Markers of rejection of a lung allograft: state of the art

Lung transplantation improves the life expectancy and quality of life in patients with progressive lung disease such as chronic obstructive pulmonary disease (COPD), pulmonary fibrosis (PF) and cystic fibrosis (CF). However, despite increased organ procurement, improved surgical techniques and perioperative care since the early lung transplantation era of the 1980s, long-term survival trends (greater than 5 years) remain similar. Transplanted lungs remain at risk of chronic rejection, also called chronic lung allograft dysfunction (CLAD) (reviewed in [1]). CLAD affects approximately 50% of all lung transplant recipients by 5 post-operative years. With a survival rate of approximately 50% at 2.5 years post-diagnosis, it remains the leading cause of death in lung transplant recipients, reducing quality of life and increasing economic burden [2].

CLAD is defined as a persistent, irreversible drop in lung function measured as forced expiratory volume in 1 s (FEV1) or forced vital capacity (FVC) to 20% or more from the post-transplant baseline after exclusion of alternate causes for reduced graft function [3]. CLAD was initially thought to result from T-cell injury; however, a growing body of evidence suggests that multiple arms of the immune system are involved, including monocytes, neutrophils and autoantibodies (reviewed in [1]). Other events such as gastroesophageal reflux, infection, reperfusion injury and air pollution can impose a higher risk of developing CLAD (reviewed in [1]). Such persistent insult activates or inflames the epithelium, combining with impaired wound healing and leading to lung fibrosis and decline in lung function (reviewed in [1]).

Heterogeneity of CLAD

Distinct phenotypes of CLAD occur with variable pathology, diverse airway neutrophilia, fibrosis, histological features and responsiveness to therapy. The recognized phenotypes are bronchiolitis obliterans syndrome (BOS) and restrictive allograft syndrome (RAS) [3]. BOS is the most common form of CLAD and is seen in 50% of patients within 5 years of transplant (reviewed in [4]). BOS is defined by a persistent 20% drop in FEV1 and an obstructive pattern in pulmonary function tests (PFTs) with no or minimal evidence of parenchymal fibrosis [5]. RAS is less common but has a worse prognosis and is classified by a 20% drop in FEV1 and reduced FVC and total lung capacity, but it has a restrictive pattern in PFTs and opacities in chest CT scans and x-rays [6]. Complicating diagnosis, CLAD may arise in mixed forms with similar mechanisms of disease progression. Reliable biomarkers that identify these phenotypes are yet to be identified (reviewed in [7,8]).

The need for CLAD biomarkers

Persistent episodes of acute rejection are a risk factor for CLAD and currently diagnosed by decline in lung function and perivascular cellular infiltrates present in transbronchial biopsies (TBBs). Bronchoalveolar lavage (BAL) is also an important tool to monitor rejection. Cellular composition (increased lymphocyte to macrophage ratio) is diagnostic of rejection, and culture of supernatant or PCR can identify infectious agents (bacteria, viruses, fungi) that impact lung function. Although biopsies are routinely performed, the procedure has only 30% sensitivity to identify chronic rejection, is invasive, imposes procedural risks (such as bleeding and pneumothorax), requires multiple sampling to reflect overall lung pathology and is subject to variability due to observational bias. Therefore, there is a need to identify biomarkers of cellular rejection preceding CLAD to stratify patients according to risk and facilitate personalized treatment options based on risk levels. Several biomarkers associated with CLAD have been evaluated, but implementation into diagnostics has not yet eventuated due to poor specificity or sensitivity, as well as failure to detect early-stage disease (reviewed in [7,8]). Early CLAD diagnosis or ideally prediction of CLAD is essential to enable early intervention before significant lung injury occurs. This review examines the current state of research on biomarkers for monitoring or detecting lung transplant rejection.

