Donor whole blood DNA methylation is not a strong predictor of acute graft versus host disease in unrelated donor allogeneic haematopoietic cell transplantation

Allogeneic hematopoietic cell transplantation (HCT) is used to treat many blood-based disorders and malignancies. While this is an effective treatment, it can result in serious adverse events, such as the development of acute graft-versus-host disease (aGVHD). This study aimed to develop a donor-specific epigenetic classifier that could be used in donor selection in HCT to reduce the incidence of aGVHD. The discovery cohort of the study consisted of 288 donors from a population receiving HLA-A, -B, -C and -DRB1 matched unrelated donor HCT with T cell replete peripheral blood stem cell grafts for treatment of acute leukaemia or myelodysplastic syndromes after myeloablative conditioning. Donors were selected based on recipient aGVHD outcome; this cohort consisted of 144 cases with aGVHD grades III-IV and 144 controls with no aGVHD that survived at least 100 days post-HCT matched for sex, age, disease and GVHD prophylaxis. Genome-wide DNA methylation was assessed using the Infinium Methylation EPIC BeadChip (Illumina), measuring CpG methylation at >850,000 sites across the genome. Following quality control, pre-processing and exploratory analyses, we applied a machine learning algorithm (Random Forest) to identify CpG sites predictive of aGVHD. Receiver operating characteristic (ROC) curve analysis of these sites resulted in a classifier with an encouraging area under the ROC curve (AUC) of 0.91. To test this classifier, we used an independent validation cohort (n=288) selected using the same criteria as the discovery cohort. Different attempts to validate the classifier using the independent validation cohort failed with the AUC falling to 0.51. These results indicate that donor DNA methylation may not be a suitable predictor of aGVHD in an HCT setting involving unrelated donors, despite the initial promising results in the discovery cohort. Our work highlights the importance of independent validation of machine learning classifiers, particularly when developing classifiers intended for clinical use