|Home | About | Journals | Submit | Contact Us | Français|
Pancreatic ductal adenocarcinoma is rarely detected early enough for patients to be cured. Our objective was to develop a monoclonal antibody to distinguish adenocarcinoma and precancerous intraductal papillary mucinous neoplasia (IPMN) from benign epithelium.
Mice were immunized with human pancreatic adenocarcinoma cells and monoclonal antibodies were screened against a panel of archived pancreatic tissue sections, including pancreatitis (n=23), IPMN grade 1 (n=16), grade 2 (n=9), grade 3 (n=13), and various grades of adenocarcinoma (n=17). One monoclonal antibody, HPC2 1-B3, was isolated that specifically immunostained adenocarcinoma and all grades of IPMN. Subsequently, HPC2 1-B3 was evaluated in a retrospective series of 31 fine needle aspiration (FNA) biopsies from clinically suspicious pancreatic lesions that had long-term clinical followup.
HPC2 1-B3 was negative in all of the chronic pancreatitis cases tested (0/31). In contrast, HPC2 1-B3 immunostained the cytoplasm and luminal surface of all well to moderately differentiated pancreatic ductal adenocarcinomas (16/16). It showed only weak focal staining of poorly differentiated carcinoma. All high grade IPMNs were positive for HPC2 1-B3. Most low to intermediate grade IPMNs were positive (66% of cases). Immunostaining a separate series of pancreatic FNA cell blocks for HPC2 1-B3 showed the relative risk (2.0 [1.23–3.26]) for detecting at least low-grade dysplasia was statistically significant (Fisher Exact test p-value=0.002).
In order to reduce the mortality of pancreatic cancer, more effective early screening methods are necessary. Our data indicate that a novel monoclonal antibody, HPC2 1-B3, may facilitate the diagnosis of early pancreatic dysplasia.
Pancreatic ductal adenocarcinoma is usually lethal, because it is often not diagnosed until after it has already metastasized.1,2 Nearly 40,000 people in the U.S. will be diagnosed with pancreatic cancer this year and most of these patients will die of disease.2 Improved patient survival may depend on detecting pancreatic cancer in the early stages of disease.
Similar to cervical cancer there are precancerous pancreatic lesions. For example, there is accumulating evidence that intraductal papillary mucinous neoplasia (IPMN) and microscopic pancreatic intraepithelial neoplasia (PanIN) are precursor lesions leading to invasive ductal carcinoma.3–7 IPMNs involve the main ducts, while PanINs involve the smaller ducts. Both have various grades of dysplasia such as PanIN 1–3 and IPMN low-, intermediate-, and high-grade. The actual prevalence of pancreatic dysplasia and long-term risk of progression to invasive adenocarcinoma is essentially unknown, but a few studies suggest that approximately one-third of these lesions may progress to adenocarcinoma within 10 years.8–11 Unfortunately, unlike cervical cancer, there is currently no reliable screening method to detect these precancerous lesions. Serological markers for invasive pancreatic cancer are in the early stages of development and so far there are no reliable markers to detect precancerous lesions.12 Radiographic features of precancerous, and early invasive pancreatic cancer are not specific.3
Endoscopic ultrasound (EUS)-guided fine needle aspiration (FNA) biopsies are becoming the standard of care to evaluate patients with pancreatic symptoms.13–19 The advantage of this evaluation strategy is that it is minimally invasive and can sample small pancreatic masses, as well as provide tissue for cytology. A challenge of this approach is that it is highly dependent on the skill and experience of both the endoscopist and cytopathologist.15 In experienced hands, EUS-guided pancreatic FNAs have good accuracy diagnosing adenocarcinoma (80%),15 but only moderate accuracy diagnosing IPMNs (50%).16 Therefore, most clinicians supplement pancreatic FNA cytology by measuring cyst fluid CEA levels (positive test is >200ng/ml) to improve negative predictive value.6,16,20
New markers with excellent positive predictive value are needed to supplement pancreatic fluid cytology. Markers like the K homology domain containing protein Overexpressed in Cancer (KOC)21–23 have shown promise in diagnosing pancreatic adenocarcinoma, but not IPMNs.21 Gene expression profiling of IPMNs has also yielded a list of potential candidate genes,24 and new commercially available genetic assays have clinical promise.20 However, these molecular tests are more complex and expensive than routine immunoassays. In this report we describe a novel monoclonal antibody, HPC2 1-B3, which appears to specifically detect both pancreatic ductal adenocarcinoma and precancerous IPMNs.
