.8% [
18,
27]. The malignancy rate for BI-RADS 3 lesions detected with MRI in recent years has ranged from 0.6% to 1.8% [
28].
We conducted this study in the absence of MBI-guided biopsy capability; therefore, we relied on other modalities for diagnostic evaluation and localization. Although biopsy capability exists for other nuclear medicine breast imaging technologies, including positron emission mammography (PEM) and single-head breast-specific gamma imaging cameras, a biopsy system compatible with the dual-head MBI gamma camera used in this study is under development [
29,
30]. Biopsy, when indicated, was performed under ultrasound guidance whenever possible. For 19 of 1585 (1.2%) patients, however, MRI was performed for imaging evaluation, and for 12 of 1585 (0.7%) patients, MRI-guided biopsy was necessary. Future availability of direct biopsy guidance for MBI may decrease the need to use MRI in diagnostic workup and thus reduce the cost associated with adjunct MBI. However, direct MBI guidance is needed because not all patients can tolerate MRI. An important observation in this study was that for any MBI finding that was not resolved with mammography or ultrasound, a corresponding finding was obtained with MRI. This is different from reported findings with PEM performed with
18F-FDG, in which different findings were seen at MRI and PEM, and direct PEM-guided biopsy was necessary [
15,
31].
Although additional costs were incurred with supplemental MBI screening, the cost per cancer detected for the combination of techniques was lower than that for screening mammography alone because of the increased cancer yield. One potential criticism of this approach is that the additional cancers detected may represent overdiagnosis, defined as the diagnosis of a breast cancer as a result of screening that would not have been clinically diagnosed in the woman's lifetime had screening not taken place. However, the cancers detected only with adjunct MBI had features of clinical importance: 2 of 14 occurred in patients with bilateral mammographically occult disease, and 11 of 14 were invasive cancers with a median size of 0.9 cm.
Data on the actual costs of supplemental screening of women with dense breasts are limited. Hooley et al. [
2] reported the cost per additional cancer detected with handheld screening ultrasound performed by technologists in women with dense breasts who had normal screening mammograms as $60,267 (2012 Connecticut Medicare reimbursement rates) [
2]. A review of six Connecticut radiology practices that offered technologist-performed handheld screening ultrasound [
32] estimated a cost per cancer detected of $50,000 using average insurance reimbursement for 2009–2010. Our results compare favorably with these, because the cost per additional cancer detected with MBI was $44,650 ($625,095 divided by 14). Although the cost of supplemental MRI screening of dense breasts has yet to be evaluated, the cost-effectiveness of MRI screening of women who are
BRCA1 or
BRCA2 mutation carriers has been found to vary widely depending on patient age [
33]. For the estimated 50% of screening-eligible U.S. women who have dense breasts [
34], the cost of MRI combined with a lower incidence of cancer than the
BRCA-positive population is likely to make MRI too expensive to use in this setting.
Reimbursement for supplemental screening when dense breast tissue is the only indication varies from state to state and between insurance providers. Cost saving per cancer detected with supplemental screening MBI is compelling evidence for future coverage of screening MBI of women with dense breasts.
There were limitations to our cost analysis. First, we considered costs of imaging tests and biopsies but did not consider costs beyond those of cancer detection. There could be potential cost savings associated with relative lead time, that is, the amount of time the breast cancer diagnosis is advanced by adding MBI compared with relying on mammographic or clinical detection. Because earlier detection may reduce the costs (in dollars and lost productivity) associated with treatments of advanced disease, the magnitude of the cost benefit for supplemental screening with MBI may be considerably higher. Second, sestamibi, which is commonly available as a myocardial perfusion agent, is off patent so may be available at a lower cost than we report. If so, the future cost of adding MBI to screening mammography in this cohort is likely overestimated. Third, we did not explicitly measure important factors in evaluation of costs, including the complexity of the examination and time of radiologist's interpretation. High interreader and intrareader agreement of MBI interpretations after a 2-hour radiologist training session has been found [
19,
35]. In our experience, a full bilateral dual-head MBI study can be rapidly assessed, with interpretation times consistently less than for concomitant mammography.
Compared with other imaging modalities under consideration for supplemental screening of women with dense breasts, MBI is associated with a relatively low benign biopsy rate and relatively high PPV for additional findings. Considering the costs of screening examinations, subsequent diagnostic imaging, and biopsies that were generated, the addition of MBI to mammography screening of dense breasts results in a high enough supplemental cancer yield to result in a lower cost per cancer detected than for mammography alone.