Donor-Derived Cell-Free DNA and Active Rejection in Renal Allografts

Sample size, patient selection, and data collection

This was a prospective study to test the diagnostic accuracy of dd-Cf-DNA. A pilot study with convenient sample size was designed, as this is the first of its kind study so far reported from India and no commercial laboratory offered this testing at the time of the study initiation. Institutional Ethics Committee approval was obtained.

Live related renal transplant recipients who presented with graft dysfunction and had been planned for a graft biopsy as per unit protocol were included as a part of the study. No patients within 2 weeks of transplant or any surgical intervention of the allograft were included. This was done as previous studies have shown that dd-Cf-DNA levels reach a peak immediately after engraftment and stabilize to low levels within 1 week.[4] After the patients gave informed consent, they were subjected to interview and examination to collect baseline data.

Ten milliliters of venous blood was drawn for estimating dd-Cf-DNA levels. All samples were drawn before graft biopsy or 24 h after the biopsy. The latter cut-off was chosen because the half-life of dd-Cf-DNA is less than 1.5 h, and this 24-h delay is sufficient to eliminate the possibility of the biopsy itself elevating the levels. Previous studies have chosen a cutoff of 12 h; we preferred a longer period to have a safer time buffer.[5] Samples were also obtained before the institution of specific therapy, such as pulse steroids or intravenous antibiotics, as these could potentially dampen the dd-Cf-DNA output from the renal allograft. All samples were stored in specialized cell-free DNA (Cf-DNA) preserving tubes (PAXgene tubes) at 4°C. All recipients also provided a saliva/buccal swab sample. Donor blood samples that had been preserved at the time of transplant were utilized or in the absence of the same, 2 ml of donor blood was freshly collected in ethylenediaminetetraacetic acid (EDTA) tube for identification of donor DNA signatures.

All graft biopsies were reported by a single experienced renal pathologist as per the revised Banff 2017 criteria for diagnosis of rejection. Sequential patients being evaluated for graft dysfunction in our unit were included, which resulted in a total of 20 patients.

As controls, 11 patients with stable graft function, who attended follow-up during the same period were chosen. Stable graft function was defined as those patients with creatinine less than 1.5 mg/dl, maintaining within 10% of baseline-best creatinine over the past 12 months, and with proteinuria less than 200 mg/d. Blood and buccal swab samples were obtained from this cohort as mentioned above. All such patients were followed up for at least 12 months post inclusion in the study to confirm stable graft function. If any developed graft dysfunction or proteinuria, they were biopsied and repeat Cf-DNA samples were obtained at the time of the new graft dysfunction.

dd-Cf-DNA estimation – Laboratory work

Ten milliliters of whole blood from each of the patients was collected in PAXgene Cf-DNA tubes (cat# 768115; Qiagen, Hilden, Germany), which were stored and transported to the laboratory at 4°C. Total Cf-DNA was extracted from 2 ml of plasma using circulating DNA isolation kit (cat#MN740300.50; Macherey-Nagel, Düren Germany), and 1 μl of the eluted fraction was quantified by Qubit fluorometer Thermo Fischer Scientific, Waltham, USA). Donor-specific DNA marker was identified by screening of peripheral blood obtained from donor and recipient from an Indian population-specific polymorphism pool (developed by Enable Biolabs India Pvt Ltd, Chennai, India). This was enhanced with simultaneous analysis of recipient-specific markers from buccal swab cells obtained from the recipient. Upon identification of the donor-specific DNA marker (80–120 single-nucleotide polymorphisms) for each of the patients, the markers were given a unique patient identification barcode, which was further used to quantify presence of the marker in the Cf-DNA fraction. The quantification was performed by comparison of the percent amplification of the donor-specific DNA markers in the Cf-DNA fraction with a standard graph and was expressed as nanograms per microliter [Figure 1].

