Clinical use of donor-derived cell-free DNA in kidney transplantation

Abstract

Traditional monitoring of kidney transplant recipients for allograft dysfunction caused by rejection involves serial checks of serum creatinine with biopsy of the renal allograft if dysfunction is suspected. This approach is labor-intensive, invasive and costly. In addition, because this approach relies on a rise in serum creatinine above historical baselines, injury to the allograft can be extensive before this rise occurs. In an effort to address this, donor-derived cell-free DNA (dd-cf DNA) is being used with increasing frequency in the clinical setting as a means of diagnosing a rejection of the renal allograft early in the course. This can potentially allow for early intervention to minimize not only injury, but the intensity of antirejection therapy needed and the avoidance of side effects. Here, we will review the available methodology for the determination and quantification of dd-cf DNA, the data supporting its use in clinical practice and the limitations of this technology.

Core Tip: Recently published data strongly supports the use of donor-derived cell-free DNA (dd-cf DNA) for routine monitoring of kidney transplants for injury and rejection. Commercial assays for dd-cf DNA have been available for more than 5 years and their use is growing however, older data supporting their use was limited to single center studies with small sample sizes. Within the last two years, newer data has emerged in larger, multi center studies and give further insight into the performance of this test and it's ability to diagnose kidney transplant rejection, predict patients at risk for rejection long before the clinical event and aid in treatment of rejection. This review aims to review the most recent literature and provide an argument for the inclusion of dd-cf DNA in the routine care of kidney transplant recipients.

INTRODUCTION

In solid organ transplantation, the first reports of donor-derived cell-free DNA (dd-cf DNA) were in the 1990’s by two publications by Lo et al[1] and Zhang et al[2]. Both describe dd-cf DNA by detection of the Y chromosomal DNA of the male donors in the serum[1] and urine[2] of female recipients. Since then, advances have allowed for development of assays which can be used commercially where the amount of dd-cf DNA is reported as a percentage of the total cell free DNA[3-5]. In addition, genotyping of the donor or the recipient is not required[6]. Elevated levels of dd-cf DNA point to allograft injury which includes acute rejection and because of this, there has been considerable interest in using dd-cf DNA as a minimally invasive biomarker for allograft health. While it is not taken the place of renal allograft biopsy for diagnosis of rejection in clinical practice, new evidence supports its clinical use to monitor for early immunologically mediated injury[7,8].

Dd-cf DNA IN KIDNEY TRANSPLANTATION

Determination of dd-cf DNA

There are currently two commercially available assays in clinical use for measuring dd-cf DNA, AlloSure (Care Dx, Inc., Brisbane, CA, United States)[3,4] and Prospera (Natera, Inc., San Carlos, CA, United States)[5]. Both detect disparate single nucleotide polymorphisms across the entire genome which then allows for separation of DNA from two individuals without prior genotyping. Comparison of these assays was done by Melancon et al[9] in a single center study involving 76 transplant recipients. Paired blood samples were drawn from each participant with each assay being allocated one sample. The result for each assay was correlated with biopsy of the corresponding patient’s allograft and scored using the Banff 2017 classification. There was no statistically significant difference between assays with respect to sensitivity, specificity and positive and negative predictor values.

Kinetics of dd-cf DNA after kidney transplantation

In the first two weeks after kidney transplantation, there is an exponential decline in the percentage of dd-cf DNA. Gielis et al[10] evaluated the change in plasma dd-cf DNA in a cohort of 42 kidney transplant recipients which included 7 living donors and 35 deceased donors. There was an exponential decline in dd-cf DNA from a median of 10.2% (2.6 to 41.9%) on post operative day (POD) 1 to a mean of 0.46% (+/- 0.21%) on POD 9.85 (+/- 5.6 days). Shen et al[11] reported similar findings in a cohort of 21 kidney transplant recipients (7 living donor and 14 deceased donor). This study included recipients with delayed graft function (DGF). There was a rapid decline in the dd-cf DNA fraction from a median of 20.69% at 3 hours after reperfusion to 1.98% by POD 2 and 0.85% on POD 7. Concentrations of dd-cf DNA were significantly higher for recipients of deceased donor allografts compared to recipients of living donor allografts 3 hours post reperfusion (44.99% vs 10.24%; P < 0.01) and this persisted at POD 7 (1.1% vs 0.59%; P < 0.05). DGF also resulted in higher concentrations but this did not meet statistical significance. Failure to follow this rapid decline in dd-cf DNA fraction can indicate allograft injury in the early post operative period. An abnormal non-exponential decline has been observed not only in rejection but in other causes of allograft injury such as infection and acute tubular necrosis[10].

