An early-biomarker algorithm predicts lethal graft-versus-host disease and survival

Area of Science:

• Hematopoietic Cellular Transplantation (HCT)
• Immunology
• Biomarker Discovery

Background:

• Predicting nonrelapse mortality (NRM) and severe graft-versus-host disease (GVHD) post-hematopoietic cellular transplantation (HCT) remains a clinical challenge.
• Current methods lack predictive power before GVHD symptom onset.

Purpose of the Study:

• To develop and validate a predictive model for NRM and severe GVHD using early post-HCT biomarkers.
• To identify distinct patient risk groups for targeted interventions.

Main Methods:

• A multicenter cohort of 1,287 patients undergoing HCT was utilized.
• Day 7 post-HCT blood samples were analyzed for four biomarkers: ST2, REG3α, TNFR1, and IL-2Rα.
• A predictive algorithm for 6-month NRM was developed using a training set (n=620) and validated in independent test (n=309) and validation (n=358) sets.

Main Results:

• A two-biomarker model (ST2 and REG3α) effectively stratified patients into high-risk (28% NRM) and low-risk (7% NRM) groups at 6 months post-HCT (P < 0.001).
• This model demonstrated consistent performance across training, test, and validation sets.
• High-risk patients exhibited significantly higher GVHD-related mortality and severe gastrointestinal GVHD.

Conclusions:

• An algorithm utilizing ST2 and REG3α concentrations from a day 7 post-HCT blood sample can reliably identify patients at high risk for lethal GVHD and NRM.
• This predictive capability allows for early risk stratification and potential therapeutic adjustments.