While there is an increasing role of ultrasound for breast cancer screening in patients with dense breast, conventional anatomical ultrasound lacks sensitivity and specificity for early breast cancer detection. In this study, we assessed the potential of ultrasound molecular imaging using clinically translatable vascular endothelial growth factor receptor type 2 (VEGFR2)-targeted microbubbles (MBVEGFR2) to improve the diagnostic accuracy of ultrasound in earlier detection of breast cancer and ductal carcinoma in situ (DCIS) in a transgenic mouse model [FVB/N-Tg(MMTV-PyMT)634Mul]. In vivo binding specificity studies (n = 26 tumors) showed that ultrasound imaging signal was significantly higher (P < 0.001) using MBVEGFR2 than nontargeted microbubbles and imaging signal significantly decreased (P < 0.001) by blocking antibodies. Ultrasound molecular imaging signal significantly increased (P < 0.001) when breast tissue (n = 315 glands) progressed from normal [1.65 ± 0.17 arbitrary units (a.u.)] to hyperplasia (4.21 ± 1.16), DCIS (15.95 ± 1.31), and invasive cancer (78.1 ± 6.31) and highly correlated with ex vivo VEGFR2 expression [R2 = 0.84; 95% confidence interval (CI), 0.72–0.91; P < 0.001]. At an imaging signal threshold of 4.6 a.u., ultrasound molecular imaging differentiated benign from malignant entities with a sensitivity of 84% (95% CI, 78–88) and specificity of 89% (95% CI, 81–94). In a prospective screening trail (n = 63 glands), diagnostic performance of detecting DCIS and breast cancer was assessed and two independent readers correctly diagnosed malignant disease in more than 95% of cases and highly agreed between each other [intraclass correlation coefficient (ICC) = 0.98; 95% CI, 97–99]. These results suggest that VEGFR2-targeted ultrasound molecular imaging allows highly accurate detection of DCIS and breast cancer in transgenic mice and may be a promising approach for early breast cancer detection in women. Cancer Res; 73(6); 1689–98. ©2012 AACR.

Breast cancer is the most common cancer diagnosed in women in the United States and the second leading cause of death from cancer in women (1). The American Cancer Society estimates 229,060 new diagnoses of breast cancer and 39,920 deaths from this cancer in 2012 in the United States (1). Earlier detection can substantially reduce deaths from breast cancer. Screening mammography, currently the most frequently conducted imaging procedure for breast cancer screening, has been shown to decrease mortality by 22% in women 50 years or older and by 15% in women ages 40 to 49 years (2).

However, diagnostic accuracy of mammography in detecting breast cancer is reduced in women with dense breasts. More than 50% of women younger than 50 years have either heterogeneously dense or very dense breast tissue, as do at least one third of women older than 50 years (3). In women with dense breast tissue, mammographic sensitivity for cancer detection can be as low as 30% to 48% (4, 5) with much higher interval cancer rates (5, 6) and worse prognosis in subsequent clinically detected cancers (7). Therefore, recent studies have addressed the value of adding ultrasound imaging to screening mammography (4, 8–12). A multicenter study of 2,637 women at increased risk for breast cancer with dense breast tissue showed that breast cancer was diagnosed on ultrasound alone in 12 of 40 patients (30%; ref. 7). However, the positive predictive value (PPV) of ultrasound for cancer detection in that study was only 8.6%, resulting in a large number of unnecessary callbacks and biopsies (7). In addition, the sensitivity of ultrasound conducted alone in detecting invasive breast cancer was only 50% (7) and even lower (27%) in another prospective multimodality screening study of 609 asymptomatic high-risk women (13). These findings underscore that further improvements in ultrasound are critically needed to increase its accuracy in diagnosing malignant precursor or invasive breast cancer lesions, should it be included as a complementary imaging modality into breast cancer screening protocols.

Ultrasound molecular imaging combines the advantages of ultrasound with the capabilities of molecular imaging to visualize molecular signatures with high sensitivity and specificity in vivo (14, 15). Microbubble contrast agents already in clinical use with ultrasound can be further chemically modified to accumulate at vascular endothelial cells overexpressing molecular markers associated with either precursor lesions or invasive cancer (14, 16, 17). VEGF receptor type 2 (VEGFR2), a transmembrane receptor tyrosine kinase, is a well-studied molecular marker overexpressed on angiogenic vascular endothelial cells of cancer (18, 19). Several studies have shown increased angiogenesis and VEGFR2 overexpression in invasive breast cancer in human breast cancer tissues (19–21). Recent immunohistochemical studies suggest increased angiogenesis even at the precursor ductal carcinoma in situ (DCIS) stage (22). Ultrasound molecular imaging of VEGFR2 expression on angiogenic vessels in DCIS and invasive breast cancer may therefore be a promising new approach for improving diagnostic accuracy of ultrasound in earlier detection of precursor lesions and breast cancer.

In our study, we evaluated how ultrasound molecular imaging can depict different histologic stages of breast cancer development, including mammary hyperplasia, DCIS, and invasive breast cancer, in the well-established FVB/N-Tg(MMTV-PyMT)634Mul transgenic mouse model (23–25) by using a new, clinical-grade VEGFR2-targeted contrast microbubble. In a prospective screening trial of transgenic and control mice, we show that DCIS and invasive breast cancer can be detected with high diagnostic accuracy and reliability by ultrasound molecular imaging.

Transgenic mouse model

All procedures involving the use of laboratory animals were approved by the Institutional Administrative Panel on Laboratory Animal Care at Stanford University (Stanford, CA). A transgenic mouse model of breast cancer [FVB/N-Tg(MMTV-PyMT)634Mul] was used for all experiments. In this model, the mammary epithelium transforms into different histologic stages of breast cancer development, similar to disease progression in humans, including mammary hyperplasia, DCIS, and invasive carcinoma (24, 25).

