The assistant professor of neurology at the University of Pennsylvania provided insight on a new way to accurately predict chronic active lesion evolution from newly developing MS lesions using 7T MRI.
Chronic active lesions (CALs) develop as a subset of focal multiple sclerosis (MS) lesions that can be identified on MRI due to the presence of a hypointense rim on phase contact. These CALs can be associated with additional tissue damage and greater neurological disability. A newly developed predictive model may be able to identify which lesions will ultimately become CALs and thus better facilitate treatments that can target these clinically important lesions.
Senior author Matthew Schindler, MD, PhD, and colleagues conducted a study that included 14 patients with 60 incident lesions that were imaged using noncontract 7T MRIs. Incident lesion outcome was determined based on the presence of a hypointense rim observed on the phase contrast from a T2 weighted, multiecho gradient echo sequence acquired at 0.8 mm3 resolution. Quantitative qT1 maps were generated from the MP2RAGE sequence acquired at 0.7 mm3. Using a generalized linear mixed effect model fit with a subject-specific random intercept, investigators found that baseline qT1 within incident lesions was significantly different in lesions that evolved into CALs compared to those that did not (P = .03).
Schindler, who presented these findings at the Americas Committee for Treatment and Research in Multiple Sclerosis (ACTRIMS) Forum 2022, February 24-26, in West Palm Beach, Florida, sat down with NeurologyLive® to provide greater insight on the results. In a new iteration of NeuroVoices, the assistant professor of neurology at the University of Pennsylvania discussed the clinical importance of CALs to the MS community, how he went about developing this predictive model, and whether CALs are recognized for their clinical utility.
Matthew Schindler, MD, PhD: Chronic active lesions are an extremely emerging biomarker in multiple sclerosis. There’s a lot of research that’s been going into them for a number of reasons. A chronic active lesion is based on sort of a histopathological description of having activated microglia macrophages that are at the edge of some MS lesion. It’s a subset of MS lesions. In the last decade or so, we’ve developed MRI markers that have good radiological histological correlation with these CALs and paramagnetic rim lesions, or PRLs. A lot of work has gone into showing that patients with MS have the presence of these paramagnetic rim lesions and have more aggressive disease phenotypes. They, on average, have greater disability scores, and reach higher disability levels earlier than patients that do not have any evidence for PRLs in their MRI.
We’re interested in understanding how these lesions affect patient outcomes over time. It also seems to be quite specific to MS. If we look at other diseases, including other inflammatory diseases and mimics of MS, we don’t see these lesions on their MRIs. It may end up being a very good diagnostic biomarker. Much of the work is focused on after a CAL or PRL has formed, and those associations, but we don’t know about which lesions will form into them. That’s important because if we’re going to try to start to target treatments against this type of lesion or patient, particularly if they have more of these lesions, we want to be able to predict who’s going to develop them, and then target our treatments to that population. That way we’re avoiding exposing patients to medicines that wouldn’t necessarily target their lesions.
We didn’t know which lesions were going to form into PRLs or chronic active lesions. It was sort of starting a priori with knowledge that we don’t know what’s going to happen. We needed to collect a wide variety of patients and lesions, go back in time to look at those early markers, and say “All right, what was unique about this patient and this lesion that formed into a PRL or CAL?” We would extract the data from the MRI usually. Right now, we’re mostly looking at lesion-specific data. We’ll look at averages of signal intensity on various measures within the lesion and then use basic linear regression modeling to be able to identify predictors for that sort of outcome.
There are a lot of different ways you can do that. You can put in any sort of measure that you think is going to be important and then be able to text whether or not that actual marker has any predictive value for your outcome measure. The next steps as we get more lesions and more CALs is that you can then develop an automated method, something like machine learning or AI [artificial intelligence]. But usually, they perform best when you have hundreds of instances so that way you can sort of parse them out. Right now, with more of a linear regression model, we’re picking the predictors instead of just feeding in everything and having a computer tell us what’s important.
One of the challenges for us in the imaging world is how to bring these biomarkers to the clinic. Because most of the time, a lot of our markers are done with advanced imaging, maybe at 7T like we do, but that’s not something that most centers have. It becomes difficult to say, “what can we use with our 7T imaging to then help our clinical populations?” It’s not given recognition because there are just not as many centers that have that capability. One of the things we’re trying to focus on is how can we use clinical MRI to tell us the same thing that we’re seeing on our advanced imaging features. That’s where it becomes potentially interesting with things like machine learning, which are going to pick up on subtle changes within the tissue that you can’t see with your eyes, but might be there on the clinical MRI that we can then correlate with what we’re seeing on advanced imaging. Hopefully it will allow it to be a more widely applicable, imaging technique. That’s true for any sort of imaging technique, and the ultimate goal is how can you get it so that it can be used by everybody?
First of all, the fact that we have this meeting and we’re able to come together as an MS community is really great. It’s been a tough two years for everybody and the medical community in particular. To be able to come back in this setting and see all the amazing people and research that’s been undergoing for 2 years. It’s interesting, a lot of these meetings are once a year, so there are usually small, incremental changes. But now it’s been a couple of years and you start to see a larger, different things. People started to go into different areas, maybe they had more time to think or focus on an area. It’s been a good meeting for that.
What’s stuck out is the theme of the MS conference, which is biomarkers. There are so may different biomarkers in different areas and contexts. I’m sort of taken by the breadth of research that’s going into this. Now, I think the challenge is, how can we bring all these seemingly disparate areas of research together? Because they’ll probably be more powerful if you’re adding everything together instead of saying, oh this is the one biomarker that fits all. The next step is alright, we have these tools, they seem to work in these different situations, how can we bring them together and apply them across the board? Our patients will greatly benefit from that.
Transcript edited for clarity. For more NeuroVoices, click here.