Cell-free DNA in Lung Transplantation: Research Tool or Clinical Workhorse?

Abstract

Purpose of the Review

Recent evidence indicates that plasma donor-derived cell-free DNA (dd-cfDNA) is a sensitive biomarker for the detection of underlying allograft injury, including rejection and infection. In this review, we will cover the latest evidence revolving around dd-cfDNA in lung transplantation and its role in both advancing mechanistic insight into disease states in lung transplant recipients as well as its potential clinical utility.

Recent Findings

Plasma dd-cfDNA increases in the setting of allograft injury, including in primary graft dysfunction, acute cellular rejection, antibody mediated rejection and infection. Dd-cfDNA has demonstrated good performance characteristics for the detection of various allograft injury states, most notably with a high negative predictive value for detection of acute rejection. Elevated levels of dd-cfDNA in the early post-transplant period, reflecting molecular evidence of lung allograft injury, are associated with increased risk of chronic lung allograft dysfunction and death.

Summary

As a quantitative, molecular biomarker of lung allograft injury, dd-cfDNA holds great promise in clinical and research settings for advancing methods of post-transplant surveillance monitoring, diagnosis of allograft injury states, monitoring adequacy of immunosuppression, risk stratification and unlocking pathophysiological mechanisms of various disease.

Introduction

Despite advances in surgical technique, donor-recipient matching and immunosuppression, overall survival for lung transplant recipients remains the lowest compared to other solid organ transplantation, with a median survival of 6.5 years(1). Episodes of acute allograft rejection, including acute cellular rejection (ACR) and anti-body mediated rejection (AMR), and infection remain a substantial cause of morbidity and mortality. It is estimated that up to 50% of patients will experience an episode of acute rejection and 33% will experience an episode of pneumonia in the first-year post-transplant(1–4). These episodes of allograft injury serve as risk factors for the development of chronic lung allograft dysfunction (CLAD), the predominant cause of long-term mortality in lung transplantation(5–8). Since CLAD has no effective treatment, post-transplant clinical care is centered around prevention via early detection and effective treatment of these risk factors.

Traditional means of detecting and monitoring for allograft injury in lung transplant recipients includes frequent clinical assessment, pulmonary function testing (PFT), evaluating for donor specific antibodies (DSA) and bronchoscopy with bronchoalveolar lavage (BAL) and transbronchial biopsy (TBBx). However, despite this seemingly comprehensive approach, these tools possess several limitations. Lung transplant recipients often experience underlying allograft injury in the absence of clinical signs and symptoms or a decline in PFT values(9). Biopsy samples that are analyzed for histopathology (the current “gold standard” for diagnosis of rejection), demonstrate high interobserver variability(10). Furthermore, screening tools such as surveillance bronchoscopy and transbronchial biopsy may be costly, inconvenient and place the patient at risk for procedural complications, further compromising morbidity and quality of life. These issues pose a sizable dilemma for the surveillance, detection and risk stratification of lung transplant patients in both the clinical and research settings.

Plasma donor-derived cell-free DNA (dd-cfDNA) is a novel, non-invasive biomarker that detects allograft injury on the molecular level. Recent evidence in lung transplant recipients indicates that dd-cfDNA accurately detects episodes of acute rejection and infection while also providing a valuable method of risk stratification for patients at risk of adverse long-term outcomes(11–13). However, does dd-cfDNA have a role in the current clinical paradigm for the detection and monitoring of underlying allograft injury? Or is it best served in the research arena to provide further understanding of the molecular characteristics of various pathophysiological states in lung transplantation? In this review, we will cover the latest evidence revolving around dd-cfDNA in lung transplantation and its role in both advancing mechanistic insight into disease states in lung transplant recipients as well as its potential clinical utility. Selected studies are summarized in Table 1.

