Against this background noninvasive donor-derived cell-free DNA (dd-cfDNA) determination has great potential to early detect graft injury and/or rejection. The rationale for using dd-cfDNA as a quantitative biomarker (“liquid biopsy”) in transplantation is based on the fact that organ transplants are also genome transplants which opens up the possibility to use them to directly monitor allograft health (De Vlaminck et al., 2014). In the case of graft cell death nucleosomes are released into the bloodstream as cfDNA. Causes of graft injury include rejection, acute tubular necrosis, ischemia, trauma and infection. The mechanisms of cfDNA release involve necrosis yielding larger fragments (∼10,000 bp) or apoptosis resulting in the release of smaller fragments (60–500 bp). The median half-life of released DNA in the circulation is only about 30 min to 2 h (Sherwood and Weimer, 2018). In the recipient´s blood dd-cfDNA accounts for only a small fraction of the total cfDNA. About 90% of cfDNA in the recipient plasma stems from white blood cells undergoing natural apoptosis (Sun et al., 2015). dd-cfDNA fragments also appear in urine (Sigdel et al., 2013) as they are filtered through the glomerular barrier. Moreover dd-cfDNA derived from the donor-derived urinary tract is also excreted in the urine (Oellerich et al., 2021).

dd-cfDNA quantification is based on genetic markers (e.g., single-nucleotide polymorphisms or SNPs) that allow discrimination between heterologous alleles in the recipient and donor (Oellerich et al., 2021). Currently in clinical assays either random (Snyder et al., 2011; Sharon et al., 2017) or targeted (Beck et al., 2013; Grskovic et al., 2016; Sigdel et al., 2019) approaches are used. Random approaches use adapter ligation followed by next generation sequencing (NGS). Targeted approaches use preselected SNPs in droplet digital PCR (ddPCR) (Beck et al., 2013) or targeted NGS methods (Grskovic et al., 2016; Sigdel et al., 2019). PCR efficiency is affected by the length of amplicons and cfDNA fragments (Dauber et al., 2020; Oellerich et al., 2021). Recently a dd-cfDNA assay has been evaluated based on donor-recipient HLA-mismatch (human leukocyte antigen) at the HLA-DRB1 locus using optimized droplet-digital PCR (Sorbini et al., 2021). Fractional determination (%) is possible with all these methods but absolute quantification which is not affected by changes in recipient cfDNA (e.g. by infection) has only been validated for ddPCR (Oellerich et al., 2019; Whitlam et al., 2019). Most of these methods have a lower limit of quantification of ∼0.15% dd-cfDNA and an imprecision (CV) of 3%–12% (Oellerich et al., 2020). Studies in heart transplant recipients reported also lower baseline levels of 0.02% (Agbor-Enoh et al., 2021a), 0.04% (Kim et al., 2022) and 0.07% (Khush et al., 2019). In kidney transplantation a lower limit of 0.08% has recently be published (Halloran et al., 2022). Samples are stable at room temperature for ∼1 week if collected in appropriate tubes (Oellerich et al., 2020).

The clinical validity of dd-cfDNA in kidney transplantation has been documented in more than 20 studies (Knight et al., 2019; Wijtvliet et al., 2020; Filippone and Farber, 2021; Oellerich et al., 2021; Xiao et al., 2021; Steggerda et al., 2022). Over the first 2 weeks after transplantation there is typically an exponential decline to a baseline level (Oellerich et al., 2019) which can be used to discriminate graft injury. An abnormal non-exponential decline of dd-cfDNA has been observed in some patients due to urinary tract infections, hemodynamic problems, or surgical complications (Gielis et al., 2018). Most studies investigating dd-cfDNA (%) used a cut-off between 0.74% and 1.0% (Oellerich et al., 2021). Based on simultaneous maximization of sensitivity and specificity threshold values of 0.43% were found for fractional determination and 52 cp/ml for appropriate absolute quantification (Oellerich et al., 2019). Although different concepts have been used to establish threshold values for detection of rejection median dd-cfDNA values in reference populations were very similar ranging from 0.21% to 0.40% (Bloom et al., 2017; Oellerich et al., 2019; Sigdel et al., 2019; Oellerich et al., 2021).

During long-term surveillance a time-dependent increase of dd-cfDNA (%) from 0.8% to 2.1% (90th percentile) was observed as a result of a systematic concurrent decrease of total cfDNA, presumably due to a decrease of calcineurin inhibitor exposure over time (Schuetz et al., 2020). The resulting time-dependent increase of fractional dd-cfDNA (%) brings the use of a single cut-off value during long term treatment into question (Schuetz et al., 2020). In contrast to fractional dd-cfDNA (%) the threshold for absolute quantification (cp/ml) was not affected during long term surveillance (Schuetz et al., 2020).

In patients with either T-cell mediated rejection (TCMR) or antibody-mediated rejection (ABMR) elevations of dd-cfDNA were observed in the vast majority of studies (Knight et al., 2019; Sigdel et al., 2019; Oellerich et al., 2021; Halloran et al., 2022). ABMR is associated with 20%–30% allograft loss (Kim et al., 2014). Early and subclinical ABMR detection enables earlier adapted therapeutic interventions and may improve outcomes (Mayer et al., 2022).

