Abstract
Pancreatic cancer is among the most lethal of cancers in part because of our inability to detect it in early premalignant stages and because it metastasizes from small lesions. It is difficult to identify at-risk populations as a relatively low percentage of cases are associated with identifiable inherited genetic factors. Nonetheless, there are significant and increasing groups of patients that are currently being identified because of incidental clinical findings who are at risk of having or developing pancreatic cancer. For example, the widespread use of high resolution imaging identifies substantial numbers of patients with pancreatic cysts or other abnormalities. Some patients with pancreatic cancer present with adult onset type II diabetes months to years prior to the diagnosis of cancer. Improved diagnostic tests would be of benefit in evaluating these and other at risk populations.
Small and inaccessible very early lesions of the pancreas are not likely to produce a sufficient quantity of any biomarker to be detected in serum or plasma, although it may be possible to identify biomarkers in biopsied material. It is possible to detect compounds produced from more advanced lesions in serum or plasma, evidenced by the fact that assays for CA19-9 have long been used to follow patients with advanced disease. The point at which a lesion achieves a sufficient size and degree of insult to produce material that can be detected in circulation or body fluids has not been established. It remains difficult to study this question in humans because of our inability to identify appropriate patients with disease and to accurately stage early lesions in pancreatic cancer. Mouse models of early pancreatic cancer are useful for identifying some early markers of transformation and for evaluating these in circulation, but mice do not accurately recapitulate many aspects of the specific biology of human neoplasia, especially in the area of posttranslational modifications such as glycosylation. There are several other commonly used clinical tests for tumor markers are serum-based immunoassays for different types of adenocarcinoma including DUPAN-2, CA50, CA125 and CA242, which are comprised of oligosaccharide structures that are present on heavily glycosylated high molecular weight mucins. The molecular structure recognized by the CA19-9 antibody is the sialyl Lewis A oligosaccharide, which can be attached to a number of different core proteins. These molecules are synthesized and expressed by epithelial cells of the gastrointestinal, respiratory, and genitourinary tracts. The structure of epithelial mucins includes a protein backbone (core protein) with tandemly repeated units that each carry numerous carbohydrate side chains. This structure enables the use of antibodies for both capture and detection of the antigen because the repeated nature of the mucin structure provides a multivalent target (you can have hundreds to thousands of oligosaccharrides per molecule), so the same antibody can be used for capture and detection.
Efforts to combine existing assays for oligosaccharide epitopes on tumor mucins that are detected in serum typically led to improvements in sensitivity, but with a concomitant loss of specificity. Retrospectively, this finding is not surprising, as these assays were combined somewhat randomly, according to availability of the assays and access to clinical samples. Many of the oligosaccharide structures that were detected by these antibodies are related and would be predicted to show overlapping reactivities, although this was not known when the assays were developed.
The CA19-9 test has been much maligned for lack of performance characteristics as a standalone biomarker. For detecting suspected and known cases of pancreatic cancer, different studies have reported that the sensitivity of the CA19-9 serum assay ranges from 69% to 93%, and the specificity varies between 46% and 98%. This lower specificity occurs because elevated levels of the CA19-9 antigen are found in chronic pancreatitis, benign liver disease, and other cancer conditions. Nonetheless, the CA19-9 test remains the benchmark for diagnostic tests for pancreatic cancer, and is widely used to follow the progression of disease in patients with moderate to advanced disease, in concert with other clinical signs and symptoms.
