An Assessment of the Value of Donor-derived Cell-free DNA Surveillance in Patients With Preserved Kidney Allograft Function : Transplantation

Journal Logo

Original Clinical Science—General

An Assessment of the Value of Donor-derived Cell-free DNA Surveillance in Patients With Preserved Kidney Allograft Function

Huang, Edmund MD1; Haas, Mark MD, PhD2; Gillespie, Matt PharmD1; Sethi, Supreet MD1; Peng, Alice MD1; Najjar, Reiad MD1; Vo, Ashley PharmD1; Jordan, Stanley C. MD1

Author Information
Transplantation 107(1):p 274-282, January 2023. | DOI: 10.1097/TP.0000000000004267
  • Free

Abstract

Introduction

Donor-derived cell-free DNA (dd-cfDNA) is a novel biomarker that is gaining traction in kidney transplantation as a predictor of allograft rejection and injury and is now used by over 160 transplant centers across the United States.1 However, its clinical utility in kidney transplantation remains undefined. Early data suggested that dd-cfDNA could be used to detect the presence of rejection on histology,2,3 but its inability to discriminate T cell–mediated rejection raised concern that reliance on dd-cfDNA alone could miss cases of rejection.4 Despite its limitations, many transplant programs have implemented routine dd-cfDNA surveillance protocols to screen for subclinical rejection and injury without sufficient data to support its use in this fashion. Of additional concern, the practice has extended beyond transplant programs to community providers which could lead to overutilization of dd-cfDNA testing in settings without proven clinical benefit.5

The performance of dd-cfDNA in the surveillance setting, where graft function is stable and there is low suspicion of rejection, has not been well-described. In this setting, because most patients will not have rejection when the pretest probability of rejection is low, a screening test with high sensitivity is desirable to detect the occasional patient with rejection. Early validation studies derived from cohorts tested “for-cause” in the setting of graft dysfunction or against archived biopsies described a sensitivity for rejection at the dd-cfDNA ≥1.0% threshold ranging from 47% to 89%,2-4,6 indicating that up to 53% of cases of rejection could be missed by relying on dd-cfDNA alone. A recent study assessing routine dd-cfDNA surveillance against indication (68/113; 60%) and surveillance kidney transplant biopsies (45/113; 40%) described similar performance at the dd-cfDNA ≥1.0% threshold with a sensitivity of 64%, although the sensitivity could be raised to 80% by lowering the dd-cfDNA threshold for a positive test to 0.5% (albeit with a reduction in specificity from 73% to 59%).7

In addition to a limited sensitivity to detect rejection, the positive predictive value of dd-cfDNA ranges from 52% to 61%,2,3,7 suggesting that routine dd-cfDNA surveillance may lead to a number of false-positive results and unnecessary kidney transplant biopsies. However, it is possible that an elevated dd-cfDNA in the absence of rejection may be clinically significant. There is emerging thought that “false-positive” dd-cfDNA results might not be erroneous and could be related to the insensitivity of histology for allograft injury. In one study of dd-cfDNA in heart transplantation, 2/3 of “false-positive” dd-cfDNA results with negative histology from endomyocardial biopsies were associated with concurrent or later allograft dysfunction or future rejection suggesting that dd-cfDNA has the potential to be a better marker of allograft injury than histology.8 This concept has also been reported in kidney transplantation, where elevated dd-cfDNA was associated with future rejection, de novo donor-specific antibodies (DSAs), and later allograft dysfunction.7,9,10 These studies have included a mixture of patients with kidney allograft dysfunction and rejection, making it unclear whether future adverse events can be predicted by dd-cfDNA alone or are explained by known injury at the time of assessment. In the absence of studies looking specifically at the significance of dd-cfDNA in the surveillance setting, where there is little clinical suspicion of rejection or injury, the appropriate interpretation and management of a dd-cfDNA result is not yet defined.

Because up to 40%–50% of patients with an elevated dd-cfDNA will not have corresponding rejection on histology based on previously reported performance characteristics, the prognostic significance of dd-cfDNA should be described further to justify its use as a surveillance test and to characterize whether elevated dd-cfDNA in the absence of rejection is a harbinger of future events or simply represents a false-positive test result. In this study, we investigated the association of dd-cfDNA in patients with preserved allograft function with graft outcomes, including future rejection, graft loss, DSA development, and estimated glomerular filtration rate (eGFR) trajectory.

