Intraoperative Resection Guidance with Photoacoustic and Fluorescence Molecular Imaging Using an Anti–B7-H3 Antibody-Indocyanine Green Dual Contrast Agent Free

Breast cancer remains the second leading cause of cancer-related deaths in women. In the United States in 2017, 255,180 new cases will be diagnosed and 41,070 patients will die of the disease (1). Breast cancer mortality is substantially decreased if detected early, as survival largely depends on the tumor stage at the time of detection (1). With more advanced screening methods, earlier stages and smaller tumors can be detected, allowing for less aggressive treatment including breast-conserving surgical resection. However, surgical resection is challenging at multiple stages during the operation including visualization of the primary tumor and its required resection margins, as well all lymph node metastasis. A complete resection with negative margins [defined as “no ink on tumor” and no involvement of ductal carcinoma in situ (DCIS; ref. 2)] optimizes patient benefit by minimizing local and distant recurrence. Without intraoperative frozen-section analysis, which greatly increases surgical time and has low sensitivity (65%–78%; ref. 3), almost one quarter of patients will undergo further surgical procedures due to positive margins or to extend disease-free margins (2). Therefore, a rapid, high resolution, and highly specific method for intraoperative margin assessment is critically needed.

Using combined photoacoustic (PA) and ultrasound imaging could increase the accuracy of intraoperative margin assessment. PA images are created when tissues undergo thermoelastic expansion after pulsed laser irradiation and subsequently emit ultrasonic waves that can be reconstructed similarly to ultrasound images, representing a spatial map of optical absorption by endogenous photoabsorbers (e.g., hemoglobin) or exogenous contrast agents. Optical absorption is wavelength dependent, and multi-wavelength imaging allows for “spectral unmixing” of the generated PA signals based from measured optical absorption spectra of hemoglobin and other photoabsorbers, known as spectroscopic PA (sPA; refs. 4–9). Therefore, information about the relative concentrations of endogenous photoabsorbers or exogenous contrast agents can be determined allowing enhanced signal to background ratios. The low optical absorption and US scattering of breast tissues makes it optimal for sPA imaging at high resolution and depth (3–5 cm deep; refs. 10–15), much further than fluorescence imaging, which has limited resolution at only a few millimeter penetration depth (16). In addition, dense breast tissue does not reduce the sensitivity of optical spectroscopy-based imaging methods (17–19). Currently, newly developed combined US/PA imaging systems are undergoing clinical testing, showing PA imaging can image sufficiently deep within tissues (8, 20–23). Therefore, sPA may be a promising method to guide intraoperative resections.

To improve imaging specificity of sPA, exogenous contrast agents targeted to specific cancer markers can be used. Recently, B7-H3 (CD276), a member of the B7 family of immunoregulators (24), has been shown to be upregulated in DCIS and breast cancer compared with normal breast tissue and benign lesions both in preclinical animal models and human tissues (7, 24). B7-H3 may, therefore, be an attractive molecular imaging target for intraoperative guidance during breast surgery.

Gold and silver plasmonic noble metal nanoparticles (25–27) are common preclinical PA contrast agents due to their excellent optical absorption properties. However, despite extensive preclinical evaluations, clinical advancement of those nanoparticles has been limited due to potential long-term toxicity concerns with retention in the mononuclear phagocyte system following systemic injection (28). Indocyanine green (ICG) is an FDA-approved and widely clinically used dye that shows excellent near-infrared optical absorption and fluorescence with distinct spectral peaks around 800 nm and is rapidly cleared from the blood circulation (29–31). ICG is in clinical use as an intravascular contrast agent, for example, with ophthalmic angiography (32), as it readily binds to serum albumin. However, when ICG is coupled to targeting ligands such as antibodies, it no longer binds to albumin and can be used as a molecular imaging agent for cancer imaging (33, 34). In our previous work (7), we coupled ICG to a B7-H3–targeted antibody and confirmed this change in pharmacokinetics of ligand-coupled ICG with strong accumulation in murine breast cancer. We showed that ICG maintains both fluorescent and optically absorbing abilities with photobleaching of little concern due to the rapid (single pulse) image acquisition time. Furthermore, we have shown that antibody-coupled ICG (B7-H3-ICG), when binding to B7-H3 on breast cancer cells, becomes internalized into cells and released from the antibody, resulting in a distinct change in the ICG optical absorption spectrum. The detectable shift in absorption spectrum can be leveraged to differentiate imaging signal from molecularly bound and freely circulating contrast agent (7).