Once dd-cf DNA reaches a baseline level, elevation beyond this can indicate allograft injury but this correlation is not always true. Schütz et al[12] described time dependent increases in dd-cf DNA from 0.8% to 2.1% (90^th centile) in a cohort of 303 clinically stable kidney transplant recipients where samples were obtained at 12 to 60 months after transplant. This was attributed to a decrease in recipient cf-DNA thought to be modulated by the effect of lower calcineurin exposure overtime and its effect on recipient lymphocytes. Dd-cf DNA amounts remained constant but represented a greater fraction of total cf-DNA.

Dd-cf DNA and acute rejection in kidney transplantation

Traditional biomarkers, such as serum creatinine, used to detect renal allograft dysfunction and rejection have limitations. Changes in these biomarkers often lag behind allograft injury and frequently only come to attention after significant injury has already occurred[13,14]. Other biomarkers such as donor specific antibodies (DSA) may indicate risk of future rejection but do not indicate the onset of rejection[15]. Elevation of dd-cf DNA has been observed in both acute T cell mediated rejection (TCMR) and acute antibody mediated rejection (ABMR)[3-8,16,17]. However, false negatives have occurred and seem to be more common with TCMR[3,17]. Elevations of dd-cf DNA are more consistent and greater in magnitude for ABMR in the published literature[3-8,16,17].

Early literature on dd-cf DNA in the diagnosis of acute rejection was confounded by small sample sizes and inconsistent methodologies. A systematic review of studies published through June 2018 included 739 kidney transplant recipients and 509 of these patients were from papers published in abstract form only[16]. There are also multiple techniques for measuring dd-cf DNA included in these papers. Now, commercial assays have become available so there is more consistency in newer literature.

The overall diagnostic performance of dd-cf DNA for detection of rejection after kidney transplantation seems to be adequate on the data reported to date. Elevated dd-cf DNA increases the likelihood of a rejection by a factor of 2.4 to 6.64[18,19]. In addition, negative predictor values for the exclusion of rejection with dd-cf DNA fractions within normal range are excellent. In order to be clinically useful, studies should specify the assays used, threshold percentage of dd-cf DNA deemed to be clinically significant with clear reporting of sensitivity, specificity and negative and positive predictor values. These studies are summarized in Table 1. As can be seen, the cohort size in early publications was small but recently, two studies with larger sample sizes were published.

Table 1.

Studies assessing plasma donor-derived cell-free DNA for the diagnosis of rejection using commercially available assays

Ref.	Assay used	Patient number	Threshold (%)	Sen/Spec	PPV/NPV
Bloom et al[3]	AlloSure	102	1	59/85	61/84
Sigdel et al[5]	Prospera	193	1	88.7/72.6	52/95
Huang et al[17]	AlloSure	63	0.74	79/72	77/75
Bu et al[7]	AlloSure	1092	0.5	78/71	50/90
Bromberg et al[8]	Prospera	424	1	79/85.3	32.6/97.9

Sen: Sensitivity; Spec: Specificity; PPV: Positive predictive value; NPV: Negative predictive value.

The Assessing Donor-derived cell free DNA Monitoring Insights of kidney Allografts with Longitudinal surveillance (ADMIRAL) study by Bu et al[7] prospectively collected data on 1092 kidney transplant recipients from 7 transplant centers in the United States. Patients were monitored for 3 years using dd-cf DNA measured by the AlloSure assay as part of routine care. Elevation of dd-cf DNA fraction of > 0.5% significantly correlated with clinical and subclinical rejection for both ABMR and TCMR greater than borderline by Banff 2019 classification. In addition, in median dd-cf DNA elevation of 149% from baseline was associated with allograft injury. A dd-cf DNA fraction of < 0.5% strongly correlated with allograft quiescence defined as an absence of clinical and subclinical injury.

Another study by Bromberg et al[8] using the Pro Active registry was recently published. These authors reported data collected from a cohort of 4902 kidney transplant recipients from 54 transplant centers who had routine measurement of dd-cf DNA as part of their care using the Prospera assay. Ultimately, 424 patients met the inclusion criteria, and those patients were then matched with the original cohort across 55 covariates. Not only was an elevated serum dd-cf DNA fraction associated with acute rejection, but there was also data to suggest that elevations could predict clinical rejections months in advance. In some patients, dd-cf DNA fractions were significantly elevated five months before ABMR and two months before TCMR found on biopsy even when biopsies done at the time of the elevation were negative for rejection. In addition, the presence of two or more elevated dd-cf DNA results was associated with a lower glomerular filtration rates (GFR) with time.