MATERIALS AND METHODS

This is a retrospective observational study approved by the Cedars-Sinai Medical Center Institutional Review Board (Pro00020945) that included all kidney transplant recipients at Cedars-Sinai Medical Center who had an assessment of dd-cfDNA (Allosure, CareDx, Inc.) in the setting of preserved allograft function between August 1, 2017, and November 3, 2020, and had at least 6 mo of follow-up. Because this study was considered to involve no more than minimal risk to subjects, the requirement for informed consent was waived by the Institutional Review Board. Preserved allograft function was defined as having a creatinine ≤1.5 mg/dL, no current DSA, and no prior history of rejection. For study inclusion, the most recent creatinine value to dd-cfDNA measurement was used and the vast majority (88%) had creatinine measured on the same day that dd-cfDNA was drawn. Patients with a prior history of DSA were included as long as DSA was absent at the time of dd-cfDNA assessment. The following exclusion criteria were used for dd-cfDNA testing per manufacturer guidelines: (1) age <18 y old; (2) <2 wk posttransplant at time of assessment; (3) pregnant status; (4) recipient of a transplant from an identical twin; (5) recipient of a multiorgan transplant; and (6) prior recipient of an allogeneic bone marrow transplant. Because the manufacturer recommends that dd-cfDNA among patients with a prior allograft in situ be interpreted similarly to those with a single kidney transplant, repeat kidney transplant patients with a prior allograft in situ were not excluded. Patients were followed until death, graft loss, or the end of the study period on May 24, 2021. All patients who maintained a functioning graft throughout the study period had a minimum of at least 180 d of follow-up.

Because dd-cfDNA was used for routine surveillance, most patients had multiple measurements of dd-cfDNA during the follow-up period. This study was interested in the prognostic significance of an isolated dd-cfDNA result, and therefore, we chose the first dd-cfDNA assessment as the dd-cfDNA value used for the study. Patients were categorized according to dd-cfDNA levels as follows: low (dd-cfDNA <0.5%), moderate (dd-cfDNA 0.5 to <1.0%), and high (dd-cfDNA ≥1.0%). This categorization was based on thresholds for dd-cfDNA previously suggested.2,9 Baseline characteristics at the time of dd-cfDNA assessment were compared between dd-cfDNA categories using the chi-square test or Kruskal-Wallis test, as appropriate.

Study outcomes included the occurrence of rejection, DSA, graft loss, death, and eGFR slope over time after dd-cfDNA assessment between dd-cfDNA categories. The decision to pursue a biopsy was made by the treating physician and was performed either to investigate allograft dysfunction or in response to an elevated dd-cfDNA result. Biopsies were scored by trained renal pathologists using Banff 2017 criteria. Luminex single-antigen testing is performed at 1, 3, 6, 9, and 12 mo and annually on patients with calculated panel reactive antibodies (cPRA) ≥ 30% or recipients of a prior transplant and annually on all other patients. Graft loss was defined as death, return to dialysis or retransplantation. eGFR was calculated with the Chronic Kidney Disease Epidemiology Collaboration creatinine equation. The slope of eGFR over time was modeled using a linear mixed-effects model with random slopes and intercepts and an unstructured covariance matrix to compare slope differences between dd-cfDNA categories. Covariates included in the model included age at dd-cfDNA assessment, donor type (living versus deceased donor), history of DSA, and the dd-cfDNA category × time (y) interaction.

P values were 2-tailed, and a P <0.05 was considered statistically significant. Statistical analyses were performed using Stata, version 14.2 (College Station, TX).

RESULTS

Baseline Characteristics

The study flowchart is shown in Figure 1. A total of 317 kidney transplant recipients were included in the study. Baseline clinical and serologic characteristics of the patients are summarized in Table 1. Of these, 239 (75%) were categorized as low (dd-cfDNA <0.5%), 43 (14%) as moderate (dd-cfDNA 0.5 to <1.0%), and 35 (11%) as high (dd-cfDNA ≥1.0%) levels of dd-cfDNA. There was no difference in the time posttransplant at dd-cfDNA assessment between the low, moderate, and high dd-cfDNA groups (Table 1 and Figure 2). Additionally, there was no difference in the distribution of eGFR between the 3 dd-cfDNA groups. The median follow-up duration after dd-cfDNA assessment was 590 d (interquartile range [IQR]: 376–763 d).