Biomarkers in blood

As blood is easily accessible, several studies have examined blood biomarkers and their association with lung rejection and CLAD. Although blood provides insight into systemic changes in the body, blood biomarkers may not necessarily reflect the lung microenvironment. This is an important consideration in lung transplantation, in which monitoring the underlying systemic allo-immune response to the transplanted lung is required to detect low-grade rejection. Several studies have investigated various constituents of blood, including cells [9–19], cytokines and chemokines [20–23], in search of biomarkers for chronic rejection in lung transplant patients.

Of the immune cells found in blood, Tregs are the most intensively studied cell type in lung transplant rejection due to their key role in immune regulation. Multiple Treg subpopulations have been studied in lung transplantation and are distinguished based on the presence of markers such as CD25 and FoxP3. Several studies report increased Tregs in patients experiencing CLAD compared with stable patients [10,12,14,15], whereas other studies show a decrease in Tregs [13,24]. Other than Tregs, Budding et al. investigated variations in other cell types in the blood of lung transplant recipients with and without BOS [16]. In their study, CD8⁺ T lymphocytes, B lymphocytes and monocytes were decreased, and natural killer cells increased, in BOS patients. Eosinophils in blood were increased in patients developing CLAD [17,18]. Coiffard found blood neutrophils elevated in patients with CLAD [19]. Bergantini et al. also showed reduced B lymphocytes during BOS [25]. Table 1 summarizes studies investigating immune cells in blood in relation to rejection and CLAD.

A number of pro-inflammatory cytokines and chemokines involved in cell recruitment and activation correlate with lung transplant rejection: IL-1 and IL-12 increased in BOS [20]; IFN-γ and MCP-1 increased with lung injury associated with BOS/CLAD [20,26]; the inflammatory marker CRP was elevated in the blood of patients with BOS [23]; and the anti-inflammatory cytokines IL-10 and IL-4 were decreased in patients with BOS [20,21].

Biomarkers in BAL

BAL is a routine technique performed at the time of bronchoscopy to collect respiratory secretions for diagnosis and research purposes [27]. Compared with TBB, BAL is safer and samples a relatively larger area of the lung, enabling repeated access to transplanted lungs [28]. Despite its lower invasiveness, the procedure itself has several limitations: different procedures in collecting BAL and sampling size result in variability [27], and the composition and volume of the instilled fluid together with dilution factors related to the BAL technique, alveolar permeability at the time of sampling and the contamination of BAL with blood affect the BAL fluid composition [29]. In order to correct for dilution factors in BAL fluid, the coefficient of urea-in-plasma/urea-in-BAL is used for standardization [30]; however, infections can increase urea in the lung and plasma, leading to inaccurate quantitation [31]. To minimize inaccuracies, The International Society for Heart and Lung Transplantation has taken steps to standardize BAL techniques among centers [27], enabling the comparison of findings across studies. Notwithstanding these limitations, BAL fluid remains an important window into the biomarker status of the lung following lung transplantation.

Like blood, cells collected from BAL fluid have been analyzed to investigate their relationship to lung transplant rejection (summarized in Table 2). Tregs have been shown to decrease in the BAL of patients experiencing CLAD [10,11,25,32]. Multiple studies also show that there is an elevation of total lymphocytes in the BAL in CLAD patients compared with that of stable lung transplant recipients [33–35]. Studies report discordant results with respect to macrophages in BAL with increasing proportions in CLAD (RAS and BOS) [33] or a decrease in proportion in BOS patients [34]. Multiple studies have reported an increase in neutrophils in the BAL with CLAD [33–36]. Furthermore, increased eosinophil counts in patients developing CLAD [17] and decreased Bregs during BOS have been reported [25].

Cytokines and chemokines in BAL fluid have been analyzed in relation to lung transplant rejection (summarized in Table 3). A consistent finding is that pro-inflammatory markers correlate with rejection: IL-1 and IL-6 are increased in BAL from patients with BOS [38–40]; however, no differences were observed in IL-6 between acute rejection and BOS [41]. Multiple studies show IL-8 is increased in BAL from patients with BOS compared with stable patients [38,42,43]. Both pro-inflammatory cytokines, IFN-γ and TNF-α and fibrotic cytokines, TGF-β, are elevated in BAL of BOS [40,42,44].

