IMS 2026 | How may MRD depth and distinct clonotypic patterns below LOD predict relapse risk in myeloma?

This question really came out from patients, right? And we have, for people who use MRD at 10 to the minus 6 by NGS, we get a lot of these results that say detected, but it was below the limit of detection or below LOD. And that’s been a really frustrating result for clinicians, but especially for patients to understand, what do I do with that? Does that mean that it’s really positive? Is it negative? Should I ignore it? Am I going to relapse? So we really wanted to understand better how we can interpret those results, particularly because now that we’re talking about cure, I think it’s really important as we’re thinking about stopping therapy for patients that we really understand these very low-level MRD states and understand their significance as we’re counseling patients. So we looked at patients, a set of 430 patients at Mount Sinai who had been in an MRD negative remission lasting for at least two years. And we looked at subsequently what were their MRD results and how did that correspond to their risk of relapse to try to understand what low-level MRD results are important and which ones can we ignore and can think about cure. So the first thing that we found is that in general that, and this has been shown previously, that NGS testing is more sensitive than flow-based testing, and that those who are NGS negative in particular have longer remissions and are less likely to relapse than those who are just flow negative with an unknown NGS result. The second thing that we showed is that looking by NGS at those samples below LOD, below the limit of detection, there are actually two different states depending on what type of sequence was detected. And there are some patients who have a very highly sensitive sequence detected. Those highly sensitive sequences are almost always reflective of disease. So if a highly sensitive sequence is detected below LOD, that showed an intermediate risk of relapse. So those patients basically were in an intermediate state between MRD negative and positive, but still at a higher risk of relapse. And you can tell that actually from the standard MRD report that when they say below LOD, they will give a range that starts like greater than zero because they’re suggesting that the LOD is more likely to be non-zero. Whereas some other patients had a sample below LOD, but the only thing that was detected was a less sensitive sequence, less sensitive or less unique sequences, excuse me, could be part of disease, but they also could be part of normal background. And this will be seen on the MRD report that they’ll say it was below LOD with a range that includes zero because that’s a less unique sequence that might actually be part of background. And those patients, if it was just a less unique sequence, those patients actually did not have an increased relapse risk compared to negative. So I think that the takeaway for clinicians really is that the highly unique sequences below LOD, so again, the LOD report that will say the range is greater than zero, those have an intermediate risk of relapse, whereas those that are less unique, so a range that includes zero, those samples probably behave just as well as true NGS negative, and probably patients can be reassured in that setting that they don’t have a high risk of relapse. I think the operative question is, and you were getting at this before, is what is the right MRD depth, and how much is enough? And it’s probably always going to be true that the more depth you have, the better it is. And one day we’ll have commercially available 10 to the minus 7. That will be even better than 10 to the minus 6. So I think that always deeper is always going to be better. I think the question is using the tools that we have, how much depth is needed and how can we understand those different low levels? And again, I think understanding the different uniqueness of sequences and how those can really stratify patients who are below the limit of detection but might have different different outcomes I think is a useful tool for clinicians to help counsel patients.

This transcript is AI-generated. While we strive for accuracy, please verify this copy with the video.