Clinical usefulness of breast-specific gamma imaging as an adjunct modality to mammography for diagnosis of breast cancer: a systemic review and meta-analysis

Review Article

DOI: 10.1007/s00259-012-2279-5

Cite this article as:
Sun, Y., Wei, W., Yang, HW. et al. Eur J Nucl Med Mol Imaging (2013) 40: 450. doi:10.1007/s00259-012-2279-5

Abstract

Purpose

The purpose of this study was to assess the diagnostic performance of breast-specific gamma imaging (BSGI) as an adjunct modality to mammography for detecting breast cancer.

Methods

Comprehensive searches of MEDLINE (1984 to August 2012) and EMBASE (1994 to August 2012) were performed. A summary receiver operating characteristic curve (SROC) was constructed to summarize the overall test performance of BSGI. The sensitivities for detecting subcentimetre cancer and ductal carcinoma in situ (DCIS) were pooled. The potential of BSGI to complement mammography was also evaluated by identifying mammography-occult breast cancer.

Results

Analysis of the studies revealed that the overall validity estimates of BSGI in detecting breast cancer were as follows: sensitivity 95 % (95 % CI 93–96 %), specificity 80 % (95 % CI 78–82 %), positive likelihood ratio 4.63 (95 % CI 3.13–6.85), negative likelihood ratio 0.08 (95 % CI 0.05–0.14), and diagnostic odds ratio 56.67 (95 % CI 26.68–120.34). The area under the SROC was 0.9552 and the Q* point was 0.8977. The pooled sensitivities for detecting subcentimetre cancer and DCIS were 84 % (95 % CI 80–88 %) and 88 % (95 % CI 81–92 %), respectively. Among patients with normal mammography, 4 % were diagnosed with breast cancer by BSGI, and among those with mammography suggestive of malignancy or new biopsy-proven breast cancer, 6 % were diagnosed with additional cancers in the breast by BSGI.

Conclusion

BSGI had a high diagnostic performance as an excellent adjunct modality to mammography for detecting breast cancer. The ability to identify subcentimetre cancer and DCIS was also high.

Keywords

Breast neoplasm Breast-specific gamma imaging Mammography Meta-analysis 

Introduction

Mammography is used as a standard breast cancer screening method because of its high sensitivity in most cases and because it leads to reduced mortality [1]. However, it has some significant limitations. As breast tissue density increases, the sensitivity of mammography deceases. Rosenberg et al. [2] found that the sensitivity in nondense breasts was 85 %, but was only 68 % in dense breasts. Breast density is strongly associated with the risk of developing breast cancer [3, 4]. Furthermore, patients with dense breasts are often young, and in this patient group breast cancers tend to be aggressive. So it is very important to find an effective modality as an adjunct to mammography.

Breast-specific gamma imaging (BSGI), also called molecular breast imaging, is a nuclear medicine breast imaging technique that uses a high resolution, small field-of-view breast-specific gamma camera. It has been significantly improved within recent years. BSGI is a functional imaging examination rather than an anatomic modality like mammography. It produces imaging based on two physiological mechanisms. One is that BSGI uses the radiopharmaceutical 99mTc sestamibi or 99mTc tetrofosmin, which specifically binds to mitochondria in cells. The density of mitochondria is a marker of cellular proliferative activity. Therefore, in the cancer cells there is a higher uptake of radiotracer than in the surrounding normal tissue [5]. The other mechanism involves neoangiogenesis in the cancer tissue, which leads to increasing pharmaceutical delivery to the lesions [6]. Therefore, in contrast to mammography, the sensitivity of BSGI is not influenced by the density of the breast tissue, implants, architectural distortion, or scars from prior surgery or radiation. In addition, in comparison to the conventional gamma camera, the dedicated gamma camera has greater intrinsic spatial resolution and accessibility to the posterior and medial areas of the breast, less radiation scatter from nearby organs on imaging, a minimal distance between the breast and the detector, and a lower amount of breast tissue between the lesion and the detector through mild compression and imaging in the positions comparable to mammography [7, 8, 9]. As a result, BSGI has better sensitivity than traditional planar scintimammography, especially in detecting subcentimetre or nonpalpable breast cancer.

