Advances in the Management of Graft Versus Host Disease - Episode 2

Challenges in Early Detection of GvHD

, , ,

Hannah Choe, MD; Nelson J. Chao, MD, MBA; and Sophie Paczesny, MD, PhD, comment on the importance of early detection of disease and the role of biomarker testing in GVHD.

Yi-Bin Chen, MD: A lot of research has gone into early detection of graft-vs-host disease [GVHD]. It’s difficult to detect graft-vs-host disease early because we’d all admit that the immune response takes weeks before we see the manifestation. Why is it difficult to detect it early?

Sophie Paczesny, MD, PhD: Hannah, do you want to take the lead on this?

Hannah Choe, MD: In terms of early detection of graft-vs-host disease, it’s a clinical diagnosis. There are biomarkers that Sophie can elaborate more on. We have biomarkers for both acute and chronic graft-vs-host disease that help us differentiate the risk of it. The actual diagnosis is hindered because even biopsies aren’t necessarily specific enough to help us diagnose it.

When we have the first initial presentation of acute graft-vs-host disease, we do it based on classical presentations, typically of the 3 main target organs that were mentioned: skin, which is usually the earliest manifestation of acute graft-vs-host disease, followed by the GI [gastrointestinal] tract and the liver, which would manifest as hyperbilirubinemia. There are risk scores and classifications using those 3 main target organs that allow us to determine the likelihood of response to therapy and predict nonrelapse mortality.

We have the revised Minnesota risk score to determine whether a patient has standard vs high risk at diagnosis, which will give us an idea of whether they’ll be responsive to corticosteroid therapy, vs the biomarker score with the MAGIC [Mount Sinai Acute GVHD International Consortium] score using ST2 and REG3-alpha. These can help us determine whether a patient should receive more aggressive therapy vs less aggressive therapy. The biomarkers and clinical presentations for risk assessments are able to be used in clinical trial design to help us risk stratify for eligibility purposes. It’s been used in the BMT CTN 1501 trial with sirolimus vs corticosteroids. It’s been used in an itacitinib frontline trial as a single agent studied through the MAGIC consortium. The biomarkers, at least in acute graft-vs-host disease, have become central and integrated into clinical trial design.

For chronic graft-vs-host disease, the risk assessment is similar in terms of clinical presentation. Instead of 3 target organs, we have 8 organ manifestations that we look at: skin, oral, eye, lung, liver, GI tract, genitourinary, and joints and fascia. A composite score was similarly created based off the NIH [National Institutes of Health] consensus criteria. That composite score will differentiate between mild, moderate, and severe and will help us differentiate whether the patient is more likely to have higher nonrelapse mortality. Biomarkers in that area, which Sophie will be able to expand on, are less integrated into clinical trials but are certainly a very interesting area of research right now.

Nelson J. Chao, MD, MBA: Before Sophie goes on, the key point is that skin rash, GI, diarrhea, and LFTs [liver function tests] happen all the time and aren’t necessarily rare to GVHD, which makes this diagnosis difficult. Because the last time I looked, most of our patients are on 21 drugs on average. They get ablative regimens or radiation. Every one of those drugs and prep regimens can cause GI symptoms. It can cause rash. You can get an infection that causes these things. The need of biomarkers is great. Because it’s a clinical diagnosis, it’s very difficult to know for sure whether it’s real GVHD or not.

Sophie Paczesny, MD, PhD: Yes. And a good marker is able to differentiate this. ST2, for instance, has been evaluated against a type of rash, drug rash, or for colitis, including CMV [cytomegalovirus] colitis and Clostridium colitis. We studied markers that are specific and sensitive not only in terms of markers themselves but also specific to GVHD. The markers you talked about were mostly predictive biomarkers. That’s the biomarker that’s measured at the beginning of GVHD, and that will tell you what the outcome will be after the treatment starts.

There are even more interesting biomarkers. These are starting and are lot of ways with the same cutoff for the biomarkers at time of diagnosis early on. With a risk biomarker, where you look before the GVHD occurs, and you find the cutoff for these, ST2 is probably one of the best markers in this field…that can tell you if you will develop actual GVHD. There has been a lot of progress in GVHD. The Joseph Pidala Blood paper that was published recently on BMT CTN 1501 showed that if you look at standard risk and usable for clinical characteristics and biomarkers, you could randomize for a corticosteroid-free treatment and achieve the same outcome. That has been a pretty neat study. Biomarkers have helped in this way. The next frontier is risk biomarkers. Can we use the biomarker to say that you will develop actual GVHD?

Chronic GVHD isn’t as well developed. We don’t have biomarkers as strong. One of the limitations has been the availability of samples at the right time. Most of the patients don’t have a sample at the time of diagnosis. There are some initiated from Stephanie Lee U54 consortium. NIH also has a longitudinal study, but they’re mostly rare. And even if you look at the sample biobank did by BMT CTN [Blood and Marrow Transplant Clinical Trials Network], they stopped with sampling at day 90. At diagnosis, we don’t have a lot of samples. We have some day 100 or day 90 samples, but it’s more difficult to find a risk biomarker than a diagnosis biomarker and then see if it’s eventually interesting as a risk biomarker. That being said, we’ve looked. It’s not published. We’ve looked at 1000 samples at day 90 and we think we have some interesting markers. We’ll submit soon.

Yi-Bin Chen, MD: We all look forward to being able to learn how to use these biomarkers, the ones in acute, which are already available commercially, and the ones in chronic, which we can hopefully develop.

Transcript Edited for Clarity