Comparison of BSGI, MRI, mammography, and ultrasound for the diagnosis of breast lesions and their correlations with specific molecular subtypes in Chinese women

Breast cancer remains the third leading cause of cancer globally and the leading cancer-related cause of death in women, with 1.7 million new cases having been diagnosed in 2016 [1]. The exact etiological basis for breast cancer is complex and remains to be fully clarified. However, as the rapid diagnosis of early-stage disease is the most reliable means of achieving a positive prognosis, the identification of optimal diagnostic approaches for the analysis of breast lesions is essential. Mammography (MMG) remains the primary approach used for breast cancer screening and detection throughout the world, with subsequent ultrasound (US) being used to confirm cancer diagnosis, for image-guided breast biopsy and localization, for axillary assessment, and to follow-up on any abnormal magnetic resonance imaging (MRI). Breast MRI is the most sensitive approach to detecting and diagnosing breast cancer. Breast-specific gamma imaging (BSGI), also known as ^99mTc-sestamibi scintigraphy, is a high-resolution molecular breast radio-imaging approach that can also be used to precisely detect breast cancer in tissues of variable density. In the present manuscript, we compared the efficacy of BSGI, ultrasound, mammography, and MRI for the differential diagnosis of breast lesions in order to better understand the relative clinical value of BSGI as a diagnostic tool, and found that such diagnostic processes can be improved by using ^99mTc-sestamibi scintigraphy as an auxiliary modality to mammography and ultrasound, thereby potentially improving the reliability of breast lesion diagnosis.

Sensitivity, specificity, and positive and negative predictive values for BSGI, ultrasound, mammography, and MRI were determined. In addition, χ² tests were used to compare the relative efficiencies of these four imaging modalities and associated indicators. The correlation between the LNR of lesions measured by malignant tumors and benign lesions and by pathology was investigated via a Pearson nonparametric correlation analysis. All statistical testing was conducted using SPSS v.22 with a significance threshold of P < 0.05.

This is the first study to our knowledge to have compared the relative diagnostic efficacy of BSGI, mammography, ultrasound, and MRI as a means of differentiating between benign and malignant breast lesions as a function of tumor molecular subtype among Chinese women. Overall, we found that BSGI was an effective approach to diagnosing suspicious lesions. Mammography-based screening is generally the first-line approach to detecting breast cancer in patients with no overt signs of disease, and as such, this approach has seen widespread clinical implementation in recent decades. However, mammography is sensitive to breast density and to the presence of scar tissue, with increasing breast density being associated with reductions in sensitivity from 85 to 68%. Breast density is also a risk factor for breast cancer development [6‐8], and upwards of 75% of Chinese women have heterogeneously or extremely dense breasts. As such, mammography-based screening approaches are of limited utility in these women, and alternative imaging modalities are thus required. BSGI is unaffected by breast density, scar tissue, structural deformities, or radiation therapy. In a retrospective analysis of 341 women that underwent BSGI prior to surgical tumor removal, Rechtman et al. [9] found BSGI to have an overall sensitivity of 95.4% for breast cancer detection, and they also determined that parenchymal breast density failed to impact such sensitivity. Consistent with these findings, in the present study we observed comparable BSGI sensitivity in dense A-D breast tissue. Chung et al. [10] previously conducted a retrospective analysis of 302 breast lesions in 266 women, and in so doing determined that BSGI was more specific than was adjunctive ultrasound US without any sensitivity loss, suggesting that adjunctive BSGI may be a valuable complementary imaging approach to detecting breast cancer in women with suspicious mammography findings. In our study, BSGI achieved the highest specificity (80.7%) of all four tested imaging modalities (Fig. 4).

Much like mammography, ultrasound is an anatomical approach to breast cancer diagnosis that is non-invasive, efficient, and convenient. As such, ultrasound is widely employed in clinical settings for the BI-RADS breast lesion classification. When women with palpable breast masses present for imaging and initial mammography findings are negative, breast cancer cannot generally be excluded. Ultrasound is thus commonly used as an adjunctive imaging modality in these women, particularly in those with dense breast tissue [11]. Tadwalkar et al. [12] retrospectively analyzed BSGI results in 129 women with invasive carcinoma, and found the efficiency of this diagnostic approach to be related to both tumor size and differentiation status. For lesions < 1 cm, BSGI exhibited sensitivity comparable to that of MRI, mammography, and ultrasound, whereas BSGI was superior to ultrasound and mammography when lesions were > 1 cm in size.

