Institutional review board approval was obtained for a retrospective review of patient data. Before analysis, all patient-identifying markers were removed. Each institution maintained a clinical registry of patients undergoing breast-specific gamma imaging. All imaging modalities, including breast-specific gamma imaging, digital mammography, ultrasound, and MRI, were conducted as deemed clinically necessary by either the referring physician or radiologist.

All patients had at least one (the majority of patients had two or more) of the following indications for recommending the breast-specific gamma imaging procedure: equivocal mammography or ultrasound findings; personal history of breast cancer; family history or other factors establishing high risk for developing breast cancer; recent positive mammogram; clinical finding such as palpable mass, breast pain, or bloody nipple discharge; and radiodense breast tissue difficult to image on mammogram.

Breast-specific gamma imaging was conducted using 555–925 MBq of ^99mTc-sestamibi and a gamma camera (model 6800, Dilon Technologies) (Figs. 1A and 1B). Imaging began 3–10 minutes after injection with a minimum of two views of each breast, including craniocaudal and medial lateral oblique. Additional views including, but not limited to, rolled craniocaudal views, 90° lateral, were conducted as deemed clinically necessary, and the imaging time for each projection ranged from 5 to 10 minutes depending on institutional protocol.

The interpreting radiologist had access to all studies and patient history at the time of interpretation. Breast-specific gamma imaging examinations were interpreted in a scoring system similar to that of the BI-RADS classification for mammography: 1, normal study with no focal abnormality; 2, broad heterogeneous distribution consistent with benign breast tissue changes; 3, regional heterogeneous distribution or patchy distribution without focal abnormality, lacking evidence of malignancy but insufficient to rule out malignant processes; 4, low-to-medium intensity focal abnormality suggesting the possibility of malignancy; and 5, high focal intensity suggesting high possibility of malignancy.

Biopsy was performed when deemed necessary by the radiologist. Imaging-guided biopsies were typically performed using ultrasound, stereotactic x-ray, or MRI; however, more recently, a stereotactic breast-specific gamma imaging–guided biopsy system has become available. In addition, follow-up imaging was conducted on an interval deemed clinically necessary. Biopsy or follow-up imaging, if biopsy was not conducted, was used as the reference standard. Biopsy results were classified as positive (i.e., malignancy or high-risk lesions, such as atypical ductal hyperplasia, atypical lobular hyperplasia, and lobular carcinoma in situ) or negative (i.e., benign conditions not requiring additional intervention). Imaging studies were classified as positive (BI-RADS category 4 or 5), negative (BI-RADS category 1 or 2), or indeterminate (BI-RADS category 0 or 3). For the purposes of statistical preparation, all indeterminate findings were classified as negative, having resulted in no definitive change of management.

Because this scoring system was used in a clinical setting, the interpreting radiologist used all imaging studies, patient history, and other available information while interpreting each study and determining the management pathway: additional imaging, biopsy, short-term follow-up, or return to annual screening. Information regarding the case, such as patient's risk factors, clinical concern, clinical history, BI-RADS rating of all imaging studies, biopsy results, and follow-up imaging, could be recorded in a spreadsheet (Excel, Microsoft). To be included in the overall analysis, each patient required a minimum of breast-specific gamma imaging findings and pathology or follow-up imaging results recorded in the patient registry. To measure the impact of breast-specific gamma imaging in patient management, the dataset of patients with breast-specific gamma imaging and biopsy or follow-up imaging used in the overall analysis was further reduced to include only those patients with results of both mammography and ultrasound recorded in the registry.