More and more studies have now investigated the potential of BSGI for detecting breast cancer. Its sensitivity ranges from 85 % to 100 % and its specificity ranges from 60 % to 95 %. To our knowledge, there is no meta-analysis of the performance of BSGI. So we performed this analysis to evaluate the diagnostic performance of BSGI as an adjunct modality to mammography for detecting breast cancer.

Materials and methods

Search strategy

We searched MEDLINE (1984 to August 2012) and EMBASE (1994 to August 2012) with the language restriction of English to identify studies evaluating the diagnostic performance of BSGI in the detection of breast cancer. The search terms used were “breast cancer” OR “breast neoplasm” OR “breast carcinoma” OR “breast tumor”; “molecular breast imaging” OR “MBI” OR “breast-specific gamma imaging” OR “BSGI” OR “scintimammography” OR “high-resolution gamma camera”; “mammography”; and “sensitivity” OR “specificity”. Corresponding medical subject headings were also used. In addition, reference lists from all relevant articles were searched to identify additional studies. At first there were no restrictions as to publication form in order to achieve a highly sensitive search. However, in the end conference abstracts were excluded because of the limited data presented.

Study selection

Eligible studies were required to fulfil the following inclusion criteria.
  1. (a)

    Patients had to have at least one of the following indications for BSGI: clinical abnormality such as a palpable mass, breast pain or bloody nipple discharge with normal mammography; suspicious mammography findings such as indeterminate asymmetry or calcifications; dense breast tissue, surgical scar or architectural distortion which was difficult to evaluate by mammography; personal history of breast cancer or a high-risk lesion; family history or other high-risk factors for breast cancer; and mammography suggestive of malignancy or new biopsy-proven breast cancer for further examination.

     
  2. (b)

    To avoid selection bias, one study had to involve at least ten patients.

     
  3. (c)

    If overlapping patient cohorts were presented among multiple studies, only the largest or the latest study was included.

     
  4. (d)

    The BSGI camera used just a single detector. Dual-head dedicated breast gamma cameras were excluded.

     
Reviews and conference abstracts were excluded. For evaluating diagnostic performance, a 2 × 2 table for true-positive, false-negative, false-positive and true-negative values for identifying breast cancer was derived from the data provided, and histopathological assessment and/or clinical and imaging follow-up was used as the reference standard. When evaluating the potential as an adjunct to mammography, the study to be selected had to include data comparing BSGI with mammography.

Two authors (Yu Sun and Wei Wei) independently screened titles and abstracts of the relevant articles based on the inclusion criteria. When an article fulfilled the criteria, the full text was reviewed. Any disagreement was resolved by a third author.

Data extraction and quality assessment

Two authors (Yu Sun and Wei Wei) independently read the eligible papers, and recorded the first author’s name, publication year, original country, number of patients and lesions, patient age, study design and reference standard. The methodological quality of the study was estimated using the quality assessment of diagnostic accuracy studies (QUADAS) tool [10]. For each item there are three grades: “yes”, “unclear” and “no” with scores of 1 for “yes” and 0 for “unclear” or “no”. There are 14 items in the QUADAS. Items 1 (representative spectrum) and 2 (selection criteria) are about the variability of the studies, items 8 (index test execution), 9 (reference standard execution) and 13 (interpretable results reported) are about the quality of the reporting, and the remaining items are about the bias of the studies. The rate of response “yes” for each question was calculated. If there were different opinions, then the problem was discussed by a third author.

Statistical analysis

A bivariate analysis was used to determine the per-lesion sensitivity, specificity, positive likelihood ratio (PLR), negative likelihood ratio (NLR) and diagnostic odds ratio (DOR) with corresponding 95 % confidence intervals (CIs) [11]. Atypical lesions were grouped with benign lesions. We evaluated only the ability of BSGI to detect breast cancer. Heterogeneity was evaluated by Q and I2 statistics. If the P value for a Q statistic was less than 0.1 or the I2 statistic was greater than 50 %, we deemed that there was statistically significant heterogeneity [12]. Then we pooled studies using a random effects model [13], and otherwise used a fixed effect model [14]. Although no absolute cut-off was used, an effective diagnostic test should have a PLR greater than 5.0 and a NLR less than 0.2 [15, 16]. DOR is the ratio of the odds of positivity in patients with disease relative to that in patients without disease. It is a single indicator of test performance [17]. The higher the DOR value, the better the diagnostic performance of the test.