Design of microbubbles

A novel clinical-grade VEGFR2-targeted microbubble contrast agent (MBVEGFR2) containing perfluorobutane and nitrogen (BR55; synthesized under GMP by Bracco Suisse SA) was prepared as described previously (26, 27). Briefly, a heterodimeric peptide (5.5-kDa) that binds to VEGFR2 with high affinity (dissociation constant, KD = 0.5 nmol/L) was conjugated with the amino group of DSPE-PEG2000-NH2 [1,2-distearoyl-sn-glycerol-3-phosphoethanolamine-N-[amino (polyethylene glycol)-2000] to form VEGFR2-binding lipopeptides (28, 29). This lipopeptide construct was purified and incorporated into the phospholipid-based shell of microbubble (mean number of heterodimeric peptides per square micrometer on the MB shell: 34,200 ± 1,300; range, 31,800–36,600). The resultant product was lyophilized and stored in septum-sealed vials. Before injection in mice, microbubbles were reconstituted in 2 mL of 5% glucose. The mean diameter of MBVEGFR2 was 1.5 μm (range, 1–3 μm), as assessed by using a cell counter and sizer (Multisizer III Coulter Counter; Beckman Coulter; ref. 26). MBVEGFR2 was previously shown to cross-react with murine, rat, and human VEGFR2 (26, 30). Nontargeted microbubbles that lack the VEGFR2-specific lipopeptide construct were used as control microbubbles (MBControl).

Ultrasound molecular imaging

Imaging protocol.

Ultrasound molecular imaging was conducted as described previously (31), and a detailed description of the imaging protocol is provided in the Supplementary Materials. Images representing signal from adherent microbubbles (molecular imaging signal) were displayed as color maps overlaid on B-mode images, automatically generated by using commercially available Vevo CQ software (VisualSonics). The scale for the color maps was kept constant for all images.

Binding specificity of microbubbles to VEGFR2 in transgenic mouse model.

In the first part of the imaging study, VEGFR2-binding specificity of MBVEGFR2 was tested in 26 transgenic mice bearing tumors (Fig. 1) in an intra-animal comparison study using MBVEGFR2 and MBControl, as well as by conducting in vivo competition experiments (details are provided in Supplementary Materials). To minimize any bias from repetitive injections in the same mice, injections were separated by at least 30 minutes waiting time to allow clearance of microbubbles from previous injections (32).

Figure 1.

Summary of overall study design, which included three parts. MBD, microbubble destruction.

Figure 1.

Summary of overall study design, which included three parts. MBD, microbubble destruction.

Close modal

Correlation between in vivo imaging and histology

Ultrasound imaging.

To determine how VEGFR2-targeted ultrasonic imaging signal varies in different histologic stages of breast cancer progression (from normal mammary gland tissue to hyperplasia, DCIS, and invasive breast cancer), in the second part of the study, the mammary glands of transgenic mice (n = 233) and glands of age-matched control littermates (n = 82) were imaged using MBVEGFR2 as described above at different stages of breast cancer development between ages 4 and 10 weeks, and the in vivo imaging signals were correlated with ex vivo histologic findings (Fig. 1). A total of 315 mammary glands were imaged, and all mammary glands were excised immediately after imaging to allow direct correlation between in vivo imaging findings and ex vivo histology. Before isolating the mammary glands, the imaging plane (position of ultrasound transducer) was marked on the skin of the mammary gland using a tissue marking dye (Cancer Diagnostics, Inc.) to ensure that the imaged plane was correlated with about the same plane and oriented in the same way on ex vivo analysis.

Imaging data analysis.

After each imaging session, the imaging data sets of all mice were analyzed offline in random order at a dedicated workstation with commercially available software (Visualsonics). Analysis was conducted in a blinded fashion by one independent reader, blinded to the type of microbubbles (MBVEGFR2 vs. MBControl), the age of the animals and the animal type (transgenic or control). Regions of interest (ROI) were drawn over the mammary gland and the magnitude of imaging signal from attached microbubbles was assessed by calculating an average for pre- and post-destruction imaging signals and subtracting the average post-destruction signal from the average pre-destruction signal, as described previously (Fig. 1; refs. 33, 34).

Ex vivo analysis of mammary glands.

Ex vivo analyses of mammary glands for quantitative immunofluorescence (QIF) and microvessel density (MVD) were conducted using standard techniques (see Supplementary Materials).

Calculation of ultrasonic molecular imaging signal thresholds

To determine thresholds of VEGFR2-targeted ultrasound imaging signals to allow discrimination of benign (normal and hyperplasia) from malignant histologic entities (DCIS and invasive breast cancer) based on imaging, receiver operating characteristic (ROC) curves were constructed, and the sensitivity and specificity for potential imaging signal threshold levels and corresponding 95% confidence intervals (CI) were calculated using the datasets from the above described imaging/histology correlation study from 315 mammary glands. In addition, a nonparametric logistic regression model was used to predict the probability of a histologic stage (normal, hyperplasia, DCIS, or invasive breast cancer) based on ultrasound imaging. On the basis of these calculations, an imaging signal threshold level of 4.6 arbitrary units (a.u.) was selected to differentiate benign from malignant histologic entities for the subsequent prospective imaging trial (see below).