Dd-cf DNA and monitoring response to treatment of rejection

Acute rejection can lead to renal allograft loss. The incidence of this is particularly high in the case of ABMR and has been reported at 20%-30%[20]. Several studies have demonstrated that early treatment of acute rejection is associated with improved outcomes[17-22]. However, results are still suboptimal. In a retrospective study by Pineiro et al[23] only 54.3% of patients with ABMR had stable or improved renal allograft function after six months following standard therapies including corticosteroids, plasmapheresis, intravenous immunoglobulin and Rituximab. Therefore, a biomarker which can predict rejection early, leading to earlier treatment and improved outcomes would be invaluable and dd-cf DNA has shown some promise in this regard.

Monitoring the response to treatment of acute allograft rejection relies on trending changes in allograft function, serum levels of DSA in the case of ABMR and in some cases, repeat allograft biopsy to determine resolution of ongoing damage. However, rates of histopathological resolution of rejection occur at different rates between individuals[24,25]. As such, histopathological findings on a biopsy performed during or after treatment may not accurately reflect ongoing injury or allow for determination of an adequate response to treatment. This can result in inappropriate further intensification of immunosuppression leading to complications such as opportunistic infections. A tool which can be used to detect real time damage of the renal allograft would allow for personalized therapeutic approaches to each patient and potentially avoid inappropriate treatment regimens.

The short half-life of cf-DNA makes it an attractive candidate for monitoring the response to treatment of rejection. Shen et al[26] reported significant reductions in dd-cf DNA after treatment of rejection. In this prospective study, 28 patients with acute rejection were recruited. The cohort included five patients with ABMR, 12 patients with either Banff IA or IB TCMR and 11 patients with Banff IIA or IIB TCMR. As a group, the percentage of dd-cf DNA declined significantly from 2.566 +/- 0.549% to 0.773 +/- 0.116% (P < 0.001) after treatment of rejection for all histological classifications of rejection. Furthermore, the change in dd-cf DNA percentage between pre and post treatment values (∆ dd-cf DNA %) demonstrated a positive correlation with estimated GFR (eGFR) at 1, 3 and 6 months with the larger the ∆ dd-cf DNA % value the higher rate of increase of eGFR. Similarly, Gupta et al[27] reported significant reductions in dd-cf DNA in eight patients following treatment of acute rejection. This reduction correlated with histological resolution of rejection. Of note, in this cohort, there was no decrease in dd-cf DNA fraction in 18 patients who did not respond to treatment.

DISCUSSION

The most recent data supports a role for routine monitoring of dd-cf DNA in the management of kidney transplant recipients. Early diagnosis of allograft injury can lead to timely biopsy, rapid intervention and tailored immunosuppression for the management of acute rejection. High negative predictor values may also aid in avoidance of biopsies with a decrease in reliance on protocol biopsies. The data reported by Bromberg et al[8] where dd-cf DNA was elevated not only long prior to clinical evidence of rejection using traditional biomarkers but also histological evidence of rejection on biopsy is of great interest. This can help identify patients who need closer surveillance allowing for more targeted allocation of available resources to patients who need it most. In addition, the data from the ADMIRAL study[7] showing a benign clinical course in patients with dd-cf DNA fractions < 5% can also assist with decisions with respect to frequency of testing and the intensity of follow up. However, whether routine use of dd-cd DNA in all kidney transplant recipients is cost effective or better utilized only in patients with higher immunological risk such as highly sensitized recipients is unclear and will require further study.

There may also be a role for dd-cf DNA and the monitoring of the response to treatment of acute renal allograft rejection by providing real time feedback of ongoing damage. This has the potential to allow for rapid modification of anti-rejection treatment to avoid inadequate or overly aggressive regimens thus preserving allograft function and avoiding complications associated with over immunosuppression.

Moving forward, there is room for further research to clarify the ideal level of cf-DNA, not only an absolute level for detection of rejection bit whether trends in levels of dd-cf DNA with time can indicate risk of future rejection and long term allograft survival.

CONCLUSION

Overall, available evidence suggests that dd-cf DNA monitoring can help assist in early diagnosis of acute rejection in kidney transplant allografts, identifying patients at high immunological risk and potentially facilitate personalized immunosuppression regimens. While dd-cf DNA cannot replace biopsy for the diagnosis of acute rejection, it has an ongoing role as a supplementary test to aid in clinical decision making.