Non-invasive Biomarkers in Heart Transplant: 2020-2021 Year in Review

Abstract

Purpose of review

Endomyocardial biopsy (EMB), the current gold standard for cardiac allograft monitoring is invasive, may have a low sensitivity and is associated with significant variability in histopathologic interpretation. Fortunately, on-going research is identifying non-invasive biomarkers that address some of these limitations. This review provides an update on non-invasive blood-based methods for rejection surveillance and diagnosis in heart transplantation.

Recent findings

Recent studies highlight good test performance to detect acute rejection for donor derived cell free DNA (dd-cfDNA) and microRNA (miR). dd-cfDNA is sensitive, non-specific, and has a high negative predictive value for acute cellular and antibody-mediated rejection. Clinical utility trials are being planned to test its role as a rule-out test for acute rejection as compared to the EMB. miRs may have an added advantage as it may phenotype the subtypes of rejection alleviating the need for an EMB or permitting the initiation of targeted therapy while awaiting results of the EMB.

Summary

In this review, we discuss the recent advances in the field of non-invasive biomarkers to detect allograft rejection after heart transplant. We provide a perspective of additional studies needed to prove their clinical utility and bring these biomarkers to clinical use.

INTRODUCTION

Heart transplantation is the standard of care for end stage heart failure refractory to medical therapy. Although significant strides have been made in long-term survival, graft dysfunction caused by acute rejection remains a leading cause of morbidity and mortality. Acute rejection has two phenotypes: acute cellular rejection (ACR) and antibody-mediated rejection (AMR). Endomyocardial biopsy (EMB) is the historic gold standard for surveillance of acute rejection and for the diagnosis of both AMR and ACR. However, EMB is invasive and causes patient discomfort, limiting the temporal frequency of assessments. EBM samples are analyzed with histopathology, which is limited by its low sensitivity and high interobserver variability(1). Non-invasive methods should permit trending of allograft health for rejection and overall graft function. Since the introduction of gene-expression profiling (GEP) as a rule out test for ACR, alternative methods are being developed with potentially better diagnostic, monitoring, and prognostic performance in heart transplantation. These methods include donor-derived, cell-free DNA (dd-cfDNA), cfDNA methylation, microRNAs (miRs), protein biomarkers, extracellular vesicles (EVs) and donor specific antibodies (DSA) (Figure). We provide a review on scientific advancements since 2020 of these non-invasive blood-based biomarkers in heart transplantation.

Figure:

Non-invasive blood-based approaches to detect transplant rejection.

DNA-BASED MARKERS

Donor derived cell free DNA

Cell-free DNA (cfDNA) is released into the circulation from dying cells, that includes cardiomyocytes, and other organs and serves as a biomarker in disease conditions(2). In transplantation, allograft cells release dd-cfDNA into the circulation, creating a genomic admixture of donor and recipient cfDNA. Since first detected in liver transplant patients(3), technologies to detect dd-cfDNA have evolved, and generally involve two principles; an approach to detect donor DNA by leveraging genetic differences such as donor-recipient sex mismatch where a Y-chromosome is detected in female recipient(3), HLA mismatch(4) or single nucleotide polymorphisms (SNPs)(5), and an approach to quantitate dd-cfDNA using quantitative polymerase chain reaction (PCR), digital droplet PCR(6), or whole genomic sequencing(7, 8). Commercial tests are now available that target a set of hypervariable genomic regions that are likely to different donor DNA from recipient DNA. This approach is then couple to sequencing (Allosure, CareDx, Inc., South San Francisco, CA or Prospera® Natera, Inc., San Carlos, CA) or quantitative PCR (MyTAI_HEART) or digital droplet PCR, leading to faster turnaround time and obviating the need to genotype transplant donors and recipients. As a result, large cohort studies have been initiated to test the performance of dd-cfDNA as a marker of transplant rejection. A comprehensive review of the studies published to date on dd-cfDNA in heart transplantation is beyond the scope of this article. The recent studies published since 2020 highlight dd-cfDNA test performance, the interval of its applicability after transplantation and its potential role as a predictive biomarker for long-term outcomes (Table).