Close

Figure 1

Identification of DNA markers – bench work

Export to PPT

Statistical analysis

Microsoft Excel and IBM Statistical Package for the Social Sciences (SPSS) software were used for analysis. Sensitivity, specificity, and positive and negative predictive values were analyzed. Receiver operating characteristic (ROC) curve analysis was performed to assess discriminatory ability. P value less than 0.05 was considered significant.

Patient characteristics, biopsies, and dd-Cf-DNA samples

Thirty-one transplant recipients were included in this analysis. None of them had been implanted or had undergone surgical manipulation of the graft in the preceding 2 weeks. Twenty patients in this group had been evaluated for graft dysfunction. All underwent graft biopsies as per unit protocol. Out of these, 12 patients had features of active rejection. Eight patients had nonrejection causes of graft dysfunction, including acute tubular injury (ATI) related to calcineurin toxicity, acute pyelonephritis, and one patient with interstitial fibrosis and tubular atrophy. An additional 11 transplant recipients with stable graft function as per the previous definition given in the “Materials and methods” section were included in the study. All such recipients were followed up for a minimum period of 12 months and continued to have stable graft function. Hence, as per our protocol, none of these 11 stable recipients underwent biopsy. This resulted in two groups for comparison – ”active rejection” in 12 patients and “no rejection” in 19 patients. The details of these two groups are presented in Table 1.

Histopathologic classification of graft pathology was done according to Banff Working Groups 2017. Among the 31 patients included, four had active antibody-mediated rejection (ABMR) with coexistent T-cell–mediated rejection (TCMR) and eight had any type of TCMR alone. The distribution of participants across the histopathologic subgroups is presented in Table 2. The diagnosis of calcineurin inhibitor toxicity was suggested based on the presence of single myocyte necrosis and isometric vacuolization in the proximal tubule.

Cf-DNA estimation and correlation with graft status

All participants had total Cf-DNA (i.e., Cf-DNA of both recipient and donor origins present in recipient’s plasma), and dd-Cf-DNA fractions estimated from their blood samples. We first attempted to analyze if total Cf-DNA levels correlated with clinical/histopathologic diagnosis. There was no correlation noted.

Further, dd-Cf-DNA estimation in each subject’s plasma was done. This was analyzed in the following ways to assess for correlation with the clinical/biopsy diagnosis:

1. as a simple yes or no test, where the detection of dd-Cf-DNA fraction was considered a positive test;

2. as a semi-quantitative test, where the absolute concentration of dd-Cf-DNA was considered; and

3. as a semi-quantitative test, where the ratio of dd-Cf-DNA fraction to the total Cf-DNA fraction was calculated as a percentage and considered.

dd-Cf-DNA was detected successfully at quantifiable levels in all four patients with ABMR. It was positive in all higher-grade TCMRs (1B). It, however, performed less accurately in TCMR 1A. It was negative (or below the detection threshold of our assay) in three out of five patients with TCMR 1A. The values for absolute concentration of dd-Cf-DNA are shown in Figure 2.

Close

Figure 2

Concentration of dd-Cf-DNA across subgroups. dd-Cf-DNA = donor-derived cell-free DNA

Export to PPT

dd-Cf-DNA was negative in five out of eight patients with nonrejection graft dysfunction (ATI, pyelonephritis). In three patients (one pyelonephritis, one ATI due to calcineurin toxicity, and one ATI with interstitial fibrosis & tubular atrophy - IFTA), dd-Cf-DNA was detected.