TABLE 1. - Baseline characteristics
dd-cfDNA category P
Low (n = 239) Moderate (n = 43) High (n = 35)
dd-cfDNA %, median (IQR) 0.26 (0.19–0.32) 0.67 (0.59–0.83) 2.3 (1.2–3.6) <0.001
 Solitary graft 0.26 (0.19–0.32) a 0.63 (0.59–0.84) b 1.7 (1.2–3.2) c a 0.93
b 0.06
c 0.26
 Prior allograft in situ 0.26 (0.19–0.35) 0.75 (0.74–0.83) 2.8 (1.3–4.8)
Time posttransplant at assessment, median d (IQR) 240 (50–1269) 80 (32–941) 117 (40–903) 0.22
eGFR at assessment, median (mL/min/1.73 m2; IQR) 58 (47–69) 54 (48–74) 62 (52–80) 0.25
Age at assessment, median y (IQR) 55 (43–64) 55 (36–67) 49 (39–59) 0.10
Gender (%) 0.001
 Male 142 (59) 19 (44) 10 (29)
 Female 97 (41) 24 (56) 25 (71)
Race/ethnicity (%) 0.11
 White 87 (36) 17 (40) 6 (17)
 Black 21 (9) 2 (5) 8 (23)
 Hispanic 84 (35) 16 (37) 14 (40)
 Asian 45 (19) 8 (19) 6 (17)
 Other 2 (1) 0 (0) 1 (3)
Cause of ESKD (%) 0.55
 Hypertension 35 (15) 5 (12) 4 (11)
 Diabetes mellitus 59 (25) 7 (16) 4 (11)
 Polycystic kidney 27 (11) 5 (12) 3 (9)
 Glomerulonephritis 65 (27) 13 (30) 12 (34)
 Other 53 (22) 13 (30) 12 (34)
Donor type (%) 0.12
 Deceased 123 (51) 26 (60) 24 (69)
 Living 116 (49) 17 (40) 11 (31)
Previous transplant (%) 28 (12) 10 (23) 17 (49) <0.001
 Prior allograft in situ 22 (9) 7 (16) 15 (43) <0.001
Calculated PRA at transplant (%) <0.001
 0%–20% 167 (70) 24 (56) 16 (46)
 21%–80% 22 (9) 7 (16) 2 (6)
 81%–98% 13 (5) 5 (12) 8 (23)
 99%–100% 11 (5) 2 (5) 6 (17)
 Missing 26 (11) 5 (12) 3 (9)
Prior DSA (%) 9 (4) 2 (5) 4 (11) 0.14
Antibody induction at transplant (%) d 0.001
 None 5 (2) 0 (0) 0 (0)
 IL-2 receptor antagonist 77 (32) 9 (21) 4 (11)
 Antithymocyte globulin 73 (31) 13 (30) 5 (14)
 Alemtuzumab 80 (33) 20 (47) 25 (71)
Immunosuppression (%)
 Tacrolimus 203 (85) 38 (88) 31 (89) 0.74
 Cyclosporine 22 (9) 3 (7) 1(3) 0.42
 Belatacept 18 (8) 2 (5) 3 (9) 0.76
 Mycophenolate 224 (94) 39 (91) 33 (94) 0.74
 Sirolimus/everolimus 7 (3) 1 (2) 0 (0) 0.59
 Prednisone 230 (96) 42 (98) 34 (97) 0.87
a,b,cP value for comparison between solitary allograft vs prior allograft in situ (a = low dd-cfDNA group; b = moderate dd-cfDNA group; c = high dd-cfDNA group).
dAntibody induction data missing for 6 patients.
dd-cfDNA, donor-derived cell-free DNA; DSA, donor-specific antibodies; eGFR, estimated glomerular filtration rate; ESKD, end-stage kidney disease; IL, interleukin; IQR, interquartile range; PRA, panel reactive antibody.

F1
FIGURE 1.:
Study flowchart. AMR, antibody-mediated rejection; CMR, cell-mediated rejection; dd-cfDNA, donor-derived cell-free DNA; DSA donor-specific antibody.
F2
FIGURE 2.:
Distribution of time posttransplant at donor-derived cell-free DNA assessment.

There were 55 patients (17% of the total population) who were recipients of a prior transplant. Of these, 44 (80%) had a prior allograft remaining in situ. Among retransplanted recipients, there was no difference in the distribution of dd-cfDNA between those with a prior allograft nephrectomy (median 0.44; IQR: 0.26–1.3) and those with a prior allograft in situ (median 0.56; IQR: 0.25–1.35; P = 0.50). Additionally, there was no difference in the distribution of dd-cfDNA by the presence of an allograft in situ in any of the dd-cfDNA groups (Table 1). Patients in the moderate and high dd-cfDNA categories were more likely to have had a previous transplant and had higher cPRA at transplant than patients in the low dd-cfDNA category. However, there was no difference between the groups in the percentage of patients with a prior DSA. Patients in the moderate and high dd-cfDNA categories were more likely to receive lymphocyte-depleting antibodies for induction at transplant. Maintenance immunosuppression was similar among the 3 groups and mostly consisted of tacrolimus, mycophenolate, and prednisone.

Rejection

A total of 62 of 372 patients overall underwent a biopsy at some point after dd-cfDNA measurement (20% of the overall population). Of these, 35 of 239 (15%) were in the low dd-cfDNA group, 9 of 43 (21%) were in the moderate dd-cfDNA group, and 18 of 35 (51%) were in the high dd-cfDNA group. The median number of days from dd-cfDNA measurement to biopsy was 68 (IQR: 27–245 d). Most biopsies (76%) were performed to assess allograft dysfunction, whereas 23% were performed in response to an elevated dd-cfDNA result (33% of biopsies in the moderate and 56% in the high dd-cfDNA groups). Twenty-one patients overall underwent a biopsy within 1 mo of dd-cfDNA measurement [9 of 35 (26%) biopsied patients in the low dd-cfDNA group, 2 of 9 (22%) in the moderate dd-cfDNA group, and 10 of 18 (56%) in the high dd-cfDNA group].

Twenty-four patients were diagnosed with rejection by Banff criteria at a median 107 d (IQR: 28–375 d) after dd-cfDNA assessment (Table 2). Compared to patients with low dd-cfDNA (13 of 239; 5%), rejections after dd-cfDNA assessment were more common among patients with high dd-cfDNA (6 of 35; 17%, P = 0.01). Similar observations were seen among patients with moderate dd-cfDNA (5 of 43; 12%), although this was not statistically significant (P = 0.13). At 1 y after dd-cfDNA assessment, rejection-free survival was 97% among patients with low dd-cfDNA, 91% among those with moderate dd-cfDNA, and 82% among those with high dd-cfDNA (Figure 3, log-rank P = 0.002).