The sensitivity and specificity of each study were used to plot a summary receiver operating characteristic (SROC) curve [18, 19]. Q* indexes (the point on the SROC curve where sensitivity and specificity are equal) were calculated. The higher the Q* value, the better the diagnostic test performance [18]. Deek’s test was performed to assess publication bias [20]. A P value less than 0.05 indicates the existence of publication bias. We also pooled the per-lesion sensitivity of BSGI for identifying subcentimetre cancer and ductal carcinoma in situ (DCIS).

Two ratios were defined as follows: the ratio of the number of patients with cancer detected by BSGI only rather than by mammography relative to the total number of patients with normal mammography; and the ratio of the number of patients with multicentric, multifocal or bilateral cancers detected by BSGI only rather than mammography relative to the total number of patients with mammography suggestive of malignancy or new biopsy-proven breast cancer. The values were also recalculated from relevant studies.

All analyses were executed using Meta-DiSc, version 1.4 (XI Cochrane Colloquium, Barcelona, Spain) and Stata, version 12.0 (Stata Corporation, College Station, TX).

Results

Literature search

The retrieval strategy and application of the eligibility criteria detailed above resulted in the selection of 40 articles from 11 institutions. Six articles were excluded because a dual-head detector had been used. Another 15 articles were excluded because that they did not have the data needed or they were not the largest or latest studies in their institution on the aspects analysed in this study. Finally 19 studies were eligible for our meta-analysis [21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39](Fig. 1).
Fig. 1

Flow diagram of the study selection procedure for the meta-analysis

Study characteristics and quality assessments

The 19 studies published between 2004 and 2012 shown Table 1 were selected for analysis. Eight studies including 2,183 lesions were evaluated for diagnostic value [23, 25, 26, 32, 33, 34, 36, 38]. Five studies including 276 lesions were evaluated for sensitivity in detecting subcentimetre lesions [26, 27, 36, 37, 39]. Six studies including 161 lesions were evaluated for sensitivity in detecting DCIS [24, 29, 33, 35, 37, 39]. In five studies in which 350 patients with normal mammography were also examined by BSGI, 47 patients were proven to have breast cancer [21, 22, 28, 34, 36]. In five studies, 904 patients with mammography suggestive of malignancy or new biopsy-proven breast cancer were also examined by BSGI, and 64 patients were found to have multicentric, multifocal or bilateral breast cancer [25, 29, 30, 31, 36]. The studies used 99mTc sestamibi or 99mTc tetrofosmin as radiopharmaceutical.
Table 1

Characteristics of the studies

Reference

Year

Country

No. of patients

No. of lesions

Age (years), mean (range)

Design

Reference standard

QUADAS score

[21]

2004

USA

37

37

54 (34–80)

Prospective

Histopathology or mammography

11

[22]

2005

USA

94

94

55 (36–78)

Prospective

Histopathology or imaging and clinical follow-up

12

[23]

2006

Italy

29

29

(27–77)

Prospective

Histopathology or imaging and clinical follow-up

12

[24]

2007

USA

20

22

55 (34–76)

Retrospective

Histopathology

11

[25]

2007

USA

99

114

59 (18–86)

Unclear

Histopathology

13

[26]

2008

USA

146

167

53.1 (32–98)

Retrospective

Histopathology

13

[27]

2008

USA

149

245

Unclear

Unclear

Histopathology or imaging and clinical follow-up

12

[28]

2008

USA

176

182

53 (27–86)

Retrospective

Histopathology or imaging and clinical follow-up

11

[29]

2009

USA

138

163

55 (30–81)

Retrospective

Histopathology

11

[30]

2009

USA

82

100

53 (33–83)

Retrospective

Histopathology

12

[31]

2010

USA

159

213

54 (29–93)

Retrospective

Histopathology or imaging and clinical follow-up

11

[32]

2011

USA

100

100

Unclear

Unclear

Histopathology

11

[33]