The human tissues used for immunizations and testing of hybridoma supernatants were obtained from the Oregon Pancreatic Tumor Registry and the Oregon Health and Science University (OHSU) Department of Pathology using an IRB approved protocol with informed patient consent. Mouse hybridomas were generated using standard techniques,25 and all work with mice was conducted under a protocol approved by the OHSU Institutional Animal Care and Use Committee. Balb/C mice were immunized every three weeks for three months (n=3–4 total injections) with enzyme-dissociated fresh human pancreatic cancer cells. Approximately 1 × 106 cells were administered intraperitoneally with each immunization using Imject Alum as an aduvant/carrier (Thermo Scientific; Rockford, IL). Four days after the final immunization, animals were euthanized and their spleens were removed for hybridoma generation. Splenocytes were fused with SP2/0 Ag14 myeloma cells25 and successfully fused cells were cloned using ClonaCell-HY media (Stem Cell Technologies Inc., Vancouver, Canada). No more than 600 isolated clones per fusion were transferred into liquid media and distributed into 96 well plates. After clonal expansion, hybridoma supernatants were screened by flow cytometry using Panc1 cell cultures and immunohistochemistry using a preselected panel of pancreatic tissue sections.
Flow cytometry was used to screen a commercially available human pancreas cancer cell line (Panc1 cells; ATCC, CRL-1469, Manassa, Virginia) for marker expression. Cultured cells were trypsinized for removal from flasks, washed, and resuspended in 100 µl DMEM with 5% FBS (DMEM/FBS). Resuspended cells were combined with an equal volume of hybridoma supernatant and incubated at 4°C for 25 minutes. After washing with DMEM/FBS (4°C), the cells were resuspended in 100 µl DMEM/FBS containing 5% normal rat serum, propidium iodide (1:500) and a 1:200 dilution of R-Phycoerythrin (PE) conjugated goat-anti-mouse IgG (H+L) (Jackson ImmunoResearch, West Grove, PA). An isotype-matched mouse monoclonal antibody was used as a negative control.
Seventy-eight surgical resections of the pancreas were identified in the OHSU Pathology Archives from the time period of 2002–2008. This was a retrospective analysis of archived tissue specimens and informed patient consent was not required. Indications for surgery included obstructive pancreatitis (n=23), invasive ductal adenocarcinomas (n=17), and various grades of isolated IPMNs (n=38) without associated adenocarcinoma. None of the IPMNs were cystic mucinous neoplasms with ovarian stroma. Some of the IPMNs had associated PanIN, but they were rare and not separately graded for this study. The IPMNs were graded by two independent surgical pathologists as low- (n=16), intermediate- (n=9), and high-grade (n=13) according to accepted criteria.1, 5, 26 Low-grade lesions were characterized by flat or papillary epithelial lesions composed of tall columnar cells with small, round, basally located nuclei and abundant mucinous cytoplasm. Intermediate-grade lesions had similar architecture with low mitotic index, but moderate nuclear atypia. High-grade lesions had complex or solid architecture with nuclear atypia, including nucleoli, frequent mitoses, and necrosis, but no evidence of invasion into surrounding pancreatic tissue. Formalin fixed and paraffin-embedded pancreatic tissue sections underwent antigen retrieval in citrate buffer solution, pH 6.0 (Target Retrieval Solution, Dako, Carpinteria, CA). They were then incubated with hybridoma supernatant at a 1:10 dilution (10ug/ml) in PBS for one hour at room temperature, followed by ImmPress Reagent Peroxidase Anti-Mouse link-antibody (Vector Laboratories, Burlingame, CA) for 30 minutes. They were washed and incubated with DAB chromagen yielding a brown positive signal. Sections were counterstained with hematoxylin. Lesional cells were considered positive if more than 10% of the neoplasm immunostained for HPC2 1-B3. Positive cells may have cytoplasmic staining (cancer), or show only luminal brush border staining (low to intermediate grade IPMNs).
