Development of a MUC16-Targeted Near-Infrared Fluorescent Antibody Conjugate for Intraoperative Imaging of Pancreatic Cancer

Pancreatic ductal adenocarcinoma (PDAC) remains a highly lethal and aggressive disease, characterized by its poor response to therapy, and advanced stage of disease at diagnosis. Surgical resection is currently the only curative option for patients with this disease (1, 2). However, difficulties in differentiating between cancerous and noncancerous tissue result in high rates of R1 resections, or resections with positive margin involvement, and thus high rates of recurrence (3, 4). Pathologic determination of resection status is defined by a minimum of a 1 mm tumor-cell free margin between the closest cancer cell and all resection margins to achieve a complete resection, or an R0 resection (5). Patients who undergo a complete R0 resection have improved survival compared with R1 resections, which exemplifies the importance of comprehensive intraoperative tumor detection and resection (6). Although preoperative imaging modalities play a critical role in initial staging, detection, and surgical planning, translating these images into an intraoperative setting can be challenging, leading to missed lesions (7). Thus, the 5-year survival rate for patients that undergo surgical resection is only approximately 25% (1). Fluorescence-guided surgery (FGS) has emerged as a method to reduce irradical resections, and improve intraoperative assessment of lesions (8). FGS relies on the preferential accumulation of a contrast agent in tumors, to differentiate malignant tissue from adjacent noncancerous tissue with fluorescence. The implementation of FGS has the potential to improve outcomes of pancreatic cancer resections by improving the detection of the primary tumor, surgical margins, and residual disease (9). Although FGS has clear potential for improving surgery in resectable cases, image guidance could also be employed to prevent unnecessary surgeries in unresectable cases, by highlighting metastatic lesions that have been missed in initial staging.

Near-infrared (NIR) dyes are preferentially employed for FGS because they exhibit low autofluorescence, reduced light scattering, and improved depth of penetration (10, 11). These properties are critical, particularly in malignancies like pancreatic cancer, where the tumor is often deeply seated in the abdominal cavity. FGS systems required to visualize fluorophores intraoperatively have been extensively reviewed here (12). A wide variety of targeting vehicles have been developed preclinically and in clinical trials to direct contrast agents to specific tumor targets, or tumor phenotypes, resulting in improved contrast agent specificity (13). Antibodies have been frequently used, particularly in clinical trials, because the high stability and target specificity, potentially low toxicity, and well-established conjugation strategies are favorable for probe development (14, 15).

Current clinical trials in pancreatic cancer are investigating fluorescent antibody conjugates that target EGFR, VEGF-A, and carcinoembryonic antigen (CEA) for FGS (16, 17). These initial clinical trials have demonstrated feasibility in delineating primary and metastatic disease. However, these studies also suggest the need to identify and expand alternative targeted probes to address the heterogeneity of biomarkers present in pancreatic tumors and improve intraoperative imaging (17). To address this problem, we investigated a new potential target for FGS in pancreatic cancer.

MUC16, also known as CA125, is overexpressed in 60% to 80% of pancreatic cancer (18). MUC16 is overexpressed in many different cancers, including colon, stomach, and esophageal, but has been most widely studied for its aberrant expression in ovarian cancer (18, 19). Recent studies in pancreatic cancer have demonstrated that MUC16 expression is positively correlated with disease progression and poor prognosis (20, 21). Although MUC16 has shown promise as a prognostic and diagnostic biomarker for pancreatic cancer, it has not yet been investigated for FGS. Thus, the goal of this work was to develop a MUC16-targeted antibody probe that could achieve optimal delivery to identify pancreatic cancer intraoperatively with FGS. To achieve this, we utilized AR9.6, a mAb that targets isoforms of MUC16 (either fully glycosylated or aberrantly glycosylated). We hypothesized that targeting MUC16 with the AR9.6 fluorescent antibody probe would improve the detection of pancreatic cancer and provide strong contrast against surrounding tissue. In this study, we employ the antibody AR9.6, conjugated to the near infrared dye, IRDye800 to target pancreatic cancer. Our findings demonstrated strong fluorescence enhancement of pancreatic tumor xenografts with the AR9.6-IRDye800 probe and suggested the potential utility of this fluorescent probe for improved intraoperative imaging of pancreatic cancer.

