2008 Scientific Report

VARI | 2008
The key to developing improved markers is the ability to reproducibly measure specific glycans on specific proteins. Many of
the carbohydrate structures on proteins in normal and cancer tissues have been characterized using mass spectrometry and
enzymatic methods. Those methods are valuable for defining structures, but they do not have the precision or throughput
necessary to look at changes in levels between samples, which is necessary to assess biomarker potential. A new method
developed in our laboratory provides the means to obtain more detailed information on glycan variation. We use lectins—
proteins that bind specific glycan structures—and glycan-binding antibodies to probe the levels of particular glycans on the
proteins captured by antibody arrays. This method provides the important feature of allowing comparison between samples of
the levels of particular glycans on specific proteins so that we can assess their diagnostic potential. A product based on this
technology is now available from GenTel Biosciences (Madison, WI).
The class of proteins called mucins shows particularly high levels of glycan alteration in pancreatic cancer. Mucins are longchain,
heavily glycosylated proteins on epithelial cell surfaces that have roles in cell protection, interaction with the extracellular
space, and regulation of extracellular signaling. Altered carbohydrates on mucins can affect critical processes in cancer such
as cell migration or extracellular signaling to the immune system. We have extensively characterized the glycan variations on
mucins secreted into the blood of pancreatic cancer patients. In some cases, the levels of certain mucin glycans are altered in
cancer patients more often than the levels of the core proteins (Figure 1a). As a result, detection of the glycans performed better
as a biomarker than detection of the core proteins (Figure 1b). The efficient analysis of many samples and glycan structures
was made possible by the ability to run dozens of samples on a single microscope slide. A device based on that technology,
which partitions microscope slides for efficient sample processing, is available from The Gel Company (San Francisco, CA).
Our work shows the promise of this approach and points to key directions for further developing biomarkers of pancreatic
cancer. Our research now focuses on the goals of identifying the protein carriers of cancer-associated glycans, of identifying
the most important cancer-associated glycans and the reagents to detect them, and of applying these discoveries to pancreatic
cancer diagnostics (Figure 1c).
Figure 1
In addition, we are seeking to better
understand the origins of glycan alterations
and the functional contribution of
these molecules to pancreatic cancer
development and progression.
Figure 1. Pancreatic cancer biomarker development. a) Comparison of glycan versus protein detection. The level of the MUC5ac
core protein in serum samples from cancer patients and healthy subjects, determined using monoclonal antibody (mAb) sandwich assays,
is indicated along the vertical axis. The level of glycan CA 19-9 on MUC5ac, determined using a mAb to capture MUC5ac and another
antibody to detect CA 19-9 on the captured protein, is indicated along the horizontal axis. b) Receiver-operator characteristic curve
analysis comparing the biomarker performance of core protein versus glycan detection. Each curve gives the sensitivity (rate of true
positive detection) and the specificity (rate of true negative detection) for discriminating cancer subjects from control subjects at various
thresholds of discrimination. “AUC” is area-under-the-curve, indicating the total discriminating ability of each marker. c) Cluster analysis.
The glycan measurements along the vertical axis were taken in the samples indicated along the horizontal axis; the color of each square
is the level of each measurement (see the color bar). The rows and columns were ordered (clustered) by similarity, showing consistently
increased levels in the cancer patients.
27