To test the diagnostic predictive value of HPC2 1-B3, we immunostained a retrospective series of 31 EUS-guided FNA biopsies of the pancreas. These cases represented the only pancreatic FNAs in the OHSU Cytopathology Archives (2001 to 2007) with cell block material sufficient for immunohistochemical analysis. No other exclusion criteria were used. All 31 cases had either surgical followup, or at least four years of negative clinical followup. Indication for FNA was pancreatitis, or symptoms of duct obstruction, with an identifiable pancreatic mass. The cytologic diagnosis was either negative for malignancy (n=8), positive for mucinous neoplasm (n=17), or positive for ductal adenocarcinoma (n=6). Notably, low-grade mucinous cases biopsied from the pancreatic body or tail had been previously stained for alcian yellow and alcian blue to exclude gastric contamination. Long-term clinical followup confirmed pancreatitis in all eight negative cases and adenocarcinoma in all six malignant cases. The surgically resected mucinous neoplasms were classified into IPMNs with high-grade dysplasia (n=4), or into low- to intermediate-grade dysplasia (n=13) for statistical analysis. Paraffin-embedded sections of the archived FNA biopsy cell block were then immunostained for HPC2 1-B3 using a 1:10 dilution (10ug/ml). Staining was considered positive if the ductal epithelium showed either cytoplasmic or cell surface signal. Notably, the background debris was often strongly positive when ductal cells were positive. Histologic sections of the cell blocks were also immunostained for KOC21–23 using a commercially available antibody (monoclonal anti-human IMP3; DAKO clone 69.1) according to manufacturer's instructions. KOC positivity highlights the cytoplasm. The followup surgical resections of these FNAs were also immunostained and agreed with the FNA results, but were not included in the panel data presented in Table 1.
Western blot analyses were used to define the molecular mass of the HPC2 1-B3 antigen and to determine if Panc1 and HeLa cells (ATCC, CCL-2, Manassa, Virginia) shed or secrete the antigen. Panc1 cells and cervical cancer HeLa cells were grown in DMEM with 10% FCS until they were approximately 80% confluent. The media was removed and replaced with DMEM with 3% FCS. After a 24 hour incubation, the supernatants from these cultures was harvested to test for the presence of HPC2 1-B3 antigen. Supernatants were centrifuged at low speed to remove non-adherent cells and cellular debris and then concentrated approximately 25-fold using an Amicon Ultra Centrifugation filter with a molecular weight cut off of 10 kDa (Ultracel-10K MWCO; Amicon). The concentrate was then microfuged at high speed for two minutes to remove any remaining debris and supernatant was collected. The samples were then separated by electrophoresis using a Criterion XT Precast Gel 12% Bis-Tris and electrophoretically transferred to Immobilon PVDF. The blot was blocked with PBS containing 10% skim milk and 1% BSA. It was then incubated overnight at 4°C in a 1:500 dilution of HPC2 1-B3 hybridoma supernatant. After washing in PBS + 0.1% Tween 20, HRP-conjugated anti-mouse was added at 1:2000 dilution for 1 hour at room temperature. The membrane was washed and developed using Western Lightning Plus ECL enhanced Luminol Reagent and visualized using Kodak X-OMAT Blue XB film.
Immunohistochemical data were compared by Chi-Square analysis using the Fisher Exact test and SAS software (version 9.1.3; SAS Institute Inc. NC). Test performance of HPC2 1-B3 and KOC immunostaining, including sensitivity (SN), specificity (SP), positive predictive value (PPV), negative predictive value (NPV), and relative risk (RR) for at least low-grade dysplasia, were calculated using 2× 2 Contingency tables with binomial 95% confidence intervals.
Hybridoma supernatants from approximately 900 colonies were screened for reactivity by flow cytometry and our preselected panel of pancreatic tissue sections. Screening yielded one monoclonal antibody, HPC2 1-B3, that stained Panc1 cells by flow cytometry (Figure 1), was negative in chronic pancreatitis, and positive in pancreatic ductal adenocarcinoma and precancerous IPMNs by immunohistochemistry (Figure 2).