Mouse AR9.6 (Quest PharmaTech, Inc.) and mouse IgG1 isotype (Thermo Fisher Scientific, 02-6100) antibodies were reconstituted to 1 mg/mL in PBS. A Zeba spin desalting column (Thermo Fisher Scientific, 89891) was used to remove 0.1% sodium azide from the IgG1 antibody. One hundred microliters of 1M potassium phosphate was added to each milligram of antibody to raise the pH to 8.5. IRDye800 N-hydroxysuccinimide (NHS) ester (0.5 mg, LI-COR Biosciences, 929-70020) was dissolved in 50 μL of nanopure water, and vortexed briefly. Ten microliters of the dye was added to 1 mg of antibody, and vortexed briefly to mix. Dye and antibody were incubated for 2 hours at room temperature. Zeba spin desalting columns were used to remove excess dye, according to the instructions from the manufacturer. A 1 cm cuvette (Eppendorf, E0030106300) was loaded with a 1:10 dilution of the antibody in a 1:1 ratio of PBS and methanol. A Thermo Scientific Evolution 220 UV-visible spectrophotometer was used to determine dye: protein ratio (D/P), such that |${{D} \over {P}}=\Big[{{A_{780}}\over {\epsiv_{Dye}}}}\Big]\div \big[{{A_{280}-(0.03\times A_{780})}\over {\epsiv_{protein}}}\big]$|⁠, where ϵ_Dye = 270,000 M⁻¹ cm⁻¹ and ϵ_protein = 203,000 M⁻¹ cm⁻¹. The concentration of the antibody after conjugation was determined such that |${{mg} \over {ml}}=\big[{{A_{280}-(0.03\times A_{780})}\over {\epsiv_{protein}}}\big]$|× MW protein × Dilution factor. Conjugation resulted in an average of 3 dyes per antibody. The emission of the antibody conjugates was determined using a FluoroMax 4 spectrofluorometer (Horiba Scientific).

Pancreatic cancer cell lines acquired 2/2018 including HPNE (CVCL_C466), Colo357 (CVCL_0221), T3M4 (CVCL_4056), Capan1 (CVCL_0237), CFPAC (CVCL_1119), and HPAC (CVCL_3517) were obtained from Dr. Michael A. Hollingsworth, and OVCAR3 (CVCL_0465) cells acquired March 2018 were obtained from Dr. Adam R. Karpf. Cells were last tested for mycoplasma on March 21, 2018, using a PCR Mycoplasma Detection Kit (ABM, G-238). All cells were grown in RPMI1640 (Corning, 10-040-CV), supplemented with 10% FBS, 100 I.U. penicillin, and 100 μg/mL of streptomycin (P/S; Corning, 30-002-CI). Cells were maintained at 37°C in a humidified incubator with 5% CO₂. In general, all cells were passaged fewer than 20 times, or 1 month after thawing before experimental use.

Cells were lysed with RIPA buffer (Thermo Fisher Scientific, 89900) supplemented with Halt protease inhibitor (Thermo Fisher Scientific, 78440). Twenty-five micrograms of protein were separated on a 4% to 15% polyacrylamide gel (Bio-Rad, 4568084) and transferred onto a nitrocellulose membrane (Bio-Rad, 1620115). The membrane was blocked with 5% Blotting-Grade Blocker (Bio-Rad, 170-6404) in TBST and incubated with AR9.6 (1:1,000, 1 mg/mL stock solution) and GAPDH antibody (1:2,000; Cell Signaling Technology, 2118S) overnight at 4°C. The membrane was incubated with goat anti-mouse HRP secondary antibody (1:2,000; Jackson Immunoresearch, 115-035-003), or goat anti-rabbit HRP secondary antibody (1:2,000; Jackson Immunoresearch, 111-035-003) in 5% Blotting-Grade Blocker for 1 hour on a rocker and was visualized with enhanced chemiluminescent (ECL) substrate (Bio-Rad, 1705060S).

For fluorescent Western blotting, the membrane was blocked with 5% blotting grade blocking buffer in TBS. AR9.6-IRDye800 (1:5,000 1 mg/mL stock solution) and IgG-IRDye800 (1:2500 0.5 mg/mL stock solution) were incubated with the membrane overnight at 4°C in 5% blotting grade blocker in TBST. After washing, the membrane was incubated with goat anti-mouse IRDye680 (1:20,000; LI-COR Biosciences, 926-68070) in 5% blotting grade blocker in TBST, washed, and imaged on Odyssey CLx (LI-COR Biosciences).

HPNE, Colo357, T3M4, and OVCAR3 cells were seeded at 30,000 to 40,000 cells per chamber of an eight chamber slide (Thermo Fisher Scientific, 154534), and allowed to adhere overnight. Cells were washed 3× with 1× PBS and fixed in ice-cold methanol at −20°C for 20 minutes. After fixation, cells were washed with 1× PBS and blocked with 3% BSA in TBS for 1 hour at room temperature. 5 μg/mL of AR9.6 and 5 μg/mL of IgG1 were incubated with cells for 1 hour at room temperature in 3% BSA in TBST. Cells were washed 3× with PBS. Fluoroshield mounting medium with 4′,6-diamidino-2-phenylindole (DAPI; Abcam, ab104139) was added to cover each chamber, and chambers were covered with a glass coverslip. Cells were imaged at 400× magnification on an Olympus DP80 Digital Camera and cellSens Dimension software.

