NEWPORT BEACH, CALIF. – Researchers say they have identified biomarkers that may help guide early treatment decisions in patients with acute graft-versus-host disease (GVHD).
The biomarkers, ST2 and REG3-alpha, were measured during the first month of GVHD treatment and proved more accurate than clinical response for predicting 6-month nonrelapse mortality (NRM). In fact, biomarker assessment revealed patients who responded to treatment but had a high risk of NRM and nonresponders who had a low risk of NRM.
The researchers also found that biomarkers changed over the first month of treatment but remained significant predictors of NRM. This suggests that modifying treatment according to biomarker findings at various time points could result in better outcomes for patients.
“We think this is going to transform the way we treat graft-versus-host disease,” said, of the Icahn School of Medicine at Mount Sinai, New York.
Dr. Ferrara and Hrishikesh Srinagesh, along with their colleagues at Mount Sinai, have conducted extensive research with these biomarkers and presented some of their findings at the Acute Leukemia Forum of Hemedicus.
Comparing biomarkers and response
In one study, the researchers evaluated 355 patients who had undergone allogeneic hematopoietic stem cell transplant at 1 of 20 Mount Sinai Acute GVHD International Consortium (MAGIC) centers between January 2016 and February 2018. All patients developed acute GVHD and received systemic steroids as treatment.
Patients provided blood samples weekly for the first month of treatment, and concentrations of ST2 and REG3-alpha were measured in each sample. Both biomarker concentrations were used to calculate the biomarker probability of NRM.
“The concentration of those two biomarkers are put into a computer, and we get … a single number, and that gives us the probability of mortality,” Dr. Ferrara said. “[W]e call this the MAGIC algorithm probability, or MAP. And when a MAP is low, the patient has a very low chance of dying from graft-versus-host disease, when it’s intermediate, they have an intermediate risk, and when it’s high, they have a high risk.”
The researchers then compared the MAP and clinical response for their ability to predict 6-month NRM throughout the first month of therapy for acute GVHD.
MAP bests response
After 1 month of therapy, the MAP was more accurate than clinical response for predicting 6-month NRM. The area under the curve was 0.84 and 0.65, respectively (P less than .001).
Likewise, the MAP after 1 week of therapy was more accurate than clinical response at 1 month for predicting 6-month NRM. The area under the curve was 0.80 and 0.65, respectively (P less than .001).
“[T]he clinical responses were good, but not great, at predicting long-term outcome, where the biomarker, the MAP, was significantly better,” Dr. Ferrara said. “[A]t every time point we tested, the biomarkers were better than the clinical responses.”
The researchers also identified subgroups of clinical responders and nonresponders for whom MAP more accurately predicted 6-month NRM.
The team found that 61% of clinical nonresponders were actually low risk according to MAP. And the incidence of 6-month NRM was significantly lower in the MAP-designated low-risk patients than in MAP-designated high-risk patients – 22% and 56%, respectively (P less than .001).
On the other hand, 10% of clinical responders were high risk according to MAP. The incidence of 6-month NRM was significantly higher in the high-risk patients than in the low-risk patients – 40% and 13%, respectively (P less than .001).