Out of 11 patients with clinically stable graft function, no dd-Cf-DNA was detected in eight patients. All three patients who had detectable levels were found to have stable graft function over a period of at least 1 year on subsequent follow-up. One was 5 years posttransplant, with one rejection episode (TCMR 1A) immediately posttransplant. She has had stable graft function since then, with a creatinine of 1.1 mg/dl. The other two patients were 1 year posttransplant, with a creatinine of 1 and 1.2 mg/dl, respectively, with no prior biopsies. There was no indication to biopsy these patients otherwise at the time of the study. A cursory view of the graphical depiction of absolute concentration of dd-Cf-DNA in Figure 2 shows the clustering of detectable levels among higher grades of rejection (TCMR 1B, ABMR).

dd-Cf-DNA was also calculated as a percentage of the donor-derived fraction to the total Cf-DNA content (donor plus recipient Cf-DNA) measured in plasma and is represented in Figure 3. It is evident that even though the absolute concentration of dd-Cf-DNA was similar among many of the subgroups, when it was expressed as a percentage of the total Cf-DNA, discriminatory ability improved. All patients with ABMR had dd-Cf-DNA detection at higher percentages (>0.25%). At a cutoff of 0.25%, we could predict ABMR (four out of four) with a few false-positive hits (three out of 27). All but one of the patients with stable graft function had either undetectable or very low signals (<0.06%). TCR 1B patients had detection in the range of 0.06%–0.2%. One patient each with ATI and graft pyelonephritis also had positive signals in this range. Three outliers with values greater than 1% included one TCR 1A, one ATI with significant IFTA, and one patient with stable graft function.

Close

Figure 3

% dd-Cf-DNA to total Cf-DNA across subgroups. Cf-DNA = cell-free DNA, dd-Cf-DNA = donor-derived cell-free DNA

Export to PPT

Diagnostic accuracy of dd-Cf-DNA as a marker of rejection

We analyzed the diagnostic performance of dd-Cf-DNA (yes/no test) as a marker of rejection. In our first analysis, we chose to keep “any rejection” as the end point, including active ABMR or any TCMR 1A and higher. We had a total of 12 cases of “any rejection” and 19 in the “no rejection” group. A 2 × 2 table was constructed and is presented in Table 3.

The above analysis revealed modest sensitivity (75%) and specificity (68%) for predicting any rejection. As it was apparent that the dd-Cf-DNA performs less robustly at predicting milder rejection, we analyzed the same for higher grade rejections of TCR 1B and ABMR as shown in Table 4. In diagnosing higher grade TCMR or ABMR, this test had 100% sensitivity with a lower specificity (67%). In short, this will serve as a good rule out test for higher grades of rejection. When dd-Cf-DNA detection was analyzed as a yes/no test for ABMR alone [Table 5], it had high sensitivity and negative predictive values (both 100%), with a trade-off being a lower specificity (70%).

When the percentage of dd-Cf-DNA to total Cf-DNA was utilized, the discriminatory ability of the test improved. At a cutoff of 0.25%, the ability to predict ABMR was superior [Table 6] with 100% sensitivity and good specificity (89%). ROC curve analysis provided a 0.889 area under the curve (P = 0.013) [Figure 4]. If a 0.05% cutoff was used to identify any rejection, then it offered a sensitivity of 67% against a specificity of 74% [Table 7]. In diagnosing TCMR 1B or ABMR, the sensitivity improved to 100% with 75% specificity [Table 8].

Close

Figure 4

% dd-Cf-DNA and ABMR prediction – ROC curves. ABMR = antibody-mediated rejection, dd-Cf-DNA = donor-derived cell-free DNA, ROC = receiver operating characteristic

Export to PPT

Limitations

Given the high cost and time taken for standardization of the test procedure, we have included only a small, convenient sample size for this analysis. As the platform for Cf-DNA estimation has been developed independently and the sensitivity of our test varies from commercially available platforms, the cut-offs expressed in this study are not comparable with previously published literature.[16]

Future directions

As a pilot study, this biomarker seems promising and actionable even in a resource-limited setting. Further analysis with more participants and subgroups would be useful and is planned.

Declaration of patient consent

The authors certify that they have obtained all appropriate patient consent forms. In the form, the patients have given their consent for their images and other clinical information to be reported in the journal. The patients understand that their names and initials will not be published and due efforts will be made to conceal their identity, but anonymity cannot be guaranteed.

Financial support and sponsorship

Nil.

Conflicts of interest

None.