Colo357 cells were seeded as indicated above. Cells were incubated with AR9.6-IRDye800 for 1 hour at 37°C, followed by incubation with 60 nmol/L of LysoTracker Deep Red (Thermo Fisher Scientific, L12492) for 1 hour, and Hoescht nuclear stain (Thermo Fisher Scientific, 62249) for 5 minutes. Cells were imaged as described above.

All animal work was performed under a protocol approved by the University of Nebraska Medical Center (UNMC) Institution of Animal Care and Use Committee (IACUC). Subcutaneous tumor models were generated by injecting 1 × 10⁶ T3M4 or Colo357 cells suspended in 100 μL of 1:1 growth media and Matrigel (Corning, 356234) into the left flank of 6- to 8-week-old female NU/J mice (Jackson Laboratories, 002019). T3M4 tumors were allowed to grow for 11 days, and Colo357 tumors were allowed to grow for 30 days. Orthotopic tumor models were established by injecting 2.5 × 10⁵ T3M4 cells suspended in 30 μL of PBS into the pancreas of 6- to 8-week-old female NU/J mice as described previously (22). Surgery was conducted on a sterile drape, and a TP700 TPump Professional Core Warming and Cooling System (Stryker Corp.) maintained at 37°C. The surgical field was prepped by alternating swabs of povidone iodine and alcohol swabs. Mice were allowed to recover and were monitored after surgery. Buprenorphine (Bupranex, Reckitt Benckiser Pharmaceuticals, 955531; 0.1 mg/kg) was administered twice per day for three days after surgery.

Two nmol of IRDye800 (free dye control; LI-COR Biosciences, 929-08972), IgG-IRDye800 isotype control or AR9.6-IRDye800 were injected via tail vein into female NU/J mice bearing subcutaneous tumors. Images were taken at 4, 24, 48, 72, 96, 120, and 144 hours postinjection with the 800 nm channel and white channel on the Pearl Trilogy small animal imaging system (LI-COR Biosciences). Full necropsies were performed at 144 hours. Digestive organs were perfused with PBS, and all organs were imaged on the Pearl Trilogy. Images were analyzed using Image Studio software version 5.0 (LI-COR Biosciences). Briefly, a region of interest (ROI) was drawn around the tumor, and a secondary ROI was drawn around the unaffected flank. Tumor-to-background ratios (TBR) were calculated as follows: (mean fluorescent signal in tumor/mean fluorescent signal in unaffected flank). Signal-to-noise ratios (SNR) were calculated as follows: (mean fluorescent signal in the tumor/SD of the background; ref. 23).

T3M4 orthotopic tumors were implanted and allowed to proliferate for 20 days. Mice were injected with 1 nmol (based on dye) of AR9.6-IRDye800 or IgG-IRDye800 via tail vein injection. Mice were euthanized 144 hours after antibody conjugate administration, and images were collected on the Pearl Trilogy, Fluobeam 800 (Fluoptics), and the Lab Flare RP1 (Curadel) FGS systems. Tumor was resected with image guidance using the Fluobeam and Lab Flare imaging systems, and complete necropsies were subsequently conducted. AR9.6-IRDye800 and IgG-IRDye800 groups were resected using the same exposure time (500 ms). Fluobeam images were analyzed to calculate intraoperative TBR using Image J software (version 1.52a, NIH). Briefly, intraoperative TBR (mean fluorescent signal in tumor/mean fluorescent signal in normal pancreas or adjacent peritoneal tissue) was calculated by manually defining a ROI around the suspected tumor and adjacent background, and analyzing the mean signal in each identified region. Dissected organs were imaged on the Pearl Trilogy. Organ signal was quantified with Image Studio software by manually defining ROI around each tissue and obtaining the mean fluorescent signal, and SNR was calculated as described previously. Necropsied organs were spectrally analyzed using a wavelength-resolved semiquantitative surgical detection system (24). The handheld probe was positioned 1 cm above resected tissues. Points across the length of the tumor and pancreas were excited at 785 nm using 8 mW (low) laser power. Emission spectra were collected from 800 to 950 nm, consistent with previous reports (22, 23).

