Some patients with multiple sclerosis have a relatively benign course of disease while others experience a more severe disease. "We can't tell the difference when they first present to us," says Philip De Jager, an associate professor at Harvard Medical School. "Clinically, it is impossible to say what course a patient will take. Similarly also for the drugs that we have, they work to some extent, but we can't predict who is going to respond to a given drug."
His lab, though, has recently found that transcriptional profiles from patients' peripheral blood indicate that there are two subsets of MS patients. As the researchers reported in Science Translational Medicine at the end of September, patients belonging to subtype A have increased expression levels of genes in the lymphocyte signaling pathway and are more likely to experience a disease relapse.
To tease those subgroups out, De Jager and his colleagues examined blood samples from about 360 patients treated with glatiramer acetate or interferon as well as untreated patients and generated transcriptome profiles.
Then, using an unsupervised clustering approach, the researchers tested a number of models. "The unsupervised approach basically just says, well, we're not making any assumptions as far as what's important in terms of which gene may be important. We're just going to let the data speak to us and to say what's the best model that fits the data," De Jager says.
For each of the treatment groups, the two-subset model fit the data the best. Further, the subsets found in the three groups were the same. "The same structure to the population is present whether you are untreated, treated on GA, or treated on interferon," he adds.
Of course, De Jager notes that there could be further subgroups of MS patients that weren't picked up due to the study's size.
One unanswered question, he adds, is whether or not patients who are subtype A stay in the group or can move between subtype A and subtype B. "That we don't know yet. I think it is probably going to be the latter because we have some early data that suggests that people can fluctuate, but I think that remains to be clearly demonstrated," he says.
Indeed, as part of the validation study the researchers are undertaking, they are including a longitudinal component to address that question. "So we will have more than one measure per person, so we can see for a large number of people: Do they tend to stay in same group or do they go back and forth?" he adds.
While not yet ready for the clinic, De Jager says that such a signature, in conjunction with other markers, may eventually help clinicians gauge disease activity. "If you saw that the patient consistently had an A profile, you may say well, the drug that they are on doesn't seem to be working well," he says. "They remain active, so let's change them to a different drug."