論文ID: CJ-21-0591
Despite the overall success of heart transplantation as a definitive treatment for endstage heart failure, cardiac allograft rejection remains an important cause of morbidity and mortality. Endomyocardial biopsy has been the standard of care for rejection monitoring, but is associated with several diagnostic limitations and serious procedural complications. The use of molecular diagnostics has emerged over the past decade as a tool to potentially circumvent some of these limitations. We present an update on novel molecular approaches to detecting transplant rejection, focusing on 4 categories: microarray technology, gene expression profiling, cell-free DNA and microRNA.
Heart transplantation is the standard of care for advanced, endstage heart failure refractory to medical therapy. Although significant strides have been made in long-term survival, rejection remains an important cause of significant morbidity and mortality. The incidence of acute rejection has declined with advances in immunosuppressive regimens, but remains around 12% in the first year following heart transplantation.1 Rejection can be subdivided into 2 main categories: acute cellular rejection (ACR) and antibody-mediated rejection (AMR). Both ACR and AMR are associated with an increased risk of early death, as well as further downstream effects, including increased rates of cardiac allograft vasculopathy. The current standard of care in surveillance for rejection is histopathologic diagnosis via serial endomyocardial biopsies that are graded by cardiac pathologists per the International Society for Heart and Lung Transplantation (ISHLT) criteria.2 However, there are several limitations to this approach, including low rates of concordance between pathologists, with significant interobserver variability found even in the diagnosis of advanced rejection.3 In addition, sampling error and the presence of artifact in biopsy samples may reduce the diagnostic yield. The invasive nature of endomyocardial biopsies is associated with inherent procedural risk, including tricuspid valve injury, cardiac tamponade, and bleeding.4 Finally, in the modern era of immunosuppression, relatively low incidence reduces the yield of routine surveillance biopsies.5 These limitations can result in both underdiagnosis and treatment of rejection, as well as overtreatment in patients without actual rejection, both of which have associated patient risks. The use of molecular diagnostics has emerged over the past decade as a tool to potentially circumvent some of these limitations.
In this review, we summarize 4 categories of molecular diagnostics being used in practice today: microarray technology, gene expression profiling, cell-free DNA and microRNA. In addition, we review emerging concepts in molecular diagnostics. We discuss the utilization of molecular diagnostics in modern clinical practice and propose future directions in this field.
Microarray technology allows for molecular examination of endomyocardial biopsy samples, and aims to increase the accuracy of the diagnosis of both ACR and AMR.6 The aim of precise molecular examination of endomyocardial biopsies is to overcome limitations in histopathologic diagnosis, including interobserver variability. This is done by identifying mRNA expression using rejection associated transcripts (RAT) to assign a molecular probability of ACR, AMR and injury. The use of molecular biopsy techniques was initially developed in kidney transplants. Halloran et al found that RAT expression was distributed similarly in heart and kidney biopsy specimens, allowing for extension of this technology to heart transplantation.7 At present, microarray technology is completed through the Molecular Microscope Diagnostic System (MMDx), which utilizes an unsupervised archetypal analysis to assign archetype scores that give a probability of a normal, ACR, AMR or injury pattern in endomyocardial biopsy samples.7 MMDx utilizes machine learning to create predictions of molecular diagnosis. Figure 1 demonstrates the patterns achieved through archetypal analysis, demonstrating AMR, ACR and injury patterns.8 This has been found to be highly useful in accurately predicting molecular rejection.9 Interestingly, compared with kidney transplantation, there is a higher degree of discordance in the diagnosis of ACR when comparing MMDX with traditional histologic measures.7 For example, Parkes et al found that MMDX was more sensitive in diagnosing AMR, whereas histology was more likely to diagnose ACR; however most of the cases were ISHLT Grade 1R, known as mild ACR.9 This is likely reflective of both the interobserver variability noted in the diagnosis of ACR, as well as the questionable significance of 1R rejection in heart transplantation.
Principal component analysis of 889 heart transplant biopsies based on their expression of rejection-associated transcripts (RATs). Samples in A–C are colored according to their highest archetype score (white=S1normal, red=S2TCMR, blue=S3ABMR, orange=S4UV) in the 4-archetype model trained on RAT expression. The large ghosted points labeled A1 (normal), A2 (TCMR), A3 (ABMR), and A4 (unexplained variance, UV) mark the positions of the theoretical archetypes to which each sample is compared. Reproduced with permission from Halloran et al.8
The molecular microscope may be particularly useful in the diagnosis of AMR, which otherwise presents a major clinical challenge. Currently, diagnosis depends on the presence of either immunopathological features, including C4d or C3d capillary deposition or CD68+ macrophages intravascularly, or histological changes notable for microvascular injury.2 However, similar to ACR, there is a need to develop a more precise, quantitative diagnostic tool for the diagnosis of AMR. Afzali et al compared gene expression in patients with AMR as compared with ACR or no rejection, and was able to identify AMR-related gene transcripts and utilize gene expression to accurately discriminate AMR from non-AMR cases.10 Loupy et al found a unique genetic profile of AMR, characterized by endothelial cell, macrophage, natural killer cell and interferon-gamma transcripts, that can be used to accurately identify AMR and correlates with degree of endomyocardial injury.11 In addition to adding diagnostic accuracy, these findings provide a window into the potential mechanisms of AMR. In summary, the use of intragraft mRNA transcripts as an adjunct to pathology reads of endomyocardial biopsies may lead to a paradigm shift in how we diagnose cardiac transplant rejection.