Fluorescence imaging has been broadly studied for guiding intraoperative tumor removal (35–37). Fluorescence imaging provides a highly sensitive imaging method with a large field of view. Combined sPA and fluorescence molecular imaging of an antibody-ICG (ICG is both fluorescent and optically absorptive) agent could provide complementary information by providing the enhanced sensitivity and field-of-view of fluorescence with the specificity, resolution, and imaging depth of photoacoustics imaging. Together, these methods could aid in intraoperative tumor margin assessment and resection, reducing the occurrence of positive margins, thereby improving patient outcomes.

The purpose of this study was to assess the sensitivity and specificity of photoacoustic and fluorescence molecular imaging of B7-H3-ICG in a transgenic murine model of breast cancer development to guide surgical resection of small foci of disease.

All animal experiments were approved by the Institutional Administrative Panel on Laboratory Animal Care (APLAC). To test the ability of sPA and fluorescence molecular imaging combined with a B7-H3-ICG contrast agent to guide surgical resection, a well-characterized transgenic mouse model of breast cancer development (FVB/N-Tg(MMTVPyMT)634Mul) was used. The MMTV-PyMT model has 10 mammary glands that independently and spontaneously progress to breast cancer through distinct pathologic stages that closely recapitulate human disease in an age-dependent manner: normal breast tissue, hyperplasia, DCIS, and invasive breast cancer (36–38). First, B7-H3 expression was verified in the progressing disease stages of murine mammary tissue with immunofluorescent staining to determine whether B7-H3 is an appropriate target for identifying small foci of early-stage disease with high specificity. Next, animals (n = 10) between the ages of 5 to 7 weeks, which have mammary glands that range in pathologic disease stage from normal to small (<5 mm) foci of early invasive cancers, were injected intravenously via the tail vein with 33 μg (100 μL) of B7-H3-ICG contrast agent, 3 to 5 days prior to surgical resection of mammary glands to allow for maximum accumulation of agent at the target site while minimizing freely circulating agent. On the day of surgery, mice were humanely euthanized immediately before imaging and surgery to comply with APLAC requirements and underwent sequential surgical resection and imaging of pieces of the mammary glands to simulate step-wise intraoperative resection with image guidance. Fluorescence and sPA molecular images were acquired both before and after each sequential removal of quarters (n = 80) of the combined 4th and 5th mammary glands, which corresponded to a 7 to 10 mm long and 1 to 2 mm wide piece of tissue, large enough for piece-wise surgical resection (Fig. 1). Resected tissues were immediately fixed in 4% paraformaldehyde for subsequent histopathologic assessment. Finally, sPA and fluorescence molecular images were analyzed and compared with the results of the histopathologic analysis.

B7-H3 expression in the four histologic stages of disease in the MMTV mice was assessed through immunofluorescence staining and confocal microscopy. Five to 10-week-old MMTV mice (n = 3) were humanely euthanized, and 10 mammary glands each (total of 30 glands) were excised and immediately frozen in optimal cutting temperature (OCT) compound on dry ice. OCT tissue blocks were sectioned in 10-μm slices onto glass slides. OCT was rinsed in PBS for 5 minutes. Next, tissues were permeabilized for 15 minutes in 0.5% Triton X-100 in PBS. Tissues were blocked with 5% BSA, 5% goat, and 5% donkey serum in PBS for 1 hour at room temperature. Primary antibodies [rabbit anti-mouse B7-H3 and rat anti-mouse CD31 antibodies (Abcam Inc.)] were applied at 1:50 and 1:100 dilutions, respectively, and left to incubate overnight at 4°C. Secondary antibodies [AlexaFluor-488 conjugated goat anti-rabbit and AlexaFluor-546 anti-rat (Invitrogen)] were applied at 1:300 dilution, covered, and incubated for 1 hour at room temperature. Confocal microscopy was used to collect fluorescent images (LSM 510 Meta confocal microscope, Carl Zeiss) at a magnification of ×200.

