Biomarkers: Foretelling the Future
American Academy of Neurology meeting report
SAN DIEGO—Predicting whether patients will progress to disability—and how they will respond to treatments—are among the greatest challenges for neurologists treating MS. At the annual meeting of the American Academy of Neurology last month in San Diego, California, many researchers presented work that will help refine those predictions.
“We have multiple medications for MS now, and it’s really not clear to physicians which drugs to use when,” said Howard Weiner, an MS researcher at Brigham and Women’s Hospital (BWH) and Harvard Medical School in Boston, Massachusetts. For example, a patient predicted to debilitate rapidly might be treated more aggressively than one who will maintain capabilities for years to come. But the current tools fall short of making that projection accurately.
“One of the next major frontiers in MS really is biomarkers,” Weiner said. Biomarkers, sought after in many diseases, are molecules or other signals that give clues about disease processes in living patients. The best biomarkers available today for MS, he said, are pathological features, such as active lesions, tracked by magnetic resonance imaging. MRI falls short, though, because it detects signals that don’t predict future disease course well; it’s also expensive and impractical for frequent clinical use.
Despite its shortcomings, MRI still provides a window into the diseased brain. Gioacchino Tedeschi and colleagues at the Second University of Naples, Italy, wanted to determine which parameters measured with MRI might best predict patients’ likelihood of progressing to disability. In 2002, the group conducted an MRI study of nearly 600 MS patients (Tedeschi et al., 2005). About half the patients from that initial study were available for the current investigation. Using MRI, the researchers examined a standard set of parameters—including gray and white matter volume, total brain volume, and lesion load—in each patient and compared them to baseline values recorded 9 years earlier. The researchers also collected details about patients’ age and disease progress, including their level of disability as measured by the Expanded Disability Status Scale (EDSS). Their analysis of 241 patients with relapsing-remitting MS (RRMS) revealed that gray matter atrophy and disability status at baseline were the best predictors of who would progress from RRMS to secondary progressive (SPMS) disease. Tedeschi summed up the findings: “The more [the patients] were atrophic in gray matter and the higher their EDSS at baseline, the more likely they were to convert to progressive disease.” White matter atrophy, in contrast, had little impact on disease progression.
Other groups attempted to foresee the future by tracking molecules found in patients’ blood. “I envision a day,” Weiner said, “when you can draw a blood sample from a patient and take [information from] that into account when deciding what medication to use.” Weiner hopes that biomarkers of MS progress might one day be tracked the way doctors use blood pressure measurements to adjust patients’ medications. Biomarkers from blood would be ideal, because blood samples can be easily collected, frozen and stored, and shared.
In a different approach, Tom Aune and his colleagues at Vanderbilt Medical Center in Nashville, Tennessee, used microarray technology to get a snapshot of MS patients’ gene-expression profiles at different disease stages. They first split their patient population into subgroups according to disease stage. Some subjects had had a clinical event but did not develop MS; another group showed early signs of MS and were later diagnosed with this condition; a third group had an initial diagnosis of MS but had not yet received treatment; and the remaining patients had had MS for a year or longer. After their initial microarray studies suggested that people’s gene-expression profiles indeed differed with disease stage, the researchers went on to look more closely at specific messenger RNAs (mRNAs) that encode cytokines and other immune-related proteins, reasoning that they would be the most likely candidates to undergo changes during disease progression. After isolating RNA fragments from patients’ blood samples, the researchers used those to synthesize complementary DNA—a more stable form of the code—and then used quantitative polymerase chain reaction to determine the levels of mRNAs found in blood.
“We saw that disease progression reveals a highly dynamic gene-expression signature,” Aune said. The expression of one cytokine in particular—interleukin 33 (IL-33)—changed profoundly with disease progression. Patients who eventually developed MS but were tested before their diagnosis had “extremely high levels of IL-33”—30 times higher than those seen in healthy control subjects—“that drifted downward as the disease progressed,” he said. In patients with either newly diagnosed or established MS, IL-33 expression remained elevated but to only about three times control levels. IL-33 levels were not higher than usual in patients with other neurological diseases or, importantly, in the patients who never progressed to MS. The researchers concluded that the increase in IL-33 might serve as an early “danger signal” to the immune system and possibly as a diagnostic indicator of MS. (Although the team did not measure protein levels directly, the increase in mRNA that they detected most likely indicates an upsurge in IL-33 protein.)
Biomarkers for MS might also help determine who might benefit from which medications. To address this question, Weiner and colleagues, including first author Francisco Quintana, used a tool called antigen microarrays, which Weiner called “a very well-developed way to study immune responses.” These arrays consist of 420 antigens, including viral- and brain-derived antigens, lipids, and other potentially autoantigenic human proteins, each dotted on a glass slide. Patients’ samples are then applied to the array; antibodies that recognize a specific antigen are “caught” and can be detected. (Because only a few drops of serum are required, banked blood can be tested.) The researchers aimed to determine whether the presence of certain antibodies identified patients who would respond to a particular drug. “We’re not looking for pathogenic antibodies but rather an immune signature,” Weiner said.