2012

USA

416

416

Unclear

Retrospective

Histopathology or imaging and clinical follow-up

12

[34]

2012

USA

329

329

Unclear

Retrospective

Histopathology or imaging and clinical follow-up

12

[35]

2012

USA

18

18

51

Prospective

Histopathology

12

[36]

2012

Italy

467

554

57 (26–81)

Prospective

Histopathology

13

[37]

2012

Italy

33

33

56.8 (41–81)

Retrospective

Histopathology

12

[38]

2012

Korea

471

474

49.63 ± 10.43

Retrospective

Histopathology or imaging and clinical follow-up

12

[39]

2012

Korea

121

228

45 ± 8.1

Retrospective

Histopathology

13

The QUADAS tool was used to assess the quality of studies (Fig. 2). For six studies the response was “yes” to 11 of 14 items [21, 24, 28, 29, 31, 32]. For nine studies the response was “yes” to 12 of 14 items [22, 23, 27, 30, 33, 34, 35, 37, 38]. For four studies the response was “yes” to 13 of 14 items [25, 26, 36, 39]. For Items 1 and 2, 68 % and 95 % of the responses were “yes”. For items 8, 9 and 13, 95 %, 100 % and 100 % of the responses, respectively, were “yes”. For the remaining items which assessed bias, the rates for the response “yes” were relatively high, except for item 11 (index test results blinded). In 18 studies, the results of reference standard were interpreted with knowledge of the index test results, and for 95 % of the studies the response to question 11 was “no”.
Fig. 2

Study design characteristics based on the QUADAS tool

Data synthesis

In eight studies the diagnostic performance of BSGI was evaluated [23, 25, 26, 32, 33, 34, 36, 38]. All the patients in these eight studies had histopathology or imaging and clinical follow-up, and all these studies provided true-positive, false-negative, false-positive and true-negative values for BSGI. Because the heterogeneity was significant, a random effects model was used. The result showed a pooled sensitivity and specificity of BSGI of 95 % (95 % CI 93–96 %) and 80 % (95 % CI 78–82 %), respectively (Figs. 3 and 4). The pooled PLR was 4.63 (95 % CI 3.13–6.85), and the NLR was 0.08(95 % CI 0.05–0.14; Figs. 5 and 6). The pooled DOR was 56.67 (95 % CI 26.68–120.34; Fig. 7). The area under the SROC curve was 0.9552, and the summary point Q* was 0.8977 (Fig. 8). Publication bias in the literature was evaluated using Deek’s test. The result was T = −0.3, P = 0.775, indicating that there was no publication bias (Fig. 9).
Fig. 3

Forest plot of sensitivity of BSGI for detecting breast cancer

Fig. 4

Forest plot of specificity of BSGI for detecting breast cancer

Fig. 5

Forest plot of PLR of BSGI for detecting breast cancer

Fig. 6

Forest plot of NLR of BSGI for detecting breast cancer

Fig. 7

Forest plot of DOR of BSGI for detecting breast cancer

Fig. 8

SROC and Q* index of BSGI for detecting breast cancer

Fig. 9

Deek’s test plot of the studies evaluating the diagnostic performance of BSGI

In five studies the sensitivity of BSGI for detecting subcentimetre breast cancer was evaluated [26, 27, 36, 37, 39]. Because the heterogeneity was significant (Q = 19.39, P = 0.0007, I2 = 79.4 %), a random effects model was used. The pooled sensitivity was 84 % (95 % CI 80–88 %; Fig. 10).
Fig. 10

Forest plot of sensitivity of BSGI for detecting subcentimetre breast cancer

In six studies the sensitivity of BSGI for detecting DCIS was evaluated [24, 29, 33, 35, 37, 39]. Because there was no heterogeneity (Q = 8.86, P = 0.1146, I2 = 43.6 %), a fixed effects model was used. The pooled sensitivity was 88 % (95 % CI 81–92 %) (Fig. 11).
Fig. 11