Prospective imaging trial in screening setting

In the third part of the study, the diagnostic accuracy of ultrasonic molecular imaging in differentiating the 4 different histologic tissues was assessed in a prospective screening setting (Fig. 1). For this purpose, an additional 63 mammary glands from transgenic and wild-type littermates at random ages between 4 and 10 weeks were scanned. This study design was chosen to simulate a screening scenario with “patients” with and without increased risk for breast cancer (transgenic and wild-type littermates) that present with different mammary gland pathologies at the time of screening ultrasound. The imager was blinded to the age and group of mice (transgenic vs. normal) at the time of screening ultrasound. After each imaging exam, image analysis was conducted as described above by 2 independent readers. Both readers were blinded to the age of the animals and the animal type (transgenic or control). Both readers were asked to render a diagnosis for each mammary gland (malignant vs. benign) based on the ultrasonic molecular imaging signal obtained from the mammary gland and using the imaging signal threshold level of 4.6. a.u. as calculated above. The mammary glands were excised, and ex vivo analysis of the mammary glands was conducted as described above. The diagnosis at pathology was used as reference standard to calculate diagnostic accuracy of ultrasound molecular imaging in diagnosing malignant versus benign disease in the prospective screening setting. One of the two readers conducted a repeat analysis 8 weeks after the first imaging analysis to assess intra-operator agreement.

Statistical analysis

All data are represented as mean ± SD. For details on statistical analysis, please refer to Supplementary Methods.

Clinical-grade MBVEGFR2 binds to murine VEGFR2 in transgenic mice with high specificity

As a first step, we tested binding specificity of clinical grade MBVEGFR2 to murine VEGR2 in 26 breast cancers with different sizes in transgenic mice (mean diameter, 7 mm; range, 3–12 mm). Following intravenous administration of MBVEGFR2, in vivo ultrasound imaging signal obtained from breast cancers was significantly higher (P < 0.001) than the signal obtained after MBControl administration (Fig. 2). In the same mice, binding specificity of MBVEGFR2 to VEGFR2 was further verified by in vivo blocking of the VEGFR2 receptor with anti-VEGFR2 antibody, which significantly (P < 0.001) decreased the signal by 87% on average (Fig. 2). Overall, these results confirmed in vivo binding specificity of clinical-grade MBVEGFR2 to murine VEGFR2 in transgenic mice and showed presence of increased VEGFR2 expression over a broad range of breast cancer tumor sizes.

Figure 2.

Summary of ultrasound molecular imaging signals obtained from intra-animal comparison after intravenous administration of MBVEGFR2, MBControl, and MBVEGFR2 after blocking with anti-VEGFR2 antibodies (*, P < 0.001). Representative transverse B-mode images with overlaid VEGFR2-targeted ultrasound molecular imaging signals are shown from one mammary gland harboring invasive breast cancer. Green line represents ROIs. Scale bar, 1 mm.

Figure 2.

Summary of ultrasound molecular imaging signals obtained from intra-animal comparison after intravenous administration of MBVEGFR2, MBControl, and MBVEGFR2 after blocking with anti-VEGFR2 antibodies (*, P < 0.001). Representative transverse B-mode images with overlaid VEGFR2-targeted ultrasound molecular imaging signals are shown from one mammary gland harboring invasive breast cancer. Green line represents ROIs. Scale bar, 1 mm.

Close modal

Molecular ultrasound imaging signal increases as the mouse breast tissue progresses from normal to cancer

We then assessed the in vivo ultrasound molecular imaging signal in 315 mammary glands randomly scanned at different ages between weeks 4 and 10. Figure 3 summarizes the distribution of the 4 histologic types (normal, hyperplasia, DCIS, invasive cancer) among all mammary glands. At 4 weeks of age, many mammary glands already showed histologic features of hyperplasia or DCIS; 30% of the mammary glands showed normal mammary tissue at that age. Presence of hyperplasia peaked in week 5 (29% of mammary glands). DCIS was the predominant histologic feature between weeks 4 and 8 with a peak at week 7. Invasive breast cancer was first noted at week 7 and, by week 10, 94% of mammary glands showed histologic features of invasive breast cancer.

Figure 3.

Bar graph shows percentage distribution of 4 different histologic findings (normal tissue, hyperplasia, DCIS, invasive breast cancer) in mammary glands of FVB/N-Tg(MMTV-PyMT)634Mul transgenic mouse model evaluated between age 4 and 10 weeks.

Figure 3.

Bar graph shows percentage distribution of 4 different histologic findings (normal tissue, hyperplasia, DCIS, invasive breast cancer) in mammary glands of FVB/N-Tg(MMTV-PyMT)634Mul transgenic mouse model evaluated between age 4 and 10 weeks.

Close modal

As the mammary glands progressed through the different histologic types, ultrasound molecular imaging signal significantly increased (P < 0.001) and positively correlated with histology (P < 0.001; Fig. 4A and B). Normal breast tissue showed only background imaging signal (mean, 1.65 ± 0.17 a.u.); imaging signal significantly increased in hyperplasia (mean, 4.21 ± 1.16 a.u.; P = 0.021). When breast tissue further progressed from hyperplasia to DCIS, mean ultrasound molecular imaging significantly increased to 15.95 ± 1.31 a.u. (P < 0.001). Imaging signal further significantly increased (P < 0.001) when DCIS progressed to invasive breast cancer (78.1 ± 6.31 a.u.; Fig. 4A and B). Imaging signal in breast cancer was 47.3-fold higher than normal breast tissue, 18.6-fold higher than hyperplasia, and 4.9-fold higher than DCIS.

Figure 4.