Although these studies highlight CLAD-specific biomarkers, to date no point of care tests using blood or BAL biomarkers have been developed to monitor lung transplant rejection.

Epigenetic markers

DNA methylation is a reversible gene regulatory process that is modifiable by aging, environmental influences and lifestyle factors such as smoking or diet. In particular, oxidative stress can induce promoter CpG methylation by recruiting MECP2 and DNA methyltransferases (reviewed in [45]). DNA methylation, like telomeres, can provide an approximate age of cells (‘epigenetic clock’) (reviewed in [46]). The difference between DNA methylation predicted age (DNAm age) and chronological age is the accelerated epigenetic age and is associated with several diseases [47]. DNAm age provides a reliable estimate of cell age at different time points of life. DNAm age has been successfully used in the identification of high-risk individuals prone to accelerated aging in a number of diseases such as cancer, diabetes, cardiovascular diseases, dementia and post-traumatic stress syndrome (PTSD) (reviewed in [46]). Recent reports show that idiopathic pulmonary fibrosis (IPF) lungs have increased DNA methylation status in gene promoters [48,49]. Differential DNA methylation may provide a potential novel epigenetic biomarker for smoking-related lung diseases, including COPD and lung cancer (reviewed in [50]). A recent epigenome-wide association study identified 59 differentially methylated regions (DMRs) in cord blood associated with childhood lung function, as well as 15% associated with COPD in adults [51]. Another study of peripheral blood mononuclear cell (PBMC) methylation using Illumina Infinium^® MethylationEPIC (EPIC) arrays identified 28 differentially methylated sites, with 26 associated with lung function and two associated with COPD (cg18181703 in SOCS3 and cg18608055 in SBNO2) [52]. In transplant patients, epigenetic markers responsible for the regulation and expression of genes important in rejection have not been identified. Genome-wide epigenetic analysis of stable and CLAD patients is needed to identify biomarkers for rejection. A handful of studies have examined epigenetic biomarkers predicting rejection episodes in heart (reviewed in [53]) and kidney transplantation (reviewed in [54]) but, to the authors' knowledge, epigenetic markers in lung transplantation are yet to be identified. DNA methylation of genes activated during rejection episodes could provide valuable information on the underlying immune and other cell pathways that are active and may predict post-transplant complications such as ischemia injury and rejection.

Gene-expression markers

Because acute cellular rejection (ACR) pathologies have been associated with increased CLAD risk, methods to detect cellular rejection are important for clinicians to monitor and treat rejection episodes prior to lung function decline. The gold standard continues to be TBB and BAL using histological scoring and lymphocyte counts, respectively, to approximate T-cell numbers [55]. However, histological assessment of T-cell-mediated rejection is difficult to reproduce, and failure to diagnose rejection can lead to insufficient or excessive immunosuppression [56]. Thus, there has been an effort to identify biomarkers of cellular rejection. Lung cell profiles translate to unique transcriptome signatures, providing potential biomarkers to aberrant pathways in lung transplant rejection. Methods to measure differential gene expression in cells and tissue include RNA micro-array profiling and RNA sequencing, which offers increased dynamic range and enhanced the sensitivity/detection of low abundance transcript. Both approaches have led to the identification of biomarkers associated with rejection and the use of PCR-based assays for specificity and sensitivity in detecting rejection. An early study using micro-array compared 32 TBB samples from patients whose concurrent histology showed acute rejection (n = 14) or no rejection (n = 18) and identified over 100 genes differentially expressed [57]. Consistent with the paradigm of ACR and CLAD pathobiology of activated T cells leading to cytotoxicity, causing allograft injury and CLAD, genes enriched in T-cell function, cytotoxic CD8 activity and granulocyte degranulation were identified. Translating this to a potential diagnostic qPCR test, ten of the 52 genes showed increased expression in samples from patients with acute rejection [57]. Although these studies present promising evidence showing a T-cell molecular signature in rejection, no diagnostic test has replaced TBB and BAL.

