Background. Acute cellular rejection (ACR) remains a major cause of morbidity after heart transplantation despite advances in immunosuppression. Whole genome transcriptomic profiling offers a systems-based, unbiased approach to elucidate the molecular mechanisms underlying ACR. However, noninvasive, longitudinal biomarker assessments capable of capturing the temporal dynamics of rejection biology remain scarce. Methods. RNA sequencing of peripheral blood from heart transplant recipients before, during, and after ACR was compared with nonrejection controls. Pathway analysis was conducted using differentially expressed genes (DEGs), and a machine learning approach was applied to assess gene-based prediction of ACR. Results. A total of 235 rejection-specific significant DEGs and 863 postrejection DEGs (false discovery rate < 0.05) were identified. During ACR, DEGs were enriched for T-cell activation/differentiation, apoptosis, and B-cell receptor signaling pathways. By combining the 2 sets of DEGs, a panel of 71 common genes was identified that reflected the significant, longitudinal transcriptomic dynamics of ACR. In an elastic net machine learning–based classifier, DYNLL1 and SERF2 were identified as ACR predictive genes, and achieved a cross-validated area under the receiver operating characteristic curve of 0.63. Conclusions. Peripheral blood transcriptomics identify dynamic temporal responses in ACR including T- and B-cell pathways with potential ACR predictive genes that warrant further investigation.
Ghazal et al. (Tue,) studied this question.