TABLE 2. - Distribution of rejection cases detected after dd-cfDNA assessment
Donor-derived cell-free DNA (dd-cfDNA) category P
Low (n = 239) Moderate (n = 43) High (n = 35)
Any rejection (%) 13 (5) a 5 (12) a 6 (17) 0.03
Cell-mediated rejection (%) 12 (5) 5 (12) 5 (14) 0.06
 Borderline 3 (1) 1 (2) 2 (6)
 1A 1 (0.4) 1 (2) 0 (0)
 1B 4 (2) 0 (0) 2 (6)
 2A 0 (0) 1 (2) 0 (0)
 2B 2 (0.8) 0 (0) 1 (3)
 Chronic active 1A 2 (0.8) 2 (5) 0 (0)
AMR,  Banff-diagnostic (AMR, %) a 2 (0.8) 1 (2) 1 (3) 0.48
 Active AMR (Banff-diagnostic) 0 (0) 1 (2) 0 (0)
 Chronic active AMR  (Banff-diagnostic) 2 (0) 0 (0) 0 (0)
 Chronic AMR (Banff-diagnostic)/ suspicious active AMR b 0 (0) 0 (0) 1 (3)
 Suspicious active AMR c 0 (0) 0 (0) 4 (11)
 Suspicious chronic active AMR c 1 (0.4) 0 (0) 3 (9)
aOne patient each in the low dd-cfDNA and moderate dd-cfDNA groups had mixed TCMR and AMR.
bBiopsy met criteria for chronic AMR but did not meet diagnostic criteria for active AMR given a historical donor-specific antibody without current donor-specific antibody.
cBiopsy did not meet criteria for chronic or active AMR given absence of historical or current DSA.
AMR, antibody-mediated rejection; dd-cfDNA, donor-derived cell-free DNA; DSA, donor-specific antibody; TCMR, T cell–mediated rejection.

F3
FIGURE 3.:
Comparison of rejection-free survival between patients with low, moderate, and high levels of donor-derived cell-free DNA (dd-cfDNA).

Most patients with rejection had cell-mediated rejection (CMR) (n = 22). Of these, 6/22 CMR cases (27%) were borderline by Banff criteria.Treatment consisted of pulse corticosteroids in 12, antithymocyte globulin in 6, and no treatment in 4 (3 cases of borderline CMR and 1 case of chronic active CMR). Excluding borderline cases and considering CMR Banff grade 1A or higher only, there was no difference in the frequency of CMR between the 3 groups [low: 9/242 (4%); moderate: 4/43 (9%); high: 3/35 (9%); P = 0.19].

Only 8 patients developed DSA after dd-cfDNA measurement, all in the low and moderate dd-cfDNA groups (low: n = 7; moderate: n = 1). There was no difference in the percentage of patients who developed DSA among the 3 groups (low: 4%; moderate: 3%; high: 0%; P = 0.52). Seven of 8 patients underwent kidney biopsy to assess for antibody-mediated changes on biopsy of whom only 3 were found to have active features of antibody-mediated rejection (active or chronic active AMR) (low dd-cfDNA: n = 2; moderate dd-cfDNA: n = 1). An additional patient in the high dd-cfDNA group had historical DSA and was diagnosed with chronic AMR based on early glomerular basement membrane duplication on electron microscopy (Banff cg1a). This biopsy had glomerulitis and peritubular capillaritis but was considered suspicious and not diagnostic of active AMR given the absence of current DSA. Of the 4 patients, 2 patients were treated with intravenous immunoglobulin plus an anti-CD20 monoclonal antibody, 1 patient was treated with intravenous immunoglobulin and tocilizumab, and 1 patient was not treated. Overall, there was no difference between the groups in the frequency of AMR diagnosed by Banff criteria (low: 0.8%; moderate: 2%; high: 2%; P = 0.48).

In addition to the 4 patients above who met Banff criteria for AMR, 8 patients had histology suspicious for AMR but did not meet Banff criteria given the absence of DSA or positive peritubular capillary C4d staining. This included 1 patient who received an ABO-incompatible transplant (blood type A→O) and was found to have very high dd-cfDNA on routine screening (dd-cfDNA: 11%). A repeat dd-cfDNA test was repeated 18 d after the first to exclude the possibility of a spurious test result or transient elevation in dd-cfDNA and remained elevated at 12%. Although clinical suspicion for rejection was low given that the creatinine was unchanged and anti-A titers were low (1:16), a kidney transplant biopsy was performed to assess for subclinical rejection. This biopsy was performed 35 d after the initial dd-cfDNA assessment and had moderate peritubular capillaritis with diffuse peritubular capillary C3d staining suggestive of acute AMR. Because Banff does not address C3d staining, this biopsy was classified as suspicious, but not diagnostic by Banff criteria, for AMR. Overall, 12 patients in total had histologic signs of AMR on biopsies performed after dd-cfDNA measurement. Most of these cases were observed in the high dd-cfDNA group [low: 3 of 236 (1%), moderate: 1 of 43 (2%), high: 8 of 35 (23%); P < 0.001].

