University of Michigan researchers have used a proteomics approach to identify at least one protein marker that could help identify and risk-stratify patients suffering from gastrointestinal graft-versus-host disease, an adverse reaction to hematopoietic stem cell transplant.
Starting in early 2012, the group plans to offer clinical testing for the protein REG3-alpha, which they were able to validate in a group of more than 1,000 samples, reporting in the journal Blood this month that concentrations of the single protein were predictive of therapy response, one-year non-relapse mortality, and one-year survival.
Work on several other proteins identified in their discovery study has been "promising," and the team also hopes one or several of the markers can also be used to predict the reaction before symptoms begin, said James Ferrara, one of the study's lead authors.
"Having markers that are both diagnostic, and particularly if they have prognostic significance, would allow us to risk-stratify our patients in a much better way," Ferrara told ProteoMonitor this week.
"At the moment about half the patients that get gastrointestinal GVHD die in the first year," he said, explaining that clinical evaluations, essentially based on the severity of symptoms and biopsy, are the only tools currently used to identify the disease and stratify risk by clinical stage.
Before the team's study, "there were no biomarkers yet, none for GI GVHD," he said.
Because the treatment for GVHD — high-dose steroids — can cause serious side effects, clinicians wait to give more intensive therapy. But this delay can lead to worse outcomes and death, Ferrara said.
Biopsy also takes a relatively long time to interpret accurately, he added. And even then there is "disagreement among pathologists as to the key features in trying to grade the severity of the disease and what that means."
"If we knew [at the outset] the ones who wouldn't respond to steroids, we could add more immunosuppressant therapy right away," he said. "But because there are lots of infectious risks in doing that, we tend to wait to see what happens before we give them more intensive therapy, but often it's too late.
"Having something that could both identify disease and tell us whether it's likely to be bad or not would greatly improve treatment," he said.
According to the group's study, REG3-alpha offers this diagnostic and prognostic information. In their report, they noted that the protein distinguished GVHD from non-GVHD with an area under the receiving operator curve of 0.80.
Ferrara said the group was lucky to have access to a repository of samples taken throughout the course of GVHD in a large number of patients.
"For ten years we have collected samples from patients on a regular basis after [hematopoietic stem cell] transplant, not only at weekly intervals in the first month when a lot of these complications arise, but also at the onset of new symptoms …That was really key," he said.
He added that the team now has about 15,000 samples from 1,500 patients.
Under the direction of Samir Hanash — now head of the molecular diagnostics program at the Fred Hutchinson Cancer Research Center, but a Michigan colleague at the time of the study — the team evaluated 20 total samples via tandem mass spec; ten samples from patients who never developed GI GVHD and ten from patients at the onset of the disease.
This analysis identified 562 proteins, of which 74 were increased at least two-fold in patients with GVHD. Five of these were preferentially expressed in the GI tract based on a review of previous literature, and one — REG3-alpha — had commercially available antibodies sufficient to quantify plasma concentrations by ELISA in the validation cohort of 1,014 patient samples.
About 850 patients in the validation cohort were from the University of Michigan sample group, and another independent set of 150 were from the group's collaborators in Regensburg, Germany, and Kyushu, Japan, Ferrara said.
The group measured REG3-alpha concentrations in plasma from these patients and found that concentrations were three times higher in patients at the onset of GI GVHD than in all other patients, including those with enteritis not associated with GVHD.
The group also evaluated several previously reported diagnostic markers of systemic GVHD – IL2Ra, TNFR1, IL-8 and HGF – but these did not add to the AUC of REG3-alpha alone in distinguishing between GVHD and non-GVHD.
They also investigated the prognostic significance of the marker, Ferrara said, finding that concentrations were three-fold higher at the time of diagnosis in patients who would go on to have no response to therapy at four weeks than in those who experienced either full or partial response to treatment.
Non-relapse mortality was also twice as high in patients with high REG3-alpha concentration, the group reported.
Interestingly, Ferrara said, the identification of REG3-alpha in this disease opens up some surprising new questions about how GVHD works. "This was like the tip of a wonderful scientific iceberg we didn't know was going to be there," he said.
The researchers hypothesized in their study, based on recent cell biology research, that GVHD causes microscopic breaches in the mucosal epithelia barrier of the GI tract, which allow REG3-alpha—a natural antibiotic normally produced near stem cells and secreted into the mucous layer — to move into the bloodstream — a mechanism that would explain the marker's presence and the association of high concentrations with the disease.
"So now there are going to be a whole set of new experiments designed around the GI microbiome, REG3-alpha, and other natural antibiotics like defensins, the adaptive immune system, T-cell activation, antigen presentation, cytokine release, and the intestinal stem cells," Ferarra said.
Meanwhile, Ferrara said the team is now working to validate four other proteins that were identified in the discovery study. "The ratio for REG3-alpha was, I think, something like 2.5, but there were others up to four or seven, so we're working to make some antibodies to look at these."
"I think it's going to wind up in the end to be two or three or four biomarkers, and this is just the beginning of this process," he said.
Focusing on REG3-alpha, the researchers are also planning to test if the marker can predict the onset of GVHD before a diagnosis via biopsy. They are also interested in investigating whether the marker can be used to measure disease activity and progression.
"If we could use the biomarker as an index of disease activity, that would also help us," Ferrara said.
"This has to be repeated," he cautioned, "but my guess is, though we are very careful to say that it doesn’t replace biopsies, once this becomes clinically available – and it will become clinically available here in Michigan starting in the spring of 2012 – it may start to reduce the need for biopsies."
With the goal of offering clinical testing for REG3-alpha concentrations next year, the team has been working with colleagues at Michigan's clinical chemistry department to find the best way to do tests on a routine basis, Ferrara said.
"In addition to REG3-alpha, we now have three biomarkers that we are looking at, and maybe a fourth, that we think give us a pretty good idea of both prognosis as well as perhaps prediction. This is preliminary data, but it looked pretty strong," he said.
"In doing these panels of biomarkers [labs] like to use robots to do multiple ELISAs, and the robots like to have the same incubation times for each of the assays … but for optimal sensitivity, some of the assays require longer incubations, so we're actually kind of troubleshooting that at the moment," he said.
One approach, he said, would be to "try to go after some more biomarkers that are not cross-reactive, so to speak."
Adding previously identified markers to REG3-alpha in the study didn’t raise the sensitivity or specificity of the single protein, Ferrara reiterated. But the researchers would still like to achieve a higher AUC if possible, hopefully with new markers identified in their continuing research.
"If we had a 0.85 or 0.9, that would be much better, and that's what we're looking for," he said.
The group has a patent pending on the work, and is currently looking for a licensing partner to help commercialize the assay, the university said.
Have topics you'd like to see covered in ProteoMonitor? Contact the editor at abonislawski [at] genomeweb [.] com.