In another study, of RNA profiling of blood cells from lung transplant recipients, Danger et al. identified three downregulated genes, POU2AF1, TCL1A and BLK, that predicted BOS 6 months before diagnosis compared with stable lung transplant recipients. Levels of gene expression stratified patients in survival analysis based on BOS risk [63]. A similar study identified 40 genes that were differentially expressed in BAL cells between CLAD and CLAD-free patients. Genes were related to recruitment, retention, activation and proliferation of cytotoxic lymphocytes such as CD8⁺ T cells and natural killer cells, consistent with episodes of repeat acute cellular rejection and CLAD [28]. Meta-analysis of datasets from 236 graft biopsy samples (kidney, heart, lung and liver) identified a common rejection module (CRM) consisting of 11 genes, overexpressed in acute rejection, across all transplanted organs [64]. Applied to patient biopsies 6 months post-lung transplant, CRM was upregulated in four TBB rejection positive samples compared with ten TBBs without rejection. A difference was also observed with CLAD phenotypes, with CRM higher in RAS explanted lungs compared with donor lungs. Two genes in the CRM (ISG20 and CXCL9) were upregulated in RAS or BOS lungs compared with donor lungs [65]. Although promising, the CRM was only applicable to TBB and not BAL cells, limiting the applicability of the assay.

An alternate approach to PCR is NanoString, a clinical platform superior to TBB and BAL. NanoString technology has several benefits over other methods, including no requirement for amplification, high sensitivity and specificity. This approach has potential for an automated platform for routine diagnostic workflows across pathology laboratories [66]. Banff Molecular Diagnostics Working Group has taken the initiative to pioneer the Banff Human Organ Transplant (B-HOT) panel, using NanoString technology to include validated genes from previous studies on kidney, heart, lung and liver allograft biopsies. The panel consists of 770 genes associated with organ rejection, tissue damage and immune responses. In practice, this panel would be useful to detect and monitor rejection with work ongoing to validate the accuracy and sensitivity of the panel [66]. This approach is promising, as both rejection-associated tissue damage and immune reactivity are measured simultaneously.

With advances in technologies, whole transcriptomic analysis of cells has become affordable and rapid and is becoming an essential tool in biomarker discovery. Generating a vast amount of useful data in a short time, genes co-regulated with disease are identifiable and pathways assembled linked to disease. In transplantation, transcriptomics will identify molecular mediators and mechanisms of tissue damage leading to rejection,, including genes and pathways predictive of pathological changes in the allograft and likelihood of graft outcome, for example, time to CLAD. Cost–effectiveness and reproducibility across multiple centers make transcriptomics a potential game changer for the diagnosis, prognosis and management of transplant patients.

In addition to gene expression, single-nucleotide polymorphisms (SNPs) associate with post-lung transplant mortality. Notably, donor SNPs in human leukocyte antigen (HLA) G gene (HLA-G*01:04 haplotype) increase expression of HLA-G and exacerbate CLAD risk and mortality [67]. Soluble HLA-G concentrations in BAL correlate with the number of acute rejection episodes in the first 12 months after lung transplantation and may be a biomarker of rejection [68,69]. Similarly, HLA-E*01:03 allele is associated with the development of CLAD [70]. Other SNPs associated with CLAD or mortality include Fc Gamma Receptor IIa [71,72], IL-17R and IL-23R [73] and Dectin-1 (CLEC7A gene) [74]. A common theme of these SNPs is immune modulation, which impacts graft survival and tolerance. SNP typing of patients or measurement of impacted protein biomarkers will likely be useful in the stratification of patients for risk and the implementation of personalized targeted therapeutics.

