Foreseeing the Future?
Molecular signature might predict likelihood of relapse
Two women age 30 with a loss of vision in one eye can each learn that they have MS in back-to-back appointments, yet their doctor will be unable to predict which one will dodge subsequent disability and which one will be using a wheelchair within 10 years. Tools to predict an individual MS patient’s disease course or help select her optimal therapy do not exist, despite a major ongoing search for relevant biomarkers.
Now, in one of the largest studies of its kind, researchers have used blood-borne molecules to distinguish among early-stage patients who were subsequently more or less likely to relapse. The work, published September 26 in Science Translational Medicine, raises hopes that researchers will one day identify helpful biomarkers for MS (Ottoboni et al., 2012). The findings could help guide treatment if they are replicated independently and validated for clinical use.
"It's very exciting," says Timothy Coetzee, chief research officer of the National Multiple Sclerosis Society, which helped fund the study. "This begins to take us down the road of being able to understand the molecular architecture of multiple sclerosis on the immune system side and to use that information to … develop tools for predicting how a person's disease course will go and what treatment might work. We are beginning to see a vision of personalized and targeted treatment for MS."
The new analysis began with frozen samples of immune cells from 363 participants in the Comprehensive Longitudinal Investigation of MS at Brigham and Women's Hospital, a long-term study to understand how MS begins and progresses. Neurologist Philip De Jager of Brigham and Women's Hospital in Boston and his colleagues selected patients that were either untreated, treated with glatiramer acetate (GA), or treated with any of the four existing versions of interferon β (IFN-β; see interferon beta-1a and interferon beta-1b). The immune cells came from standard peripheral blood samples, often from a patient's arm, which are easy to obtain. These cells are clinically relevant because they storm the central nervous system during MS attacks and play a key role in the destructive inflammation that marks the disease.
Beginning without any preconceived notions about the potential disease mechanism, the investigators used mRNA to measure expression of about 20,000 genes. This assessment showed distinctive transcriptional changes in patients treated with IFN-β compared with those who were untreated, but almost no changes in GA-treated patients compared to untreated patients. At first glance, this observation might seem surprising, but the authors offer several possible explanations: The perturbations might be too subtle to show up in these experiments, for instance, or they might not occur in the cells studied.
For each of the next phases of the study, the researchers made the initial observation in the data from the untreated patients and validated it in the treated sets, a method that increases the reliability of the results, according to Seth Blackshaw, a developmental neurobiologist at Johns Hopkins University School of Medicine in Baltimore, Maryland, who has expertise in microarray study design and who was not involved in the study. In a technique known as unsupervised clustering, the researchers asked the computer to mathematically place the patients—separately for each group—into subsets based on their gene-expression similarities and differences. The computer spit out several possibilities, but the data fit best into two subsets.
The team named the subsets MSa and MSb and winnowed the distinguishing transcriptional signature down to 98 genes. Many of the expression differences mapped to immune cell–signaling pathways, which were apparently much more active in MSa.
The same divide showed up in the untreated and the GA and IFN-β groups. It is striking that the signature persists “even in the presence of [interferon] therapy that elicits change in the activity of thousands of genes,” says Sergio Baranzini, a geneticist at the University of California, San Francisco, who was not involved in the work.
Once the computer did the sorting, the researchers explored whether patients with MSa and MSb gene activity patterns showed any clinical differences. Turning to the medical records of the treated patients, they looked for evidence of subsequent disease activity, defined by a clinical relapse, new lesions visible on an MRI, or a sustained increase in the Expanded Disability Status Scale. This clinical information covered 3 years for most individuals. Data for the untreated patients were not analyzed because most of those people began therapy soon after they provided the blood samples used for this study.
The two treatment groups did not significantly differ in their relapse rates, but treated patients with the MSa profile were about 40% more likely to relapse than were treated patients with the MSb profile. Put another way, by the end of year two after providing the blood sample, roughly 45% of MSb subjects remained relapse-free in contrast with about 30% of MSa subjects, De Jager says.
"What you have now is a cross-sectional study looking at all the people at one time," says Thomas Aune, an immunologist at Vanderbilt University Medical Center in Nashville, Tennessee, who measures mRNA activity to develop diagnostic tests for MS and other autoimmune diseases. "What you need next is a longitudinal study to see how the signatures change over time." Such information could reveal whether the MSa or MSb signature stays constant in a given individual or whether a person converts back and forth, with MSa predicting an imminent relapse.
Richard Rudick, a neurologist at The Cleveland Clinic in Ohio, underscores the potential clinical utility of a biomarker that predicts degree of disease activity if it is more accurate and precise than the current MRI imaging biomarkers. "Now we treat everyone the same with chronic disease-modifying therapy, even people with mild disease who might not need it," he says. A milder prognosis might need watchful waiting, he speculates, whereas a more severe prognosis might justify powerful medications that also carry a risk of side effects.
With such ideas in mind, De Jager and his colleagues are further analyzing the results to determine whether patients’ signatures persist or alternate between the two states. The group has “early evidence for the second case, … but we have to confirm the original observation first," De Jager says.
The researchers are beginning a larger study to test whether they can replicate their findings. Of the numerous proposed biomarkers of MS activity suggested by more than 100 studies, none have been validated for clinical use, according to a review from last year (Graber and Dhib-Jalbut, 2011).
De Jager also wants to probe what gives rise to the gene-activity differences. "It could be more of a certain cell type, such as a particular kind of lymphocyte, or it could be that the lymphocytes are more activated," he says.
The study is “carefully done and well controlled,” Blackshaw says. The large number of patients and unsupervised clustering contribute to its strength, he says, as does the rigor with which the researchers collected and analyzed the data.
Blackshaw, De Jager, and others emphasize the preliminary nature of the results, which need verification by other research groups in different MS populations. "The main challenge is to create a reliable source of data from independent groups," says Baranzini, who several years ago published a similar smaller study that found (in one cell type) a distinctive molecular signature that correlates with development of MS after clinically isolated syndrome (Corvol et al., 2008). The MSa signature is consistent with the earlier signature associated with more severe disease, De Jager says, but he and his colleagues are analyzing the details and will report them in a subsequent paper. Additional large studies are needed, he says, and such projects are under way.
Key open questions
- Will the two distinct molecular types predict disease activity in larger validating studies?
- Over time, does an individual patient fluctuate between the two states or remain in a single state as the disease relapses and remits?