Lack of a universal site-specific conjugation methodology for antibodies limits their potential to be developed as tumor-specific imaging agents or targeted therapeutics. A potential mechanism for site-specific conjugation involves utilization of the conserved N-glycosylation site in the CH2 domain. We sought to develop an antibody with an altered azido-sugar at this site whereby site-specific label could be added. The HB8059 hybridoma was cultured with peracetylated N-azidoacetlymannosamine (Ac4ManNAz). The resulting azido-sugar antibody was conjugated to phosphine-polyethylene glycol (PEG3)-biotin via a modified Staudinger reaction. Biochemical and functional characterization of the biotinylated antibody was performed. The azido-sugar antibody was also labeled with DyLight-650-Phosphine and injected into mice harboring pancreatic cancer xenografts. The tumors were dissected and imaged utilizing an IVIS fluorescent camera. The antibody was successfully produced in 100 μM Ac4ManNAz. The biotinylated antibody demonstrated a 50 kDa heavy and 25 kDa light chain on sodium dodecyl sulfate-polyacrylamide gel electrophoresis, but demonstrated a single band at 50 kDa on western blot. Treatment with a N-linked glycosidase extinguished the band. Flow cytometry demonstrated antigen-specific binding of CA19-9-positive cells and the antibody localized to the antigen-positive tumor in vivo. We successfully produced an antibody with an azido-sugar at the conserved CH2 glycosylation site. We were able to utilize this azide to label the antibody with biotin or fluorescent label and demonstrate that the label is added in a site-specific manner to the heavy chain, N-linked glycosylation site. Finally, we demonstrated functionality of our antibody for in vitro and in vivo targeting of pancreatic cancer cells.

Purpose

Intact antibodies have a long serum persistence resulting in high background signal that inhibits their direct translation as imaging agents. Engineering of antibody fragments through the introduction of mutations in the fragment crystallizable (Fc) region can dramatically reduce serum persistence. We sought to develop a Fc-mutated, anti-CA19-9 antibody fragment (anti-CA 19-9 scFv-Fc H310A) to provide micro-positron emission tomography (microPET) imaging of pancreatic cancer xenografts.

Procedures

The anti-CA19-9 scFv-Fc H310A was successfully expressed and purified. Biochemical characterization included size exclusion chromatography, sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS-PAGE), Western blot, and flow cytometry. The antibody fragment was labeled with iodine-124 ((124)I) and injected into mice containing human pancreatic cancer xenografts. MicroPET/CT images were then obtained. Blood, organ, and tumor radioactivity was measured and expressed as the percent of injected dose per gram of tissue (%ID/g).

Results

Biochemical characterization was consistent with the creation of a 105 kD dimer containing a human Fc region. Flow cytometry demonstrated antigen-specific binding, and cell-based ELISA further established a dissociation constant (K D) of 10.7 nM. (124)I-labeled scFv-Fc H310A localized to the antigen-positive tumor xenografts as detected by microPET. Objective confirmation of targeting was demonstrated by higher %ID/g in the antigen-positive tumor compared to the blood, antigen-negative tumor, and liver.

Conclusions

We successfully engineered and produced an anti-CA19-9 scFv-Fc H310A antibody fragment that retains similar affinity when compared to the parental intact murine antibody. Additionally, our engineered and mutated fragment exhibited antigen-specific microPET imaging of both subcutaneous and orthotopic pancreatic cancer xenografts at early time points secondary to decreased serum half-life.

Background

Pancreatic ductal adenocarcinoma (PDAC) patients demonstrate highly variable survival within each stage of the American Joint Committee on Cancer (AJCC) staging system. We hypothesize that tumor grade is partly responsible for this variation. Recently our group developed a novel tumor, node, metastasis, grade (TNMG) classification system utilizing Surveillance Epidemiology and End Results (SEER) data in which the presence of high tumor grade results in advancement to the next higher AJCC stage. This study's objective was to validate this TNMG staging system utilizing single-institution data.