Analysis of B7-H3 expression in various pathologic stages that were previously confirmed through H&E staining (described later in the histopathology methods), was performed as follows: To provide a definable region of interest (ROI), B7-H3 expression on vessels, colocalized with CD31 staining, was measured for quantitative analysis. Random fields of view within each immunostained section were imaged by using a 20× objective on an LSM510 metaconfocal microscope (Zeiss). CD31-associated B7-H3 immunofluorescent intensities were quantified by using ImageJ software. CD31 staining was used to identify blood vessels, and ROIs were automatically drawn around each vessel in the field by first converting the images to 8 bit, then applying a threshold to create a binary image of the vessels and using that to define the ROIs. The coordinates of each ROI were copied and applied to the B7-H3 channel, where the intensity was then quantified by measuring the mean fluorescent intensity (in arbitrary units) of B7-H3 within each ROI. The associated value used for quantitative immunofluorescence comparisons was calculated as the mean of the B7-H3 mean fluorescent intensity for all blood vessels within the images.

Preparation and characterization of the B7-H3-ICG agent was performed as described previously (8). In brief, NHS ester labeling of the amine groups on Lysine peptides was used to conjugate ICG to the anti–B7-H3 antibodies. After anti–B7-H3 mAbs [Abcam Inc., (EPNCIR122; ab134161)] were purified using a Protein A Purification Kit (Abcam Inc.), 100 μg of antibody per batch was incubated with a 20 molar equivalent of succinimidyl ester modified ICG (ICG-NHS, Intrace Medical, Co.) in 200 μL 0.1 mol/L PBS buffer (pH 7.4). The reaction was carried out at room temperature for 2 hours protected from light. B7-H3-ICG was purified using a Zeba Spin Desalting Column with a 7K molecular weight cutoff (Thermo Fisher Scientific, Co.) and concentrated to 0.33 mg/mL of PBS using 30,000 Da molecular weight cut-off filters centrifuged at 2,500 × g for 5 minutes. B7-H3-ICG concentration was determined through spectrophotometric analysis at 280 and 800 nm with correction factors (8).

sPA and fluorescence molecular images were acquired both before and after each subsequent resection of quarters (n = 80) of the mammary tissues corresponding to the combined 4th and 5th glands (n = 20) from MMTV-PyMT mice (n = 10). This approach was meant to mimic intraoperative image-guided surgical resection with direct comparison of imaging signal with the histologic findings of resected pieces of mammary tissue. For this purpose, the surgical field was opened and mice were placed into a warmed saline water bath for ultrasonic coupling. The VevoLAZR (FUJIFILM VisualSonics) with a 21 MHz transducer (lateral and axial resolution of 165 and 74 μm, respectively) and an average 10 mJ/cm² average fluence laser pulse (10 ns pulse width, 20 Hz pulse repetition frequency, that has been optimized and calibrated) was used to obtain single-plane, multiwavelength (680–900 nm, 10 nm increments, persistence of 4) photoacoustic images with wavelength-dependent fluence compensation. Each US/sPA imaging cross section required approximately 5 seconds to obtain. Coregistered B-mode ultrasound images were also acquired to provide anatomic registration and gland localization. In the same imaging session, fluorescence images were collected both before and after subsequent tissue resections with the IVIS Spectrum Preclinical Imaging System (PerkinElmer) in epifluorescence mode equipped with 710/30 nm (ex) and 820/20 nm filters (em). After each individual resected piece of mammary gland, B-mode ultrasound images were used to realign mammary glands to maintain similar imaging planes of the mammary glands as accurately as possible for the repetitive imaging sessions after each resection step. Each complete resection with imaging took approximately 15 minutes to complete. Imaging parameters were maintained through various imaging sessions.