Table 1 –

Summary of ddcfDNA studies to detect lung allograft injury or predict CLAD

Cohort Studies Evaluating Performance Characteristics of dd-cfDNA to detect Allograft Injury in Lung Transplant Recipients
Authors and Date Published	Study Design	# of Patients	Approach to detect dd-cfDNA	dd-cfDNA Threshold	Primary Endpoint	Performance Characteristics
De Vlaminck et al. 2015 27	Single Center Prospective Cohort Study	51	Two Genome Genotyping	> 1%	ACR ≥ 2R	Sensitivity: 100% Specificity: 73% AUC: 0.9
Sayah et al. 2020	Multicenter Prospective Cohort Study	26	Targeted SNPs	> 0.85%	ACR ≥ 1R	Sensitivity: 73% Specificity: 53% AUC: 0.72
Khush et al. 2021	Single Center Prospective Cohort Study	38	Targeted SNPs	> 0.85%	ACR ≥1R or AMR or BOS	Sensitivity: 56% Specificity:76% NPV: 84% AUC: 0.67
Jang et al. 2021	Multicenter Prospective Cohort Study	148	Two Genome Genotyping	> 1%	ACR ≥1R or AMR	Sensitivity: 77% Specificity: 84% NPV: 90% AUC: 0.89
Performance of ddcfDNA to monitor for acute rejection or infection in routine clinical care
Authors and Date Published	Study Design	# of Patients	Approach to detect dd-cfDNA	dd-cfDNA Threshold	Primary Endpoint	Performance Characteristics
Keller et al. 2021 *	Multicenter Retrospective Cohort Study	157	Targeted SNPs	> 1%	ACR ≥1R, AMR or Infection	Sensitivity: 74% Specificity: 88% NPV: 97% AUC 0.81
Association of CLAD and plasma dd-cfDNA
Authors and Date Published	Study Design	# of Patients	Approach to detect dd-cfDNA	Primary endpoint	Main findings
Keller et al. 2021	Multicenter Prospective Cohort Study	99	Two Genome Genotyping	CLAD	PGD patients who developed CLAD showed higher dd-cfDNA on Day 3 of transplant compared to PGD patients who did not develop CLAD.
Bazemore et al. 2021	Multicenter Prospective Cohort Study	51	Two Genome Genotyping	CLAD	CLAD associated pathogens showed higher dd-cfDNA levels compare to non-CLAD-associated pathogens
Agbor-Enoh et al. 2019	Multicenter Prospective Cohort Study	108	Two Genome Genotyping	Composite endpoint of CLAD or death	Patients with a higher dd-cfDNA levels in the first 3 months of transplant showed a 7-fold higher risk of allograft failure compared to patients with lower levels

^*Stable patients in this study not all evaluated with bronchoscopy and were considered stable by absence of signs or symptoms of allograft dysfunction over the course of the study period with at least 1–3 months follow up.

Background: Detection and Quantitation of dd-cfDNA

Cells undergoing apoptosis and necrosis release fragments of double-stranded DNA into the bloodstream. These fragments, known as cell-free DNA (cfDNA), have an estimated half-life of 16 minutes to 2.5 hours and are excreted in urine or degraded by resident macrophages in the liver or spleen(14–16). In quiescent states, plasma levels are generally low, however, levels increase in the setting of tissue injury. Methods for detecting tissue-specific cfDNA have evolved significantly. Solid organ transplant creates a unique scenario, producing an admixture between donor and recipient genomes. Thus, it is possible to genetically distinguish cfDNA originating from the donor allograft tissue vs cfDNA coming from recipient tissue. In this manner, levels of dd-cfDNA provide insight into the degree of allograft injury on the molecular level.

Assays to measure plasma dd-cfDNA typically rely on 2 components: 1) a method of identifying donor vs recipient genetic material and 2) a method of quantifying or semi-quantifying the amount of dd-cfDNA present. While initial approaches utilized sex-specific chromosomal differences or human leukocyte antigen (HLA) mismatch to identify dd-cfDNA(17, 18), recent methods leverage differences in specific single nucleotide polymorphisms (SNPs) between donor and recipient. SNP-based approaches may also vary, and include: 1) whole-genome genotyping of pre-transplant donor and recipient DNA to compare donor-recipient pairs for identification of informative SNPs (two-genome approach)(19), 2) genotyping only the recipient and using computational algorithms to estimate informative donor and recipient SNPs (one-genome approach)(20) and 3) leveraging publicly available population genomic data to establish sets of common and informative population-based SNPs which are mapped onto a donor-recipient pair for comparison (eliminating the need for donor-recipient genotyping)(21, 22).

After identifying differences in donor vs recipient genetic material, methods of amplifying and quantifying cfDNA include whole-genome sequencing, targeted genome sequencing, quantitative PCR (qPCR) and digital droplet PCR (ddPCR). Given that allograft size (and therefore, absolute amounts of dd-cfDNA) varies between patients, dd-cfDNA is typically reported as a percentage of donor-derived to total cfDNA (dd-cfDNA plus recipient cfDNA). The latter approach (ddPCR) quantitates absolute amount of donor and recipient cfDNA in the sample. The utility of absolute amount versus percentage of dd-cfDNA to detect allograft rejection in lung transplantation remains unknown.