Invasive adenocarcinoma stained strongly and diffusely for HPC2 1-B3. The antigen predominantly localized to the cytoplasm of moderately differentiated cancer (Figure 2) and the apical border of well differentiated ductal adenocarcinoma (Figure 2H inset). HPC2 1-B3 staining of poorly differentiated adenocarcinoma showed only weak focal cytoplasmic signal, although all six of these cases stained strongly for the KOC marker (Figure 2) (Table 1).
Both HPC2 1-B3 and KOC immunostained EUS-guided FNAs of pancreatic cancer (5/6 and 4/6, respectively), but not chronic pancreatitis (Table 2). HPC2 1-B3 was observed in carcinoma cells and in the surrounding mucinous debris (Figure 3). In contrast, KOC stained carcinoma cells, but not the surrounding debris (Figure 3). Pancreatitis was reproducibly negative for HPC2 1-B3 (n=8 cases), suggesting the antigen is not made by reactive pancreatic ductal epithelium (Figure 3).
Immunostaining of surgically resected IPMNs revealed diffuse strong apical border staining for HPC2 1-B3 in most low-grade precancerous lesions (10/16, 63%), nearly all of the intermediate-grade lesions (8/9, 89%), and all of the high-grade IPMNs with severe dysplasia (13/13, 100%). In addition, HPC2 1-B3 antigen was reproducibly detected in the duct lumens of IPMNs, suggesting it is also secreted by these precancerous lesions (Figure 2E). A few tissue sections also contained foci of grade 3 PanIN, which stained for HPC2 1-B3 (data not shown). Only rare foci of grade 1 PanIN were identified in our study sections; therefore, although they were negative for HPC2 1-B3, further testing will be required in a larger cohort.
HPC2 1-B3 immunostained all four high grade IPMNs (100%) and half of the FNA biopsies from low to intermediate grade IPMNs (6/13, 46%) (Table 2). Positive staining was predominantly seen along the apical border of dysplastic cells and in the surrounding mucinous debris (Figure 3). KOC stained the majority of high-grade cases (3/4, 75%), but none of the low- to intermediate-grade IPMNs (0/13) (Table 2).
HPC2 1-B3 detected more cases of severe dysplasia and invasive adenocarcinoma than the KOC marker in our series (9/10 versus 7/10, respectively). All eight cases of pancreatitis were negative (Table 2). HPC2 1-B3 was more sensitive than KOC identifying precancerous pancreatic ductal lesions (65% versus 30%, respectively). The PPV of both markers was 100%, although the confidence interval range was wide given the relatively limited sample size. The relative risk (RR) and Fisher Exact test of detecting at least low-grade dysplasia using HPC2 1-B3 was significant: RR = 2.0 [1.23–3.26]; p-value = 0.002. In contrast, KOC did not detect low-grade dysplasia and had an overall RR of 1.5 [1.13–1.99] and Fisher Exact test p-value = 0.15.
Western blot analysis of cultured Panc1 and HeLa tumor cells revealed HPC2 1-B3 protein in both the cell lysates and culture supernatant (Figure 4). Human cervical cancer HeLa cells were chosen because we identified high levels of HPC2 1-B3 protein in this cell line. Our data suggest HPC2 1-B3 is approximately 55–65 kDa in size and is secreted or shed by both tumor cell lines into culture supernatant. Panc1 cells secrete less HPC2 1-B3 than HeLa cells, but together with our immunohistochemical analyses of IPMNs and pancreatic adenocarcinomas (Figures 2 and and3),3), this Western blot data support the hypothesis that the HPC2 1-B3 antigen is likely secreted by dysplastic and neoplastic pancreatic ductal cells.
In order to reduce the mortality of pancreatic cancer, early screening methods similar to those employed for cervical cancer will be necessary. Our data indicate that a novel monoclonal antibody, HPC2 1-B3, may facilitate the diagnosis of early pancreatic ductal dysplasia, which we suspect could be more likely to progress to invasive adenocarcinoma. The detection of these early lesions is vital to reducing patient mortality.
