Unencapsulated or ‘‘cell-free’‘ DNA was first discovered in human plasma by Mandel and Metais in 1948[27]. Following a resurgence of interest into its clinical potential in the 1990s[28], the scientific community has since learnt much about the biology of cell-free DNA. The majority originates from haematopoetic cells such as leukocytes[29,30], and is released into the circulation during apoptosis and necrosis[31-33]. These fragments of DNA are then rapidly cleared from plasma by the liver, spleen and kidneys[34,35]. As a result, cell-free DNA has a short half-life of approximately 1.5 h[36,37]–rendering it a ‘‘real-time’‘ marker of cellular injury. Subsequently, scientists identified that lower levels of this circulating free DNA were also being released during normal physiological turnover[38-40].

Given these characteristics, cell-free DNA has emerged as a useful biomarker in multiple clinical settings. This was particularly notable in those where a genetic difference could be exploited, such as oncology, obstetrics or solid-organ transplantation. In cancer patients, researchers isolated circulating free DNA characterised by mutations specific to particular malignancies[41-43]. This gave rise to the notion of a ‘‘liquid biopsy’‘ for diagnostic and management purposes[44-47]. Similarly, in the plasma of pregnant women, researchers detected fragments of DNA unique to the foetus[28], and subsequently analysed these for genetic conditions[48]. Today, ‘‘non-invasive pre-natal testing’‘ has replaced the need for chorionic villus sampling with a simple blood test[49], which is commercially available throughout the world[50]. In solid-organ transplantation, genetic differences become fundamentally intertwined. With the exception of an identical twin donor-recipient pair, this procedure places a unique genome within the recipient–theoretically creating the ideal environment for detecting circulating free donor DNA via minimally-invasive blood sampling. Moreover, this biomarker could plausibly reflect graft integrity at low levels, and cellular death when elevated. A particular focus has emerged regarding the dynamics of this DNA during rejection, given it is this element of solid-organ transplantation that currently necessitates invasive biopsies. This is particularly the case in LT, where routine biopsies are considered controversial–and often only performed if clinically indicated[51,52]. Clearly, a liquid biopsy could be revolutionary in this setting.

In order to critically appraise studies examining the clinically utility of donor-specific cell-free DNA in LT, it is important to understand the scientific advancements that have enhanced its detection.

The first group to detect circulating free donor DNA in transplant recipient plasma were Lo et al[53] in 1998. In their landmark study, they isolated fragments of donor DNA in the plasma of 36 liver or kidney transplant recipients–including six females who had received livers from male donors. In this subset of participants, the authors isolated genetic sequences unique to the Y-chromosome, which they amplified using polymerase chain reaction (PCR) and visualised using gel electrophoresis. In so doing, they provided ground-breaking data proving the concept of donor-specific cell-free DNA, depicted in Figure 1. However, this methodology was limited to male-to-female engraftments only–just as a subsequent Rhesus (Rh) gene quantitative PCR (qPCR) assay was restricted to positive-to-negative transplantations[54]. As such, a focus on identifying other genetic targets that differed more broadly between individuals subsequently emerged.

Open in New Tab   Full Size Figure   Download Figure

Figure 1 The concept of donor-specific cell-free DNA in liver transplantation.

The following decade, the advent of next generation sequencing (NGS) completely revolutionised gene discovery. By enabling massive genetic throughputs[55], multiple genetic loci that were highly heterogeneous within the population could now be identified. The most common of these were ‘‘single nucleotide polymorphisms’‘ (SNPs)–where DNA sequences differed by one adenine, thymine, guanine or cytosine molecule between individuals[56]. By using NGS to analyse multiple SNPs, researchers could now detect genetic sequences likely to differ between the vast majority of donor-recipient pairs. The first group to achieve this were Snyder et al[57] in 2011, who analysed blood samples from heart transplant donors and recipients, and detected circulating free donor DNA using a genome-wide SNP assay[57]. Since then, three other groups have published more targeted NGS methodology in this field[58-60], two of which circumvented this need for baseline donor blood sampling by using computational techniques[59,60]. However, in clinical practice, NGS assays have several key limitations. Not only are they highly complex and expensive, but they can take up to seven days to process[57]–rendering them potentially futile as a real-time transplantation biomarker.