dd-cfDNA as a Marker of Allograft Injury in Lung Transplant Recipients

In an early study investigating the association of dd-cfDNA with allograft injury, De Vlaminck et al. performed a prospective single-center cohort study collecting serial plasma samples at several defined timepoints after lung transplant in 51 consecutively enrolled patients, including at the time of bronchoscopy(23). Using a two-genome approach and whole-genome sequencing to identify and quantitate dd-cfDNA, respectively, they demonstrated that dd-cfDNA (expressed as a percentage) increased at the time of moderate or severe ACR compared to stable controls. Notably, they reported that a threshold value of >1% dd-cfDNA showed 100% sensitivity and 73% specificity for detection of moderate to severe ACR. Since then, additional studies have also established an increase in plasma dd-cfDNA with lung allograft injury, with several studies estimating the performance characteristics of dd-cfDNA to detect various forms of allograft injury (11, 13, 24–27).

A recent multicenter prospective cohort study, performed by Jang et al., evaluated levels of dd-cfDNA in the setting of acute rejection and positive BAL microbiology vs stable controls(11). The study enrolled 148 lung transplant recipients and collected serial plasma samples at the time of all surveillance and clinically indicated bronchoscopy and PFTs. Using a two-genome approach with whole-genome sequencing to identify and quantitate dd-cfDNA, respectively, this study demonstrated median levels of dd-cfDNA of 1.95% in the setting of acute rejection (n = 87) vs. 0.30% in stable controls. At an optimum threshold value of >1%, dd-cfDNA demonstrated 77% sensitivity and 84% specificity with an area under the receiver-operating characteristic curve (AUC) of 0.89 for detection of acute rejection, a composite endpoint of both ACR and AMR. The positive predictive value (PPV) was 60% and negative predictive value (NPV) was 90%. Levels of dd-cfDNA were significantly higher in AMR compared to ACR. A key observation in this study was that levels of dd-cfDNA were often elevated weeks-to-months prior to the diagnosis of acute rejection, particularly in AMR, where levels often began to rise at the time of identification of DSA. This finding suggests that molecular evidence of allograft injury in the setting of acute rejection may begin significantly earlier than clinical and histopathological signs. While levels of dd-cfDNA in patients with positive microbiology were not significantly elevated in comparison to stable controls, those with positive microbiology and a concomitant decline in PFT values (or abnormal histopathology) did demonstrate higher levels of dd-cfDNA, suggesting that dd-cfDNA may be elevated in the setting of clinical infection rather than simply microbiological colonization.

In another single-center prospective cohort study conducted by Khush et al., 38 lung transplant recipients underwent serial measurement of plasma dd-cfDNA, including at the time of bronchoscopy with TBBx(25). Using a commercially available targeted next generation sequencing technique based on a set of previously established informative SNPs from population-based genomic data, this study demonstrated that median levels of dd-cfDNA were higher in episodes of ACR (n = 28) vs stable controls (0.91% vs 0.38%; p = 0.02). Patients with Bronchiolitis Obliterans (BOS) phenotype CLAD also demonstrated higher median levels of dd-cfDNA vs healthy controls (2.06% vs 0.38%; p = 0.02). Median levels of dd-cfDNA were not higher in patients with AMR vs stable controls (1.34% vs 0.38%; p = 0.07), however, the sample size for AMR was small (n = 9). Likewise, patients with positive microbiology alone did not demonstrate higher levels of dd-cfDNA, potentially reflecting the effects of colonization vs clinical infection.

dd-cfDNA as a Predictor of Allograft Failure

In addition to its potential for detecting underlying allograft injury, dd-cfDNA is emerging as a promising risk stratification tool to predict downstream allograft failure in lung transplantation. A recent multicenter prospective cohort study performed by our group enrolled 108 lung transplant recipients with a median post-transplant follow up 36.4 months and demonstrated that elevated levels of dd-cfDNA in the first 90 days post-transplant were associated with an increased risk of allograft failure (a composite of CLAD and death)(12). After stratifying subjects into terciles based on average levels of dd-cfDNA in the first 90 days post-transplant, we also observed 3 distinct patterns of dd-cfDNA decay kinetics among individuals that further distinguished the risk of allograft failure. While all 3 groups demonstrated elevated levels of dd-cfDNA in the post-transplant period, there was variability between the groups in the rate of dd-cfDNA decline. The lowest tercile group displayed a rapid decline in dd-cfDNA to low, baseline levels within 1 month of transplantation while the middle tercile experienced a slower decline to baseline levels by 90 days post-transplant. The upper tercile group, however, demonstrated the slowest decline, with persistently elevated levels of dd-cfDNA compared to the other groups that did not return to low baseline levels. Patients in the upper tercile had a 6.6-fold higher risk of CLAD or death compared to the lowest tercile group with a median time to allograft failure of 25 months vs. 45 and 42 months in the lower and middle tercile groups, respectively. A follow-up study by our group also demonstrated that levels of dd-cfDNA were elevated in the setting of Primary Graft Dysfunction (PGD) at 72 hours post-transplant and correlated with the clinical severity of PGD(13). Likewise, levels of dd-cfDNA in patients with PGD were associated with an increased risk of developing CLAD, providing further insight into the role of early allograft injury in the development of downstream allograft failure.