Table:

Blood-based biomarkers in heart transplantation, a summary of cell-free DNA and microRNA studies published 2020 - 2021

Test performance of dd-cfDNA to detect acute rejection
PubMed ID	First author/date published	Study type	# of patients	# of samples	Primary endpoints	Approach to detect donor DNA	Major findings
							AUROC	ddcfDNA cutoff	NPV
31929557	North, 01/2020	Pilot	76	156	ACR	Targeting highly polymorphic SNPs	0.84	0.32	100%
31983667	Richmond 05/2020	Cohort	241	624	Acute rejection (ACR & AMR)	Targeting highly polymorphic SNPs	0.78 – 0.81	0.3%	81%
33435695	Agbor-Enoh & Shah, 01/2021	Cohort	171	1834	Acute rejection (ACR & AMR)	Genotype donor and recipient	0.92	0.25%	99%
33653997	Knüttgen, 02/2021	Cohort	87	770	ACR	Targeting highly polymorphic SNPs	0.81	0.35%	97%
34134912	Sorbini, 08/2021	Pilot	19	232	Acute rejection (ACR & AMR)	Targeting polymorphic HLA-DRB1	0.72	0.11%	Not reported
dd-cfDNA as part of routine clinical care
PubMed ID	First author/date published	# of patients		# of dd- cfDNA test	Protocol	Approach to measure dd- cfDNA	Major findings
33606316	Gondi, 05/2021	153		495	GEP + dd-cfDNA in place of EMB	Targeting highly polymorphic SNPs	A reduction of 83.9% of EMB without a reduction in acute rejection episodes detected
34420244	Feingold, 08/2021	58		495	dd-cfDNA in place of EMB	Targeting highly polymorphic SNPs	A reduction of 81% of surveillance EMB without short-term adverse outcome.
Other cell-free DNA publications
31825144	Zangwill, 02/2020	Cohort	17	Not reported	Survival	Targeting highly polymorphic SNPs	Early post-transplant total cfDNA predicted survival.
33039362	Scott, 10/2021	cohort	87	257	Survival	Targeting highly polymorphic SNPs	Early post-transplant total cfDNA, but not dd-cfDNA predicted survival.
33368611	Holzhauser, 03/2021	Cross-sectional	65	65	CAV	Targeting highly polymorphic SNPs	Higher dd-cfDNA associated to CAV
33756005	Patter, 06/2021	Pilot	6	24	ACR	Cardiac-specific DNA methylation	High cardiac-specific cfDNA with acute rejection compared to controls
miRNA publications
PubMed ID	First author/date published	Study type	# of patients	# of samples	Primary endpoints	Approach to measure miRNA	Major findings
							AUC	NPV
32654912	Constanso-Conde, 9/2020	Cohort	121	153	ACR	RT-PCR	80	98%
33713567	Kennel, 06/2021	Case-control	43	43	Acute rejection (ACR & AMR)	miRNA sequencing	0.63 – 0.96	Not applicable

Recent Clinical Studies highlighting dd-cfDNA test performance

Sorbini and colleagues conducted a single-center, prospective pilot study to evaluate an optimized droplet digital PCR assay(4). Their approach targets the hypervariable HLA-DRB1 gene, which accounts for >85% of the genetic variability in White patients, to detect donor DNA. The prospective cohort study included 19 patients and 232 serial plasma samples evaluated ACR, but not AMR. They showed higher dd-cfDNA levels with ACR compared to no rejection controls(4). Using receiver operator curve (ROC) analysis, dd-cfDNA showed an area-under-the curve (AUC) of 0.72 for detecting ACR. This assay is easy to use and potentially broadly applicable if additional gene loci with adequate variability in non-White populations are included.

North and colleagues reported the analytic properties of a commercial test that targets a panel of 94 highly informative and broadly applicable SNPs to distinguish donor and recipient cfDNA(9). Richmond and colleagues used the same assay in a cohort of 174 pediatric and adult patients. Dd-cfDNA showed AUC of 0.83 to 0.87 for detecting ACR or AMR. A threshold of 0.3% had a NPV of 90 to 98% for these endpoints(10). Only a few ACR events were included in this study. Further, a large proportion of samples were compromised due to nuclear DNA contamination. This study therefore highlighted that delayed sample processing could be a limitation for reliable dd-cfDNA measurement due to nuclear DNA contamination from cell lysis, even when specialized blood collection tubes are used. Additional considerations prior to dd-cfDNA testing were reviewed recently(11).