Rejection Outcomes Stratified by Time of Assessment Posttransplant

Among the 317 patients in the study population, 118 (59%) were assessed with dd-cfDNA within the first year of transplant. Of these, 137 (73%) were in the low dd-cfDNA group, 27 (14%) were in the moderate dd-cfDNA group, and 24 (13%) were in the high dd-cfDNA group. This distribution was similar to those assessed after the first year [low: n = 102 (79%); moderate: n = 16 (12%); high: n = 11 (9%); P = 0.40]. Among those who were assessed beyond the first year, the number of days posttransplant at dd-cfDNA assessment ranged from 367–11 027 d. Nearly all rejection cases (21 of 24, 88%) were observed among patients assessed with dd-cfDNA in the first year. These rejections were identified at a median 77 d (IQR: 27–358 d) after dd-cfDNA assessment. Rejections were identified in 10/137 (7%) patients in the low dd-cfDNA group, 5 of 27 (19%) in the moderate dd-cfDNA group, and 6 of 24 (25%) in the high dd-cfDNA group (P = 0.02).

Three additional cases of rejection were identified among 129 patients (low, n = 102; moderate, n = 16; high, n = 11) assessed with dd-cfDNA beyond the first year posttransplant. All 3 patients were in the low dd-cfDNA group and had dd-cfDNA measured at 382, 658, and 4483 d posttransplant. These rejections were identified at 171, 180, and 395 d after dd-cfDNA assessment. None of the 11 patients in the high dd-cfDNA group underwent a transplant biopsy during the follow-up period because of ongoing clinical stability. Although the possibility of subclinical rejection cannot be excluded, no patient showed signs of clinical worsening throughout the study period. At a median 696 d follow-up after dd-cfDNA assessment (IQR: 658–784 d), the eGFR remained stable in all 11 patients [median change in eGFR from baseline to the last follow-up: 11 mL/min/1.73 m2 (IQR: −2 to 13 mL/min/1.73 m2)].

Death and Graft Loss

Ten deaths occurred during the study period at a median of 612 d (IQR 412–748 d). All grafts were functioning at the time of death in each of these cases. There were 8 deaths (3%) in the low dd-cfDNA group, 1 death due to liver failure in the moderate dd-cfDNA group (2%), and 1 death due to infection (3%) in the high dd-cfDNA group. The causes of death in the low dd-cfDNA group were attributed to infection in 6 cases (4 coronavirus disease 2019–related), malignancy in 1 case, and unknown in 1 case. Only 1 graft loss developed otherwise in a patient with low dd-cfDNA, which was due to CMR refractory to treatment. This rejection developed >1 y after dd-cfDNA assessment.

Comparison of eGFR Over Time Between Low, Moderate, and High dd-cfDNA Categories

All patients had a follow-up eGFR measurement at least 6 mo after dd-cfDNA assessment. Over 6 mo, eGFR was stable among all 3 dd-cfDNA categories and there were no differences in the change in eGFR from baseline to 6 mo between the groups (Table 3). This observation was seen both among patients assessed with dd-cfDNA within and after the first posttransplant year.

TABLE 3. - Comparison of eGFR over time by dd-cfDNA category
Low (n = 239) Moderate (n = 43) High (n = 35) P
Overall
 Change in eGFR over 6 mo, mL/min/1.73 m2  [median (IQR)] 1.6 (−4.5 to 11.2) 0 (−7.2 to 7.6) 0 (−6.5 to 11.3) 0.60
 Slope of eGFR, a mL/min/1.73 m2/y (95% CI) 1.6 (0.31 to 2.89) −3.2 (−6.3 to −0.1) 5.1 (1.9 to 8.3) Moderate vs low: 0.005
High vs low: 0.048
Assessed within the first y
 Change in eGFR over 6 mo, mL/min/1.73 m2  [median (IQR)] 3.6 (−3.6 to 14.7) 2.5 (−4.2 to 13.3) 0 (−2.9 to 12.6) 0.89
 Slope of eGFR, a mL/min/1.73 m2/y (95% CI) 2.1 (0.3 to 3.8) −4.8 (−8.9 to −0.8) 4.4 (0.2 to 8.6) Moderate vs low: 0.002
High vs low: 0.31
Assessed after the first y
 Change in eGFR over 6 mo, mL/min/1.73 m2  [median (IQR)] 0 (−4.7 to 6.1) −0.26 (−6.9 to 3.0) −3.2 (−19.5 to 9.4) 0.56
 Slope of eGFR, a mL/min/1.73 m2/y (95% CI) 0.7 (−1.3 to 2.6) −0.6 (−5.6 to 4.4) 7.1 (1.9 to 12.3) Moderate vs low: 0.64
High vs low: 0.02
aDerived from linear mixed effects model adjusted for age, donor type (living vs deceased donor), historical DSA, and the dd-cfDNA category × time interaction.
CI, confidence interval; dd-cfDNA, donor-derived cell-free DNA; DSA, donor-specific antibody; eGFR, estimated glomerular filtration rate; IQR, interquartile range.