Forest plot of sensitivity of BSGI for detecting DCIS

In five studies patients with normal mammography were also examined by BSGI [21, 22, 28, 34, 36]. The rates of newly diagnosed breast cancer were recorded. Because the heterogeneity was significant (Q = 123.74, P = 0.000, I2 = 96.8 %), a random effects model was used. The pooled rate was 18 % (95 % CI 6–30 %) (Fig. 12). The rate of newly diagnosed breast cancer found in the study by Spanu et al. [36] was much higher than in the other studies. If this study by Spanu et al. was omitted, the heterogeneity decreased (Q = 8.36, P = 0.039, I2 = 64.1 %) the pooled rate was 4 % (95 % CI 1–8 %; Fig. 13).
Fig. 12

Pooled proportion of patients with normal mammography in whom breast cancer was detected by BSGI

Fig. 13

Pooled proportion of patients with normal mammography in whom breast cancer was detected by BSGI after omitting the study by Spanu et al. [36]

In five studies patients with mammography suggestive of malignancy or new biopsy-proven breast cancer were also examined by BSGI [25, 29, 30, 31, 36]. The rates of multicentric, multifocal, and bilateral cancer were recorded. Because there was no heterogeneity (Q = 6.61, P = 0.158, I2 = 39.5 %), a fixed effects model was used. The pooled rate was 6 % (95 % CI 5–8 %; Fig. 14).
Fig. 14

Pooled proportion of patients with mammography highly suggestive of malignancy or new biopsy-proven breast cancer in whom additional cancers were detected by BSGI

Discussion

The results of this meta-analysis indicated that BSGI has an excellent diagnostic performance as an adjunct modality to mammography for identifying breast cancer with high sensitivity and moderate specificity. The overall PLR was 4.63 and the NLR was 0.08. This means that the likelihood of a positive BSGI result in patients with breast cancer is 4.63 times higher than in patients without the disease, and if BSGI is negative, the probability that the patient has breast cancer is 8 %. The overall DOR was 56.67. This means that the odds ratio of BSGI positivity in a patient with breast cancer relative to the odds of positivity without the disease is 56.67. The area under the curve of the SROC was 0.9552 and point Q* was 0.8977. This reveals the discriminatory power of BSGI for detecting breast lesions.

Fibrocystic changes, fibroadenoma and benign breast tissue were the most common false-positive lesions detected by BSGI [23, 25, 26, 32, 33, 36, 38, 39]. Most of the false-negative lesions were subcentimetre invasive ductal carcinoma and DCIS [26, 33, 36, 37, 38, 39]. Just a few studies reported the poison of the false-negative lesions [25, 26, 33, 36]. Lesions located deep in the breast close to the chest wall were not readily shown by BSGI. Improving detector and breast positioning techniques may reduce the false-negative rate.

Tumor diameter is an independent prognostic indicator. With increasing diameter of the breast cancer from less than 20 mm to more than 50 mm, the 5-year survival rate decreased from 96.3 % to 82.2 % in patients with node-negative disease [40]. So it is important to find an effective examination technique for detecting small cancer to achieve early detection and early treatment. In our analysis, BSGI showed a sensitivity for detecting subcentimetre breast cancer of 84 %, and the smallest carcinoma identified by BSGI was 1 mm [26, 33]. Mammography is the standard screening tool for the diagnosis of DCIS [41, 42]. Most of these lesions show microcalcification [43, 44], but not all DCIS show microcalcification. The sensitivity of mammography in detecting DCIS ranges from 27 % to 82 % [24, 45, 46]. Based on our analysis, the pooled sensitivity of BSGI for detecting DCIS was higher than that of mammography at 88 %.