A, box plot summarizes VEGFR2-targeted ultrasound molecular imaging signals obtained in normal, hyperplasia, DCIS, and invasive breast cancer of 315 mammary glands. Imaging signal increased with mammary gland tissue progressing through different histologic stages and imaging signals significantly increased compared with preceding stage. *, P = 0.021; **, P < 0.001). Note: each box in the plot represents the 25th and 75th quartiles, the line inside each box identifies the median, and the vertical lines outside indicate the largest and smallest values in the series of measurements, excluding the outliers. The 5th and 95th percentile of the data are represented as outliers (dots). B, representative ultrasound and hematoxylin and eosin (H&E) images of mammary glands with 4 different histologic stages. Transverse B-mode with overlaid VEGFR2-targeted ultrasound molecular imaging signal (second row) shows substantial increase of ultrasound molecular imaging signal in DCIS and invasive breast cancer compared with hyperplasia and normal mammary gland tissue (scale bar, 1 mm). Note that B-mode images alone (first row) do not allow characterization of 4 different histologic types. H&E staining of the corresponding mammary glands is shown in third row (scale bar, 1 mm). Ten-fold magnified images from insets are shown in fourth row (scale bar, 0.1 mm).

Figure 4.

A, box plot summarizes VEGFR2-targeted ultrasound molecular imaging signals obtained in normal, hyperplasia, DCIS, and invasive breast cancer of 315 mammary glands. Imaging signal increased with mammary gland tissue progressing through different histologic stages and imaging signals significantly increased compared with preceding stage. *, P = 0.021; **, P < 0.001). Note: each box in the plot represents the 25th and 75th quartiles, the line inside each box identifies the median, and the vertical lines outside indicate the largest and smallest values in the series of measurements, excluding the outliers. The 5th and 95th percentile of the data are represented as outliers (dots). B, representative ultrasound and hematoxylin and eosin (H&E) images of mammary glands with 4 different histologic stages. Transverse B-mode with overlaid VEGFR2-targeted ultrasound molecular imaging signal (second row) shows substantial increase of ultrasound molecular imaging signal in DCIS and invasive breast cancer compared with hyperplasia and normal mammary gland tissue (scale bar, 1 mm). Note that B-mode images alone (first row) do not allow characterization of 4 different histologic types. H&E staining of the corresponding mammary glands is shown in third row (scale bar, 1 mm). Ten-fold magnified images from insets are shown in fourth row (scale bar, 0.1 mm).

Close modal

VEGFR2 expression increases with breast tissue progression from normal, hyperplasia, DCIS to invasive breast cancer

Following ultrasound imaging, mice were sacrificed and tumors excised for ex vivo analysis. VEGFR2 expression was significantly increased (P < 0.001) in hyperplasia, DCIS, and invasive breast cancer compared with normal breast tissue and was highest on the angiogenic vasculature of invasive breast cancer (Fig. 5). Quantitative VEGFR2-based immunofluorescence signal strongly correlated with in vivo VEGFR2-targeted ultrasonic molecular imaging (R2 = 0.84; 95% CI, 0.72–0.91; P < 0.001). In addition to higher VEGFR2 expression levels, MVD significantly (P ≤ 0.001) increased in hyperplasia, DCIS, and invasive breast cancer compared with normal tissue, suggesting that increased in vivo VEGFR2-targeted imaging signal during breast tissue progression from normal to invasive breast cancer was caused by a combination of higher target expression per vessel and an increase in the overall number of vessels expressing the target VEGFR2. Using a linear regression model, the fitted equation reflecting the association between in vivo ultrasonic imaging signal, QIF, and MVD was: ultrasound signal = 0.11 × MVD + 3.55 × QIF − 0.47. The quantitative value generated by this equation is referred to as the angiogenesis score and represents a combination of the number of vessels expressing the VEGFR2 and VEGFR2 expression per vessel. The angiogenesis score significantly increased (P ≤ 0.001) from normal to hyperplasia, DCIS, or invasive subtypes of breast cancer in this transgenic mouse model (Supplementary Fig. S1). The adjusted coefficient of correlation was 0.74 (95% CI, 0.51–0.89; P ≤ 0.001), suggesting good correlation between in vivo ultrasonic molecular imaging signal and both MVD and VEGFR2 expression levels on the vasculature.

Figure 5.

Representative photomicrographs of mouse mammary glands with 4 different histologic stages of breast cancer development. Mammary gland sections were stained for vascular endothelial cell marker CD31 (green, first column) and for VEGFR2 (red, second column). Merged images (yellow, third column) show expression of VEGFR2 on vascular endothelial cells. Note that number of microvessels and the VEGFR2 staining increased with breast cancer development (scale bar, 0.1 mm).

Figure 5.

Representative photomicrographs of mouse mammary glands with 4 different histologic stages of breast cancer development. Mammary gland sections were stained for vascular endothelial cell marker CD31 (green, first column) and for VEGFR2 (red, second column). Merged images (yellow, third column) show expression of VEGFR2 on vascular endothelial cells. Note that number of microvessels and the VEGFR2 staining increased with breast cancer development (scale bar, 0.1 mm).

Close modal

Ultrasound molecular imaging allows detection of DCIS and invasive breast cancer with high diagnostic accuracy

Figure 6 shows the ROC curve to differentiate benign histologic entities (normal and hyperplasia) from diagnosing those that need further clinical work-up (DCIS and invasive breast cancer) based on ultrasound molecular imaging signals obtained in 315 mammary glands. Overall, ultrasound allowed the differentiation of benign from malignant diseases with high diagnostic accuracy (area under the curve, 0.94). Sensitivities, specificities, and PPV and negative predictive (NPV) values to differentiate benign from malignant cases using different imaging signal threshold levels were calculated and are summarized in Supplementary Tables S1 and S2. At a threshold level of 4.6 a.u., benign and malignant histologic entities were predicted to be diagnosed with a sensitivity of 84% (95% CI, 78–88) and a specificity of 89% (95% CI, 81–94; Supplementary Table S1).