Organs and tumors resected at 144 hours after injection of antibody conjugates were embedded in optimal cutting temperature compound (OCT) blocks and stored at −20°C until processing for histology. Tissues were cut into 10-μm-thick sections using a cryostat (Leica) and mounted on charged microscope slides (Thermo Fisher Scientific, 22-037-246). Slides were dried at room temperature for 20 minutes and fixed with chilled acetone for 10 minutes, and allowed to dry at room temperature for 20 minutes. To detect the presence of AR9.6-IRDye800 or IgG-IRDye800 at 144 hours, slides were imaged with an Olympus DP80 Digital Camera and cellSense Dimension Software in the FITC channel (tissue autofluorescence), and the 800 nm channel (AR9.6-IRDye800 or IgG-IRDye800). Hematoxylin and eosin (H&E) staining was conducted according to standard procedures and imaged under brightfield microscopy.

Tumors resected at the time of surgery were flash-frozen and embedded in OCT. IHC was conducted using standard procedures for frozen tissues, and the AR9.6 antibody (1:125 dilution, 1 mg/mL stock solution). Tumors resected at autopsy were obtained from the University of Nebraska Medical Center's Tissue Bank through the Rapid Autopsy Pancreatic (RAP) program in compliance with IRB 091-01. To ensure specimen quality, organs were harvested within three hours postmortem, and the specimens placed in formalin for immediate fixation. Sections were cut from paraffin blocks of formalin-fixed tissue into 4 μm thick sections and mounted on charged slides. Slides were stained with OC125 (Roche, 760-2610) according to standard protocols for paraffin-embedded tissues. Slides were imaged on an Olympus DP80 Digital Camera and cellSense Dimension software.

All statistical analyses were conducted using GraphPad Prism software version 7.03 (GraphPad Software). Two-way ANOVA was used, followed by Tukey test for multiple comparisons to compare differences in SNR and TBR between AR9.6-IRDye800, IgG-IRDye800, and unconjugated IRDye800. Multiple t tests were used to compare the biodistribution of AR9.6-IRDye800 to IgG-IRDye800. The SNR of AR9.6-IRDye800 and IgG-IRDye800 were compared with an unpaired t test. All values are reported as the mean ± SD.

MUC16 has been implicated as a potential target for pancreatic cancer. To investigate the expression of MUC16 in patients, IHC was conducted on patient samples. Human pancreatic cancer samples shown in Fig. 1A–C were obtained from the UNMC rapid autopsy pancreas program, and samples shown in Fig. 1D and E were collected during surgical resection. These specimens were stained for MUC16 expression with mAb AR9.6. All patient samples were analyzed by a board-certified pathologist (G.A.T.). Normal pancreas samples obtained at surgery and autopsy and pancreatitis samples obtained at autopsy showed no expression of MUC16 (Fig. 1A, B, and E). Conversely, pancreatic cancer samples collected at autopsy showed strong positive apical staining as well as diffuse cytoplasmic and membrane staining of MUC16 (Fig. 1C). Precursor lesions termed pancreatic intraepithelial neoplasms, or PANIN lesions were also present in the pancreatic cancer sections. These lesions displayed weak MUC16 staining. Samples collected at surgical resection showed MUC16 positive cytoplasmic staining in the stromal fibroblasts, and apical cytoplasmic staining in the tumor (Fig. 1D). These results indicate that MUC16 is a feasible target for FGS in the key patient population of interest.

To develop a targeted FGS probe, AR9.6 and isotype-specific IgG control antibody were conjugated to the NIR dye, IRDye800 NHS Ester. AR9.6-IRDye800 and IgG-IRDye800 antibody conjugates were synthesized by reacting NHS ester dye with free amines on the antibody to form stable amide bonds (Fig. 2A). IgG was used as a nonspecific isotype control throughout this study. Conjugation reactions resulted in an average of three dyes per protein as determined by absorbance spectroscopy (Fig. 2B). Fluorescence spectra of the antibody conjugates demonstrated that fluorescence was not quenched upon conjugation to the protein.