miRNA markers

miRNAs are non-coding, single-stranded RNA 22–25 nucleotides in length, with the ability to repress translation and/or degrade mRNA as a post-transcriptional regulatory point in gene expression by binding to the 3′ UTR of target genes. Their origin is unclear but they may arise from damaged or apoptotic cells. In lung transplantation, miRNA expression profiling of PBMCs and BAL in stable lung transplant recipients, recipients with donor-specific antibodies (DSA) and no BOS and recipients with DSA and BOS found different miRNA profiles with respect to differentially expressed genes in signaling pathways of B cells, Toll-like receptors, TGF-β, chemokine receptors and cytokine–cytokine interactions [75]. Interestingly, in BOS, two miRNAs (miR-10a and miR-133b) were differentially expressed in both BAL and blood. miR-10a is involved in the differentiation of hematopoietic cells specifically in Treg stability and functioning [76] and miR-133b co-regulated with IL-17 production [77]. Variations in the expression of these miRNAs preceded BOS development and most of the differently expressed miRNA was involved in vital cellular processes such as cell cycle signaling pathways [75]. These findings highlight a likely role of miRNA in the development of BOS and the utility of miRNA as a biomarker for early detection of rejection. However, studies in larger cohorts are required to validate the accuracy and reproducibility of these results.

cfDNA markers

cfDNA is DNA released into the bloodstream by dying cells and is a direct measure of organ injury. Increased cfDNA occurs with sepsis, trauma and cancer (reviewed in [78]). Importantly in transplantation, genetic disparity between the donor and recipient enables discrimination and quantification of donor-derived cfDNA (ddcfDNA) as a marker of allograft injury [79]. Due to its short half-life, ddcfDNA can provide a time-dependent resolution to understanding ongoing immunomodulatory events occurring at the time of rejection. This has been demonstrated in kidney, heart and liver transplantation (reviewed in [80]).

A handful of studies have investigated the possibility of using ddcfDNA in lung transplantation to detect, assess and monitor rejection. A proof-of-concept study by Agbor-Enoh et al. showed that the average percentage of ddcfDNA in plasma strongly associates with the development of allograft failure and all-cause mortality in lung transplantation. Importantly, this study showed the average percentage of ddcfDNA detected clinically silent episodes of acute rejections and enabled stratification of the patient cohort based on the risk of subsequent allograft injury [81]. An independent study reported that cfDNA in BAL was elevated in patients with BOS compared with stable patients and combining cfDNA and CXCL10 quantitation in BAL facilitated the stratification of BOS, RAS and stable phenotypes [82]. These findings reinforce the idea that multiple biomarkers may offer the most sensitive and accurate approach to diagnosing and predicting lung transplant rejection.

There are, however, limitations to measuring ddcfDNA, including the need to sequence and SNP type both donor and recipient cells. Furthermore, reliance on the percentage of ddcfDNA may be influenced by changes in recipient cfDNA levels due to infection and granulocyte cell death (reviewed in [83]).

An alternative approach to measuring total ddcfDNA is to use methylation-specific PCR to quantitate only transplant rejection-associated cell cfDNA and use this as a biomarker of CLAD risk. Methylation of DNA-encoding genes and transcriptional enhancers is essential to silencing gene expression. Conversely, demethylation, which can occur throughout life, enables transcription factors to bind DNA and induce gene expression. Thus, methylation patterns are ideal for identifying cell types in mixed populations, as they reflect the transcriptome, function and ontogeny of cells [84]. Measuring DNA methylation of cfDNA for the identification of new biomarkers in lung transplant rejection may offer a sensitive method of measuring donor lung damage preceding CLAD.