The typical methods for statistical analysis of medical tests assumes that all imaging studies are either positive or negative and thus can be sorted into one of the four statistical expressions used: true-positive, true-negative, false-positive, or false-negative. However, according to the American College of Radiology BI-RADS Atlas [14], mammograms (and each subsequent diagnostic imaging modality: diagnostic mammography, ultrasound, and MRI) are interpreted on the basis of a six-category reporting system. Similarly, in clinical practice and according to several authors, it is common to use a similar rating system for reading breast-specific gamma imaging. Each category and the management recommendations provided by the American College of Radiology [14] are as follows: BI-RADS category 1, negative (return to annual screening); BI-RADS category 2, benign findings (return to annual screening); BI-RADS category 3, likely benign (less than 2% chance of malignancy; 6-month follow-up imaging is recommended); BI-RADS category 4, possibility of malignancy (biopsy should be considered); BI-RADS category 5, 95% probability malignancy likely (intervention required); and BI-RADS category 0, additional imaging or information is required to make proper management decisions. Working from these BI-RADS classifications, there are essentially three management routes for the diagnostic breast patient: follow-up imaging at 6- or 12-month intervals, biopsy, or additional imaging.

In addition, it is typical to analyze the performance of a breast imaging modality using biopsy or follow-up imaging as the reference standard and to classify the findings as positive (biopsy-proved cancer) or as negative (no evidence of malignancy). However, in clinical practice, there are a number of benign lesions that typically undergo additional interventional procedures either because of the possibility of upgrading the lesion at excision or because they are sometimes associated with malignant processes. Therefore, there are three possible clinical classifications for needle biopsy results: benign, no additional intervention required; high risk, additional intervention needed to clarify diagnosis; and malignant, additional intervention needed to treat.

To understand the clinical impact of adjunctive imaging modalities, such as breast-specific gamma imaging and ultrasound, on the population undergoing mammography, these management routes and biopsy results need to be considered in the quantification. Given that mammography is typically the primary imaging modality, the registry data were divided into three classifications: negative or likely benign findings resulting in follow-up imaging at an established time interval, indeterminate findings leading to additional imaging, or positive findings indicating the potential need for biopsy.

To quantify and compare the impact of breast-specific gamma imaging and ultrasound on patient management, the results from adjunctive imaging procedures can be sorted into one of three categories as they relate to the mammogram: concordant, no change in management (same finding as mammography or a BI-RADS category 0 result); or discordant, correct change in management (improving cancer detection or eliminating intervention for a benign lesion) or incorrect change in management (misdiagnosis or delay in cancer detection or indicating the need for intervention on a benign lesion).

Because all of these patients had a mammogram indicating follow-up imaging as the proper management pathway, the adjunctive imaging studies were recommended because of some additional clinical factors impeding the clinician's confidence in the mammogram. As part of the patient registry, the interpreting physicians were asked to choose from the following reasons for recommending the additional diagnostic examinations for each patient: questionable findings on mammography, palpable mass, dense breasts (limiting the utility of mammography), or physician referral.

A total of 2004 patients were entered into the patient registry at four institutions. There were 962 patients who did not undergo biopsy and who lacked the required 6 months of follow-up imaging at the time of this analysis. The other 1042 patients underwent pathologic analysis (n = 642) or follow-up imaging (n = 400) and were included in this study. Pathologic analysis or follow-up imaging resulted in 250 positive and 792 negative findings. Breast-specific gamma imaging was positive in 408 patients, 227 of whom had a malignant or high-risk lesion. Breast-specific gamma imaging was negative in 634 patients, and 611 of these lesions were negative on biopsy or follow-up imaging. Breast-specific gamma imaging was indeterminate in 69 lesions, all of which were benign. Breast-Specific gamma imaging had an overall sensitivity, specificity, positive predictive value, and negative predictive value of 91%, 77%, 57%, and 96%, respectively.

Mammography and ultrasound findings were reported for 329 patients, and the results from each of the modalities are compared in Table 1. When categorized by age, there were 42 patients younger than 45 years, 99 patients 46–55 years old, 97 patients 56–65 years old, and 91 patients older than 65 years. In this group of patients, 48 (14.6%) had one indication for breast-specific gamma imaging, whereas 222 (67.5%) had two indications and 59 (17.9%) had three or more indications. Overall, breast-specific gamma imaging provided a statistically significant improvement in sensitivity and negative predictive value (p < 0.000001, McNemar test).