BSGI has also shown other particular advantages in the detection of breast cancer. Invasive lobular carcinoma (ILC) spreads through diffuse infiltration of single rows of malignant cells. It does not readily form a discrete mass because of its incohesive histological growth pattern [47]. The low rates of suspicious calcification and opacity on mammography lead to difficulties in detecting ILC [48]. The sensitivity of mammography for identifying ILC ranges from 34 % to 81 % [46, 48, 49, 50]. Brem et al. [50] found that the sensitivity of BSGI in detecting ILC was 93 %, which is higher than that of mammography. Women with atypical lesions, such as atypical ductal hyperplasia and lobular neoplasia, are at high risk of breast cancer. Ling et al. [51] reviewed 15 patients in whom the most aggressive pathology on surgical excision was atypical ductal hyperplasia or lobular neoplasia. The sensitivity of BSGI for the detection of atypical breast lesions was 100 %. BSGI was also used to evaluate the response to neoadjuvant chemotherapy or hormone therapy. Spanu et al. [52] assessed 15 patients with locally advanced breast cancer treated by neoadjuvant chemotherapy or hormone therapy. BSGI correctly classified all patients. Two patients had minimal residual disease after treatment, which was not accurately evaluated by clinical examination and other imaging modalities. However, BSGI was able to accurately monitor tumour response. Therefore BSGI could help in the planning of appropriate surgical treatment. The number of patients examined using BSGI is limited. Further studies with a large number of patients are needed to assess the use of BSGI for evaluation of breast cancer.

The results of our analysis revealed that BSGI showed breast cancer in 4 % of patients with normal mammography (BI-RADS categories 1–3), and BSGI showed multicentric, multifocal and bilateral cancers in 6 % of patients with mammography suggestive of malignancy (BI-RADS categories 4 or 5) or new biopsy-proven breast cancer. These new findings could greatly help are vital for both in planning the surgical strategy and in prognosis. The need for a second surgical procedure or a high risk of local recurrence may occur with preoperatively undetectable additional cancers.

The use of BSGI can also reduce the rate of unnecessary biopsies. In the study by Kessler et al. [32], there were 93 mammographies in patients with BI‐RADS category 4, and the positive biopsy rate was 14 %. Before the biopsy they all had BSGI and the results showed 67 lesions were negative. If they carried out biopsy according to the BSGI, 65 lesions would avoid unnecessary biopsies and only 2 lesions with cancer were not biopsied. So if BI‐RADS category 4 lesions negative on BSGI were excluded from biopsy, the positive biopsy rate would have been 42.3 %. The American College of Radiology suggest that the positive biopsy rate should be 25–40 % [32]. In patients with undetermined mammography (BI-RADS category 0), BSGI can correctly categorize the lesions. In the study by Weigert et al. [32], 119 patients were BI-RADS category 0 on mammography. However, 90 patients were correctly categorized by BSGI, in 15 of whom the lesion was malignant and in 75 benign.

Our analysis showed BSGI to have a high diagnostic performance. However, BSGI is considered just as an adjunct rather than as an alternative to mammography. First, mammography screening has been proved to reduce mortality in breast cancer [1], but there is no such study concerning the use of BSGI. Second, the radiation exposure from BSGI is higher than that from mammography, which suggests that BSGI is not suitable for screening breast cancer [53]. So it is better to use BSGI as a complement to mammography.

In order to reduce breast thickness and limit movement artefacts, BSGI uses light pain-free compression forces of 15 lb in contrast to 35–45 lb in standard mammography [54]. O’Connor et al. [25] evaluated the tolerability of patients using a pain score (on a scale of 0 to 10, with 0 indicating no pain). The average pain score for BSGI was 0.8 ± 1.5, and that for mammography was in the range 3.8–4.7 [55]. So BSGI is more comfortable than mammography.

MRI is an established adjunct physiological breast imaging modality. Two studies [24, 56] comparing BSGI with MRI indicated comparable sensitivity and improved specificity of breast cancer diagnosis by BSGI over MRI. In addition, BSGI is performed in a sitting position as against a prone position inside a small chamber in MRI, which makes patients uncomfortable especially those with claustrophobia or an endomorphic body habitus. In MRI the administration of gadolinium causes renal complications which is not an issue with BSGI. Patients with other MRI contraindications such as ferromagnetic implanted devices can also undergo BSGI. Furthermore, the cost of BSGI is lower than that of MRI. Zhou et al. [28] reported the total costs of BSGI, MRI and mammography in their institution were $1,259, $3,400 and $340, respectively. Compared with the four to ten images obtained with BSGI, hundreds images can be obtained with MRI. So the interpretation time for BSGI is shorter. Another advantage of BSGI is the use of the same position as used in mammography, leading to the possibility to compare directly corresponding images from the two modalities.

