The director of neuroimmunology at Johns Hopkins shared his insight into the validation of neurofilament light as a biomarker for multiple sclerosis.
Peter A. Calabresi, MD
One of the hottest topics in multiple sclerosis (MS) has been the potential of a new biomarker, neurofilament light, to assess the efficacy of disease-modifying treatments. Although most of the conversation has centered around the possibilities that arise with it, it has yet to be fully clinical validated.
The research required to get that validation is ongoing, with many investigators, including Peter A. Calabresi, MD, the director of neuroimmunology and a professor of neurology at Johns Hopkins University. To share some of his experience with the biomarker, as well as what specifically the research has found, he sat with NeurologyLive in an interview.
Peter A. Calabresi, MD: Basically, we've been able to show that a level of 16 pg/mL in the blood or greater is associated with having worse outcomes for that patient. On an individual patient level, if you look at a negative predictive value, or positive predictive value, that there were significant associations with gadolinium-enhancing lesions, new T2 lesions, and as I mentioned, ultimately, changes on the disability scale or the proportion of patients who reach a certain level of disability.
In this case, we used the EDSS rating scale and looked at the number of patients who had gotten up to 3.5—it’s on a 1 to 10 scale, so these are mildly impaired patients, but it had some predictive value for telling you if they were going to get to that level. We also found that less than 8 pg/mL seems to be normal, and people in that category did very well for a period of time.
This is useful because MS is a heterogeneous disease and not everyone has the same outcomes. It's not like cancer, where if your doctor says you have cancer and you're going die in 6 or 12 months, you're going to bring out the big guns and treat it aggressively. In MS, we know that some people will end up in a wheelchair in 5 or 10 years and other people can go 40 years and you wouldn't even know they have MS by looking at them walking across the street. So when you're dealing with the disease like that, that presents in the 20s and 30s, you would really benefit from having some prognostic biomarkers to say either you're having damage in your nervous system and need to go on aggressive therapy and take some risk here, or no, maybe not, it looks like you had an inflammatory event but it really seems to cause a spike in your neurofilament so it's not causing tissue damage, maybe we can wait.
This is what we're thinking and the data, so far, suggests that if your levels are less than 8 pg/mL that there's probably not a lot of ongoing tissue damage and maybe you have a milder form of the disease. Whereas the people, as I said, with the more aggressive form, maybe they should be on what we call high efficacy but somewhat higher risk therapy.
To support that we had access to the clinical trial data from the phase 3 registration studies, and it appeared that the patients on interferon ß, which is the first-line drug, did have reductions in the level of neurofilament. We looked at the time course of that, and it showed that you can see the neurofilament levels drop down to a normal range in a significant proportion of patients. It wasn't a home run though, and we know that interferon ß is not a home run drug. It just reduces disease activity, so we're still left with some patients having residual elevation of their neurofilament.
The story was even better with our one of our high efficacy therapies called natalizumab, and in that case, we were left with only 4% of the patients having elevated neurofilament. That is consistent with what we believe is the case clinically with that drug, that it's a high efficacy drug and it seems to work well on clinical parameters in MRI. Now, this corroborates that it actually is protecting patients from the release of the neurofilament—or what we think is brain tissue damage. It's going to be useful, but we need more information.
For example, we'd love to know how frequently to use the test. In 1 of the trials, the samples were collected every 3 months over the course of the year and that did appear to enrich for people who are either doing really well or really badly, but you're also left with an intermediate group that sometimes is up and sometimes down. What do you do with those people? Every time you do a test, you get more information, then you have to wrestle with the fact that life isn't always black and white or as clean as we would like it to be.
The final issue with developing a biomarker that's clinically applicable is to have an assay that everyone can have access to, that’s relatively inexpensive and is reproducible across centers. What's happening right now with neurofilament is that there's 1 company that has the Simoa assay that will run these tests in-house, and that's very reproducible. They also sell what's called a “home-brew kit,” where you can make the assay yourself, and if you have a really sophisticated immunology lab, it turns out you could pretty much reproduce the in-house Quanterix data. But if you're new, or not paying attention to how the assay is run, there's potential for a little bit more noise.
One of the concerns, when new biomarkers are being developed, is that one group gets a different result from another, and then you don't know which is actually correct. Another one is just getting the result back quickly because you want a big company that could handle thousands of samples. There are hundreds of thousands of people with MS around the world, and so you really need a platform that's automated, not something where someone is pipetting samples into a dish and over the holidays they may not be there, so you don't get the result for two weeks. That's not going to work. One of the neat things that Biogen is doing, is they're partnering with Siemens, and are trying to develop a platform where this assay could be as sensitive as the present one, but available in a more automated fashion that could be scalable to provide a service, globally, so that everyone could be using the same test and interpret the results in the same way.
Transcript edited for clarity.