Future Perspective

Given the emerging data surrounding the use of dd-cfDNA in lung transplantation, what role can it play in advancing clinical paradigms and research practices? There appears to be great promise in both regards.

Clinical Applications of dd-cfDNA

From a clinical perspective, the strongest evidence lies with the ability of dd-cfDNA to detect episodes of acute allograft injury, particularly acute rejection. Specifically, given the reported high negative predictive values of dd-cfDNA for acute rejection, it appears well suited as a non-invasive surveillance monitoring method in the post-transplant setting. This approach is further supported by the results of a recent multicenter retrospective cohort study documenting the first use of dd-cfDNA in the routine clinical care of lung transplant recipients(26). In this study, in an effort to reduce the risk of COVID-19 infection at the onset of the COVID-19 pandemic, 4 lung transplant centers selectively deferred surveillance bronchoscopy and clinic-based PFTs, and instead enrolled patients into a home-based monitoring program using dd-cfDNA and home spirometry to monitor for episodes of acute rejection and infection. Overall, 198 patients were enrolled in this program, receiving >400 plasma samples for dd-cfDNA during the initial six months of the study. Using the dd-cfDNA threshold value of ≥ 1% to trigger bronchoscopy and in-clinic PFTs, this approach identified 23 episodes of asymptomatic acute rejection and infection that required treatment while reducing the expected amount of surveillance bronchoscopies performed by 81%. Furthermore, >96% of patients with dd-cfDNA values <1% remained stable, without the development of signs or symptoms of allograft dysfunction over the course of the study. Although a key limitation of this study was the inability to definitively identify episodes of sub-clinical acute rejection or infection (that either spontaneously resolved or persisted sub-clinically) in patients with dd-cfDNA levels < 1% due to the absence of routine surveillance bronchoscopy, the study demonstrated the feasibility and relative safety of incorporating a dd-cfDNA based monitoring approach in a real-world setting. This study also further supports the potential for a randomized controlled trial comparing a dd-cfDNA based monitoring approach with surveillance bronchoscopy.

Aside from its potential role in post-transplant surveillance monitoring, dd-cfDNA may offer several additional clinically relevant applications. As an early predictor of downstream long-term adverse outcomes, dd-cfDNA may convey important prognostic information to more precisely risk stratify lung transplant recipients and identify patients that warrant increased vigilance or modified treatment in the post-transplant setting. In addition, as levels of dd-cfDNA increase in the setting of underlying allograft injury, dd-cfDNA may serve as a useful tool to monitor the response to treatment and guide therapeutic decision making. Likewise, dd-cfDNA may serve a role in tailoring maintenance immunosuppressive therapy to the degree of subclinical allograft injury on the molecular level, as suggested by recent literature in kidney transplant populations(28). These potential clinical applications warrant further investigation (Figure 1).

Figure 1: Potential for individualized therapy in lung transplant patients using dd-cfDNA –

This figure illustrates potential clinical applications of dd-cfDNA at different time points in a lung transplant patient undergoing serial measurements. In combination with other established standards of care, plasma dd-cfDNA has the potential to move care toward precision medicine, with personalized management strategies geared toward treating and preventing molecular evidence of allograft injury. AMR: anti-body mediated rejection; CLAD: chronic lung allograft dysfunction; dd-cfDNA: donor-derived cell-free DNA; DSA: donor specific antibodies; PFTs: pulmonary function tests.

dd-cfDNA as a Research Tool

Beyond the clinical landscape, dd-cfDNA holds great promise for advancing our understanding of several aspects of lung transplantation. As a molecular marker of allograft injury, dd-cfDNA has potential to guide further mechanistic study into the pathophysiology of various disease states in lung transplant. Identification of the timing and degree of molecular evidence of allograft injury, and its association with other immunologic and histopathologic alterations, may provide insight into mechanisms of various forms of allograft injury. This may be further augmented by the recent development of epigenetic advances to identify tissue-specific cfDNA patterns that identify the cfDNA tissue of origin. These techniques include identifying varying cfDNA tissue-specific methylation patterns(29–32), cfDNA fragment size (fragmentomics) and chromatin footprinting topological differences(33) to determine the tissue of origin. This opens the door for not only additional mechanistic understanding of pathophysiological disease states, but also more precise profiling and phenotyping of the underlying cause given that distinct tissue types are involved in different disease states. The ability to model the trajectory of underlying molecular allograft injury in patients until the development of CLAD, coupled to other deep molecular approaches to define mechanisms of injury, may allow for further understanding of the relationship between prior episodes of allograft injury and the development of CLAD.