In DSA + patients early detection of ABMR is important and dd-cfDNA may also be helpful to identify patients with “indolent” DSA from those with graft injury due to DSA (Kataria et al., 2021). A clinical trial to prove this hypothesis (NCT04897438) is under way. Measuring dd-cfDNA is also of interest for monitoring the potential therapeutic effect on organ injury of novel ABMR treatments, such as anti-CD38 monoclonal antibody Felzertamab (Mayer et al., 2022).

In the multicenter DART study (Bloom et al., 2017) acute and chronic ABMR could be very well detected by measuring dd-cfDNA (%). For TCMR dd-cfDNA elevations were insufficiently detected with the assay employed. This has been confirmed in an independent study (Huang et al., 2019). The observed false negative TCMR results might have been due to an insufficient detection of short DNA fragments due to the use of relatively long amplicons (100–130 bp) in the particular test used in these two studies (Grskovic et al., 2016). It is possible that in TCMR, which is primarily a process of the kidney tubulointerstitial space, there is more extensive fragmentation of dd-cfDNA compared to ABMR. This hypothesis needs to be confirmed by further investigations.

It has been demonstrated using INDEL q PCR that dd-cfDNA quantification can be substantially improved using smaller amplicon sizes (Dauber et al., 2020). In a further retrospective study using a SNP- based multiplex PCR NGS test TCMR (tubulitis score >2, interstitial inflammation score >2) could be very well detected (Sigdel et al., 2019). Using the same method, in the recent prospective Trifecta multicenter study Halloran et al. (Halloran et al., 2022) compared fractional dd-cfDNA levels to indication biopsy results obtained by a microarray-based Molecular Microscope Diagnostic System (MMDx) and conventional histology. Dd-cfDNA was strongly associated with both molecular and histologic rejection results suggesting the potential of such testing to reduce unnecessary biopsies. Absolute quantification also discriminated very well between patients with TCMR, ABMR, borderline TCMR and acute tubular necrosis from stable and biopsy negative patients (Table 1). Diagnostic accuracy (AUC-ROC) regarding detection of acute rejection was superior for absolute quantification (83%) compared to fractional dd-cfDNA determination (73%) (Oellerich et al., 2019).

A problem with fractional dd-cfDNA determination is that it is influenced by variations in total cf-DNA levels due to leukopenia, leukocytosis and inflammatory illness; leading to falsely elevated or decreased results (Whitlam et al., 2019; Osmanodja et al., 2021). Of particular importance is the systematic increase of dd-cfDNA percentages when CNI are tapered, due to an influence of CNI on leucocyte stability (Schuetz et al., 2020). In clinical practice a combination of absolute quantification and fractional abundance of dd-cfDNA with consideration of total cfDNA as well seems to provide the most comprehensive diagnostic information (Oellerich et al., 2019; Whitlam et al., 2019; Oellerich et al., 2021; Osmanodja et al., 2021).

The overall diagnostic performance of dd-cfDNA for detection of rejection after KTx seems to be adequate based on data reported to date (Oellerich et al., 2021). Based on pooled data from published studies the diagnostic accuracy (ROC AUC) was on average 0.81 (seven studies) for dd-cfDNA (%) and 0.87 (two studies) for dd-cfDNA (cp/ml) (Oellerich et al., 2021). Mean values for sensitivity were 80% for dd-cfDNA (%) and 79% for dd-cfDNA (cp/ml). For specificity, mean values were 76% for both dd-cfDNA (%) and (cp/ml). Positive predictive values (that depend on the variable prevalence of rejection in the study populations) were 56% for dd-cfDNA (%) and 33% for dd-cfDNA (cp/ml). The very high negative predictive values of 90% for dd-cfDNA (%) and 97% for dd-cfDNA (cp/ml) indicate that this test is useful to exclude rejections with high probability. Still, other entities that can cause elevations such as recurrent primary disease may be excluded with high probability only by kidney allograft biopsy. (Oellerich et al., 2021).

Dd-cfDNA detects rejection also in liver (Schuetz et al., 2017; Levitsky et al., 2022), heart (Snyder et al., 2011; De Vlaminck et al., 2014; Khush et al., 2019; Richmond et al., 2020; Agbor-Enoh et al., 2021a; Sorbini et al., 2021), lung (Rosenheck et al., 2022a) and simultaneous pancreas-kidney (Ventura-Aguiar et al., 2022) transplant patients with good accuracy (Table 2). Dd-cfDNA fraction was also significantly elevated for CLAD (De Vlaminck et al., 2015). The high negative predictive value of 97% in heart transplant recipients (Khush et al., 2019; Kim et al., 2022; Knuettgen et al., 2022) has the potential to reduce the number of endomyocardial biopsies that have a serious complication rate of 0.5%–1% (Miller et al., 2013; Knuettgen et al., 2022).