Although Table 1 provides an overall comparison of the clinical performance of the three modalities, in practice, mammography is the primary screening and diagnostic imaging modality for the majority of patients, whereas ultrasound and breast-specific gamma imaging are adjunctive imaging procedures used when the results of mammography are not sufficient to identify the proper management pathway (i.e., return to screening, short-term follow-up, or biopsy). In Table 2 and the following sections, patients are subcategorized by their mammographic findings to compare the effectiveness of breast-specific gamma imaging and ultrasound in their role as adjunctive imaging modalities to contribute toward the management of patients. The majority of patients (38%) were referred because of questionable findings in mammography. These findings included nonspecific asymmetric changes, vague calcifications, potential findings obscured by overlaying breast tissue, or discordant results between multiple imaging studies, such as screen and diagnostic mammography or diagnostic mammography and ultrasound. Another 20% of patients were reported to have breast density sufficient to impede the interpretation of mammogram, and 11% had a palpable mass that was negative on mammography. The remaining 31% of patents were either referred for additional imaging studies by the primary physician (11%) or had no specific reason provided for the additional imaging (20%).

The role of adjunctive imaging for breast diagnostic patients is to improve management when prior imaging studies have failed to provide conclusive confident results. For the purposes of quantifying the impact on management, if the findings of an adjunctive imaging study are concordant with the mammogram or if the adjunctive study results in a BI-RADS category 0 result or equivalent, then the study has resulted in no net change to management. However, if the adjunctive study is discordant, then it indicates that a change in management should be considered.

For patients with a negative mammogram and remaining diagnostic concern, only a positive finding from the adjunctive procedures indicates a change in management. In this group of patients, breast-specific gamma imaging and ultrasound were positive in 29 and 27 patients, respectively. Breast-specific gamma imaging was slightly more likely to be positive in patients with malignant or high-risk lesions than ultrasound (15 vs 11) and slightly less likely to be positive in benign lesions (14 vs 16), although the improvements were not statistically significant (p = 0.69). The resulting positive predictive values for breast-specific gamma imaging and ultrasound were 52% and 41%, respectively (p = 0.69).

In patients with a positive finding on mammography, a negative finding by adjunctive imaging indicates a change in management. Breast-specific gamma imaging and ultrasound were negative in 21 and 15 patients, respectively. Ultrasound was less likely to be negative in malignant lesions than breast-specific gamma imaging (six vs eight), but breast-specific gamma imaging was more likely to be negative in benign lesions (13 vs nine); these differences were not statistically significant (p = 1.0). Because the need for biopsy has already been indicated by mammography, a negative adjunctive imaging procedure may contribute only by obviating biopsy. To maintain the standard of care, the false-negative rate of the imaging procedure must be equal to or greater than that provided by biopsy procedures (∼ 3%) [9]. The false-negative rate for ultrasound and breast-specific gamma imaging was 6% and 8%, respectively, indicating that neither procedure had adequate performance to obviate biopsy when indicated by mammography. There may still be a role for these adjunctive imaging procedures in treatment planning, such as determining the extent of the primary lesion and detecting additional disease occult by mammography, but these roles are beyond the scope of this work.

Adjunctive imaging procedures play a greater role for patients who have a BI-RADS category 0 finding on mammography because the management pathway has not been defined. Discordant findings may be positive, indicating the need for biopsy, or negative, indicating either a return to screening or follow-up imaging at a later date. Breast-specific gamma imaging provided a change in management in a larger number of patients (109/119 [92%]) compared with ultrasound (71/119 [40%]). For patients with positive findings, breast-specific gamma imaging had a positive predictive value of 50% (17/34) compared with 26% (10/38) for ultrasound, which was a statistically significant improvement over ultrasound (p < 0.0003). Breast-specific gamma imaging also performed better than ultrasound in terms of false-negative rate (0% vs 9%; p < 0.000001) and it provided better accuracy than ultrasound (84% vs 56%).

