Protein Expression Profiling: At the Crossroads Between AMR and ACR?

Heart transplantation is still the only treatment markedly changing the disease course in individuals with end-stage heart failure.1 However, despite an increase in the use of molecular methodologies to detect cardiac allograft rejection, about 40% of adult heart transplant recipients are hospitalized for acute rejection within the first 5 y posttransplant. In addition, within 10 y posttransplant, nearly 50% develop coronary artery vasculopathy, a manifestation of chronic rejection.2 As such, monitoring and diagnosing rejection of cardiac allografts remains challenging: despite being the “gold standard,” endomyocardial biopsy (EMB) has shown several limitations due to its invasiveness and interobserver variability.3 To overcome EMB limitations, several noninvasive methods based on molecular biology assays from peripheral blood have been developed.4-6 However, none of these methods have yet achieved enough validation to be fully recommended in clinical practice. In this issue of Transplantation, Goldberg et al7 presented early results from the Genomic Research Alliance for Transplantation study, which was designed to investigate if protein expression profiling (PEP) has the potential to delineate specific alloimmune responses and cardiovascular injury phenotypes during rejection—in particular relevant differences in protein expression during acute cellular rejection (ACR) and antibody-mediated rejection (AMR). In this multicenter patient cohort of 104 adult heart transplant patients, 74 had no rejection, 18 developed ACR, and 12 developed AMR. During ACR, differential expression was found for 2 proteins, CXCL10 and CD5. In contrast, significantly more proteins (n = 73) were differentially expressed during AMR, the most abundant being N-terminal pro-B-type natriuretic peptide (NT-proBNP) and suppression of tumorigenicity 2 (ST2). These observations were confirmed by univariate logistic regression analysis showing a 2-fold increase of CXCL10 and CD5 concentrations in ACR and an increase in odds of AMR with a doubling of NT-proBNP and ST2 concentrations. After multivariable analysis with clinical covariates, these proteins showed similar odds of ACR or AMR outcomes. The authors conclude that cardiac injury-associated biomarkers were more pronounced in AMR, whereas inflammatory biomarkers were more pronounced in ACR, and that in this patient collective, proteomic analysis revealed specific biomarkers and pathway expression in ACR and AMR. An interesting finding of this study is that protein expression in ACR patients did not look very different from samples from nonrejection patients, and the noted differences were not nearly as striking as those of patients having AMR. This may be related to AMR having a more significant effect on the measured protein biomarkers of cell injury and myocardial dysfunction, whereas other protein groups may be more dysregulated in ACR. Because AMR has a direct vascular injury, the circulating proteome could have been more affected than in ACR. This association with “injury”-related genes has been shown previously, as in particular, donor-derived cell-free DNA (dd-cfDNA) is associated with macrophage genes, natriuretic peptide genes, and myocyte de-differentiation and is relevant for the “injury” hypothesis that was identified in heart allografts.8 In contrast, most of the correlations in ACR are modest, with the highest being the one for CXCL9. One explanation for this can be found in the methodology used for this study, as the authors only analyzed one biopsy per patient, resulting in low numbers of ACR and AMR events. In addition, a significant percentage of these patients had missing dd-cfDNA data for comparison, which was a shortcoming of this study, resulting in an even more complex interpretation of these data. For further studies, it would be very useful to compare myocardial tissue-based proteomics from ACR and AMR patients with healthy controls to be able to specify which proteins are indeed dysregulated in patients with ACR. For wider spread clinical use and to validate their results, it would have been helpful to measure and analyze routine clinical biomarkers such as NT-proBNP blood levels for the patient cohort and correlate these with the observed protein signatures seen in AMR and ACR. Unfortunately, NT-proBNP was not routinely collected for the patients included in this study and therefore not available for analysis. In addition, for the prediction of rejection in clinical routine, the application of these biomarkers would require a validation of identified proteomic signals in a larger patient population with simultaneous investigation of proteomic and analogous single-protein assays, such as NT-proBNP and ST2. This would help to characterize this biomarker in AMR in more detail. In addition, investigation of its results in concert with other molecular biomarkers of rejection, including circulating dd-cfDNA and microRNA as well as gene expression in EMB samples, would be desirable. These data from the Genomic Research Alliance for Transplantation study show that PEP reveals unique biomarkers and pathway expression in ACR and AMR. Cardiac injury-associated biomarkers were more pronounced in AMR, whereas inflammatory biomarkers were more pronounced in ACR. The limited sample size and low event rates do represent a significant warning against any firm conclusion, and I look forward to further analysis of this relevant project. Therefore, in my opinion, PEP is an additional item in the “toolbox” for monitoring rejection, but as a single biomarker in clinical routine surveillance of heart transplant patients, it is not yet “at the Crossroads between AMR and ACR.”