Figure 6.

ROC curve for differentiating benign histologic entities (normal and hyperplasia) from malignant entities (DCIS and invasive cancer) at molecular ultrasound imaging threshold level of 4.6 a.u. Scale bar on right represents VEGFR2-targeted ultrasound molecular imaging signal in arbitrary units.

Figure 6.

ROC curve for differentiating benign histologic entities (normal and hyperplasia) from malignant entities (DCIS and invasive cancer) at molecular ultrasound imaging threshold level of 4.6 a.u. Scale bar on right represents VEGFR2-targeted ultrasound molecular imaging signal in arbitrary units.

Close modal

Using the threshold level of 4.6 a.u., 2 readers analyzed an additional 63 mammary glands in a prospective screening setting (20 glands each with normal, DCIS, and invasive cancer; 3 hyperplastic glands). In this trial, malignant lesions were differentiated from benign lesions with a sensitivity of 100% (95% CI, 91–100) and a specificity of 87% (95% CI, 66–97) for both readers, respectively (Table 1). Both readers overcalled 3 benign findings (1 normal mammary gland and 2 glands with hyperplasia) as malignant (Table 2). Discrepancies between both readers were observed in only 2 of 63 cases (Table 2), which corresponded to an excellent interobserver agreement [intraclass correlation coefficient (ICC) = 0.98; 95% CI, 97–99]. Similarly, reproducibility in repeated analysis of ultrasound molecular imaging data sets after 8 weeks conducted by one of the readers was high (ICC = 0.96; 95% CI, 90–98).

Table 1.

Sensitivity, specificity, PPV, and NPV in diagnosing malignant disease (DCIS and invasive cancer) using an ultrasound imaging threshold level of 4.6 a.u. in prospective imaging trial in 63 mammary glands analyzed by two independent readers

Ultrasound imaging threshold level (a.u.)Sensitivity (95% CI)Specificity (95% CI)PPV (95% CI)NPV (95% CI)
4.6 100 (91–100) 87 (66–97) 93 (81–99) 100 (83–100) 
Ultrasound imaging threshold level (a.u.)Sensitivity (95% CI)Specificity (95% CI)PPV (95% CI)NPV (95% CI)
4.6 100 (91–100) 87 (66–97) 93 (81–99) 100 (83–100) 

NOTE: Values are identical for both readers as both readers misdiagnosed the same number of cases (see Table 2).

Table 2.

Summary of 4 (cases A–D) of 63 cases with discrepancies between prospective ultrasound imaging diagnosis provided by two independent readers and histologic findings

Misdiagnosed casesHistologic findingUltrasound signal measured by reader 1Diagnosis by reader 1Ultrasound signal measured by reader2Diagnosis by reader 2
Benign (normal) 4.16 Benign 4.75 Malignant 
Benign (normal) 4.96 Malignant 0.57 Benign 
Benign (hyperplasia) 8.81 Malignant 4.67 Malignant 
Benign (hyperplasia) 11.87 Malignant 8.48 Malignant 
Misdiagnosed casesHistologic findingUltrasound signal measured by reader 1Diagnosis by reader 1Ultrasound signal measured by reader2Diagnosis by reader 2
Benign (normal) 4.16 Benign 4.75 Malignant 
Benign (normal) 4.96 Malignant 0.57 Benign 
Benign (hyperplasia) 8.81 Malignant 4.67 Malignant 
Benign (hyperplasia) 11.87 Malignant 8.48 Malignant 

NOTE: Both readers independently misdiagnosed 3 of 63 cases compared with histology using the ultrasound imaging threshold value of 4.6 a.u., corresponding to an error rate of 4.8%. Two cases (C and D) were overcalled as malignant by both readers; there was discrepancy between the two readers in two cases (A and B).

Our study showed that VEGFR2-targeted ultrasound imaging signal increases when mammary glands progress from normal tissue to hyperplasia, DCIS, and invasive breast cancer in a transgenic mouse model. At an ultrasound molecular imaging signal threshold level of 4.6 a.u., DCIS and invasive breast cancer can be differentiated from normal and hyperplastic mammary glands with high sensitivity and specificity. Our results suggest that VEGFR2-targeted contrast microbubbles can increase the diagnostic accuracy of ultrasound in earlier detection of both breast cancer and its precursor lesion DCIS.

Ultrasound (7, 35, 36) and other imaging technologies such as MRI (37, 38) and positron emission tomography (PET; ref. 39) are currently being evaluated as adjunct screening modalities in patients with mammographically dense breasts, as screening mammograms can miss breast cancer in more than 50% of patients with dense breast tissue (4, 5). In fact, in several states in the United States, dense breast laws have recently been passed that mandate radiologists to specifically inform patients about the presence of dense breast tissue on mammograms, acknowledging the limited sensitivity of mammography in detecting breast cancer in these patients. However, the diagnostic accuracy of current ultrasound screening techniques in breast cancer detection is low (40).

We explored whether the diagnostic accuracy of ultrasound in earlier breast cancer detection could be improved by molecular imaging of tumor angiogenesis in a mouse model. To simulate the apparent progression of ductal breast adenocarcinoma from normal to hyperplasia, DCIS, and invasive breast cancer as observed in patients and to quantify the ultrasound molecular imaging signal associated with these 4 histologic stages, we used a transgenic mouse model that recapitulates this progression in murine mammary glands. We found that the angiogenic switch, the stage at which angiogenesis occurs in tumor progression, can already be observed to some extent in hyperplasia. Presence of angiogenesis further increased significantly when mouse mammary glands developed DCIS and invasive cancer; the highest values of MVD and VEGFR2 were observed in mammary glands with invasive breast cancer. In accordance with our findings in transgenic mice, recent immunohistochemical studies in human breast samples showed an angiogenic initiation at the hyperplasia stage with a further increase of angiogenesis with the development of DCIS and invasion (22, 41).

