Glycoforms of the protein haptoglobin could aid in the early detection of pancreatic cancer, a study by researchers at the University of Michigan Comprehensive Cancer Center reported.
Focusing specifically on fucosylation, one of the forms of glycosylation commonly associated with cancer, the researchers developed a mass spectrometric blood serum assay for analyzing haptoglobin fucosylation patterns that distinguished between pancreatic cancer cases and healthy controls with a sensitivity of 94 percent and a specificity of 100 percent.
"We are developing a discovery path to look for the glycoproteins that change during cancer, and it was known from other studies by our group and other groups that fucosylation is often a good marker for these changes," David Lubman, an author on the paper and a researcher at the cancer center, told ProteoMonitor.
In addition to identifying patterns of fucosylation that distinguished cancers from non-cancers, the researchers also identified a glycan structure that had not been seen in pancreatic cancer before, a bifucosylated triantennary structure. This, as well as another bifucosylated structure, was present only in cancer samples and not in any of the controls, suggesting the two could serve as markers to distinguish early-stage pancreatic cancer from benign conditions.
"The problem in pancreatic cancer is we want to detect earlier stages where they can actually treat the cancer or do surgery… Once you get to later stages it's much harder to treat," Lubman said. And it is especially hard to distinguish early-stage pancreatic cancer from other conditions like diabetes or chronic pancreatitis, he added.
In the study, published in March in the Journal of Proteome Research, Lubman and his colleagues analyzed 31 samples: 15 with benign conditions (normal serum, chronic pancreatitis, and type II diabetes), and 16 with pancreatic cancer ranging from stage IA to stage IV.
Their process included isolating haptoglobin from the blood serum samples, removing the glycans, or sugar molecules, from the haptoglobin, concentrating and processing the glycans, and cataloguing and analyzing the fucosylation patterns those isolated glycans represent using Shimadzu Biotech's Axima MALDI-QIT, a MALDI-QIT-TOF MS instrument originally developed in the team's lab in 1992.
According to Lubman, mass spectrometry offered a more precise view of the glycans' individual structural changes than would be possible with other tools.
"Mass spec is a very accurate way of looking at these glycan units," Lubman said. "We can use antibody or lectin arrays of various sorts, but then you look at total fucosylation content, and in this case we wanted to very accurately look at changes in structure — changes in individual glycan peaks that would be difficult with other technology."
Of eight glycan structures identified, fucosylated triantennary and tetra-antennary structures were the best at distinguishing pancreatic cancer from non-cancers, the study found. The team also found that bifucosylated triantennary and tetra-antennary glycans were present in 12 out of 16 cancer samples and none of the non-cancer samples, highlighting these structures in particular as promising cancer markers.
Comparing the relative abundance of these glycans in cancer and non-cancer, the researchers created an index that allowed them to distinguish cancer or non-cancer samples by their degree of fucosylation.
The fucosylation index correctly identified every non-cancer sample as well as cancer samples at stage IIB and above. It missed one out of four of the early-stage samples. Overall, the sensitivity of the index as a predictor for cancer was evaluated at 94 percent, and its specificity at 100 percent. "Out of 31 samples we looked at, there was only one early stage cancer that had a slight overlap with the diabetes set," Lubman said.
Lubman cautioned however, that these degrees represent only the discovery set. He and his colleagues next plan to test their method on a blinded set of 20 samples from the University of Pittsburgh. "We found that we could characterize all the cancers correctly, but they might throw in some other control groups that might confound the disease we are looking at — for example, something like acute pancreatitis," Lubman said.
He also said they will have to study larger numbers of samples to refine and validate these fucosylation patterns as a diagnostic marker. Unfortunately, samples, especially of early-stage pancreatic cancer, are rare Lubman said, because the disease is so often diagnosed in its later stages.
According to Lubman, when the test is up against a blinded set its performance may go down, so his team is also looking for additional markers based on other changes in glycosylation that could be included to further refine the test.
"We may have several markers for pancreatic cancer that will work by the time we are done and we have a couple in the works that are promising," he said.
He said the team is also looking at fucosylation changes in haptoglobin and other molecules in the case of other cancers, and has already identified a potential marker for hepatocellular carcinoma.
With other cancers that are more frequently diagnosed than pancreatic cancer, there are better data and samples available, according to Lubman, and so the process of creating a commercial test might progress more quickly.
"With HCC liver cancer, there is a standard blinded set to test our markers against," he said. This process is still at an early stage, but he called the research "very promising" and noted the possibility of a second stage of validation with samples from the National Cancer Institute's Early Detection Research Network.
Lubman added that because pancreatic cancer is so rare, a fucosylation test for the disease would not, if fully developed, be used to screen the general public.
"What you'd really be focused on are people with family histories, people who are obese, heavy smokers, people who have long-term diabetes, especially type II. Those are some of the main categories," he said.
Have topics you'd like to see covered in ProteoMonitor? Contact the editor at mashford [at] genomeweb [.] com.