Methods

All patients with PDAC who underwent resection at UCLA between 1990 and 2009 were identified. Clinicopathologic data reviewed included age, sex, node status, tumor size, grade, and stage. Grade was redefined as a dichotomous variable. The impact of grade on survival was assessed by Cox regression analysis. Disease was restaged into the TNMG system and compared to the AJCC staging system.

Results

We identified 256 patients who underwent resection for PDAC. Patients with low-grade tumors experienced a 13-month improvement in median survival compared to those with high-grade tumors. On multivariate analysis, tumor grade was the strongest predictor of survival with a hazard ratio of 2.02 (p = 0.0005). Restaging disease according to the novel TNMG staging system resulted in improved survival discrimination between stages compared to the current AJCC system.

Conclusions

We were able to demonstrate that grade is one of the strongest independent prognostic factors in PDAC. Restaging with our novel TNMG system demonstrated improved prognostication. This system offers an effective and convenient way of adding grade to the current AJCC staging system.

BACKGROUND:Occult metastatic tumors, below imaging thresholds, are a limitation of staging systems that rely on cross-sectional imaging alone and are a cause of the routine understaging of pancreatic ductal adenocarcinomas (PDACs). We investigated circulating tumor cells (CTCs) as a preoperative predictor of occult metastatic disease and as a prognostic biomarker for PDAC patients. EXPERIMENTAL DESIGN:A total of 126 patients (100 with cancer, 26 with benign disease) were enrolled in our study and CTCs were identified and enumerated from 4 mL of venous blood using the microfluidic NanoVelcro assay. CTC enumeration was correlated with clinicopathologic variables and outcomes following both surgical and systemic therapies. RESULTS:CTCs were identified in 78% of PDAC patients and CTC counts correlated with increasing stage (ρ = 0.42, p < 0.001). Of the 53 patients taken for potentially curative surgery, 13 (24.5%) had occult metastatic disease intraoperatively. Patients with occult disease had significantly more CTCs than patients with local disease only (median 7 vs. 1 CTC, p < 0.0001). At a cut-off of three or more CTCs/4 mL, CTCs correctly identified patients with occult metastatic disease preoperatively (area under the receiver operating characteristic curve 0.82, 95% confidence interval (CI) 0.76-0.98, p < 0.0001). CTCs were a univariate predictor of recurrence-free survival following surgery [hazard ratio (HR) 2.36, 95% CI 1.17-4.78, p = 0.017], as well as an independent predictor of overall survival on multivariate analysis (HR 1.38, 95% CI 1.01-1.88, p = 0.040). CONCLUSIONS:CTCs show promise as a prognostic biomarker for PDAC patients at all stages of disease being treated both medically and surgically. Furthermore, CTCs demonstrate potential as a preoperative biomarker for identifying patients at high risk of occult metastatic disease.

To understand the potential and limitations of circulating tumor cell (CTC) sequencing for molecular diagnostics, we investigated the feasibility of identifying the ubiquitous KRAS mutation in single CTCs from pancreatic cancer (PC) patients. We used the NanoVelcro/laser capture microdissection CTC platform, combined with whole genome amplification and KRAS Sanger sequencing. We assessed both KRAS codon-12 coverage and the degree that allele dropout during whole genome amplification affected the detection of KRAS mutations from single CTCs. We isolated 385 single cells, 163 from PC cell lines and 222 from the blood of 12 PC patients, and obtained KRAS sequence coverage in 218 of 385 single cells (56.6%). For PC cell lines with known KRAS mutations, single mutations were detected in 67% of homozygous cells but only 37.4% of heterozygous single cells, demonstrating that both coverage and allele dropout are important causes of mutation detection failure from single cells. We could detect KRAS mutations in CTCs from 11 of 12 patients (92%) and 33 of 119 single CTCs sequenced, resulting in a KRAS mutation detection rate of 27.7%. Importantly, KRAS mutations were never found in the 103 white blood cells sequenced. Sequencing of groups of cells containing between 1 and 100 cells determined that at least 10 CTCs are likely required to reliably assess KRAS mutation status from CTCs.