Biomarkers of allo-immune & auto-immune responses

Allo-immune responses impact CLAD with mismatched HLAs increasing risk [85–90]. HLA mismatching of donor and recipient results in anti-HLA donor-specific antibodies (DSAs). Pre-transplant DSAs are a prominent risk factor for CLAD and mortality [91,92] and, in organ allocation, potentially reactive HLA mismatches in the donors via virtual crossmatching of the recipient are avoided [93,94]. Similarly, post-transplant DSAs increase CLAD [91,95–103] and DSA measurements are routinely used in recipient follow-up and risk stratification [104]. The combined use of HLA antibodies and CXCL9 levels has recently enabled the identification of recipients at high risk of CLAD [105]. Interestingly, lung transplant recipients with antibodies to the ubiquitous self-antigens, collagen V (ColV) and K-α 1 tubulin (Kα1T) have increased the risk of chronic rejection [106–110].

Many of the previously mentioned immune modulatory biomarkers are present in circulating exosomes. Exosomes are released from donor and recipient cells and contain biomarkers relevant to solid organ transplantation tolerance and rejection [111,112]. In lung transplantation, mismatched donor HLA and lung antigens, including ColV and Kα1T; immune modulatory molecules, including miRNAs responsible for inducing inflammation; TH-17 differentiation; and endothelial activation have been identified in exosomes during acute or chronic rejection [108,113–118]. These and other biomarkers are reflective of allo- and auto-immune modulation, and their abundance may reveal lung transplant recipients with a heightened risk of CLAD. Combined with additional novel biomarkers, including lung microbiome characteristics [119] and nutritional assessment scores such as prognostic nutrition index (PNI) [120], the accuracy of CLAD prediction may be improved. Combinatorial biomarker assessment would increase pre-operative risk management in lung transplantation and efficient organ procurement when used together with recently developed virtual crossmatch-based strategies [94]. Utilization of these novel approaches with multi-omic profiling [121] would aid in the effective prediction of CLAD and lung allograft outcome.

Biomarkers of biological aging

Aging in general is the progressive decline in homeostasis after the reproductive phase of life, leading to increasing risk of disease or death. Though biological aging is often linked to chronological aging, it can also occur earlier in life due to malfunctions in organ or cell maintenance and repair (reviewed in [122]). Such accelerated biological aging has been emphasized as a strong risk factor for a number of human diseases such as cardiovascular diseases, metabolic diseases, neurological diseases and cancer (reviewed in [123]). Markers of biological aging may be reliable clinical biomarkers to detect, predict and monitor age-related diseases. Markers of aging include cellular senescence, genomic instability, telomere attrition and mitochondrial dysfunction (reviewed in [123]). Lung function declines with age [124] and the correlation of lung capacity with a biomarker of accelerated aging of the lung may be a good predictor of lung transplant rejection.

Telomere length is a strong predictor of biological aging in cells. Telomeres are nucleotide repeats of TTAGGG at the ends of chromosomes that provide protection during cell replication. Telomeres shorten over time and extreme shortening can lead to cell senescence via the activation of DNA damaging pathways. Telomere length differs inversely with age and short telomeres trigger cell senescence. Several syndromes have been identified with accelerated telomere shortening [125], and inherited and sporadic mutations in the telomerase genes are associated with early aging [126]. These genes include TERT, TERC, PARN, RTEL1 and NAF1 and are associated with familial PF, but not with sporadic and non-familial PF [127]. PF is also known to result from immune rejection leading to CLAD [5], and telomere length may be useful as a biomarker of lung transplant-associated PF. Transplantation offers a unique opportunity to study PF outcome of mismatched telomere length in donor and recipient cells, in turn informing which compartment, immune or parenchymal, drives transplant-associated PF. In renal and stem cell transplantation, telomere shortening correlates with poor graft survival [128,129].