After surgical resection and image collection, data were processed to determine the photoacoustic molecular imaging signal (in arbitrary units; a.u.) for the resected tissue portions. The B7-H3–specific molecular signal was computed using a linear least squares error regression-based classification method to compare collected signal (proportional to absorption) in spectroscopic PA images to known optical absorption spectra of the molecularly bound B7-H3-ICG agent. This method has been described in detail previously (5–8). Briefly, collected B-mode images coregistered with the PA images were used to determine ROI selection of to be resected tissue pieces. These ROI maps were applied to the spectroscopically resolved PA molecular images, and molecular signal was averaged over the ROI within resected tissue. Molecular imaging signal results in a qualitative comparison of molecularly bound B7-H3 with a.u. For all tissues, fluorescence images were analyzed using the Living Image Software 4.0 (PerkinElmer) and a circular ROI with diameter of 4 mm was placed over each tissue portion to be removed guided by coregistered photographs and the total radiant efficiency (p/s)/(μW/cm²) was measured.

After surgical excision, mammary sections were formalin-fixed and embedded in paraffin for sectioning. Tissues were sectioned with 10 μm thickness and stained with hematoxylin and eosin (H&E) via standard protocol. Histologic analysis was evaluated in random order by one reader experienced with pathologic analysis of the breast tissues from the MMTV-PyMT model and who was blinded to the imaging results. Sections were scored histologically using the following definitions: 1, normal breast tissue (primarily fatty tissues with well-organized, ductal epithelial cells); 2, hyperplasia (still predominantly fatty tissue with an increased volume of organized ductal epithelium); 3, DCIS (foci of complete expansion of ductal proliferation in acinar cell clusters); and 4, invasive carcinoma (proliferation of epithelial cells into foci of solid mass with loss of structural organization) (36, 37). Furthermore, the sections were also scored 1–4 based on the percent area of the tissue containing the assigned disease stage with 1, <10% of gland, 2, ≥10% but less than 50%, 3, ≥50% but less than 75%, and 4, ≥75%. A composite histology score was determined by multiplying the histology score by the area covered score, giving a score range from 1 to 16 (7, 24). Figure 1 summarizes the surgery, imaging, and analysis methods used.

Statistical analysis was performed in GraphPad Prism 7. Computed molecular PA and fluorescence imaging signals were organized into boxplots following Tukey rules. The continuous measures were summarized by mean and SD. After histologic analysis, tissue signals were grouped by histology composite scores into normal and hyperplasia (scores 1–2), focal DCIS (score 3), and all DCIS and invasive carcinomas (scores 3–16). Group comparisons were performed with unpaired, two-tailed t tests with statistical significance if P < 0.05. Means were reported ±SEM. ROC curves were plotted for imaging signals belonging to disease free tissue versus tissues containing positive margins and AUC and associated 95% confidence intervals (CI) were calculated.

First, the expression of the B7-H3 marker in progressive disease stages that develop within the MMTV-PyMT mouse model of breast cancer development used in this study was assessed. Mammary glands pathologically assessed as normal or hyperplasia (n = 7) were compared with tissues classified as DCIS (n = 7) and invasive carcinoma (n = 7) and quantitative immunofluorescence was used to assess the expression levels in each disease stage. The B7-H3 protein was found to be expressed on the vascular endothelium and tumor epithelium as shown in Fig. 2. Normal/hyperplasia tissues (clinically nonactionable) showed a mean fluorescence intensity of 1.3 ± 0.8 a.u. This was significantly lower (P < 0.005) than in DCIS (46.0 ± 4.8 a.u.) and invasive carcinoma (91.7 ± 21.4 a.u.).