GEP of peripheral blood leukocytes is a useful tool for noninvasively assessing rejection in stable patients more than 2 months following heart transplantation.
Horwitz et al initially described that peripheral GEP could feasibly discriminate between rejection and no rejection, and correlated with biopsy-proven allograft rejection.12 The Cardiac Allograft Rejection Gene Expression Observational (CARGO) study then utilized real-time PCR to identify 11 genes that were able to discriminate between quiescence as compared with moderate or severe rejection in stable heart transplant recipients.13 This led to the development of Allomap (CareDx), the current gene expression profile test utilized in clinical practice (Figure 2).14 Allomap integrates these gene expression profiles into a single score of 0–40, with a score ≥34 associated with a higher likelihood of ACR.
The AlloMap test score reflects the differential gene expression in diverse regulatory pathways. The AlloMap test classifier for acute cellular rejection (ACR) consists of 20 genes, including 11 that inform a variety of regulatory pathways and contribute to its 7 terms, which correspond to the expression of individual genes or the average expression levels of coordinately expressed genes termed metagenes. The classifier also includes 9 genes used for normalization of gene expression and quality control. Low AlloMap scores on a scale from 0 to 40, below the clinician-specified threshold, can be used to rule out ACR. Upregulated (gray) and downregulated (black) expression of genes, as assayed in peripheral blood mononuclear cells by qPCR, contribute to the AlloMap score. Reproduced with permission from Fang.14 ACR, acute cellular rejection.
Allomap is associated with a high negative predictive value (NPV) of 99.6% for moderate to severe rejection, making it useful as a screening tool for rejection in stable patients. The Invasive Monitoring Attenuation through Gene Expression (IMAGE) trial compared Allomap with endomyocardial biopsy in patients more than 6 months following transplant and demonstrated similar clinical outcomes, with survival of 94% in the GEP group, and freedom from the composite clinical endpoint, including hemodynamic compromise, graft dysfunction, death or re-transplantation of 85%, which was noninferior to patients monitored by routine endomyocardial biopsy.15 The CARGO II trial established a score of <34 to identify patients at low risk of rejection.16 Finally, the E-IMAGE (Early- IMAGE) single-center study examined the feasibility of using Allomap to rule out acute rejection in patients 2–6 months following transplant, also finding a high NPV and acceptable clinical outcomes with no difference in the composite endpoint, rates of rejection, between patients monitored by surveillance biopsies as compared with GEP.17 The E-IMAGE study also demonstrated that Allomap could be used to guide early steroid weaning in appropriate patients. At present, the utilization of Allomap has been incorporated into the guidelines for stable outpatients >55 days following heart transplantation. Moayedi et al18 performed a prospective observational study examining short- and long-term clinical outcomes in a real-world population utilizing GEP for routine rejection surveillance in the Outcomes AlloMap Registry. They found that, in clinical practice, utilization of GEP was associated with a similarly high NPV, thus confirming its utility in ruling out moderate to severe rejection in clinical practice. Patients monitored by GEP notably had good clinical outcomes, including high rates of survival, freedom from rejection and graft dysfunction.18 However, GEP is limited by low positive predictive values. In addition, it is limited to ACR, therefore excluding AMR.
In current practice, Allomap is an established technique to use in lieu of routine surveillance endomyocardial biopsies for stable outpatients at low risk of rejection. Patients with scores of <20, <30 or <40 and >2 months, >6 months or >1 year, respectively, are able to forego routine endomyocardial biopsy.19 In addition, by identifying patients at low risk for rejection, it can be used as a tool for early steroid weaning.
Another area of interest is high-throughput transcriptomics profiling assays applied to whole-blood samples acquired during endomyocardial biopsy. In the Biomarkers in Transplantation initiative, investigators utilized GEP to identify mRNA transcripts associated with acute rejection.20 In the HEARTBiT trial, Shannon et al21 used NanoString nCounter technology to profile and quantify these nine mRNA transcripts to assess for early ACR events. They achieved an area under the curve (AUC) of 0.79 in identification of >2R rejection in patients <55 days following transplantation.21 This is clinically important because most episodes of acute rejection occur in the first 3 months following heart transplantation. These findings will hopefully be further validated in an ongoing, larger prospective trail.
