This story originally ran on May 13.
By Tony Fong
Using a 2D gel approach, researchers in Australia have identified a set of serum biomarkers that may help clinicians determine whether a patient with multiple sclerosis will respond to interferon-beta therapy.
In doing so, they optimized a DIGE protocol for plasma "to obtain cost-effective and high-resolution gels for effective spot comparison," the researchers said.
The work, published May 5 in PLoS One, also adds to a growing body of proteomics studies directed at developing companion diagnostics for drug efficacy.
The PLoS One study tackles a subtype of multiple sclerosis called relapsing-remitting multiple sclerosis, or RRMS, which affects up to 90 percent of patients in the initial stages of the disease. RRMS is typified by a cycle in which periods of remission are followed by periods of exacerbations or relapses, differing in their severity and duration.
About 65 percent of RRMS patients progress to the secondary progressive phase, or SPMS, in which neurologic decline progresses.
One treatment for RRMS and SPMS is interferon-beta, which mimics the natural interferon manufactured by the body in order to fight off disease. While interferon-beta has been shown to reduce the frequency of relapses, lesion load, and disability in RRMS and SPMS patients, up to one-third of patients do not respond to interferon-beta treatment.
According to the study's authors, among patients who are resistant to interferon-beta treatment, "a number" develop antibodies to the drug that prevent the protein from binding to the receptor. As a result, "it is important to identify people who do not respond clinically to [interferon-beta] promptly, so they can be treated with less immunogenic interferon-beta or alternate therapies at an early stage in the disease course," they added.
While other studies have indicated that RRMS patients who respond to interferon-beta treatment show a more inflammatory and less neurodegenerative disease at the start of treatment, compared to interferon-beta non-responders, no specific biomarkers exist that can differentiate the two groups of patients, the researchers said.
The aim of their research was to find such stratification biomarkers in order to improve patient outcomes, Jonathan Arthur, an associate professor of medicine at the Sydney Medical School, and the corresponding author on the PLoS One study, said in an e-mail.
"A biomarker could be used to create a test for either predicting or monitoring whether a patient was responding to treatment with interferon-beta, allowing those who are not responding to be moved to alternative treatments as soon as possible, and thus, hopefully, to have a better prognosis," Arthur said.
Using an approach that combined both discovery-driven and targeted work, the researchers from the University of Sydney identified three biomarkers as putative clinical response markers — apolipoprotein A1, alpha1-macroglobulin, and fibrinogen B.
Previous studies have used discovery proteomics and targeted proteomics to generate a multiple sclerosis-specific protein profile from cerebrospinal fluid. 2D DIGE has also been used to identify CSF markers in multiple sclerosis and to differentiate the CSF proteome of patients who experienced a clinically isolated syndrome and who later developed multiple sclerosis from those who had a CIS but did not develop multiple sclerosis.
But this study, the authors said, is the first to use both discovery and targeted proteomics to investigate clinical response to interferon-beta treatment for the disease.
"We hypothesized that proteins which differ in abundance between the plasma proteome of clinical responders and non-responders to [interferon-beta] could serve as clinical response markers for treatment with [interferon-beta]," they wrote.
[ pagebreak ]
One key step they included in their work, they said, was the inclusion of a power calculation in the discovery phase in order to determine the correct sample size needed to detect biological changes.
According to Arthur, "Power calculations allow us to determine the ability of an experiment to detect changes in protein expression large enough to be deemed interesting with a reasonable probability. … If we had not done the power calculation, we would have no way of estimating whether what we found was likely to be the full set of differentially expressed proteins or whether a larger number of samples would be likely to find more proteins."
The calculation method they used is based on a protocol developed by Sybille Hunt formerly of Proteome Systems, now at Sapphire Bioscience, which "estimates biological and technical variance between the groups using a mixed effects linear model involving a fixed term for the difference between groups and random deviations representing differences between samples and gels," the authors wrote.
The "power analysis tool" was used to detect the minimal detectable difference, "defined as the size of effect required to give a chosen statistical power at a specific significance level," they wrote.
They selected 100 spots matched across all six gels used in their experiment. The spots were randomly selected across the entire gel to "remove any bias toward a particular region of the gel," the researchers said.