Furthermore, investigation into the population of patients with molecular evidence of allograft injury in the absence of clinical or histopathological signs of allograft injury may encourage the application of other novel methods of molecular phenotyping in this patient population to further identify the mechanism of subclinical allograft injury. Utilizing the precepts of dd-cfDNA as a risk stratification tool for patients at risk of poor long terms outcomes, dd-cfDNA may identify at-risk patients for consideration in the enrollment in clinical trials evaluating measures to prevent the development of downstream allograft failure.

Mounting evidence suggests that cfDNA may directly inform or contribute to disease states. In a recent study, Sadeh and colleagues used chromatin immunoprecipitation to analyze plasma cfDNA in healthy controls and various disease states(34). In healthy controls, this approach identified constitutionally active gene sets. In various disease states, this approach then identified the underlying molecular mechanisms correlating with tissue RNA sequencing signals. In another study, Andargie and colleagues observed significant elevations of cfDNA in patients with COVID-19. Plasma from these patients, as well as the isolated cfDNA, triggered production of mitochondrial reactive oxygen species in a time and concentration dependent manner - a process that was inhibited by toll-like receptor 9 inhibitors(32). The toll-like receptor and other molecular pathways through which cfDNA could directly trigger cell and tissue injury was recently reviewed(35). These studies suggest that in addition to being a sensitive biomarker, cfDNA may identify, contribute and/or amplify the pathology of the underlying disease state. The application of these new insights in lung transplantation deserves further investigation.

Limitations

While dd-cfDNA possesses several potential clinical and investigational applications in lung transplantation, some limitations warrant further study. First, to account for differences in the relative sizes of recipient and donor allografts, dd-cfDNA is typically reported as a percentage, reflecting the fraction of donor to donor + recipient cfDNA. Because recipient cfDNA lies in the denominator of this fraction, significant changes in levels of recipient cfDNA may impact levels of dd-cfDNA in the absence of changes in the degree of allograft injury. Levels of recipient cfDNA may fluctuate in several situations including rises in the setting of sepsis, leukocytosis, multiorgan failure or extreme exercise, therefore affecting interpretation of dd-cfDNA levels. While measurement of absolute levels of dd-cfDNA are not dependent on recipient cfDNA levels, variations in allograft size may impact interpretation and render the application of specific threshold values for identification of various disease states challenging. The performance characteristics of absolute levels of dd-cfDNA in comparison to measures as a percentage in lung transplant patients is an active area of investigation. Similarly, the impact of single vs. double lung transplant on dd-cfDNA levels warrants further investigation. Several studies have doubled the value of dd-cfDNA in single lung transplant patients to account for differences in lung tissue mass, however, other studies have not accounted for single vs double lung transplant. Significant differences in dd-cfDNA levels between single vs double lung transplant in both stable patients and across varying disease states may affect the interpretation of dd-cfDNA values and performance characteristics at specific threshold values. This is also an area of ongoing research. Lastly, while dd-cfDNA is a sensitive marker of allograft injury, it is not specific to particular disease states and may increase with various forms of allograft injury including ACR, AMR, infection and CLAD. As previously stated, advances in epigenetic techniques to identify tissue of origin and provide a map of allograft injury may further clarify this issue.

Conclusion

Advances in molecular approaches of identifying, risk stratifying, phenotyping and treating various forms of allograft injury are essential to advancing the field of lung transplantion in a direction geared toward precision medicine. As a non-invasive, quantitative biomarker of allograft injury, dd-cfDNA holds significant promise in realizing this goal, both through its broad potential clinical applications and as a valuable research tool.

Key Points.

• Levels of plasma dd-cfDNA rise in the setting of allograft injury, including in episodes of acute rejection, infection and primary graft dysfunction.

• Elevations of dd-cfDNA in the early post-transplant period are associated with an increased risk of chronic lung allograft dysfunction and death.

• Ongoing research is being performed using dd-cfDNA to further define pathophysiological mechanisms of various allograft injury states.