The expression of MUC16 was assessed by Western blot analysis in five pancreatic cancer cell lines: T3M4, Capan1, Colo357, CFPAC, and HPAC (Fig. 2C). An immortalized normal pancreas cell line (HPNE) served as a negative control, and the ovarian cancer cell line, OVCAR3, which has well-documented MUC16 expression, served as a positive control (25, 26). A range of moderate to high expression of MUC16 was seen across all pancreatic cancer cell lines. To confirm that conjugation of AR9.6 to IRDye800 did not drastically impact antigen recognition and cell binding, a fluorescent Western blot analysis and fluorescence microscopy were conducted, as shown in Fig. 2D and E. Fluorescent Western blotting in the 800 nm channel demonstrated that AR9.6-IRDye800 could still recognize MUC16 after dye conjugation. Secondary antibody binding (700 nm channel) confirmed that the fluorescence seen in the 800 nm channel was because of the presence of AR9.6-IRDye800 binding, as shown by colocalization between the 700 and 800 nm channels in Fig. 2D. As expected, the nonspecific IgG-IRDye800 conjugate did not bind to MUC16, and secondary antibody staining confirmed that AR9.6 was not present. Fluorescence microscopy showed strong fluorescence signal from AR9.6-IRDye800 in MUC16 expressing pancreatic cancer cell lines, which was consistent with OVCAR3 cells (positive control). AR9.6-IRDye800 did not bind to MUC16 negative HPNE cells, and the IgG-IRDye800 control did not bind to cells, regardless of MUC16 expression levels (Fig. 2E).

To monitor the tumor accumulation of the antibody conjugates over time, AR9.6-IRDye800, IgG-IRDye800, and unconjugated IRDye800 were assessed for 6 days in a subcutaneous T3M4 xenograft model of pancreatic cancer. Figure 3A depicts representative images of tumor accumulation from the three groups over 144 hours based on images acquired daily on the Pearl Trilogy. Strong fluorescence signal was observed throughout the mouse at 4 hours, whereas robust enhancement of the tumor was observed within 24 hours after injection of AR9.6-IRDye800, and signal was retained in the tumor at 144 hours. Diffuse signal was observed with IgG-IRDye800 at 24 hours, whereas unconjugated IRDye800 was cleared within 24 hours as expected (27). TBRs were highest for AR9.6-IRDye800 at 144 hours after injection (4.47 ± 1.43), as compared with 2.12 ± 0.12 for the IgG-IRDye800 control and 0.89 ± 0.11 for the unconjugated IRDye800 (Fig. 3B). The TBR for AR9.6-IRDye800 was significantly higher (P < 0.0001) than IgG-IRDye800 at 144 hours (the peak TBR for the IgG control). SNRs were also calculated for the three groups (Fig. 3B). At 144 hours, AR9.6-IRDye800 still showed significantly higher (P = 0.0492) SNR (94.35 ± 33.66) than IgG-IRDye800 (48.18 ± 6.68) and IRDye800 (5.56 ± 1.24). Tumor accumulation of AR9.6-IRDye800 in Colo357 xenografts (Supplementary Fig. S1) resulted in comparable TBR (N = 4, TBR = 5.63 ± 0.70) to the T3M4 xenografts. The function of both the TBR and SNR metrics for imaging has been reviewed previously (28). In vitro analysis of the cellular kinetics of AR9.6-IRDye800 binding shown in Supplementary Fig. S3 suggests that antibody internalization, and lysosomal degradation, may contribute in part to the decrease in fluorescence signal over time (29).

To further assess the specificity of the AR9.6 conjugate for the tumor, all T3M4 xenograft tumors and clearance organs were collected at 144 hours postinjection and imaged on the Pearl Trilogy to analyze fluorescence signal in key clearance organs as shown in Supplementary Fig. S2. Quantification of tumor signal demonstrated a significant 2.18-fold increase in mean fluorescence intensity in the AR9.6-IRDye800 tumors compared with the IgG-IRDye800 (P < 0.0001). AR9.6-IRDye800 also showed significantly lower signal in the liver at the 144-hour time point when compared with the IgG control (P = 0.0120). AR9.6-IRDye800 displayed a 2.65-fold increase in tumor signal compared with AR9.6-IRDye800 liver signal (P < 0.0001). Conversely, the IgG-IRDye800 control showed higher liver signal than tumor signal. These results demonstrate the specificity and retention of the AR9.6 conjugate for tumor detection. In addition, higher fluorescent signal was observed in the tumor compared with the liver, suggesting that AR9.6-IRDye800 could potentially highlight metastatic lesions on key background tissues in future studies.

FGS of orthotopic pancreatic cancer xenografts was conducted on multiple FGS systems, as depicted in Fig. 4. The FDA-approved Fluobeam 800 imaging system provided real-time intraoperative feedback via a single 800 nm channel display system (12). Using Fluobeam 800 guidance, AR9.6-IRDye800 provided robust tumor enhancement against background organs, including the liver (Fig. 4A). Conversely, there was variable and negligible fluorescence in the IgG-IRDye800 group, and the signal in the liver was brighter than tumor signal (Fig. 4B). As shown in Fig. 4C, the mean intraoperative TBR for AR9.6-IRDye800 (3.75 ± 0.29) was significantly higher than the mean intraoperative TBR for IgG-IRDye800 (1.89 ± 0.55; P = 0.0010). Videos of AR9.6-IRDye800 and IgG-IRDye800 tumor enhancement during resection can be viewed in Supplementary Videos S1 and S2. The Lab Flare RP1 image-guided surgery system (currently restricted to preclinical studies) was used to confirm AR9.6-IRDye800 signal with real-time NIR-color channel display. Images on this system also showed strong contrast-enhancement with AR9.6-IRDye800. Tumor was clearly delineated against healthy pancreas, and additional lesions could be seen on the posterior surface of the pancreas (Fig. 4D).