To assess the feasibility of using sPA and fluorescence molecular imaging of B7-H3-ICG to guide intraoperative tumor resection, a total of 10 mice (5–7 weeks of age) underwent consecutive breast tissue resections after injection of B7-H3-ICG 3 to 5 days prior as depicted in Fig. 1. In all animals, intraoperative image guidance with step-wise resection was successful. Figures 3 to 5 show representative examples of consecutive resections of a mammary gland with and without breast cancer at the resection margin. Figure 4 also shows a disease-free intramammary lymph node in a gland with breast cancer confirmed by both imaging and histology.

Overall, in 17 resected tissue pieces, a composite histology score between 1 and 2 (normal or hyperplasia) was given with an average sPA imaging signal of 3.17 ± 0.48 a.u. and an average fluorescence imaging signal of 6.83E07 ± 2.00E06 (p/s)/(μW/cm²). The remaining tissue pieces (n = 63) were assigned composite histology scores between 3 and 16 (DCIS and invasive cancer) with an average sPA molecular imaging signal of 23.98 ± 4.88 a.u. and an average fluorescence imaging signal of 7.56E07 ± 1.44E06 (p/s)/(μW/cm²). Both imaging modalities showed a statically significant difference (P values < 0.05) in their signal intensity mean values between resected sections containing normal or hyperplastic mammary tissue and those with the disease states DCIS or invasive carcinoma. Tissues sections assigned a histology composite score of 3 (n = 9), which represent small (< 1 mm) foci of DCIS, also showed a statistically significant (P < 0.05) difference in imaging signal mean values compared with clinically nonactionable tissues (scores 1–2) with and average sPA molecular imaging signal of 7.10 ± 0.87 a.u. and an average fluorescence signal of 7.85E07 ± 4.27E06 (p/s)/(μW/cm²). The quantified imaging signals are represented in the boxplots in Fig. 6A and C.

Next, ROC curves were plotted to assess diagnostic accuracy of sPA and fluorescence molecular imaging at various imaging signal thresholds to differentiate between disease free (scores of 1 and 2) and positive resection margins (scores 3–16). For molecular sPA imaging, the AUC was 0.93 (95% CI, 0.87–0.99). For fluorescence imaging, the AUC was calculated to be 0.71 (95% CI, 0.57–0.85).

The purpose of this study was 2-fold. First, we assessed whether the recently described specific breast cancer marker B7-H3 is present in both DCIS and invasive breast cancer in mice to allow demarcation with intraoperative B7-H3–targeted molecular imaging methods. Second, the feasibility of using complementary sPA and fluorescent molecular imaging in combination with a dual-modality contrast agent, B7-H3-ICG, was assessed in an intraoperative-like scenario with consecutive imaging-guided resections of remaining margins.

With improved screening capabilities provided by mammography, ultrasound, and MRI, increased numbers of earlier stage, local breast cancer disease are detected. Surgical resection remains the primary treatment for breast cancer, although without intraoperative guidance, almost one quarter of patients undergo additional excision surgery due to positive margins (2). Intraoperative ultrasound for intraoperative guidance has been shown to reduce the rate of positive margins, thereby improving morbidity and mortality, decreasing resected surgical volumes, and improving patient satisfaction with improved postoperative cosmetic appearance (38, 39). However, ultrasound as a screening methodology has limited positive predictive value and cannot differentiate between benign and malignant lesions (40, 41). Therefore, an imaging modality with an increased specificity and molecular capabilities would ultimately be more useful for intraoperative margin analysis (6, 7).