Using a linear mixed-effects model (median follow-up 590 d; IQR 376–763 d) to model eGFR over time, there was a significant interaction between dd-cfDNA category and time. This indicates that there was a significant difference in eGFR slope over time between groups. After adjusting for age at dd-cfDNA assessment, donor type (living versus deceased donor), and history of DSA, the change in eGFR over time was 1.6 mL/min/1.73 m2/y for patients in the low dd-cfDNA group, -3.2 mL/min/1.73 m2/y in the moderate dd-cfDNA group, and 5.1 mL/min/1.73 m2/y for patients in the high dd-cfDNA group (P for interaction: moderate versus low = 0.005; high versus low = 0.048). Similar trends were seen in the subgroups of patients assessed within and after the first posttransplant year aside from no difference in the eGFR slope between the low and moderate dd-cfDNA groups observed among patients assessed after the first posttransplant year (Table 3).

Approximately half of the patients with high dd-cfDNA (18 of 35; 51%) underwent an allograft biopsy, of whom 6 were found to have rejection and were treated. There were 17 patients with high dd-cfDNA who did not undergo allograft biopsy. Their eGFR over time was similar to the overall high dd-cfDNA group and demonstrated clinical stability despite not undergoing a biopsy nor being treated. The median change in eGFR over 6 mo in nonbiopsied patients was 3.3 (−0.8 to 21.0) mL/min/1.73 m2 and the eGFR slope over a median 543 (IQR 385–657) d of follow-up was 5.1 (95% CI, 1.1-9.1) mL/min/1.73 m2/y.

DISCUSSION

In this study, we observed that elevated levels of dd-cfDNA in patients without clinical suspicion of rejection were more commonly observed in patients with higher immunologic risk. These patients were more likely to have had a previous transplant and generally had higher cPRA at transplant than patients with low dd-cfDNA. When dd-cfDNA was measured for surveillance of subclinical injury, more cases of rejection were detected over the follow-up period among patients with higher levels of dd-cfDNA. However, most patients with elevated dd-cfDNA did well on follow-up without subsequent clinical rejections, graft loss, or significant deterioration in eGFR, although longer-term follow-up will be necessary to validate these findings.

Our data suggest that dd-cfDNA is better characterized as a predictor of rejection rather than a marker of injury. The distinction between predictor and marker is subtle but important for consideration of the role of dd-cfDNA for routine surveillance of clinically stable allografts. A predictor characterizes patients who are at risk for adverse outcomes whereas a marker indicates the presence of injury. Because cell-free DNA arises as a byproduct of cell death in most circumstances where intracellular DNA is released into the circulation, dd-cfDNA can, in theory, be a marker for the presence of allograft injury11 and is attractive as a surveillance marker because it is more sensitive and specific for rejection than creatinine.2,3 However, the association between elevated dd-cfDNA and concomitant rejection is imperfect, with reported sensitivities ranging from 47% to 89% and specificities ranging from 73% to 88% across clinical studies.2-4,6 Because allograft injury can result from both immunologic and nonimmunologic factors, studies have suggested that dd-cfDNA has broader applications for allograft surveillance to detect injury beyond rejection. However, the association of dd-cfDNA with other forms of injury, such as BK virus nephropathy, bacterial infections, glomerulonephritis, and drug toxicity is inconsistent and therefore dd-cfDNA appears to be unreliable as a universal marker of allograft injury.3,12,13

Although there is a theoretical rationale for the use of dd-cfDNA as a screening marker for allograft injury, its ability to identify subclinical injury has not been fully validated. A central problem in the literature is that the definition of injury is nebulous and inconsistent across studies. Injury can in theory be defined at the cellular or molecular level, by the presence of histologic abnormalities, allograft dysfunction, or a combination of any of these factors. In the recently published Assessing Donor-Derived Cell-Free DNA Monitoring Insights of Kidney Allografts with Longitudinal Surveillance study, elevated dd-cfDNA was associated with a composite measure of allograft injury defined by out-of-range tacrolimus levels, BK viremia, presence of DSA, urinary tract infection, proteinuria, rejection, or recurrent focal segmental glomerulosclerosis.7 However, the significance of this association is unclear, as not all of these conditions are consistently associated with pathologies, such as out-of-range tacrolimus levels, BK viremia without nephropathy, DSA without AMR, or acute cystitis without pyelonephritis. Furthermore, cell-free DNA can be released into the circulation by mechanisms other than cell death and its origins and underlying pathology are critical to understanding the implications of an elevated dd-cfDNA.14 Therefore, it cannot be assumed in all cases that elevated dd-cfDNA indicates an underlying injurious process in the allograft.