Using the calculation tool, the researchers determined that a minimum of 10 biological variants from each patient group were necessary to be confident "of a one-and-a-half-fold (50 percent effect size) and a two-fold (100 percent effect size) change in abundance" between clinical responders and clinical non-responders at their chosen levels of statistical significance.
The main goal of the discovery stage was to detect and identify differentially abundant proteins from their samples of blood from three responders to interferon-beta treatment, and blood from three non-responders to interferon-beta treatment.
Using the DeCyder biological variation analysis module from GE Healthcare, they matched spot maps and performed a statistical analysis to identify differentially expressed spots "based on the standard abundance value for each spot."
They identified three spots, A2M, ApoA1, and FIBB, using MALDI-TOF mass spectrometry, and confirmed them using MS/MS matching three peptides for each protein.
While other research has at least suggested A2M and ApoA1 as potential clinical biomarkers for interferon-beta treatment, "to our knowledge, there is no known [prior] association between FIBB and [interferon-beta] treatment or FIBB and [multiple sclerosis]," the researchers said in their study.
In their targeted proteomics work, the authors sought to identify lower-abundance clinical response proteins that discovery-based approaches such as 2D-DIGE are unable to detect, and in particular hypothesized that biomarkers previously associated with interferon-beta treatment, such as chemokines and cytokines, would be markers of clinical response.
In previous studies, IL-8, IP-10, MCP-1, MIP-1a, and MIG levels were found to be differentially expressed in sera from patients who were treated with interferon-beta, compared to those who weren't, and were chosen by the Australian researchers for their target analysis. In addition, they chose eotaxin. Changes in the serum cytokine are not known to be associated with interferon-beta treatment, but eotaxin "is regulated by Th2 cell-regulated cytokines, which play a significant role in the [interferon-beta] treatment mechanism." It also is found to be reduced in the CSF of people with multiple sclerosis, compared to those without.
The research team also included Th2-induced IL-6 as the only anti-inflammatory cytokine in the study, because treatment with interferon-beta increases its levels in the serum of patients with multiple sclerosis.
But using Cytometric Bead Arrays from Becton Dickinson, the researchers found no significant differences in the serum concentration of the seven proteins in their analysis.
[ pagebreak ]
Both eotaxin and IL-6 levels, however, were high in the serum of clinical responders compared to non-responders — though not statistically significant because some samples were below the levels of detection — and require continued research as potential clinical biomarkers, the researchers said.
Arthur said that he and his colleagues have not generated any additional data on eotaxin or IL-6 since submitting the paper.
The research, the authors noted, could be just the tip of the iceberg and larger samples would be needed to "fully characterize clinical response markers using a proteomic approach. Furthermore, heterogeneity in multiple sclerosis itself, as well as the approaches to treatment with [interferon-beta], also suggests the need for larger sample sizes to help eliminate other sources of biological variability not directly related to the effect of [interferon-beta] treatment."
From Disease to Drug Efficacy
The work by Arthur and his colleagues is another example of the use of proteomics in the personalized medicine arena. While the science has been traditionally used for the detection of diseases, proteomics is now branching out and being explored as a technology to evaluate drug efficacy.
As part of the I-SPY2 screening trial to develop new drugs to treat high-risk, fast-growing breast cancers, proteomic technology developed by Lance Liotta and Emanuel Petricoin at George Mason University, will be used to help discover and identify new biomarkers to evaluate new investigational drugs (PM 03/26/10).
Separately, the reverse-phase protein microarray technology from Liotta and Petricoin is also being used in a trial testing Gleevec for colorectal cancer (PM 01/22/10), and 13 FDA approved targeted cancer therapies for breast cancer (PM 03/12/10).
Also, researchers from the University of Oxford and the MD Anderson Cancer Center in March demonstrated that a protein biomarker could predict which patients responded to a new class of cancer drug, called HDAC inhibitors, for a rare kind of non-Hodgkin lymphoma (PM 03/26/10).
In his e-mail Arthur said that while more attention has been paid to DNA-based companion diagnostics, "proteins are responsible for much of the biological function, so proteomic approaches bring us extra information about the molecular mechanism driving response, or lack of response to a particular treatment."
Together, DNA and proteins "give two inter-related sets of information about the biological system under study," he added.