A basic limitation of the CA19-9 test (and these other similar tests) is that there are not good tests to differentiate between expression of the CA19-9 epitope on different mucin-type core proteins. Recent knowledge of the biology of mucins suggests to us that it is possible to improve these assays by identifying both oligosaccharide structures and the types of core proteins on which the oligosaccharide structures are expressed. Elevated serum CA19-9 levels occur in many different adenocarcinomas, but also in cases of benign diseases of the liver, pancreas, gall bladder, and other organ sites. The serum tests typically capture and detect proteins that have multiple CA19-9 epitopes attached, without discriminating among the core proteins. There is now substantial evidence that different organ sites produce different core proteins, many of which have CA19-9 attached. Thus, there will be differences in the overall concentration and composition of core proteins carrying the CA19-9 epitope that arise from different sites and end up in serum as a result of a pathological process in that organ site. Thus, it should be possible to refine the CA19-9 test by also determining the amounts and types of core proteins on which it is found. It is our tenet that a systematic analysis of both oligosaccharide structures and the core proteins of circulating mucins would be predicted to improve the specificity of diagnosis and might lead to improved sensitivity and early detection of a number of cancers, including pancreatic cancer.
Early studies to characterize the molecular nature of mucins were complicated by their biophysical properties: relatively large mass (well over 106 Daltons), complex biochemical composition (50-80% O-linked oligosaccharides) and tendency to form higher order structures through polymerization. To date, genes encoding 20 different mucin core proteins have been described partially or completely. Studies of the patterns of mucin core proteins (MUC) that are overexpressed by adenocarcinomas derived from different tissues indicate that those derived from different organs of origin produce distinct patterns of mucin core proteins, because tumors often operate under the differentiation program of the organ site from which they arise. Some tumors express mucins that are not normally found in that organ site. Early pancreatic cancer lesions have been shown to express novel proteins and importantly in humans to express specific glycoproteins that have novel posttranslational modifications, especially glycosylation, that are specific to those premalignant and malignant lesions. For example, MUC4 is expressed in premalignant and malignant lesions of the pancreas, but is not commonly expressed in normal pancreatic ductal epithelial cells. Mucin type glycoproteins have a high carbohydrate content (>50%), and are expressed by normal epithelial cells, but some are upregulated in the carcinomas that develop from them. MUC1 for example is a membrane mucin that is expressed and aberrantly glycosylated in a high proportion of pancreatic carcinomas. Previous studies have documented changes in glycosylation of MUC1 seen in cancer: shorter O-glycans are added at increased density. These changes result in dramatic changes in the profile of epitopes expressed on mucin molecules, which in turn present unique tumor specific glycopeptide structures. The expression of novel proteins and glycoproteins, especially those carrying Tn, sialyl-Tn, T and sialyl T antigens, can provoke the production of autoantibody responses in humans and mice, a response that is amplified significantly by virtue of the immune response. Antibodies are stable and can be easily detected in serum and plasma. As malignant lesions progress and enlarge, there comes a point where specific antigens detected by autoantibodies are produced in sufficient quantities to bind circulating antibodies, which would block their detection (and cause them to be cleared from circulation). Ultimately, in advanced carcinoma, high levels of tumor burden and circulating antigen prevail (e.g., high CA19-9 levels in late stage disease), and often autoantibodies cannot be detected because they are bound to antigen or have been cleared from circulation. It is also possible that high concentrations of some antigens or other immunosuppressive effects from the tumor provoke high-dose tolerance or other forms of immunosuppression that eliminates further production of autoantibodies in patients with advanced disease.
It is our premise that the earliest premalignant and malignant lesions in the pancreas express MUC1, MUC4, MUC5ac, MUC16, and MUC17 with truncated O-linked structures attached (Tn, sialyl Tn, T). Although small lesions that produce these aberrantly glycosylated mucins may not produce a sufficient amount to be detected in serum or plasma, autoantibodies to these structures will be produced at sufficiently high concentrations to be detected. As lesions progress and achieve a size and degree of tissue disruption, mucins will be detected in circulation, initially as immune complexes with the autoantibodies, and in later stages (as antibody responses are overwhelmed or immunosuppressed) in the absence of immune complexes. This overall hypothesis is being tested by developing a set of three integrated tests for serum: a test that quantifies autoantibodies, a test that quantifies immune complexes, and a test that quantifies circulating mucins for both oligosaccharide and core protein (on the same molecule).
Citation Information: Cancer Prev Res 2010;3(12 Suppl):CN09-04.
RSS Feeds