We then confirmed our hypothesis that in vivo ultrasound molecular imaging signal using a contrast agent targeted at tumor angiogenesis significantly increases during breast cancer development by correlating in vivo ultrasonic molecular imaging findings with ex vivo histopathologic findings in 315 mammary glands. QIF showed that both the number of angiogenic vessels and the ex vivo expression levels of VEGFR2 per vessel strongly correlated with in vivo ultrasound molecular imaging signal across the four different breast histologies. Collectively, these findings indicate that tumor angiogenesis including overexpression of angiogenic markers such as VEGFR2 may play a significant role in breast cancer progression in this mouse model and that imaging and quantification of tumor angiogenesis using VEGFR2-targeted ultrasound contrast agents could be leveraged to aid in the earlier detection of clinically relevant findings such as DCIS and invasive breast cancer by defining imaging signal threshold levels.

This was further tested in a blinded prospective screening trial by using newly defined ultrasound molecular imaging signal threshold levels. To simulate a clinical scenario with “patients” at risk to develop breast cancer, transgenic and wild-type mice were randomly selected at various ages and prospectively screened with ultrasound molecular imaging, followed by direct ex vivo correlation with the in vivo imaging results. Our results in 63 prospectively analyzed mammary glands showed that clinically actionable findings such as DCIS and invasive breast cancer can be detected with a sensitivity of up to 100%, which was higher than the predicted value of 84% obtained from the previous part of our study. An explanation of this high sensitivity is the observation that all 20 DCIS and 20 invasive breast cancer cases included in this prospective screening trial happened to show high levels of VEGFR2 and MVD and thus, the ultrasound molecular imaging signal was measured above the preselected threshold level value of 4.6 a.u.. Both readers highly agreed between each other in differentiating malignant from benign disease and the false-positive rate was low, resulting in a PPV as high as 93%. This suggests that ultrasound molecular imaging has the potential not only to increase detection rates of DCIS and breast cancer in a screening setting but also to eventually decrease the currently high call back rate and high number of unnecessary biopsies.

We acknowledge several limitations of our study. First, we acknowledge that the fast progression of mammary tissue over several weeks in the transgenic mouse model may not reflect the usually slow development of breast cancer in patients. This rapid transition may have artificially overstimulated angiogenesis and subsequently VEGFR2 expression in DCIS and invasive cancer in our study. Second, this mouse model does not develop other benign lesions beyond hyperplasia, and additional studies are needed to assess whether ultrasound allows differentiation of those lesions from DCIS and breast cancer based on the magnitude of the molecular imaging signal. Third, the prevalence of DCIS and invasive breast cancer in our mouse screening trial was higher than that in a clinical screening population; this may have overestimated PPV and NPV for ultrasound molecular imaging. Fourth, the injected microbubble dose was used according to previous experience in ultrasound molecular imaging of mice (26, 33, 42). The contrast agent dosing when scaling to future first-in-man clinical trials need to be assessed and may differ compared with the dosing used in mice. Finally, we acknowledge that with the current imaging equipment available, we scanned only a 2-dimensional imaging plane from the center of the breast lesions. Ongoing research explores volumetric ultrasound molecular imaging approaches to assess the molecular signature from the entire 3-dimensional extent of the target tissues.

In conclusion, our results suggest that ultrasound molecular imaging using a novel clinical-grade VEGFR2-targeted ultrasound contrast agent allows highly accurate and reliable detection of DCIS and breast cancer in a screening trial of transgenic mice. This study lays the foundation for further development of this promising imaging approach for earlier breast cancer detection and paves the way for translational clinical trials in the future.

K.C. Jensen has served as a consultant for Clarient/GE. F. Tranquart is an employee of Bracco. S.S. Gambhir and J.K. Willmann have served as consultants for Bracco. No potential conflicts of interest were disclosed by the other authors.

Conception and design: S.V. Bachawal, J.K. Willmann

Development of methodology: S.V. Bachawal, J.K. Willmann

Acquisition of data: S.V. Bachawal, K.C. Jensen, A.M. Lutz, J.K. Willmann

Analysis and interpretation of data: S.V. Bachawal, K.C. Jensen, A.M. Lutz, L. Tian, J.K. Willmann, MD

Writing, review and/or revision of the manuscript: All authors

Administrative, technical, or material support: F. Tranquart, S.S. Gambhir

Study supervision: J.K. Willmann

The authors thank Kira Foygel, PhD, for her assistance in staining tissue sections and the support by Timothy Doyle, PhD, in the Small Animal Imaging Facility at Stanford University.

This research was supported by the R01 CA155289-01A1 (J.K. Willmann) and by a Developmental Cancer Research Award from the Stanford Cancer Center (J.K. Willmann).