The Genomic Research Alliance for Transplantation (GRAfT) published the clinical validity of dd-cfDNA using a rigorously conducted, multicenter, prospective cohort study. The investigators enrolled 171 heart transplant patients and performed over 1800 dd-cfDNA measurements. The dd-cfDNA approach genotyped transplant donors and recipients DNA to identify informative SNPs (~400,000). Post-transplant levels of dd-cfDNA were elevated and then followed a 1-phase logarithmic decay to reach 0.13% on day 28 after transplantation(8). Based on this decay, the GRAfT study only considered dd-cfDNA measurements beyond day 28. Knüttgen and colleagues studied 87 patients and 770 samples using a well validated digital droplet PCR approach(6) and showed that dd-cfDNA reached baseline earlier by one week post-transplant, similar to other studies (6, 12-14). In the GRAfT Study, dd-cfDNA showed an AUC of 0.92 to detect acute rejection, a composite endpoint of AMR and ACR(8). A strategy of screening for a dd-cfDNA ≥0.25% from post-transplant day 28 onwards would have safely eliminated 81% of all EMBs. Recent studies report a similar reduction of surveillance EMB when dd-cfDNA was integrated in routine clinical care for surveillance in place of EMB (15, 16). In the GRAfT Study, if dd-cfDNA was used as a gold standard for the diagnosis of acute allograft injury, the positive predictive value of EMB was only 19.6%; although 76% of patients with a positive dd-cfDNA and negative EMB subsequently developed rejection or showed other clinical signs of allograft dysfunction. Knüttgen and colleagues reported higher levels of dd-cfDNA levels with pericardial effusion compared to controls. These findings highlight the high sensitivity of dd-cfDNA to detect any cause of allograft injury, but equally highlight its low specificity for acute rejection.

Cell-free DNA as a prognostic indicator of long term outcomes

Studies by Zangwill and colleagues(14) and by Scott and colleagues(17) examined dd-cfDNA in relation to survival. These studies showed that higher early post-transplant total cfDNA predicted death. Holzhauser and colleagues showed a similar association with cardiac allograft vasculopathy.

DNA-Methylation

A pilot study by Patter and colleagues analyzed 24 samples from 6 patients using a panel of left ventricle-specific cfDNA methylation signatures to measure cardiac-specific cfDNA. They showed that cardiac-specific cfDNA was increased with acute rejection compared to no rejection controls. Advances in cfDNA methylation analysis allows for the concurrent mapping of cfDNA from multiple tissues sources(18-20). These approaches may be used to differentiate AMR and ACR.

Cell-free DNA as a mediator

Recent evidence suggests that dd-cfDNA itself is triggers allograft injury(21, 22). In COVID-19 patients, cfDNA triggers release of reactive oxygen species via toll-like receptor 9(23). Studies are needed to define the biological consequence of dd-cfDNA as a potential trigger of allograft injury.

RNAs

Gene expression profiling (messenger RNA)

Gene expression profiling (GEP) is the most extensively studied and clinically adopted non-invasive approach to monitor heart allograft rejection(24-27). The commercial test, AlloMap (CareDx, Brisbane, CA) measures the expression of 11 informative genes and 9 housekeeper genes from peripheral blood mononuclear cells. The test has an excellent NPV of 99%, but poor PPV of ~10% for ACR and has no ability to detect AMR.

More recently, Shannon and colleagues assessed the performance of 9-gene panel to detect ACR early after transplant in the HEARTBiT study(28). The HEARTBiT study included 160 patients and tested over 1600 samples using the NanoString nCounter. This approach detected ACR at a median of 42 days after transplant with AUC of 0.69.