Monitoring of dd-cfDNA may also be helpful to guide immunosuppression minimization. Increased dd-cfDNA values were associated with subtherapeutic tacrolimus concentrations after kidney transplantation in patients with variable tacrolimus levels (Oellerich et al., 2019). Therefore dd-cfDNA might be helpful to detect subclinical graft damage caused by immune activation due to under-immunosuppression (Oellerich et al., 2021). This would be very useful in kidney recipients with a high epitope mismatch burden, high immune competence or non-adherence.

In a recent study, fractional dd-cfDNA was tested during subclinical graft injury after liver transplantation confirmed by surveillance biopsies (Baumann et al., 2022). The lower limit of quantification was 0.15% and the interassay CV ranged from 3% to 5.4%. Dd-cfDNA increased stepwise significantly from patients without histological signs of rejection (median, 2.2%) over subclinical TCMR (median, 3.5%) to clinical overt TCMR (median, 23%). The data confirmed the 10% cut-off for liver (Schuetz et al., 2017) yielding a similar sensitivity and specificity for acute rejection. Before immunosuppression minimization subclinical TCMR should be excluded. The usefulness of longitudinal dd-cfDNA monitoring of subclinical injury should be assessed in future prospective studies.

A majority of studies have found that dd-cfDNA was clearly superior compared with eGFR in the discrimination of patients with or without acute rejection (Bloom et al., 2017; Sigdel et al., 2019; Whitlam et al., 2019; Oellerich et al., 2021). Dd-cfDNA release, however, is not rejection-specific and elevations can also be observed due to other causes of graft injury (Bloom et al., 2017; Oellerich et al., 2021). Elevated dd-cfDNA levels have also been found in patients with pyelonephritis and biopsy-proven BKV nephropathy (Sigdel et al., 2013; Bloom et al., 2017; Oellerich et al., 2021). In particular, urinary dd-cfDNA concentration was significantly elevated in patients with BKV nephropathy (Sigdel et al., 2013; Chen et al., 2022). In patients with interstitial fibrosis/tubular atrophy (IFTA) only a small increase of dd-cfDNA was observed (Sigdel et al., 2013; Bloom et al., 2017; Oellerich et al., 2019; Whitlam et al., 2019) potentially due to lower viable cell numbers. Due to the slow fibrosing process IFTA may not be associated with a relevant release of DNA from slowly dying kidney cells and the resulting gradual decrease in viable cells over time. During stages of active disease progress, however, dd-cfDNA values might be increased in IFTA.

A previously published value proposition concept (Price et al., 2016) can be used to characterize the benefits of dd-cfDNA-testing to stakeholders involved in organ transplantation (Oellerich et al., 2020).

Various stakeholders are involved in delivering and receiving care in transplantation. There are the transplant patients whose care could be altered by the less invasive graft injury detection, clinicians who manage solid organ transplant patients, laboratory medicine specialists who analyze and interpret test results and hospital management, insurance companies, public payers and policy makers who are involved in providing value-based health care. Value in this context is defined as outcomes relative to costs. For patients the following benefits are expected relative to outcome: detection or exclusion of graft injury or rejection, earlier transplant injury detection and subsequent intervention, biopsy alternative, early diagnosis of subclinical antibody-mediated rejection, detection of under-immunosuppression, evaluation of infectious complications of kidney allografts and personalized immunosuppression. Benefits for transplant physicians include better personalized immunosuppression guidance; for example during tapering, enhanced biopsy interpretation, less trial and error changing of immunosuppression, less time dealing with complications, and indication of response to rejection treatment. Laboratory medicine specialists would have increased involvement in molecular diagnostics regarding the use and interpretation of tests. Hospital management, insurance companies, public payers and policy makers can expect cost savings due to a decreased burden for care-givers; for example due to fewer re-transplantations or less returns to dialysis in KTx.

Dd-cfDNA has great potential for cost savings. Acute rejection is associated with a significant increase in the cost of care (17,000–28,000 USD/per year) (First et al., 2017). TCMR and ABMR contributed to graft loss in more than 60% of graft failures (Mayrdorfer et al., 2021). A graft failure with return to dialysis causes an average annual expense of about 75,000 USD. If the patient is retransplanted, the average cost is about 111,000 USD. The annualized cost of transplantation is less than 25% of the cost of dialysis (16,844 USD vs. 70,581 USD) (First et al., 2017). Compared to these costs dd-cfDNA testing by ddPCR is reasonable with about 401 USD per test based on the German Code of Medical Fees (GOÄ) (Oellerich et al., 2021). As these data show, improvement of outcome in transplantation could save substantial costs.

For implementation of dd-cfDNA in clinical practice the following points are relevant. Depending on the required turnaround time, transplant centers can either set up the test themselves or utilize a qualified external central laboratory. Using ddPCR results can be obtained in about 1 working day and by the use of NGS in 2-3 working days (Oellerich et al., 2021). Implementation within a laboratory requires the availability of appropriate instrumentation (e.g., Illumina NGS instrument or Biorad ddPCR). Qualified laboratory medicine specialists and a quality management program are also necessary. Assessment of clinical utility in prospective randomized outcome studies and the development of clinical practice recommendations are desirable. Metrics for monitoring the impact of implementing the test are summarized in Table 3.