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

1.
Siegel
R
,
Naishadham
D
,
Jemal
A
. 
Cancer statistics, 2012
.
CA Cancer J Clin
2012
;
62
:
10
29
.
2.
Humphrey
LL
,
Helfand
M
,
Chan
BK
,
Woolf
SH
. 
Breast cancer screening: a summary of the evidence for the U.S. Preventive Services Task Force
.
Ann Intern Med
2002
;
137
:
347
60
.
3.
Stomper
PC
,
D'Souza
DJ
,
DiNitto
PA
,
Arredondo
MA
. 
Analysis of parenchymal density on mammograms in 1353 women 25–79 years old
.
AJR Am J Roentgenol
1996
;
167
:
1261
5
.
4.
Kolb
TM
,
Lichy
J
,
Newhouse
JH
. 
Comparison of the performance of screening mammography, physical examination, and breast US and evaluation of factors that influence them: an analysis of 27,825 patient evaluations
.
Radiology
2002
;
225
:
165
75
.
5.
Mandelson
MT
,
Oestreicher
N
,
Porter
PL
,
White
D
,
Finder
CA
,
Taplin
SH
, et al
Breast density as a predictor of mammographic detection: comparison of interval- and screen-detected cancers
.
J Natl Cancer Inst
2000
;
92
:
1081
7
.
6.
Boyd
NF
,
Guo
H
,
Martin
LJ
,
Sun
L
,
Stone
J
,
Fishell
E
, et al
Mammographic density and the risk and detection of breast cancer
.
N Engl J Med
2007
;
356
:
227
36
.
7.
Berg
WA
,
Blume
JD
,
Cormack
JB
,
Mendelson
EB
,
Lehrer
D
,
Bohm-Velez
M
, et al
Combined screening with ultrasound and mammography vs mammography alone in women at elevated risk of breast cancer
.
JAMA
2008
;
299
:
2151
63
.
8.
Buchberger
W
,
Niehoff
A
,
Obrist
P
,
DeKoekkoek-Doll
P
,
Dunser
M
. 
Clinically and mammographically occult breast lesions: detection and classification with high-resolution sonography
.
Semin Ultrasound CT MR
2000
;
21
:
325
36
.
9.
Crystal
P
,
Strano
SD
,
Shcharynski
S
,
Koretz
MJ
. 
Using sonography to screen women with mammographically dense breasts
.
AJR Am J Roentgenol
2003
;
181
:
177
82
.
10.
Gordon
PB
,
Goldenberg
SL
. 
Malignant breast masses detected only by ultrasound. A retrospective review
.
Cancer
1995
;
76
:
626
30
.
11.
Leconte
I
,
Feger
C
,
Galant
C
,
Berliere
M
,
Berg
BV
,
D'Hoore
W
, et al
Mammography and subsequent whole-breast sonography of nonpalpable breast cancers: the importance of radiologic breast density
.
AJR Am J Roentgenol
2003
;
180
:
1675
9
.
12.
Kaplan
SS
. 
Clinical utility of bilateral whole-breast US in the evaluation of women with dense breast tissue
.
Radiology
2001
;
221
:
641
9
.
13.
Weinstein
SP
,
Localio
AR
,
Conant
EF
,
Rosen
M
,
Thomas
KM
,
Schnall
MD
. 
Multimodality screening of high-risk women: a prospective cohort study
.
J Clin Oncol
2009
;
27
:
6124
8
.
14.
Pysz
MA
,
Willmann
JK
. 
Targeted contrast-enhanced ultrasound: an emerging technology in abdominal and pelvic imaging
.
Gastroenterology
2011
;
140
:
785
90
.
15.
Kiessling
F
,
Bzyl
J
,
Fokong
S
,
Siepmann
M
,
Schmitz
G
,
Palmowski
M
. 
Targeted ultrasound imaging of cancer: an emerging technology on its way to clinics
.
Curr Pharm Des
2012
;
18
:
2184
99
.
16.
Deshpande
N
,
Needles
A
,
Willmann
JK
. 
Molecular ultrasound imaging: current status and future directions
.
Clin Radiol
2010
;
65
:
567
81
.
17.
Klibanov
AL
. 
Ligand-carrying gas-filled microbubbles: ultrasound contrast agents for targeted molecular imaging
.
Bioconjug Chem
2005
;
16
:
9
17
.
18.
Hicklin
DJ
,
Ellis
LM
. 
Role of the vascular endothelial growth factor pathway in tumor growth and angiogenesis
.
J Clin Oncol
2005
;
23
:
1011
27
.
19.
Kranz
A
,
Mattfeldt
T
,
Waltenberger
J
. 
Molecular mediators of tumor angiogenesis: enhanced expression and activation of vascular endothelial growth factor receptor KDR in primary breast cancer
.
Int J Cancer
1999
;
84
:
293
8
.
20.
Nakopoulou
L
,
Stefanaki
K
,
Panayotopoulou
E
,
Giannopoulou
I
,
Athanassiadou
P
,
Gakiopoulou-Givalou
H
, et al
Expression of the vascular endothelial growth factor receptor-2/Flk-1 in breast carcinomas: correlation with proliferation
.
Hum Pathol
2002
;
33
:
863
70
.
21.
Price
DJ
,
Miralem
T
,
Jiang
S
,
Steinberg
R
,
Avraham
H
. 
Role of vascular endothelial growth factor in the stimulation of cellular invasion and signaling of breast cancer cells
.
Cell Growth Differ
2001
;
12
:
129
35
.
22.
Carpenter
PM
,
Chen
WP
,
Mendez
A
,
McLaren
CE
,
Su
MY
. 
Angiogenesis in the progression of breast ductal proliferations
.
Int J Surg Pathol
2011
;
19
:
335
41
.
23.
Hutchinson
JN
,
Muller
WJ
. 
Transgenic mouse models of human breast cancer
.
Oncogene
2000
;
19
:
6130
7
.
24.
Guy
CT
,
Cardiff
RD
,
Muller
WJ
. 