Dysplastic pancreatic ductal lesions include IPMNs in the larger ducts and PanINs in the smaller ducts.4,5 Similar to cervical cancer, many cases of pancreatic ductal neoplasia begin as low-grade dysplasia and progress over time into high-grade dysplasia and invasive adenocarcinoma.8–11 Unfortunately, unlike cervical cancer, there are currently no effective screening methods that are both reliable and relatively inexpensive. For example, cytology alone identifies pancreatic ductal adenocarcinoma and high-grade dysplasia, but it often misses low- to intermediate-grade IPMNs. In our experience, a potential pitfall with the cytologic diagnosis of low-grade IPMNs is gastric contamination. EUS-guided FNAs of the body or tail of the pancreas are through the gastric wall, rather than the duodenum. This is significant because duodenal contamination is easily discriminated by the presence of goblet cells in cytologic preparations, but gastric mucosa is morphologically very similar to low-grade IPMNs. Some authors suggest immunohistochemical evaluation of mucin gene expression profiles to address this issue.27–29 Gastric mucosa and duodenal mucosa are negative for HPC2 1-B3 (data not shown).
Serological markers for invasive pancreatic cancer are in the early stages of development, but markers12 and radiographic features ofprecancerous lesions are not reliably accurate. 9 Aside from aspirate CEA levels, few immunohistochemical markers have been validated to screen EUS-guided FNAs. For example, in a recent study of 36 cases (7 pancreatitis, 15 IPMNs, and 14 ductal adenocarcinomas), Toll et al (2009) showed that KOC was specific for pancreatic cancer and did not stain pancreatitis or IPMNs.21 Similarly, Zhao et al (2007) has shown KOC is positive in 88% of pancreatic ductal cancers diagnosed by EUS-guided FNA (35/40).23 Yantiss et al (2005) reported positive KOC staining in surgically resected ductal adenocarcinomas and in a subset of IPMNs with high-grade dysplasia (4/10), but not in IPMNs with low- to intermediate-grade dysplasia (n=2 and n=3 cases, respectively).22 Given the NPV provided by CEA testing, 6,16 the additional PPV for high-grade dysplasia and adenocarcinoma provided by KOC is impressive.21–23 However, cytopathologists are already reliable at identifying high-grade pancreatic ductal lesions;15 therefore, the diagnostic problem is detecting low- to intermediate-grade IPMNs that may progress to invasive ductal adenocarcinoma. This is a weakness of the KOC marker.
Our marker is only in the earliest stages of validation, but our reported sample size in this pilot study is comparable to that published by others investigating similar antibody-based markers for pancreatic cancer.21–23 The added value of using HPC2 1-B3 is not only its ability to immunostain pancreatic ductal adenocarcinoma and high grade dysplasia, but its ability to detect low- to intermediate-grade IPMNs. The marker appears to be specific, because it was negative in all 23 cases of chronic pancreatitis in our screening panel, and it was negative in all eight cases of pancreatitis identified in our series of EUS-guided FNAs. HPC2 1-B3 detected more cases of high-grade dysplasia and invasive adenocarcinoma than the KOC marker; and, HPC2 1-B3 identified half of the low- to intermediate-grade IPMNs biopsied by EUS-guided FNA. In surgically resected cases, HPC2 1-B3 identified most low- to intermediate-grade IPMNs and all high-grade cases. It did not diffusely stain 6/17 adenocarcinomas, but all six of these cases represented the poorly differentiated carcinomas inserted into this primary screening panel and they did show weak focal staining.