Previously, B7-H3 has been shown to be upregulated in human breast cancer and DCIS tissues, and not in normal and benign lesions, with AUC values ranging between 0.90 and 0.96 in differentiating breast cancer versus noncancer (7, 24). In the current study, as a first step, expression of B7-H3 was assessed on various histologic stages of breast cancer development both on the neovasculature and extravascular compartment in a transgenic mouse model. We found little to no B7-H3 staining in normal and hyperplastic glands, with increasing expression in DCIS and invasive carcinomas. Even less than 1 mm foci of DCIS showed substantially higher B7-H3 expression compared with normal and hyperplastic mammary tissue in our study. These results suggested that B7-H3 may be promising as a molecular imaging target for intraoperative margin detection for surgical guidance. For quantitative analysis, the little B7-H3 signal in normal and hyperplastic glands prevented the definition of ROIs to calculate total (endothelial and epithelial) expression of B7-H3. Therefore, blood vessels, highlighted by their CD31 expression, were used to define ROIs for B7-H3 expression. This allowed quantitative analysis of B7-H3 expression levels on endothelial cells. Qualitatively however, additional epithelial expression can be observed in DCIS and invasive carcinomas. These results suggest that B7-H3 is a promising marker for detecting early-stage disease (DCIS) within tumor margins and is an ample vascular and epithelial target for the B7-H3-ICG agent to bind.

As a next step, we assessed whether molecular imaging of B7-H3 using sPA imaging allows intraoperative guidance of resection margins, using histology as the reference standard. To simulate imaging-guided consecutive resections in a small animal model, the mammary glands were divided into four parts and imaging with subsequent resections and histologic analyses were performed. This approach allowed direct comparison between the preresection molecular imaging signal with the histologic diagnosis within the resected tissue part. Using this approach, B7-H3–targeted sPA performed well as an intraoperative technique and was able to distinguish between normal and hyperplastic tissue versus DCIS and invasive cancer with an AUC of 0.93. A similar AUC value of 0.90 to differentiate actionable from nonactionable breast lesions has been reported previously for a different imaging technique, transcutaneous ultrasound molecular imaging using B7-H3–targeted contrast microbubbles in the same transgenic mouse model (24). This further confirms that B7-H3 is a promising imaging marker for breast lesion characterization that may be used for different molecular imaging techniques.

Furthermore, sPA imaging had sufficiently high resolution and specificity to distinguish submillimeter foci of disease, including DCIS, within a background of normal tissues highlighting them for resection in our study. Using sPA may help pinpoint disease with high accuracy reducing the volume needed for complete resection. An added benefit of sPA imaging is the depth at which images can be obtained, although in this murine study, less than a centimeter of imaging depth was needed to aid in localization of diseased mammary tissues, PA imaging has been demonstrated successful in tissues at up to 5 cm deep. Finally, the calculated sPA molecular imaging signal presented here was an average of the signal over the entire ROI drawn over the resected tissue pieces. Given the very focal nature of early-stage disease surrounded by normal breast tissue in some of the glands in the transgenic mice, the ROI signal averaging actually decreased the magnitude of molecular signal within the resected tissue parts potentially minimizing the molecular signal enhancement.

Compared with sPA, fluorescence molecular imaging of the B7-H3-ICG contrast agent underperformed in differentiating normal and hyperplasia from DCIS and breast cancer in our study. In general, fluorescence imaging techniques typically have higher sensitivity to agents due to minimal background signal than photoacoustic imaging that may be confounded by high background signal from endogenous photoabsorbers. However, the sPA molecular imaging technique employed here reduced background signal from not only endogenous chromophores, but also B7-H3-ICG contrast agent that has not bound to its molecular target. It has been shown previously that sPA can detect the shifts in the optical absorption spectrum that ICG undergoes when the agent is endocytosed and degraded within lysosomal compartments after binding to the B7-H3 marker and subsequent endocytosis (7), whereas fluorescence imaging did not allow differentiation between contrast agent that had bound to its molecular target and freely circulating or passively accumulated agent, as sPA molecular imaging can, contributing to a sharp decrease in specificity of the modality, only 17.6% with a 95% sensitivity level in our current study. Fluorescence imaging also lacks the high resolution and imaging depth of sPA imaging. As shown in Fig. 4, fluorescence imaging was unable to differentiate a disease negative lymph node from surrounding diseased tissues, whereas sPA imaging allowed identification of the node. However, fluorescence imaging has a significantly larger field of view than the sPA photoacoustic imaging allowing for visualization of potential distant disease, which is also critical in a surgical setting. A complementary use of both modalities during surgery may be advantageous where fluorescence imaging would identify superficial and distant disease and then sPA imaging would be used to more closely examine the suspected disease at high resolution and depth at the surgical margin. Furthermore, both modalities are operated in real time and are cost-effective. Combined, sPA and fluorescence imaging could be used as a complementary pair of modalities that provide high sensitivity, high specificity, high resolution, increased imaging depth, and a large field of view while using the same molecularly targeted contrast agent to guide intraoperative tumor resection.