Induction of mammary tumors by expression of polyomavirus middle T oncogene: a transgenic mouse model for metastatic disease
.
Mol Cell Biol
1992
;
12
:
954
61
.
25.
Lin
EY
,
Jones
JG
,
Li
P
,
Zhu
L
,
Whitney
KD
,
Muller
WJ
, et al
Progression to malignancy in the polyoma middle T oncoprotein mouse breast cancer model provides a reliable model for human diseases
.
Am J Pathol
2003
;
163
:
2113
26
.
26.
Pysz
MA
,
Foygel
K
,
Rosenberg
J
,
Gambhir
SS
,
Schneider
M
,
Willmann
JK
. 
Antiangiogenic cancer therapy: monitoring with molecular US and a clinically translatable contrast agent (BR55)
.
Radiology
2010
;
256
:
519
27
.
27.
Tardy
I
,
Pochon
S
,
Theraulaz
M
,
Emmel
P
,
Passantino
L
,
Tranquart
F
, et al
Ultrasound molecular imaging of VEGFR2 in a rat prostate tumor model using BR55
.
Invest Radiol
2010
;
45
:
573
8
.
28.
Pillai
R
,
Marinelli
ER
,
Fan
H
,
Nanjappan
P
,
Song
B
,
von Wronski
MA
, et al
A phospholipid-PEG2000 conjugate of a vascular endothelial growth factor receptor 2 (VEGFR2)-targeting heterodimer peptide for contrast-enhanced ultrasound imaging of angiogenesis
.
Bioconjug Chem
. 
2010
Feb 19. [Epub ahead of print]
.
29.
Shrivastava
A
,
von Wronski
MA
,
Sato
AK
,
Dransfield
DT
,
Sexton
D
,
Bogdan
N
, et al
A distinct strategy to generate high-affinity peptide binders to receptor tyrosine kinases
.
Protein Eng Des Sel
2005
;
18
:
417
24
.
30.
Bzyl
J
,
Lederle
W
,
Rix
A
,
Grouls
C
,
Tardy
I
,
Pochon
S
, et al
Molecular and functional ultrasound imaging in differently aggressive breast cancer xenografts using two novel ultrasound contrast agents (BR55 and BR38)
.
Eur Radiol
2011
;
21
:
1988
95
.
31.
Deshpande
N
,
Lutz
AM
,
Ren
Y
,
Foygel
K
,
Tian
L
,
Schneider
M
, et al
Quantification and monitoring of inflammation in murine inflammatory bowel disease with targeted contrast-enhanced US
.
Radiology
2012
;
262
:
172
80
.
32.
Willmann
JK
,
Cheng
Z
,
Davis
C
,
Lutz
AM
,
Schipper
ML
,
Nielsen
CH
, et al
Targeted microbubbles for imaging tumor angiogenesis: assessment of whole-body biodistribution with dynamic micro-PET in mice
.
Radiology
2008
;
249
:
212
9
.
33.
Willmann
JK
,
Paulmurugan
R
,
Chen
K
,
Gheysens
O
,
Rodriguez-Porcel
M
,
Lutz
AM
, et al
US imaging of tumor angiogenesis with microbubbles targeted to vascular endothelial growth factor receptor type 2 in mice
.
Radiology
2008
;
246
:
508
18
.
34.
Lindner
JR
,
Song
J
,
Xu
F
,
Klibanov
AL
,
Singbartl
K
,
Ley
K
, et al
Noninvasive ultrasound imaging of inflammation using microbubbles targeted to activated leukocytes
.
Circulation
2000
;
102
:
2745
50
.
35.
Youk
JH
,
Kim
EK
. 
Supplementary screening sonography in mammographically dense breast: pros and cons
.
Korean J Radiol
2010
;
11
:
589
93
.
36.
Nothacker
M
,
Duda
V
,
Hahn
M
,
Warm
M
,
Degenhardt
F
,
Madjar
H
, et al
Early detection of breast cancer: benefits and risks of supplemental breast ultrasound in asymptomatic women with mammographically dense breast tissue. A systematic review
.
BMC Cancer
2009
; 
9
:
335
.
37.
Berg
WA
,
Zhang
Z
,
Lehrer
D
,
Jong
RA
,
Pisano
ED
,
Barr
RG
, et al
Detection of breast cancer with addition of annual screening ultrasound or a single screening MRI to mammography in women with elevated breast cancer risk
.
JAMA
2012
;
307
:
1394
404
.
38.
Saslow
D
,
Boetes
C
,
Burke
W
,
Harms
S
,
Leach
MO
,
Lehman
CD
, et al
American Cancer Society guidelines for breast screening with MRI as an adjunct to mammography
.
CA Cancer J Clin
2007
;
57
:
75
89
.
39.
Garami
Z
,
Hascsi
Z
,
Varga
J
,
Dinya
T
,
Tanyi
M
,
Garai
I
, et al
The value of 18-FDG PET/CT in early-stage breast cancer compared to traditional diagnostic modalities with an emphasis on changes in disease stage designation and treatment plan
.
Eur J Surg Oncol
2012
;
38
:
31
7
.
40.
Kuhl
CK
,
Schmutzler
RK
,
Leutner
CC
,
Kempe
A
,
Wardelmann
E
,
Hocke
A
, et al
Breast MR imaging screening in 192 women proved or suspected to be carriers of a breast cancer susceptibility gene: preliminary results
.
Radiology
2000
;
215
:
267
79
.
41.
Bluff
JE
,
Menakuru
SR
,
Cross
SS
,
Higham
SE
,
Balasubramanian
SP
,
Brown
NJ
, et al
Angiogenesis is associated with the onset of hyperplasia in human ductal breast disease
.
Br J Cancer
2009
;
101
:
666
72
.
42.
Lee
DJ
,
Lyshchik
A
,
Huamani
J
,
Hallahan
DE
,
Fleischer
AC
. 
Relationship between retention of a vascular endothelial growth factor receptor 2 (VEGFR2)-targeted ultrasonographic contrast agent and the level of VEGFR2 expression in an in vivo breast cancer model
.
J Ultrasound Med
2008
;
27
:
855
66
.