In lung transplantation, a preliminary study reported shorter recipient lung telomeres associated with a decreased rate of acute cellular rejection, but neither the telomere length of the donor nor that of the recipient was associated with survival. Both Donor lymphocyte and recipient lung tissue telomere lengths were inversely related to age [130]. Shorter telomere length was associated with reduced post-transplant survival, incidence of CLAD, shorter time to CLAD and increased infection in a further study [131]. A recent study found that short telomeres early post-transplant were associated with leukopenia and decreased CLAD-free survival [132]. Another study, by Faust et al., reported telomere shortening in airway epithelial cells, which was associated with CLAD. The study examined 175 lung transplant recipients and measured the telomere lengths of both donors and recipients using qPCR and QFISH assays. Shorter telomere length was associated with an increased risk of CLAD or death. Short PBMC telomere length was associated with significantly worse CLAD-free survival and increased the occurrence of post-transplant leukopenia [133]. In contrast, Mackintosh et al. recently showed that short telomeres strongly associate with donor age and smoking history but show no association with the future incidence of CLAD [134]. Cytomegalovirus (CMV) infection in lung transplant recipients augments the shortening of telomeres, likely as a result of increased cell turnover due to infection, suggesting that telomerase activity may be important in the regeneration and healing of airway epithelia in response to lung injury [135]. Finally, variants of recipient TERT, RTEL1 and PARN genes associate with shorter time to CLAD and explicitly restrictive CLAD (R-CLAD) [136]. These studies suggest that shorter telomeres can directly correlate with negative outcomes in lung transplantation.

Future biomarkers & single-cell RNA sequencing

A major limitation of the clinical use of biomarkers for diagnosis is the source of samples. Both blood and BAL have complex cell populations and separation of cells is costly, time consuming and not always possible. Therefore, bulk samples are the main source for biomarker testing for practical purposes. This poses challenges, as biomarkers may be present in small populations of cells that vary over time and the testing of bulk RNA may lead to poor reproducibility. In an effort to understand cellular heterogeneity and cell-specific pathways in health and disease, new techniques to analyze the transcriptome of individual cells in mixed populations are becoming available. single-cell RNA sequencing is becoming the preferred method. This technology partitions cells, adds a genetic barcode and following bulk DNA sequencing and alignment to a reference genome, the transcriptome of each cell is determined. Downstream bioinformatics generates count tables of gene expression at individual cell level. Applying dimensional reduction and machine learning, cells are clustered by gene signature and assigned a cell type based on their gene-expression profile.

Applied to healthy lung tissue, which has previously been estimated to have 40 different cell types [137], single-cell RNA sequencing was able to detect 58 cell populations [138]. In the context of lung disease, single-cell RNA sequencing has identified changes in cells from unused donor lungs compared with explant lungs from patients with PF, including IPF and interstitial lung disease (ILD). Recent findings include the discovery of distinct epithelial cell populations, including an aberrant basaloid epithelial cell increased in fibrotic lungs [139–141], increased goblet cells and Th2 cells in asthma [142] and identification of monocyte-derived macrophages with fibrotic transcriptome enriched in ILD [140,143,144]. Clearly, for diagnosis and patient monitoring, explant tissue is not possible, so single-cell RNA sequencing methods need to adapt to TBB, BAL and blood. Reyfman et al. demonstrated that cryobiopsies can be analyzed at the single-cell level [143], and Morse showed macrophage populations in BAL of healthy patients [144]. However, these studies were only exploratory demonstrating feasibility, and the cell types, transcriptome and relationship to disease remain unexplored, requiring larger comparative studies of healthy and diseased samples. In lung transplantation, single-cell RNA sequencing combined with male- and female-specific markers has been used to assess the engraftment of cells in sex-mismatched transplant recipients [145]. Although only four BAL samples were analyzed, and patients were a year or more post-transplant, a high level of replacement of donor immune cells was observed with the majority of lung cells being of recipient origin, confirming a higher-than-expected turnover of alveolar macrophages in the transplanted lung. Future studies particularly in the first 6 months following transplantation are needed to fully characterise cellular changes in donor and recipient cells and their relationship to primary graft dysfunction, an important risk factor for developing CLAD. A recent study examined BAL from a transplant recipient with CLAD and demonstrated that monocyte-derived macrophages, of recipient origin, expressed profibrotic genes, including autotaxin (ENNP2), the main enzyme able to generate LPA, implicated in fibroblastic recruitment and differentiation [146]. In the small cohort of transplant patients, ENPP2 expression in BAL cells, as assessed by PCR, correlated with time to CLAD [146], demonstrating the utility of single-cell RNA sequencing as a new tool for biomarker discovery in cells from BAL, routinely collected from lung transplant recipients for surveillance bronchoscopy. Surprisingly, to date, no studies have reported cell types and transcriptome of blood cells in lung transplant patients by single-cell RNA sequencing. Considering the large number of studies showing blood cell changes and association with lung transplant rejection (Table 1), single cell analysis should provide an opportunity to analyze all these cell types and their active pathways simultaneously, providing unparalleled insight into blood cell dynamics. Future work will hopefully provide this information and, in combination with BAL, identify correlative biomarkers useful for diagnosis and patient monitoring in blood following lung transplantation.