Donor-derived cell-free DNA (dd-CFDNA), or DNA of donor origin that is detectable in a recipient’s blood, has emerged as a potential noninvasive marker of acute rejection. Dd-cfDNA is released with cell apoptosis or necrosis, representing graft damage, occurring in both ACR and AMR. Initially, techniques used to differentiate between donor and recipient DNA relied on the presence of the Y-chromosome, limiting its utility to male donor-female recipient pairs. Current techniques utilize shotgun whole-genome sequencing, utilizing single nucleotide polymorphisms (SNPs) to differentiate between donor and recipient DNA.22 Allosure (CareDx, South San Francisco, CA, USA) is a commercially available test that uses next-generation sequencing to differentiate between donor and recipient DNA by comparing 266 SNPs, and can detect dd-cfDNA (ratio of dd-cfDNA/recipient-cfDNA) at a level of 0.16%.
The percentage of dd-cfDNA is compared with the total cell-free DNA found in the blood. Quantities of dd-cfDNA increase significantly during rejection episodes and then decline following treatment, making this a useful marker of rejection. De Vlaminck et al23 confirmed these findings and noted that dd-cfDNA was significantly higher during episodes of acute rejection, and levels correlated with severity of rejection. The AUC for ACR ≥2R was 0.83 using a threshold level of cf-dDNA of 0.25%.23 Figure 3 show the correlation of dd-cfDNA elevation with biopsy grades of ACR and AMR, as well as echocardiographic findings.24 Another such commercially available assay is Prospera (Natera, San Carlos, CA, USA). Utilizing technology that was initially created to detect fetal cfDNA in prenatal testing, this test utilizes NSP-based PCR to measure dd-cfDNA for the detection of rejection. This was demonstrated to be effective in discriminating acute vs. no rejection in kidney transplantation detected on biopsy.25 Investigation is currently underway to evaluate the Prospera dd-cfDNA test for heart transplant recipients as part of the intragraft mRNA and donor-derived cell-free DNA study.
(A) Percent donor-derived cell-free DNA (dd-cfDNA) in relation to severity of acute cellular rejection (ACR) by histopathologic interpretation of endomyocardial biopsy. Grade 0 rejection includes both ACR grade 0 and pAMR 0. (B) %dd-cfDNA in relation to severity of antibody-mediated rejection (AMR) by histopathologic interpretation of the endomyocardial biopsy. Grade 0 rejection includes both ACR grade 0 or 1 and pAMR 0. (C) %dd-cfDNA levels correlated with severity of allograft dysfunction measured by echocardiography; median %dd-cfDNA levels for no (<5% LVEF decline), mild (5–<10%), moderate (≥10–<15%) or severe (≥15%) allograft dysfunction were 0.02%, 0.06%, 0.19%, and 0.32%, respectively. P values obtained by generalized estimating equation approach comparing all categories. Adapted with permission from Agbor-Enoh et al.24 LVEF, left ventricular ejection fraction.
In addition, dd-cfDNA rises prior to changes detected in the endomyocardial biopsy, allowing for earlier detection of acute rejection in asymptomatic patients.23 Khush et al performed a prospective study to clinically validate the use of dd-cfDNA to detect acute rejection in heart transplant recipients.26 They found that the percent dd-cfDNA significantly differed in patients with and without acute rejection. Using a cutoff of 0.2%, they established 80% specificity and 44% sensitivity in differentiating acute rejection from no rejection, which yielded a high NPV of 97.1% using a cutoff of 0.2. In ACR, levels of dd-cfDNA are similar in 1R compared with no rejection, but are significantly increased in ACR ≥2. In terms of AMR, there is a significant rise in dd-cfDNA with pAMR 1 and higher grades of rejection.27 Patterns of dd-cfDNA elevation also differ in AMR as compared with ACR, which may help facilitate diagnosis. Agbor-Enoh et al found that, for similar histopathologic grades, the percent dd-cfDNA was higher in AMR as compared with ACR, rose preceding the histologic diagnosis, and had a distinct fragment length with shorter fragments of cfDNA.24 These unique insights can help facilitate more accurate diagnosis of AMR.
In addition to diagnosing rejection, detection of dd-cfDNA may have several other interesting implications. Elevations in dd-cfDNA are associated with formation of donor-specific antibodies (DSA), suggesting that subclinical graft injury may predispose to DSA formation, identifying a potential unique risk factor for sensitization.28 In addition, a recent study noted an increase in dd-cfDNA in stable patients with cardiac allograft vasculopathy.29 These findings expand the utility of dd-cfDNA beyond that of rejection and suggest potential future directions for its use.