BSGI is being developed into a standardized diagnostic modality. Clinical and research indications of BSGI have been listed by the Society of Nuclear Medicine [57], and include evaluation of patients with recently detected breast malignancy, patients at high risk of breast malignancy, patients with indeterminate breast abnormalities and remaining diagnostic concerns, patients with technically difficult breast imaging, patients with MRI contraindications and patients who require monitoring of neoadjuvant tumour response. Conners et al. [58] have provided a detailed lexicon for BSGI. The lexicon can been used to regularize BSGI reports. The study [58] also showed that for newly trained radiologists with experience in breast imaging, BSGI studies can be quickly interpreted and the interobserver agreement is high.

In spite of the advantages of BSGI described above, it also has weaknesses, such as its limited ability to detect axillary lymph nodes and to guide biopsy, and relatively high dose of radionuclide. However, researchers from various institutions are developing this technique to overcome these limitations. Concerning the limitation of BSGI in detecting axillary lymph node metastasis of breast cancer, Spanu et al. [59] found that the sensitivity of BSGI in the detection of axillary lymph node metastasis in 76 breast cancer patients was just 25 %. Fewer than three nonpalpable or metastatic nodes could not be detected by BSGI. Jones et al. [60] developed a new method to detect nodes by BSGI. They suggested that the axilla of the patient should be positioned as close as possible to the camera face and the arm hung over the camera at an angle of 90°. A lead apron should be placed across the shoulder and upper arm of the axilla to avoid artefacts from the nearby organs. They have applied this method in the clinic and found it to be a good approach, but there are as yet no detailed data on sensitivity and specificity. Further studies are needed to evaluate this technique.

Some new BSGI-guided biopsy procedures have been suggested. Coover et al. [21] described a detailed method to localize the lesion using an open biopsy paddle with a dedicated breast camera. Welch et al. [61] developed a small dedicated gamma camera-guided stereotactic breast biopsy system. These methods are very useful when the lesion is not detected by palpation or mammography and is shown only by BSGI.

The most important disadvantage is the radiation dose from 99mTc sestamibi and 99mTc tetrofosmin. According to the US Food and Drug Administration drug safety sheet and all the studies to date, 740–1,110 MBq (20–30 mCi) of 99mTc is recommended for breast imaging, which is equivalent to 6.29–9.44 mSv [56]. However, the radiation dose to the breast from a screening mammogram is 0.7–1.0 mSv [62]. Many institutions are engaged in decreasing breast uptake of tracer. Hruska et al. of the Mayo Clinic [63, 64] have put forward two dose-reduction methods: optimized collimation, widened energy window. They used a dual-head camera with low-dose BSGI performed with 148 MBq 99mTc sestamibi and obtained images with quality matching that of standard BSGI performed with 740 MBq dose in phantoms [63]. They also found that an administered dose from BSGI using the above methods of 296 MBq 99mTc sestamibi was possible in patients [64]. The approaches to solve the disadvantages of BSGI are still new. More studies are required to confirm the clinical usefulness of these methods.

Our meta-analysis had several limitations. BSGI is a new technique for identifying breast cancer. The details of the BSGI protocols, such as the camera, the dose of radionuclide and breast position, and the ability of radiologists, are different between institutions. These could all have affected the estimates of diagnostic accuracy which might have been the reasons for the heterogeneity found. This was not analysed because the number of studies included was limited and insufficient. Large and well-designed studies of BSGI are still needed.

Conclusion

BSGI will gradually become a normal procedure in the future. Current evidence suggests that BSGI is an extremely useful adjunct to mammography for its ability to identify breast cancer with a high diagnostic performance. It also has a high sensitivity for detecting subcentimetre cancer and DCIS. BSGI also has some weaknesses including its limited ability to detect axillary lymph nodes and to guide biopsy, and relatively high dose of radionuclide. Many new techniques to solve these problems have been proposed. Studies with a large number of patients are needed in the future to further improve BSGI.

Conflicts of interest

None.

Copyright information

© Springer-Verlag Berlin Heidelberg 2012

Authors and Affiliations

  1. 1.Department of Breast Surgery of Guangxi Cancer HospitalAffiliated Cancer Hospital of Guangxi Medical UniversityNanningPeople’s Republic of China

Personalised recommendations