There is a growing interest to define changes mitochondrial energetics and mitochondrial gene expression in rejection. Tarazon and colleagues investigated the role of circulating mitochondrial genes as biomarkers of ACR in a single-center cohort of 40 heart transplant patients(29). They performed RNA-sequencing of 112 mitochondrial genes and identified several genes which were differentially expressed during ACR. These genes were internally validated and identified ACR with an AUC of 0.90. Interestingly, mitochondrial genes stimulate the immune system (30); suggesting that these mitochondrial genes may not only be biomarkers of rejection, but potential mediators(31).

MicroRNAs

MicroRNAs are small, ~21-24 nucleotide, single-stranded non-coding RNA molecules. The human genome encodes for ~2,200 miRs. Many miRs either block mRNA translation to proteins or promote mRNA transcript degradation. They often regulate multiple genes simultaneously and function in miR-messenger RNA networks to control biological pathways and cellular activity(32).

In heart transplantation, initial studies by Wei and colleagues demonstrated that miR-182 was implicated in heart allograft rejection and modulated graft injury(33). A subsequent study by Duong and colleagues assessed miR expression through PCR of formalin-fixed EMB tissue and paired serum samples(34). They identified unique miR expression patterns during ACR and AMR, with certain miRs (miR-10a, miR-31, miR-92a and miR-155) differentially regulated in the both the EMB tissue and serum.

In 2020, Constanso-Conde and colleagues conducted a prospective, observational single center study to discover and validate circulating miRs in ACR. In a cohort of 121 patients, they analyzed 179 miRs by PCR. They identified 21 ACR episodes which were analyzed (a total of 63 serum samples). In the validation cohort of 90 patients, miR-185a-5p showed an AUC of 0.80 and a NPV of 98%(35). This study was limited by the relatively small number of patents and the absence of AMR patients. In 2021, Kennel and colleagues performed a case control analysis and sequenced circulating miRs in a small cohort of ACR (n=13) and AMR patients (n=13)(36), they identified 3 miRs (miR-139-5p, miR-151a-5p and miR-186-5p) which were increased during rejection, and miRs that predicted the subsequent development of ACR (miR-29c-3p) or AMR (miR-486-5p). Limitations of this analysis included a small cohort with only 23 heart transplant patients and a sequencing methodology which led to ~10% of all reads being mapped to miRs. In addition, miR-486-5p is a highly abundant in red blood cells and may reflect underlying sample hemolysis(37).

The intriguing aspect of miR research is that these molecules may serve not only as unique biomarkers of ACR and AMR, but themselves represent targets for the future development of novel therapeutics and may serve as future targets of immunosuppressant drug development(38, 39).

PROTEINS AND OTHER BIOMARKERS

High-sensitive cardiac troponin

Cardiac troponins T and I are exclusively present in cardiomyocytes and are highly sensitive and specific markers of myocardial injury.

Although initial studies evaluated the role of standard cardiac troponin assays to detect rejection, more recently, high sensitivity immunoassays can detect very low-level troponin concentrations. These studies had inconsistent findings as to whether high sensitivity cardiac troponin (hs-cTnI) can be used for acute rejection monitoring after heart transplant(40-43). Recently, Fitzsimons and colleagues studied the performance of hs-cTnI assay to detect ACR in a single center cohort of 170 patients and 883 paired serum and EMB samples. The hs-cTnI assay had no diagnostic utility in detecting ACR with an AUC of 0.51(44). These findings are inconsistent with an earlier study which suggested hs-cTnI could detect acute rejection(45), suggesting that additional work is needed to assess the performance on this assay

N-terminal pro-brain natriuretic peptide

N-terminal pro-brain natriuretic peptide (NT-proBNP) reflects increased ventricular wall stress, pressure and is a well-known diagnostic and prognostic marker in heart failure. NT-proBNP was previously assessed by several studies as a possible biomarker for acute allograft rejection (46-49). Those studies again showed the potential role of NT-proBNP in allograft rejection monitoring, but the clinical utility is limited by non-specificity of the biomarker.