Biodistribution of fluorescent antibody conjugates was analyzed in necropsied organs after FGS (Fig. 5A). Quantification of the biodistribution demonstrated significantly higher mean fluorescent signal in the tumors with AR9.6-IRDye800 (0.05 ± 0.006) compared with the IgG-IRDye800 control (0.014 ± 0.009; P = 0.0009; Fig. 5B). There were no significant differences in mean signal between any of the other organs. Variable fluorescent signal was observed in the stomach due to autofluorescence from the mouse diet. Signal in the liver and kidneys can be attributed to clearance of the antibody-dye conjugate and released dye, respectively. The mean tumor SNRs of AR9.6-IRDye800 (94.09 ±11.95) and IgG-IRDye800 (28.33 ± 18.67) were also significantly different (P = 0.0010).

Necropsied organs were also analyzed with a spectrophotometric device. Spectra were collected at defined points across the tumor. Figure 6A and B depict the integrated fluorescence emission spectra that correspond to each point excitation. Fluorescent enhancement of the tumor tissue corresponded to an increase in AUC from spectral acquisition. Furthermore, colocalization between fluorescent signal and the presence of tumor tissue can be seen in H&E staining. Resected specimens were confirmed as poorly-differentiated tumors by a board-certified pathologist. AR9.6-IRDye800 showed increased signal from fluorescence images as well as wavelength resolved measurements compared with the IgG-IRDye800 control. Tumor sections embedded in OCT were analyzed for the presence of AR9.6-IRDye800 and IgG-IRDye800 (Fig. 6C). NIR fluorescence was detected in tumors from mice that were administered AR9.6-IRDye800, whereas NIR fluorescence was not detected in tumor sections from mice administered IgG-IRDye800.

As shown in Fig. 7, micrometastatic lesions were identified during AR9.6-IRDye800-guided resection. NIR signal was readily detected in spleen, muscle, and lung metastases. H&E staining of resected OCT embedded specimens demonstrated the presence of tumor tissue in the detected metastatic lesions. Metastatic lesions were confirmed by a board-certified pathologist (G.A.T). Fluorescence microscopy demonstrated the presence of AR9.6-IRDye800 in the tumor tissue. Fluorescent signal from AR9.6-IRDye800 was consistent with the location of the metastatic lesions, whereas adjacent resected tissue had no visible fluorescence present. Fluorescence localization in the metastatic lesions demonstrated high specificity of AR9.6-IRDye800 for the tumor tissue.

MUC16 is widely expressed in the tumors of patients with pancreatic, ovarian, gastric, esophageal, and colon cancer, among others (18). Because MUC16 is natively expressed in normal bronchial, ovarian, endometrial, and corneal cells in the body, there is high potential for low background signal from key organs during pancreatic cancer imaging (19). Furthermore, in comparison to other targets in pancreatic cancer such as CEA and CA-19-9, MUC16 can better differentiate between chronic pancreatitis and pancreatic cancer (21). Thus, there is compelling evidence to support the development of MUC16 as a target for image-guided surgical intervention of pancreatic cancer.

Compared with current FGS probes in clinical trials, AR9.6-IRDye800 has several potential advantages. EGFR, the target of cetuximab-IRDye800 and panitumumab-IRDye800, has variable expression levels across pancreatic cancer (30). EGFR is also expressed in the healthy pancreas, chronic pancreatitis, and normal stomach and duodenal tissues, which could contribute to elevated background signal during intraoperative pancreatic cancer imaging (16, 31). In comparison, MUC16 is highly expressed in pancreatic cancer tissues, is not expressed in the normal pancreas, and is not expressed in key background tissues, which could improve contrast with FGS. However, results from the phase I/II cetuximab-IRDye800 trial showed promising results for using EGFR as a target for imaging (NCT02736578). One of the major strengths of this study was the use of multimodal imaging in combining FGS with photoacoustic imaging to improve the depth of penetration. EGFR is also advantageous in that it is present in a wide variety of malignancies, and can be used as a tumor-agnostic FGS target. In addition, SGM-101 is another antibody-targeted probe that has demonstrated success in a phase I clinical trial by targeting the tumor antigen CEA (NCT02973672). However, the use of a 700 nm NIR dye limited the depth of penetration of FGS and introduced problems with autofluorescence. Results have not been reported, to date, from the panitumumab-IRDye800, or bevacizumab-IRDye800 (PENGUIN) studies (NCT03384238, NCT02743975). Although MUC16 is relatively unexplored as an imaging target in comparison to EGFR and CEA, there is similar potential to utilize this probe to guide resection of a variety of solid tumors.