HPC2 1-B3 immunostaining of low to intermediate grade IPMNs raises challenging management issues. Some clinicians may chose to follow low grade IPMNs and only excise higher grade lesions that are more likely to progress to invasive adenocarcinoma5,14 In our practice, we excise any large duct IPMN, regardless of grade. In contrast, side branch PanINs without high grade atypia are excised only if they create mass lesions larger than 3 cm. Therefore, it is possible, that HPC2 1-B3 may detect side duct PanINs, which are not necessary to excise. But, we reject this scenerio as a weakness because imaging studies should be able to distinguish main duct and side branch lesions. Therefore, we think positive HPC2 1-B3 immunostaining would be a benefit solely based on increased sensitivity and specificity of the cytologic diagnosis. Moreover, we hypothesize that positive HPC2 1-B3 immunostaining may potentially discriminate between low grade IPMNs that may progress to carcinoma and those cases that do not progress. In future studies, our group intends to correlate HPC2 1-B3 staining with KRAS mutations14 and long-term progression risk. In contrast, the KOC marker cannot be used for this purpose since it does not stain low grade IPMNs.
Another potential strength of the HPC2 1-B3 antibody may be its ability to detect secreted antigen in FNA biopsy aspirates or potentially in pancreatic duct fluid by ELISA. Duct fluid screening would be preferred because it could be performed periodically as a screening test for pancreatic ductal dysplasia. Indeed, the HPC2 1-B3 antigen was observed in the supernatant of our cell cultures and was also seen in the luminal debris of IPMNs and invasive adenocarcinomas immunostained for HPC2 1-B3. The combination of ELISA and cytology would be most effective, because the cytopathologist could then distinguish low grade from high grade nuclear features. The decision to resect an IPMN or PanIN is dependent on location, size, and grade. Therefore, maximizing diagnostic information for the surgeon is best.
This may be especially important when distinguishing PanINs. Many practices resect large duct IPMNs, but may follow side branch PanINs if they are low grade. Others may follow large duct IPMNs if they are low grade, but resect if they are high grade. High grade PanINs identified in our tissue sections immunostained for HPC2 1-B3. We had few low grade PanINs, but the rare cases of PanIN 1 we had were negative. This is important, because low grade PanINs have a low risk of progression into invasive cancer.
For now the diagnostic value of HPC2 1-B3 should be viewed with caution until larger prospective studies are completed. HPC2 1-B3 appears to be specific for pancreatic ductal dysplasia and we have recently shown it discriminates between metastatic pancreatic adenocarcinoma and primary cholangiocarcinoma in liver biopsies.32 Our long-term objective is to determine whether HPC2 1-B3 and pancreatic duct fluid CEA levels will complement each other to provide improved predictive value as a relatively inexpensive screening “pap smear of the pancreas”. Unfortunately, we did not have available CEA levels in our pilot study.
We have not yet identified the antigen targeted by the HPC2 1-B3 monoclonal antibody. However, once it is isolated, it will be interesting to compare the antigen with other potential IPMN markers of similar size (55–65 kDa) suggested by cDNA microarray analysis.24 A better understanding of the antigen and its role in pancreatic dysplasia may provide significant insights into the progressive pathophysiology of this malignant disease.
The authors thank Maria Grompe, YongPing Zhong, and Pamela Canaday, for their assistance with flow cytometric data acquisition and analyses; Stephanie Abraham and Kelsea Lanxon-Cookson for support in the generation of HPC2 1-B3; Carolyn Gendron, Cara Poage, and Dornald Myles for outstanding histology support; and Miriam Douthit for management of the Oregon Pancreatic Tumor Registry. We would also like to acknowledge the Flow Cytometry and Monoclonal Antibody Shared Resources within the Knight Cancer Institute and Oregon Stem Cell Center.
Financial Support: Dr. Morgan's contribution was funded by the Office of Research on Women’s Health and the National Institute of Child Health and Human Development, Oregon BIRCWH HD043488-08. Dr. Streeter was supported by the Oregon Stem Cell Center, the OHSU Knight Cancer Institute, and a generous gift from Randy and Mary Huebner. Several of the authors (TKM, MG, CC, and PS) are listed on a patent application for the antibody HPC2 1-B3. This publication was made possible in part with the support from the Oregon Clinical and Translational Research Institute (OCTRI), grant number UL1 RR024140 from the National Center for Research Resources (NCRR), a component of the National Institutes of Health (NIH) and NIH Roadmap for Medical Research and the OHSU Knight Cancer Institute, grant number P30 CA 069533 from the National Cancer Institute.