Donor-specific antibodies

Donor-specific antibodies (DSAs) are antibodies formed by the recipient that can bind to the donor organ, triggering activation of the complement cascade and resultant allograft injury. The development of de novo DSA after heart transplantation is a critical risk factor for subsequent AMR, graft failure and death. Circulating DSA are frequently present at the time of an AMR diagnosis, and studies have shown that higher mean fluorescence intensity of DSAs correlate with AMR severity. The current ISHLT guidelines recommend DSA monitoring at 1, 3, 6 and 12-months post-transplant, followed by yearly intervals in non-sensitized patients(50).

Circulating extracellular vesicles

Extracellular vesicles (EVs) are lipid bound vesicles secreted by almost all types of cells into the extracellular space. Based on the biogenesis, release pathways and size, EVs are divided into three main subtypes: microvesicles, exosomes, and apoptotic bodies. The EV content is modulated by cell of origin and molecular targets.

Kennel and colleagues first hypothesized that exosomes are altered in acute rejection and can help distinguish rejection(51). Using liquid chromatography–tandem mass spectrometry analysis of serum exosomes, they identified 15 differentially expressed proteins during rejection(51).These proteins are implicated in specific immunologic processes. The circulating transplant heart exosomes expressed donor-specific HLA I molecules, suggesting a cardiac origin.

Hu and colleagues purified donor heart exosomes using anti-donor HLA I antibody beads and assessed complement C4d protein expression as marker for AMR. C4d protein was seen in donor heart exosome subset in one patient who developed AMR on EMB post-transplant, while C4d protein was not detected in three other patients without AMR(52). Recently, Castellani and colleagues used EVs in combination with EMB for the diagnosis allograft rejection(53). This study found that the concentration of EVs was significantly increased, and their diameter decreased in patients with ACR and AMR. Different EV surface markers were identified as discriminants between ACR and AMR. The respective receiver operating characteristics curves provided an AUC range of 0.73–0.94 for detecting acute rejection. A diagnostic model was built based on these results. In a validation study, this model showed a diagnostic accuracy of 86.5%.

CONCLUSIONS

Existing data indicate that non-invasive blood-based approaches may be used to monitor allograft health with a higher temporal resolution than traditional histopathology. dd-cfDNA shows great promise as a sensitive test with a high NPV for both ACR and AMR. Commercial entities have initiated large cohort studies that should further inform dd-cfDNA performance and provide guidance on its application. Randomized, clinical trials are also being planned to assess its clinical utility as a rule out test for acute rejection or to guide immunosuppression titration. Unfortunately, these studies use different dd-cfDNA approaches and study protocols. Cross assessment of the different dd-cfDNA approaches would be critically important to standardize the application of the findings of these studies. With its high sensitivity and non-invasiveness, adoption of dd-cfDNA testing in heart transplantation is expected to grow(54), as has been the case in kidney transplant(55).

The current dd-cfDNA approaches are unable to distinguish ACR, AMR and other causes of allograft injury. In addition, these approaches show limited utility in patients receiving multi-organ transplants. Recent advances in cfDNA epigenomic analysis can provide a molecular injury scan, which may address these limitations and discriminate AMR and ACR. Several studies suggest that miRs may be promising biomarkers to detect ACR and AMR and discriminate the two rejection phenotypes. External, rigorous validation of these miRs is needed prior to an assessment of their clinical utility. We also look forward to clinical studies that assess the performance of a combination of the above biomarkers in heart transplantation.

Review Highlights.

• Non-invasive blood biomarkers allow more frequent assessments of allograft health compared to endomyocardial biopsy

• Donor-derived cell-free DNA (dd-cfDNA) is a sensitive tool for detecting AMR and ACR.

• The high NPV and low specificity make dd-cfDNA a reliable tool to monitor patients for AMR, ACR or any other cause of allograft injury.

• miRNA is a tool in development that can detect rejection and phenotype AMR versus ACR.

Abbreviations:

ACR

Acute Cellular Rejection

AMR

Antibody-Mediated Rejection

AUC

area under the curve

CAV

cardiac allograft vasculopathy

dd-cfDNA

donor-derived cell-free DNA

EMB

Endomyocardial Biopsy

EV

extracellular vesicles

GEP

gene expression profiling

miR

microRNAs

NPV

negative predictive value

PCR

polymerase chain reaction

PPV

positive predictive value

RT-PCR

reverse transcription polymerase chain reaction