Conclusion

A molecular biomarker is any measurable biological molecule or signature associated with a particular physiological or pathological state that can be quantified as an indicator of predisposition, development or prognosis of diseases and response to pharmacological activity. Many recent studies targeting biomarker discovery in lung transplant rejection have emerged, but no reliable biomarker identified so far is able to diagnose or predict rejection successfully. However, the findings are promising and add knowledge to the field. Understanding the dynamics of immune cells and immune-related cytokines and chemokines provides insight into the immunological mechanisms acting during different stages of rejection and rejection phenotypes such as BOS and RAS. This information will likely shape personalized medicine and the lifestyles of patients to avoid or minimize rejection. Gene-expression studies likewise yield information furthering our understanding of mechanisms of rejection as well as serving as validation of previously published studies. Results of observational studies in BAL and blood and associated co-regulated genes will help clarify the pathways of rejection and the spectrum of phenotypes observed in rejection. CLAD is a cluster of heterogeneous phenotypes of rejection, and more research is needed in the field to untangle the mechanisms and pathobiology of these phenotypes, yet no biomarker has been successful at discriminating between phenotypes. Additionally, research needs to integrate environmental variations, clinical history and biomarkers to answer these questions.

CLAD is usually associated with long-term, gradual decline in lung function rather than rapid decline, suggesting that earlier stages of rejection need to be identified before an irreversible stage is reached. This needs to be taken into consideration when investigating biomarkers, and biomarkers should be identified that can predict time to CLAD. Detecting early decline in lung function will be crucial in developing treatments and interventions prior to irreversible decline in lung function. Early detection would mean early intervention to reduce rejection and improve the survival of lung transplant recipients.

Future perspective

All of the above-mentioned studies identify several potential biomarkers needing validation in large-scale, multicenter analyses. The aim would be to replace the current gold standard method, TBB, which is invasive, costly and prone to sampling error and inter-observer variation. Therefore, it should be emphasized that the need for reliable biomarkers to diagnose and monitor rejection is urgent and crucial to improve the current low survival rates observed in lung transplantation. A single biomarker may not be ideal; however, combining several reliable markers would be the ultimate approach to maximize the reliability and specificity for detecting rejection in the lung. Different stages of lung transplantation from organ procurement (as in organ matching) to immune monitoring of patients after lung transplantation require careful evaluation. We propose a hypothetical sequence of using potential biomarkers effectively, as illustrated in Figure 1. Similarly, studies need to focus on identifying risk factors based on a more holistic evaluation considering environmental factors, patient history, age, ethnic and geographic variations and different phenotypes and stages of rejection. This will enable clinicians to identify high- and low-risk patients and help them to treat patients based on risk with tailor-made therapies to suit the individual phenotype of the patient. Such studies would enable the curation of a precise group of biomarkers predicting graft survival in lung transplant patients. Though we are not quite there yet, ongoing studies are promising, and we are now on the brink of a breakthrough in the discovery of ideal biomarkers in lung transplant rejection.