While physiological nuclear imaging approaches are commonly used to diagnose tumors, using auxiliary methods in addition to these physiology-based approaches can significantly improve rates of tumor detection. BSGI is a functional imaging modality that achieves improved sensitivity in patients with dense breasts and tumors < 1 cm in size relative to traditional planar scintigraphy [13]. Following injection into patients, the radioactive tracer ^99mTc-sestamibi can enter mitochondria. As numbers of mitochondria are generally correlated with cellular metabolic activity, highly metabolically active cancer cells take up ^99mTc-sestamibi more readily than do cells in the surrounding tissue. Indeed, tumor cells take up ^99mTc-sestamibi at rates ≥50% higher than do normal cells [14]. Kim et al. [15] conducted a retrospective study of 520 patients with suspected breast cancer that were evaluated via ^99mTc-sestamibi imaging, and found that lesions from patients diagnosed with malignant disease exhibited significantly higher LNR values than did those with benign disease (2.00 ± 1.88 vs. 0.60 ± 0.70). Choi et al. [16] similarly found LNR values to be higher in malignant breast lesions relative to benign lesions (2.2 ± 1.0 vs 1.6 ± 0.5). Consistent with these past results, we found that patients ultimately diagnosed with malignant lesions exhibited an LNR of 2.66 ± 1.32, with this value being significantly increased relative to that of patients with benign lesions (1.46 ± 0.49).

^99mTc-sestamibi is a non-specific agent when used for tumor imaging, and as such it can yield false-positive results when taken up by hyperplasic benign lesions, thus reducing its diagnostic performance. Fibrocystic breast disease, fibroadenomas, and breast benign hyperplasia are the leading causes of false-positive BSGI results [17]. Furthermore, malignant breast tumors drive local angiogenesis, and certain benign growths such as intraductal papillomas, inflammatory lesions, fibroadenomas, or adenopathies can mimic this activity and are associated with abundant blood supplies. Inflammatory lesions also exhibit features including irregular infiltration of the surrounding tissue. Weight et al. [18] found that BSGI was associated with a higher incidence of altered patient management (109/119) than was ultrasound (71/119), supporting the fact that BSGI achieves higher levels of positive predictive value and accuracy than does ultrasound. In this study, we detected 31 false-positive lesions via BSGI, with this rate being lower than the 47 false-positive lesions detected via ultrasound (Fig. 5).

MRI is accepted to be the optimal imaging modality for the diagnosis, staging, and monitoring of breast cancer in patients undergoing neoadjuvant chemotherapy (NAC) treatment in order to assess therapeutic responses [19]. Bilimoria et al. [20] found that breast MRIs are the most sensitive approach to detecting breast cancer, and that they have the potential to detect tumors that are not detectable upon physical examination, mammography, or ultrasound. However, breast MRIs can nonetheless yield high false-positive rates and appear to enlarge tumors, potentially leading women diagnosed via this approach to undergo an unnecessary mastectomy. MRI can also not be used regularly owing to its high costs. Hwang et al. [21] determined that preoperative MRI assessment was not sufficient to predict the odds of achieving negative margins for lumpectomy specimens, nor did such assessment reduce subsequent re-excision rates. As preoperative MRI is not associated with benefits associated with ipsilateral breast tumor recurrence (IBTR), there is no cause at present to recommend that preoperative MRI be integrated into the routine assessment of all women with newly diagnosed breast cancer.

Other imaging approaches such as quantitative shear-wave elastography (SWE) have also been employed to enhance breast cancer diagnostic accuracy. As malignant breast lesions are harder than healthy breast tissue, SWE can both quantify tissue hardness and offer information regarding tumors. Yoon et al. [22] analyzed 199 consecutive women and determined that SWE was associated with higher false-positive than false-negative rates, found size, depth, and breast thickness all having impacted these findings. Kim et al. [23] also reported in their study of 166 total masses (118 benign, 48 malignant) that false SWE features were more commonly detected in benign masses (53% vs. 8.2%).

Recent advances in digital x-ray system development have also facilitated the design of approaches such as contrast-enhanced spectral mammography (CESM) that can overcome many of the limitations of mammography and achieve diagnostic efficacy comparable to MRI. Xing et al. [19] found that CESM achieved higher specificity than MRI (89.5% vs 80.2%) and that the same was true for its positive predictive value (94.7% vs 90.5%). However, this approach is limited by the fact that it is unable to image the entire chest wall and axilla, requires compression, utilizes ionizing radiation, and requires that patients receive iodinated contrast injection. In a prospective two-center, multi-reader study, Fallenberg et al. [24] determined that MRI and CESM exhibited comparable diagnostic performance, but that CESM had higher specificity and lower sensitivity relative to MRI. Given that radiation dosing is always a pertinent concern, it is worth noting that combination CESM + MMG imaging delivers radiation doses higher than mammography alone. In addition, the invasiveness of this approach and the need for contrast injection make it unsuitable for breast cancer screening.

In summary, the results of this study demonstrate that, relative to ultrasound and mammography, BSGI is a highly sensitive and specific tool that can be used to reliably evaluate breast lesions. BSGI achieved sensitivity that was comparable to that of MRI when used as an auxiliary diagnostic imaging modality to ultrasound and mammography, suggesting that it can further enhance diagnostic efficacy in this context. As such, our findings highlight the value of the clinical application of BSGI as a means of differentiating between benign and malignant breast lesions with the goal of reducing the incidence of misdiagnosis and unnecessary surgery.