https://orcid.org/0000-0002-9268-9655
Key Points
ST2, REG3α, and/or AREG at the time of acute GVHD diagnosis are excellent predictors of 12-month NRM
The best biomarker algorithm and threshold for risk stratification may depend on the target population
Graft-vs-host disease (GVHD) is a major cause of non-relapse mortality (NRM) following allogeneic hematopoietic cell transplant (HCT). Algorithms containing either the GI GVHD biomarker amphiregulin (AREG) or a combination of two GI GVHD biomarkers, (ST2+REG3α) when measured at GVHD diagnosis are validated predictors of NRM risk, but have never been assessed in the same patients using identical statistical methods. We measured serum concentrations of ST2, REG3, and AREG by ELISA at the time of GVHD diagnosis in 715 patients divided by date of transplant into training (2004-2015) and validation (2015-2017) cohorts. The training cohort (n=341) was used to develop algorithms for predicting probability of 12 month NRM that contained all possible combinations of 1-3 biomarkers and a threshold corresponding to the concordance probability was used to stratify patients for risk of NRM. Algorithms were compared to each other based on several metrics including the area under the receiver operating characteristics curve (AUC), proportion of patients correctly classified, sensitivity, and specificity using only the validation cohort (n=374). All algorithms were strong discriminators of 12 month NRM, whether or not patients were systemically treated (n=321). An algorithm containing only ST2+REG3α had the highest AUC (0.757), correctly classified the most patients (75%), and more accurately risk stratified those who developed Minnesota standard risk GVHD and for patients who received post-transplant cyclophosphamide-based prophylaxis. An algorithm containing only AREG more accurately risk stratified patients with Minnesota high risk GVHD. Combining ST2, REG3α, and AREG into a single algorithm did not improve performance.