Given this, an interest in developing more accessible, affordable and rapid assays arose. This coincided with the commercial availability of droplet digital PCR (ddPCR), which had a six hour turnaround time, and could more precisely quantify DNA than previous qPCR techniques[61]. Researchers began designing new ddPCR probes and primers to detect donor-specific sequences. Y-chromosome and SNP targets were revisited, but new sites included regions of the human leukocyte antigen (HLA) gene and ‘‘deletion insertion polymorphisms’‘ (DIPs). At a population level, HLA genes are characterised by high levels of heterogeneity[62]. However, as donor-recipient pairs are often HLA ‘‘matched’‘[63], this target is potentially problematic in transplantation. DIPs, conversely, remain a promising option–as these are regions of the genome characterised by the absence or presence of certain nucleotides, leading to common allelic differences between individuals[64]. Ultimately, understanding these methodologies highlights the relative complexity of genetic tests, compared to more standard biochemistry such as LFTs[65]. Accordingly, each assay for circulating free donor DNA requires validation, in order to establish its utility in the clinical setting.

A total of 12 publications have studied donor-specific cell-free DNA in LT, as summarised in Table 1. These studies differ in their size (n = 1-115), design and assay methodologies. However, they all demonstrate that this biomarker shows promise in monitoring graft health and detecting injury–especially when caused by acute rejection.

Fifteen years after Lo et al[53] first demonstrated the presence of Y-specific donor DNA fragments in LT recipient plasma, Beck et al[66] went on to establish three additional key findings. In their 2013 study, they used probe-based ddPCR to scrutinise a panel of 40 SNPs and detect donor-specific sequences in 10 newly transplanted and seven stable LT recipients. These fragments of donor DNA were then quantified in terms of relative abundance and expressed as a percentage of total cell-free DNA. Firstly, Beck et al[66] observed high levels of circulating free donor DNA post-engraftment (approximately 90%), which fell exponentially and stabilised within 10 d in recipients without complications. Secondly, this DNA was elevated (> 60%) in two newly transplanted patients with biopsy-proven acute rejection (BPAR), yet not in another with obstructive cholestasis. Notably, this DNA began to increase several days prior to LFTs in those cases with rejection. Thirdly, the authors identified a ‘‘healthy’‘ threshold of donor-specific cell-free DNA of < 10% in the stable LT recipients. Additional benefits of this assay included its same-day turnaround and lack of a need for donor blood sampling. However, its limitations included the use of PCR preamplification and post-PCR handling, which can introduce several forms of bias and pose a high contamination risk, respectively[67].

The next year, Macher et al[68] published a longitudinal study using qPCR to detect Y-specific DNA fragments in 10 gender-mismatched LT recipients. As with Beck et al[66], the authors also found that this circulating free donor DNA was elevated immediately post-LT, then rapidly decreased in recipients without complications and remained stable[68]. Macher et al[68] also identified a threshold reflective of organ health–however as their assay was one of absolute quantification, this was expressed as 150 ng/mL. The authors made the novel observation that these fragments of donor DNA were also elevated in recipients who experienced cholangitis and vascular complications. Unfortunately, this study proved too small to examine the dynamics of this DNA in acute rejection, as no patients experienced this endpoint. As such, Macher et al[54] subsequently published an additional study in 2016. This time, they measured circulating free donor DNA by using qPCR to detect Rh-positive sequences in 17 Rh-mismatched LT recipients. Here, in the patients who experienced BPAR, levels of donor-specific cell-free DNA were found to rise compared to those without complications. However, as these two qPCR assays targeted restrictive genetic differences only, they intrinsically had limited clinical utility.

Between 2014 and 2017, the Beck group published three additional studies using their more expansive SNP methodology[69-71]. The first of these was a case study, which described a LT recipient of a marginal graft, who had experienced multiple complications post-operatively–and retrospectively undergone donor-specific cell-free DNA analysis[69]. Kanzow et al[69] demonstrated that levels rapidly became elevated in the following settings: BPAR, traumatic liver haematoma and cytomegalovirus infection. They also made the pioneering observation that circulating free donor DNA subsequently fell post successful treatment of each complication. The authors concluded that this biomarker was useful for monitoring organ health.

Next, Oellerich et al[70] prospectively measured circulating free donor DNA and CNI levels in 10 receipts during the first month post-LT. They aimed to identify the minimum trough tacrolimus concentration that was associated with graft integrity. Using the pre-established healthy threshold of < 10%, the authors observed significant segregation and determined the lower limit of the therapeutic tacrolimus range to be 8 ug/L. Although larger studies with longer follow up were still needed, Oellerich et al[70] postulated the assay could be useful in monitoring for graft injury in LT recipients whose immunosuppression was being weaned.