We acknowledge several limitations of this study. First, as the animals had to be moved between the two imaging modalities as well as the surgical station, exact realignments of the sPA imaging planes for consecutive imaging exams was impossible, although by using anatomic landmarks on B-mode ultrasound imaging, every effort was made to realign the tissues imaging planes as closely as possible. In addition, the moving of living animals during surgery was not approved by the institutional APLAC; therefore, surgeries and imaging occurred on freshly euthanized animals. Second, two-dimensional imaging and histology do not allow full assessment of the disease state within the gland given the multifocal nature of the disease in this murine model. To account for the possibility of multiple pathologies, several histology sections from throughout the resected tissues were examined to ensure the most representative sections were chosen for histology scoring. Third, given the small dimensions of the mammary glands in mice and that the resected tissue pieces contained small foci of disease in a primarily normal, fatty tissue, the imaging techniques could not be evaluated for real-time margin resection of a primary tumor. Instead, we chose an experimental protocol, which allowed close correlation between small pieces of resected tissues and histology. This protocol allowed for eventual removal of all disease while highlighting the imaging modalities' abilities to detect small foci of residual disease closer in comparison with inspecting margins after primary tumor removal. Technical challenges remain for sPA molecular imaging to become a clinical reality. First, clinically designed photoacoustic imaging equipment is just now becoming available on the market. Furthermore, clinical imaging requires greater depths of imaging, which may be more difficult for PA imaging. However, clinical studies have demonstrated PA imaging at ever increasing depths, and the intraoperative application provides an optimal imaging due to reduced depth requirements and removal of interference from melanin in the skin. Finally, the use of novel contrast agents, such as B7-H3-ICG, despite their biocompatibility, requires extensive regulatory oversight. Promisingly though, antibody–dye conjugates have already made it to clinic (37), laying the foundation for this translation.

In conclusion, the results of our study suggest that intraoperative sPA B7-H3–targeted molecular imaging allows assessment of disease status in a transgenic mouse model of breast cancer development. In the future, sPA can be further developed to help guide surgeons during tumor resections minimizing positive resection margins.

No potential conflicts of interest were disclosed.

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

Development of methodology: K.E. Wilson, S.V. Bachawal

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): K.E. Wilson, S.V. Bachawal

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): K.E. Wilson, S.V. Bachawal

Writing, review, and/or revision of the manuscript: K.E. Wilson, S.V. Bachawal, J.K. Willmann

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): K.E. Wilson

Study supervision: K.E. Wilson, J.K. Willmann

The authors thank Lu Tian, PhD, for verification of the statistical analysis. This study was supported by NIH grants R01 CA155289 grant (to J. Willmann), R21EB022214 (to J. Willmann and K. Wilson), K99EB023279 (to K. Wilson), and the Stanford Molecular Imaging Young Investigator Award (to K. Wilson). The Vevo LAZR was upgraded under grant NIH 1-S10-OD01034401A1. Stanford Neuroscience Microscopy Service is supported by NIHNS069375.

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.