Regardless of any debate about the utility of dd-cfDNA as an injury marker, our study does point to dd-cfDNA as a predictor of future rejection. Future rejection was more common among patients with elevated dd-cfDNA; however, >80% of patients with dd-cfDNA ≥1% were not found to have rejection over the study period. Although one explanation could be that dd-cfDNA is more sensitive than histology as a marker of subclinical injury, the injury should result in future dysfunction. However, dd-cfDNA was not associated with later dysfunction or graft loss and affirms that the current practice of graft surveillance with creatinine and DSA measurements is reasonably good. Our findings are comparable to those reported from a post hoc analysis of the Circulating Donor-Derived Cell-Free DNA in Blood for Diagnosing Acute Rejection in Kidney Transplant Recipients study (NCT02424227) where more patients with at least 1 measurement of dd-cfDNA ≥1% in the first year after transplant experienced ≥25% reduction in eGFR by year 2 compared to those with dd-cfDNA <1% (21% versus 4%, P = 0.005), indicating that elevated dd-cfDNA is a risk factor for future dysfunction.10 On the other hand, 79% of patients with dd-cfDNA ≥1% in the first year in this study did not exhibit a subsequent decline in eGFR.

Whereas there appears to be a role of dd-cfDNA surveillance to identify patients at risk for rejection, questions remain regarding which patients to test, the appropriate time intervals posttransplant to assess dd-cfDNA, and the appropriate steps to further evaluate an elevated dd-cfDNA. A kidney transplant biopsy can be considered to exclude the presence of rejection and closer monitoring of patients with elevated dd-cfDNA may be reasonable. However, the data indicates that a fair number of biopsies prompted by an isolated elevated dd-cfDNA will not reveal pathology that will lead to an actionable intervention.2,3,7 Although these can be considered “false-positive” dd-cfDNA cases, an alternative explanation could be that these cases represent “false-negative” histology where underlying injury is missed due to interobserver variability or sampling error.15,16 It is well-known that histology can be discordant with molecular diagnoses rendered by tissue transcriptomics,17 yet it remains unclear how to incorporate non–histology-based measures of rejection or injury into the current diagnostic schema. Two recent studies assessing the correlation between dd-cfDNA, histology, and tissue transcriptomics observed a number of cases where dd-cfDNA was discrepant with histology, transcriptomics, or both.18,19 Additional biomarkers may be necessary to differentiate patients with elevated dd-cfDNA who will experience future adverse outcomes from those who will have a stable course and are an area of investigation. There is recent data showing that the combination of dd-cfDNA and blood-based transcriptomics can help improve the detection of subclinical rejection,6 and further studies investigating this combination for routine monitoring of kidney transplant patients are ongoing (NCT03326076, NCT04491552). Ultimately, future studies that correlate dd-cfDNA with both protocol biopsies and long-term outcomes are needed to define how best to use dd-cfDNA as a monitoring tool.

The strength of our study is that it only included patients who had preserved allograft function without a prior history of rejection and no clinical suspicion of rejection. In doing so, we sought to investigate the significance of an isolated elevated dd-cfDNA to better understand the prognostic capability and utility of routine surveillance with dd-cfDNA. Our study observed that elevated dd-cfDNA is more likely observed among patients with higher immunologic risk and suggests that targeted dd-cfDNA surveillance in patients at higher risk for rejection may be more fruitful than universal screening in all kidney transplant recipients.

We acknowledge some limitations of this study. First, we do not perform protocol biopsies at our center and, because we do not have a defined protocol for responding to a dd-cfDNA result, not all patients had a biopsy to correlate with dd-cfDNA measurements. Therefore, we did not assess whether dd-cfDNA can discriminate the presence of subclinical rejection and relied on outcomes to characterize the clinical relevance of dd-cfDNA. Nevertheless, we suggest that outcomes, rather than biopsy correlates, are a better measure of the clinical relevance of dd-cfDNA, particularly because prior studies have differed on the prognostic significance of subclinical rejection, especially for subclinical CMR.20-22 The reasons for biopsy varied between the groups, where patients with low dd-cfDNA were more likely to be subsequently biopsied to assess allograft dysfunction whereas patients in the moderate and high dd-cfDNA groups were more likely to be biopsied to assess an elevated dd-cfDNA result. Nevertheless, very few (5%) patients with a low dd-cfDNA result were found to have later rejection and eGFR remained stable in all 3 dd-cfDNA groups. Second, because most patients in this study who were found to have rejection were treated, we cannot rule out that treatment of rejections attenuated future deterioration in kidney function and masked a potential benefit of dd-cfDNA screening, nor can we speculate whether these cases would have fared similarly if rejections were not identified and thus not treated. Only 4 patients with rejection were not treated in this study and none had any change in kidney function over the study period. There were 17 patients in the high dd-cfDNA group who did not undergo a biopsy and maintained standard immunosuppression through follow-up. Despite the possibility that subclinical rejections may not have been detected, eGFR over a median 543 (IQR 385–657) d follow-up was stable. Third, it is possible that differences in outcomes may become more apparent with a longer duration of follow-up, although the relevance of an isolated dd-cfDNA measurement on graft events beyond the follow-up time in this study is unclear given the short half-life of dd-cfDNA in circulation (approximately 30 min).

In conclusion, most clinically stable patients monitored with dd-cfDNA for injury surveillance exhibited continued clinical stability with short-term follow-up irrespective of dd-cfDNA level. The exact role of dd-cfDNA in the prediction of longer-term outcomes and monitoring of stable kidney transplant recipients needs to be clarified further.