The study followed more than 200 patients taking glatiramer acetate (GA), a widely used immune-modifying medication for MS. The researchers collected blood samples and examined disease activity with MRI in the subjects before they began treatment with GA, and then again 6 to 12 and 18 to 24 months later. At that time, patients were classified by their response to this drug: Nonresponders were those who developed new lesions or pathology appearing on MRI, other symptoms of an attack, or worsening disability; whereas responders displayed no changes in these parameters over 2 years after starting treatment.
The researchers probed the antigen microarrays with the serum samples, revealing which antibodies were produced by the two types of patients at different times. Data from two-thirds of the patients’ arrays were used as a “training set,” and the remainder were used as a “validation set,” to determine whether the identified signature could indeed predict whether the patient was a responder or a nonresponder.
Humans make different isotypes of antibodies, the two most common being immunoglobulin G (IgG) and IgM. Two versions of the arrays were used: one designed to recognize IgG antibodies and the other to detect IgM. The antibody signatures found on the IgM arrays did not differ between future responders and nonresponders. When only the pretreatment training set arrays were examined, the IgG antibody signatures appeared to differentiate the patient types, but this pattern did not hold up with validation. That is, despite the apparent differences in the array data between the two patient types, those signatures were not strong enough to predict patients’ response to GA. When the scientists looked at IgG reactivity in the 6- to 12-month post-treatment period, however, “the training set indicated and the [validation] set confirmed that we were able to identify responders from nonresponders,” Weiner said. “The responders had a decrease in IgG reactivity to central nervous system antigens after therapy.” He went on to suggest that the findings could reflect higher levels of T-cell activity in nonresponders.
The study showed that serum biomarkers could be used to determine who was responding to therapy early in the treatment course, and that biomarkers might be able to accurately predict who will respond to treatments a priori, although these data fell short of that mark. “Although we found this result with GA,” Weiner said, “I think that we can expand this” technique to other drugs.
Some studies of biomarkers use more sophisticated cell assays. Although they may never be widely used in patient populations, Weiner said, they could be valuable for making new discoveries about the disease, for clinical trials, and for understanding drug effects. Another study presented by his group falls into this complex discovery effort.
This time, the researchers investigated another potential biomarker found in blood: microRNAs (miRNAs), which are tiny RNAs that don’t encode proteins; instead, they pair with sequences in mRNAs to regulate gene expression. miRNAs were previously studied in samples of T cells isolated from MS patients and in lesions themselves, but not from whole blood samples.
In this study, headed up by Roopali Gandhi of BWH and published last month (Gandhi et al., 2013), the researchers collected blood samples from 10 RRMS patients, nine SPMS patients, and nine healthy controls. They used a special array to profile the levels of 368 different miRNAs in these samples; from this “discovery set,” they then homed in on 23 specific miRNAs, some of which were observed to differ between the patient groups and others of which were simply selected as important immune-regulating miRNAs. Six of these miRNAs distinguished RRMS patients from controls, one differentiated SPMS patients from controls, and five others distinguished SPMS from RRMS. One miRNA in particular, called hsa-miR-92a-1, which regulates genes involved in cell division and signaling, differentiated all three patient groups from one another and was also associated with disability level and disease duration. Another miRNA, let-7, regulates genes involved in cell differentiation and T-cell activation and is also linked to neurodegeneration; it differentiated RRMS from SPMS. The study confirmed that miRNAs, which Weiner described as “sophisticated immune molecules that are stable in circulating blood,” can serve as accessible biomarkers to monitor MS.
How long will it be before biomarkers can be used to predict the disease course in individual patients? “We will ultimately get to that,” Weiner said. But just like clinical trials of new medications, reaching that goal will require studies with many patients from multiple clinical centers. Weiner has played a pivotal role in the CLIMB study, which continues to follow more than 2000 MS patients, collecting yearly MRI scans and blood samples. “I think the two keys to individualized therapy will be MRI and blood biomarkers,” Weiner said. Physicians and patients alike look forward to seeing those tools of the future materialize in the wake of more studies like those presented at this year’s meeting.
Key open questions
- Do MS patients at different stages of disease bear detectable markers present in blood that can differentiate patients or predict their long-term disease outcome?
- Will useful biomarkers consist of miRNA, mRNA, antibodies, or some other molecule?
- Might such biomarkers reveal new features about MS pathogenesis?
Image credit
Thumbnail image on landing page. “San Diego palm trees sidewalk piers” by Jon Sullivan, via Wikimedia Commons and Public Domain Images, public domain.