MicroRNA are small, 20–24 nucleotide, single-stranded noncoding RNAs that regulate gene expression at the post-transcriptional level via repression of translation, as well as functioning extracellularly in the regulation of intercellular communication. MiRNA are expressed differently by different organs, and expression profiles differ between disease states, making this a promising marker for the detection of various disease states, including heart transplant rejection. Nováková et al evaluated microRNA expression in endomyocardial biopsy samples and identified 3 miRNAs (miR-14, miR-589 and miR-182) that were significantly dysregulated in ACR.30 Duong Van Huyen et al identified 7 miRNA that were expressed differently in patients with and without biopsy-proven rejection; 4 (miR-10a, miR-31, miR-92a, and miR-155) were expressed serologically and able to strongly discriminate between rejection and no rejection noninvasively.31
In addition, evaluation of miRNA may be useful in distinguishing subtypes of acute rejection. Di Francesco et al compared miRNA expression patterns in endomyocardial biopsy samples and were able to distinguish specific patterns that distinguished ACR, AMR and mixed rejection32 (Figure 4). Kennel et al compared miRNA expression profiles in patients with AMR or ACR and all patients with rejection. In both ACR and AMR, they found 3 miRNA that were commonly elevated, but they noted different expression profiles in AMR as compared with ACR.33
Global miRNA expression and unsupervised hierarchical clustering of the 3 type of rejection and the group. The miRNA expression value shown in the map is the logarithm of the original value. Each row represents the expression level of miRNAs, and each column is a different sample. The color scale setting was done according to the MultiExperiment Viewer version 4.2 software’s indication from −0.5 to 4.8. Reproduced with permission from Di Francesco et al.30 ACR, acute cellular rejection; pAMR, pathology antibody-mediated rejection; MIX, mixed rejection; CTRL, control.
One particular miRNA of interest is mi-182, which is increased with T-cell expression, including in the setting of ACR, and is downregulated in the setting of calcineurin inhibitor use.34 mi-182 has been found in the plasma of mice during ACR, making it of interest as a potential noninvasive marker of acute rejection. Finally, Wei et al demonstrated that targeting mi-182 may be a future area of therapeutic intervention to prevent ACR.35
Long noncoding RNA influences the function of T and B cells. In a mouse model it was found that expression of long noncoding RNA was different and may be associated with T-cell responses, specifically with helper T-cells. This may serve as another potential novel marker of rejection.
Circulating EV are a novel potential biomarker for diagnosis of acute allograft rejection. EVs are small, nanospheric membranes that are released into the extracellular space to communicate information between cells, including in specific pathologic conditions such as states of enhanced immune response and inflammation.36 EVs are of particular interest because they can be easily extracted from blood via minimally invasive approaches. Castellani et al37 recently studied the application of EVs as a tool for evaluating acute rejection in heart transplant recipients. They found that the concentration of EVs rose significantly in both AMR and ACR. Furthermore, they were able to identify specific surface markers that accurately identified both ACR and AMR.37 Exosomes, or membrane-bound EVs, may also be potentially useful in specifically identifying AMR. Exosomes derived from the donor graft express donor-specific HLA molecules. Detection of antibodies directed against donor HLAs using exosomes may be a sensitive way of noninvasively evaluating AMR.38
Digital BiopsyAnother promising avenue of more precise diagnosis of allograft rejection is the digital biopsy, in which machine-learning techniques are used to evaluate endomyocardial biopsy samples. These automated approaches seek to overcome subjective differences in histopathologic interpretation, and potentially reduce interobserver variability in diagnosis.39 Preliminary results for the ability of machine learning to successfully discriminate differentiate between grades of ACR in heart transplant recipients have been promising. One such tool, the CACHE-Grader, demonstrated that automated histologic diagnosis is feasible and noninferior to evaluation by expert pathologists.40
Surveillance for allograft rejection following heart transplantation remains a major clinical challenge. Although endomyocardial biopsies remain the gold standard in diagnosis, in the past decade great strides have been made in enhancing this methodology with molecular approaches. Molecular diagnostics have the potential to increase diagnostic accuracy, aid in earlier diagnosis, and reduce the need for invasive approaches in allograft rejection diagnosis. In addition, the development of these technologies provides additional insight into the potential mechanisms underlying both ACR and AMR, leading the way for therapeutic advances. Future challenges will be to determine ways of incorporating these various technologies into a diagnostic schema in order to accurately identify patients with rejection and aid in treatment decisions.
L.B., T.S.: No disclosures. J.K.: Research grants from CareDx, Molecular Microscope. Advisory Board for CareDx.