REFERENCES

1. CareDx. Kidney AlloSure: lead with the predictive power of AlloSure. 2022. Available at https://caredx.com/products-and-services/transplant-services/kidney/allosure/. Accessed February 20, 2022.
2. Bloom RD, Bromberg JS, Poggio ED, et al.; Circulating Donor-Derived Cell-Free DNA in Blood for Diagnosing Active Rejection in Kidney Transplant Recipients (DART) Study Investigators. Cell-free DNA and active rejection in kidney allografts. J Am Soc Nephrol. 2017;28:2221–2232.
3. Sigdel TK, Archila FA, Constantin T, et al. Optimizing detection of kidney transplant injury by assessment of donor-derived cell-free DNA via massively multiplex PCR. J Clin Med. 2018;8:E19.
4. Huang E, Sethi S, Peng A, et al. Early clinical experience using donor-derived cell-free DNA to detect rejection in kidney transplant recipients. Am J Transplant. 2019;19:1663–1670.
5. Danovitch GM, Bunnapradist S, Cohen D, et al. Tests for the noninvasive diagnosis of kidney transplant rejection should be evaluated by kidney transplant programs. Am J Transplant. 2021;21:3811.
6. Park S, Guo K, Heilman RL, et al. Combining blood gene expression and cellfree DNA to diagnose subclinical rejection in kidney transplant recipients. Clin J Am Soc Nephrol. 2021;16:1539–1551.
7. Bu L, Gupta G, Pai A, et al. Clinical outcomes from the Assessing Donor-derived cell-free DNA Monitoring Insights of kidney Allografts with Longitudinal surveillance (ADMIRAL) study. Kidney Int. 2022;101:793–803.
8. Agbor-Enoh S, Shah P, Tunc I, et al.; GRAfT Investigators. Cell-free DNA to detect heart allograft acute rejection. Circulation. 2021;143:1184–1197.
9. Stites E, Kumar D, Olaitan O, et al. High levels of dd-cfDNA identify patients with TCMR 1A and borderline allograft rejection at elevated risk of graft injury. Am J Transplant. 2020;20:2491–2498.
10. Sawinski DL, Mehta S, Alhamad T, et al. Association between dd-cfDNA levels, de novo donor specific antibodies, and eGFR decline: an analysis of the DART cohort. Clin Transplant. 2021;35:e14402.
11. Jackson AM, Amato-Menker C, Bettinotti M. Cell-free DNA diagnostics in transplantation utilizing next generation sequencing. Hum Immunol. 2021;82:850–858.
12. Mayer KA, Doberer K, Tillgren A, et al. Diagnostic value of donor-derived cell-free DNA to predict antibody-mediated rejection in donor-specific antibody-positive renal allograft recipients. Transpl Int. 2021;34:1689–1702.
13. Xie WY, Kim K, Goussous N, et al. Causes of renal allograft injury in recipients with normal donor-derived cell-free DNA. Transplant Direct. 2021;7:e679.
14. Grabuschnig S, Bronkhorst AJ, Holdenrieder S, et al. Putative origins of cell-free DNA in humans: a review of active and passive nucleic acid release mechanisms. Int J Mol Sci. 2020;21:E8062.
15. Furness PN, Taub N; Convergence of European Renal Transplant Pathology Assessment Procedures (CERTPAP) Project. International variation in the interpretation of renal transplant biopsies: report of the CERTPAP Project. Kidney Int. 2001;60:1998–2012.
16. Smith B, Cornell LD, Smith M, et al. A method to reduce variability in scoring antibody-mediated rejection in renal allografts: implications for clinical trials – a retrospective study. Transpl Int. 2019;32:173–183.
17. Madill-Thomsen K, Perkowska-Ptasińska A, Böhmig GA, et al.; MMDx-Kidney Study Group. Discrepancy analysis comparing molecular and histology diagnoses in kidney transplant biopsies. Am J Transplant. 2020;20:1341–1350.
18. Gupta G, Moinuddin I, Kamal L, et al. Correlation of donor-derived cell-free DNA with histology and molecular diagnoses of kidney transplant biopsies. Transplantation. 2022;106:1061–1070.
19. Halloran PF, Reeve J, Madill-Thomsen KS, et al.; Trifecta Investigators. The trifecta study: comparing plasma levels of donor-derived cell-free DNA with the molecular phenotype of kidney transplant biopsies. J Am Soc Nephrol. 2022;33:387–400.
20. Scholten EM, Rowshani AT, Cremers S, et al. Untreated rejection in 6-month protocol biopsies is not associated with fibrosis in serial biopsies or with loss of graft function. J Am Soc Nephrol. 2006;17:2622–2632.
21. Loupy A, Vernerey D, Tinel C, et al. Subclinical rejection phenotypes at 1 year post-transplant and outcome of kidney allografts. J Am Soc Nephrol. 2015;26:1721–1731.
22. Rush D, Nickerson P, Gough J, et al. Beneficial effects of treatment of early subclinical rejection: a randomized study. J Am Soc Nephrol. 1998;9:2129–2134.
Copyright © 2022 Wolters Kluwer Health, Inc. All rights reserved.