As with all studies, there are some limitations to the data provided. First, this is a retrospective analysis without a control group. In addition, of the 2004 patients, 1042 were included in the overall statistical analysis; pathologic analysis was available for 642 patients, whereas 400 patients had negative follow-up imaging studies at a minimum of 6 months after enrollment. This length of follow-up provides limited evidence regarding the lack of malignant process; however, it should be noted that, of the 329 patients included in the comparison with ultrasound, 283 (86%) had pathologic confirmation of diagnosis. Finally, in nearly all cases, breast-specific gamma imaging was recommended after mammography and ultrasound failed to provide a confident diagnosis; thus, there is a selection bias toward patients with difficult to interpret or discordant mammographic and sonographic studies.

MRI was not evaluated in this study because most of the patients were not eligible for MRI because of insurance-related issues, personal choice, acute claustrophobia, or physical issues, such as ferromagnetic implants, pacemakers, excessive body habitus or weight exceeding the table limit, and breasts too large for the coil. In addition, MRI is not recommended in the workup of indeterminate lesions according to the 2005 guidelines established by the American Cancer Society.

In summary, for patients who had adjunctive imaging procedures with results discordant from those for mammography, breast-specific gamma imaging provided higher accuracy than ultrasound (77% vs 53%), and the group of patients with BI-RADS category 0 mammograms received the greatest benefit from the use of breast-specific gamma imaging. Pathology reports obtained from needle biopsy show the largest differences in detection between the two modalities are atypical ductal hyperplasia (breast-specific gamma imaging was positive in six cases and ultrasound was positive in four cases) and ductal carcinoma in situ (breast-specific gamma imaging was positive in 20 cases and ultrasound was positive in 13 cases).

The primary advantage of ultrasound is that it does not involve exposure to ionizing radiation. For a breast-specific gamma imaging examination, the patient undergoes an IV injection of ^99mTc-sestamibi, a radiopharmaceutical common to nuclear medicine procedures. According to the U.S. Food and Drug Administration drug data sheet for sestamibi and all published literature to date, the recommended dose for breast imaging is 740–1110 MBq. A 740-MBq injection of sestamibi exposes the patient to a whole body effective dose of approximately 6 mSv [10]. This dose is comparable with or lower than that of other diagnostic imaging procedures, and the radiation dose to the breast tissue is lower than that for standard mammography [11]. According to the National Institutes of Health [12], an exposure of 6 mSv increases the lifetime risk of fatal cancer of less than one-tenth of a percent, from 25% to 25.024%. As with all medical procedures, the risks of performing the procedure must be considered along with the benefits. In this population of 329 patients, when the adjunctive imaging study results were discordant with those of mammography, breast-specific gamma imaging detected eight more malignant lesions than ultrasound. As a result, 2% of patients experienced a benefit from breast-specific gamma imaging; therefore, the benefit-to-risk ratio is 83:1.

In addition, it should be noted that the existing dose recommendation was established in the 1990s using the general nuclear medicine imaging systems. The newer position-sensitive photomultiplier and cadmium zinc telluride technologies used in the breast-optimized cameras have nearly identical photon sensitivity that is more than 3.5 times greater than the older systems [15, 16]. With this improvement, these advanced technologies should be capable of producing adequate clinical images using an injected dose of 148–296 MBq, resulting in whole-body radiation dose of 1.2–2.5 mSv. Several prospective studies are under way to determine the clinical impact of reducing the dose; however, using a dose of less than 740 MBq at this time is an off-label use of sestamibi without clinical literature to show comparable efficacy when using the lower dose.

In conclusion, breast-specific gamma imaging significantly contributed to the detection of malignant or high-risk lesions in patients with negative or indeterminate mammographic findings and it provided improved management of patients with indeterminate mammograms when compared with ultrasound. However, neither breast-specific gamma imaging nor ultrasound should be used to obviate biopsy in patients with suspicious findings in mammography. In addition, although breast-specific gamma imaging involves radiation exposure to the patient, the benefit of using breast-specific gamma imaging outweighs the risks by a factor of 82:1. Breast-specific gamma imaging is a useful diagnostic modality to augment mammography in the management of patients with difficult to diagnose breast tissue and in cases where unresolved clinical concern remains after mammography.

Dr. Weigert is on the Speaker’s Group with Dilon Technologies, Mammotome, and GE Medical. Drs. Bertrand, Stern, and Kieper have a financial disclosure with Dilon Technologies.