This unmet need was addressed by the third study, published by Schütz et al[71] In their multicentre prospective trial, donor-specific cell-free DNA was measured in 115 LT recipients at seven timepoints during the first year post-LT, plus whenever rejection was suspected. The stereotypic exponential fall of this DNA was seen in 88 stable recipients, who had a median level of 3.3%. In 17 recipients with BPAR, median levels were elevated at 29.6%. Moreover, this circulating free donor DNA was found to be an accurate and early marker of BPAR–with a superior area under the receiver operating characteristic curve (AUC) of 0.97 compared to LFTs (0.83-0.96), and levels increasing up to two weeks prior to diagnosis on liver biopsy. In patients with infective complications, median donor-specific cell-free DNA was slightly higher than in stable recipients, but lower than in BPAR (5.3%-5.7%) – similar to patterns seen by other authors[68,69]. In patients with cholestasis alone, levels remained < 10%[71]. On multivariate logistic regression, Schütz et al[71] found that this biomarker provided independent information regarding graft integrity.

Whilst the benefits of the Beck et al[72] assay they utilised prevailed, there were several limitations to this study[71]. These were highlighted by two cases, where patients had BPAR, but circulating free donor DNA levels remained < 10%. In the first patient, who had a marked leukocytosis, Schütz et al[71] acknowledged that this factor may have ‘‘masked’‘ the percentage of cell-free DNA from the donor present in recipient plasma, due to an increase in the denominator of total cell-free DNA. Indeed, expressing circulating free donor DNA in terms of relative abundance renders it innately susceptible to this form of error–including in other circumstances where cell-free DNA increases such as infection[73], obesity[74] and exercise[75]. In the second patient with BPAR but circulating free donor DNA below the ‘‘healthy’‘ threshold, the authors attributed this to the fact that the rejection was only mild histologically, with a rejection activity index (RAI) of 1/9, and did not require treatment[71]. This case demonstrates the limited clinical utility of BPAR as an endpoint–compared to treated BPAR (tBPAR) of RAI ≥ 3, which is now widely utilised in clinical trials[76,77].

These limitations, however, were not present in the Goh et al[78] publication from 2019. This group originally validated their probe-free ddPCR assay in 2017, when they successfully targeted a panel of nine DIPs and achieved absolute quantification of circulating free donor DNA in three LT recipients[79]. Two years later, they used this technique to examine 40 recipients divided into two cohorts[78]: Longitudinal (n = 20), who had donor-specific cell-free DNA measured at five timepoints during the first six weeks post-LT; and cross-sectional, who were either undergoing a liver biopsy at least one-month post-LT (n = 16), or stable and at least one-year post-LT (n = 4). The authors demonstrated findings in keeping with the aforementioned literature. In the longitudinal group, levels of circulating free donor DNA fell exponentially and stabilised in the 14 recipients without complications. Elevated levels of this DNA were observed in three recipients with tBPAR, but not in three with cholestasis alone. In the cross-sectional cohort, elevated levels of this DNA accurately identified six patients with tBPAR, with an AUC of 0.97 that was again superior to LFTs. A healthy threshold of < 898 copies/mL was identified in the 14 cross-sectional patients without rejection and found to be reliable in the longitudinal cohort from day 14 post-LT onward. By using primer sets to hybridize across allelic breakpoints, Goh et al[78] had also eliminated the need for costly florescent probes. However, the assay called for a donor blood sample for optimal processing and the study was ultimately underpowered.

Most recently, Ng et al[80-82] pioneered the measurement of circulating free donor DNA in live donor LT (LDLT). These authors utilised different assays to detect the relative abundance of this DNA in paediatric recipients from day 0-60 post-LDLT. First, NGS was used to detect Y-specific sequences in two gender-mismatched LDLTs⁹⁶. Next, a qPCR SNP assay was examined in two additional LDLT recipients⁹⁷. In both publications, Ng et al[82] found that circulating free donor DNA exponentially fell and stabilised at < 0.1, as seen with the Beck et al[66] group. Finally, the initial NGS Y-specific assay was used in 7 gender-mismatched LDLTs to detect circulating free donor DNA, which was then profiled according to its fragment size[82]. Here, the authors made the innovative observation that donor DNA fragments were ‘‘short’‘ (105-145 bp), compared to the ‘‘long’‘ fragments of recipient DNA (> 160-170 bp). NGS and automated electrophoresis was then used to detect these short donor DNA fragments in four gender-matched LDLT recipients. The authors also noted that the ratio of short to long (S/L) fragments correlated with the circulating free donor DNA levels–and identified a healthy S/L fragment threshold of < 0.6. Interestingly, in the oncology and obstetric research settings, the fragments of DNA from tumour cells or from the foetus are also shorter (i.e. than those from non-malignant or maternal cells respectively) but the mechanism behind this is unclear[83,84]. Certainly, this Ng et al[80-82] fragment size-based assay was quicker and less restrictive than targeting the Y-chromosome. However, its methodology was still slower (24 h) and more expensive than PCR. Furthermore, these three studies were limited by their small sample size of uneventful LDLTs[80-82]–precluding insights into the dynamics of their assays during complications.