According to Geoffrey Cuvelier, MD, it can be difficult to diagnose chronic graft-versus-host disease (cGVHD) in children using the National Institutes of Health consensus criteria. Dr. Cuvelier and colleagues sought to identify novel biomarkers that could help identify cases of pediatric cGVHD. Using a combination of selected cellular and plasma biomarkers, the researchers developed a diagnostic classifier that had an area under the receiver operating characteristic curve (AUC) of 0.89.
Biomarkers for Pediatric Chronic GVHD
The ABLE study was published in Blood Advances and included 234 evaluable pediatric patients from 27 transplant centers who received hematopoietic cell transplantation. The researchers used mixed- and fixed-effect regression models on cGVHD onset blood samples to identify biomarkers that met the following criteria:
- Effect ratio ≥1.3 or ≤0.75
- AUC ≥0.60
- P-value <5.814×10-4 (Bonferroni correction) for mixed effect or <0.05 for fixed effect
The researchers then developed a machine learning-based diagnostic classifier that combined relevant cellular and plasma biomarkers with clinical factors. According to the report, the onset of cGVHD was predicted by decreases in regulatory natural killer cells, naïve CD4 helper T cells, and regulatory T cells, as well as increases in CXCL9, CXCL10, CXCL11, ST2, ICAM-1, and sCD13.
Time-based analyses determined that sCD13, ST2, and ICAM-1 differed according to the timing of cGVHD onset. Overall, the diagnostic classifier showed a positive predictive value of 82% and a negative predictive value of 80% for pediatric cGVHD diagnosis.
“Our polyomic approach to building a diagnostic classifier could help improve the diagnosis of cGVHD in children but requires validation in future prospective studies,” the authors concluded.