AR9.6-IRDye800 demonstrates comparable or improved tumor delineation in comparison to other preclinical studies. Preclinical studies conducted with panitumumab-IRDye800 (now in a phase I/II trial for pancreatic cancer) in a subcutaneous model of head and neck cancer showed TBRs of 2.9 using an IRDye800 optimized system. Comparatively, with AR9.6-IRDye800 in a subcutaneous pancreatic cancer model, an IRDye800 optimized imaging system yielded a TBR of 4.7 and 5.6 across two different cell types. Both AR9.6-IRDye800 and panitumumab IRDye800 had significantly higher tumor signal using the targeted agent as compared with an IgG control (32). In an orthotopic model, panitumumab-IRDye800 had a mean intraoperative TBR of 4.2 compared with a mean intraoperative TBR of 3.75 with AR9.6-IRDye800 in the orthotopic model. Preclinical studies with SGM-101 (completed phase I clinical trial) performed similarly to AR9.6-IRDye800 with intraoperative TBRs to of 3.5 in an orthotopic pancreatic cancer model. Preclinical studies with a humanized CEA antibody (hM5A) conjugated to IRDye800 also showed a mean TBR of 3.5 in a patient-derived orthotopic xenograft model (33). Similar results have also been shown in studies using targeting agents other than antibody conjugates in preclinical pancreatic cancer FGS studies. An αvβ6 integrin targeting peptide yielded TBRs of 2.7 intraoperatively in an orthotopic model (34). An antibody fragment, ssSM3E/800CW, targeting CEA, displayed a mean TBRs of 2.37 in an orthotopic model of pancreatic cancer (35). TBRs achieved under AR9.6-IRDye800 guidance in this study met the recommended TBR of >3.0 for preclinical studies, suggesting that this probe is a potential candidate for clinical translation (28).

Pancreatic cancer surgeries are complex and often result in recurrence. Despite the challenges of targeting this aggressive tumor model, there is an unmet clinical need to improve surgical outcomes. Because of the heterogeneity of pancreatic tumors, initial clinical trials with existing targeted probes have seen only modest tumor enhancement under FGS (17). Future clinical trials in pancreatic cancer will likely incorporate a panel of FGS probes to more accurately capture the tumor heterogeneity and improve intraoperative contrast and enhancement for improved resections (36). However, the feasibility of incorporating multiple probes needs to be investigated to determine if multiple targets improve overall contrast. Because of the well-documented role of mucins in pancreatic cancer, AR9.6-IRDye800 has strong potential to improve current FGS in pancreatic cancer. Thus, we have identified an additional target for FGS that is widely expressed in patients with pancreatic cancer.

Identification of occult peritoneal metastases is critical in FGS. Preoperative imaging modalities can accurately predict unresectability in pancreatic cancer but fall short in detecting small metastatic lesions, identifying vascular involvement, and differentiating benign conditions (37). Thus, FGS can be beneficial in identifying metastases that may have been missed on initial imaging scans. This will impact the course of treatment for patients with pancreatic cancer, as the presence of metastatic lesions determines the resectability of the tumor (38). In this study, AR9.6-IRDye800 identified multiple metastatic lesions in the muscle, lung, and spleen. These suggest that AR9.6-IRDye800-guided resection could potentially improve the identification of missed lesions during surgery. Further studies are needed to investigate the detection capabilities of micrometastases in patient samples.

This study demonstrates the feasibility and proof-of-concept for MUC16 as a potential target and warrants further investigation. The AR9.6 antibody used here was a murine antibody that can recognize both mouse and human MUC16 and is thus limited in its translational potential, because of potentially high levels of immunogenicity. However, this antibody was chosen for investigation because a humanized AR9.6 antibody is under development, and future studies will investigate the humanized antibody for clinical translation. In addition, because of variations in tumor size, not all of the orthotopic xenografts had sufficient unaffected healthy pancreas remaining at the time of FGS. Thus, TBR was calculated using either remaining adjacent normal pancreas or adjacent peritoneal tissue as background signal. Future studies will standardize tumor size and calculate TBR in the adjacent healthy pancreas or inflammatory tissue. In addition, the optimal dye:protein ratio for AR9.6-IRDye800 imaging of pancreatic cancer, as well as methods for site-specific conjugation, will also be investigated. Furthermore, this study does not address the role of tumor heterogeneity on intraoperative fluorescence contrast. Because dense stroma and heterogeneous cell populations can impact the deposition of antibody probes as well as drugs, particularly in pancreatic cancer, further studies are needed to explore FGS in the context of varying cell populations. Discrepancies between experimental studies and clinical results in pancreatic cancer can be attributed in part to inaccurate or ineffective recapitulation of the tumor stroma and microenvironment (39). Genetically, engineered mouse models or patient-derived xenograft models must be employed to more accurately depict the complexity of both the tumor and stroma in pancreatic cancer, although these models have limitations as well. The presence of dense stroma may also impede the delivery of antibodies, and smaller antibody fragments should be investigated to optimize tumor penetration and intratumoral distribution (40–42).

In addition, the identification of metastatic lymph nodes during surgery is a key component of treating patients with pancreatic cancer (43). Lymph node status is an important predictor of survival, and identification of regional lymph node involvement (N1 disease) may influence the course of adjuvant therapy (44). Thus, robust identification of N1 disease and the extent of node involvement with FGS could be beneficial in determining treatment, and sufficient nodal resection. This study did not specifically investigate the presence of MUC16 in metastatic lymph nodes. Future studies are needed to explore the prevalence of MUC16 in positive lymph nodes, and the potential for AR9.6-IRDye800 to identify positive lymph nodes during surgery. Finally, the ectodomain of MUC16 can be cleaved and has been found to circulate in the bloodstream in several cancers, which could impact the feasibility of MUC16 as a target in cancer. Cleavage and subsequent circulation of the antigen may impact the amount of the antibody that actually reaches the tumor. However, a large body of work has demonstrated feasibility for targeting and imaging various cleaved antigens (45–48).

Herein, we developed a MUC16-targeted NIR fluorescent antibody probe and demonstrated its efficacy in delineating pancreatic cancer intraoperatively in vitro and in vivo by FGS. AR9.6-IRDye800 imaging resulted in significantly higher tumor signal compared with an IgG-IRDye800 control. Enhanced tumors could be distinguished from healthy tissue and key background organs relevant to pancreatic cancer. AR9.6-IRDye800 guidance also enabled the detection of micrometastatic lesions in an orthotopic xenograft tumor model. These data suggest that AR9.6-IRDye800 has the potential for clinical translation as a probe for surgical resection of pancreatic cancer.

P. Radhakrishnan has ownership interest (including patents) in OncoCare Therapeutics. M.A. Hollingsworth has ownership interest (including patents) in OncoCare Therapeutics. A.M. Mohs has ownership interest (including patents) in Spectropath, Inc. No potential conflicts of interest were disclosed by the other authors.

Conception and design: M.T. Olson, N.E. Wojtynek, P. Radhakrishnan, Q.P. Ly, M.A. Hollingsworth, A.M. Mohs

Development of methodology: M.T. Olson, N.E. Wojtynek, T.C. Caffrey, Q.P. Ly, A.M. Mohs

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): M.T. Olson, N.E. Wojtynek, G.A. Talmon, A.M. Mohs

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): M.T. Olson, N.E. Wojtynek, M.A. Hollingsworth, A.M. Mohs

Writing, review, and/or revision of the manuscript: M.T. Olson, N.E. Wojtynek, G.A. Talmon, T.C. Caffrey, P. Radhakrishnan, Q.P. Ly, M.A. Hollingsworth, A.M. Mohs

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): T.C. Caffrey, A.M. Mohs

Study supervision: A.M. Mohs

The authors thank Dr. Denis Svechkarev for his assistance with optical probe development, Dr. Madi Madiyalakan (OncoQuest, Inc.) for his assistance with antibody acquisition, A.J. Crawford for her assistance with orthotopic surgical techniques, Paul Grandgenett and the Rapid Autopsy Program for providing patient samples, and Constance McKee (Manzanita Pharmaceuticals, Inc.) for providing the Fluobeam imaging system. This work was supported by NIH grants, U01CA210240 (to M.A. Hollingsworth), P30CA036727 (Fred and Pamela Buffett Cancer Center at UNMC), CA208108 (to P. Radhakrishnan), P50CA127297 (SPORE in Pancreatic Cancer), 5R50CA211462 (NCI Research Specialist), T32CA009476 (to M.T. Olson), and the Nebraska Research Initiative (to A.M. Mohs).

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.