Serial Peripheral Blood Perforin and Granzyme B Gene Expression Measurements for Prediction of Acute Rejection in Kidney Graft Recipients

Authors


* Caner Süsal, caner.suesal@med.uni-heidelberg.de

Abstract

In the present study we investigated whether peripheral blood gene expression measurements may serve as an early and non-invasive tool to predict renal allograft rejection. Peripheral blood was collected twice weekly after transplantation and gene expression was measured using real-time polymerase chain reaction (PCR). Recipients with acute rejection (n = 17) had higher levels of perforin and granzyme B transcript on days 5–7, 8–10, 11–13, 17–19, 20–22, and 26–29, as compared to patients without rejection (n = 50, p < 0.05 in all cases). Rejection diagnosis using gene expression criteria, determined with receiver operating characteristic (ROC) curves, was possible 2–30 days before traditional diagnosis (median 11 days). The best diagnostic result was obtained from samples taken on days 8–10, with a specificity of 90% and a sensitivity of 82% for perforin, and a specificity of 87% and sensitivity of 72% for granzyme B. Decreases in perforin (p < 0.01) and granzyme B expression (p < 0.05) were observed after initiation of anti-rejection therapy. Our data indicate that gene expression measurement is a useful tool for the recognition of graft rejection in its earliest stages. Serial measurements could be implemented as a monitoring system to highlight patients at higher risk of rejection, making them candidates for biopsy or pre-emptive anti-rejection therapy.

Introduction

Rejection remains the major cause of graft failure in kidney transplantation. Although precise mechanisms of chronic pathology remain unknown, studies show that early acute rejection events, not successfully treated with immunosuppressive therapy, are associated with increased risk of chronic rejection (1,2). In addition, surveillance biopsies displaying features of subclinical rejection have been associated with chronic pathology and eventual renal transplant loss (3,4). These findings suggest that a sensitive method for diagnosis of imminent rejection is important during early postoperative care.

No single completely satisfactory method of diagnosing acute rejection is currently available. Instead, clinical methods are used in concert for post-transplantation monitoring. Duplex Doppler ultrasonography detects rejection with widely varying sensitivity of 9–76% and is used as an adjunct to the measurement of serum creatinine (5). Increased creatinine levels signal kidney dysfunction and are used as an indication for biopsy. Unfortunately, rising serum creatinine levels are only evident after graft damage has occurred and do not always correlate with acute rejection, but instead may be related to cyclosporine nephrotoxicity, infections, or other complications (6). Additionally, biopsy and histological analysis of the tissue for infiltrating cells, does not guarantee conclusive results (7).

The analysis of molecular changes during the early post-transplantation period provides insights into the in vivo activity of the immune system in response to transplantation. We have recently shown that gene expression is strongly affected by anti-thymocyte globulin (ATG) induction therapy in the first weeks post transplantation (8). Past studies on pancreatic islet allograft rejection have shown increased cytokine gene expression levels, which reached a high steady-state before significant graft damage occurred (9). Li et al. used conventional competitive polymerase chain reaction (PCR) in 17 samples from 8 renal allograft recipients rejecting their graft and 83 samples from 29 non-rejecting patients, to show that perforin and granzyme B expression is higher in the urine of rejecting patients during the first 9 days post transplantation (10). Thus, molecular correlates may not only be a more sensitive tool for detecting acute rejection, they may also predict an impending rejection before extensive damage has been done to the graft. Pre-emptive anti-rejection therapy could potentially be used to preserve graft integrity and reduce the risk of chronic rejection.

In the present study, we analyzed the diagnostic usefulness of measuring perforin and granzyme B gene expression in the peripheral blood using real-time PCR. Characteristic cytotoxic lymphocyte (CTL) effector genes were chosen as targets, since cytotoxicity is considered to play an important role in the first stages of tissue destruction during allograft rejection (11,12). Peripheral blood was chosen as the sample source, as it provides the advantages associated with a non-invasive sampling procedure and has previously been shown to correspond to intragraft gene expression (13,14).

Materials and Methods

Patients

Blood samples were collected twice weekly from adult patients who received a kidney transplant at the Heidelberg center. This study was restricted to patients with an available sample for gene expression calibration, within the first 4 days post transplantation. Five patients could not be conclusively grouped as rejectors or non-rejectors and were excluded. Expression levels of the genes perforin and granzyme B were measured in 364 samples from 67 patients during the first month post transplantation (Table 1). Clinical rejection diagnosis was determined from biopsy, or as clinical response to anti-rejection therapy in one patient who did not undergo biopsy.

Table 1. : Demographic characteristics of patients
 RejectorsNon-rejectors
Parameter(n = 17)(n = 50)
  • a

    Mean ± SE.

  • b

    p < 0.001 (Mann–Whitney U-test).

  • c

    MMF = mycophenolate mofetil.

Recipient gender
 Female53%40%
 Male47%60%
Donor agea50 ± 345 ± 2
Recipient agea49 ± 344 ± 2
Panel reactivitya3 ± 3%3 ± 2%
Cold ischemia time (h) a,b20 ± 113 ± 1
Living transplants014
Number of transplants
 182%84%
 >118%16%
HLA-A,-B,-DR mismatchesa2 ± 0.42 ± 0.2
Immunosuppression
 Cyclosporine + MMF + steroids88%84%
 Other combination therapy12%16%
 Induction therapy (ATG/OKT3)0%0%

Immunosuppression and anti-rejection therapy

No induction therapy with ATG or OKT3 was administered. Anti-rejection therapy consisted of steroid-pulse therapy (3 × 250 mg) in nine patients. One patient received a reduced dose of steroids (2 × 150 mg) in combination with ATG (2 × 100 mg; Fresenius Bad Homburg, Germany).

Sample collection and processing

Peripheral EDTA-blood samples were collected and stabilized using RNA Stabilization Reagent for Blood and Bone Marrow (Roche Molecular Biochemicals, Mannheim, Germany). RNA was isolated with the Roche High Pure RNA kit, including DNase treatment, according to the manufacturer's instructions with two exceptions: stabilized samples were directly added to the filter in place of the initial lysis step with RLT Buffer, and an additional wash with inhibition removal buffer (Roche Molecular Biochemicals) was performed. RNA was reverse-transcribed to cDNA, using random hexamers and the TaqMan Reverse Transcription Kit (Applied Biosystems, Foster City, CA, USA), and stored at − 20 °C.

Real-time quantitative PCR and primers

Oligonucleotides were designed using Primer Express Software (Applied Biosystems) for perforin (forward primer: TGG AGT GCC GCT TCT ACA GTT, reverse primer: GCC CTC TTG AAG TCA GGG TG, FAM probe: CCA TGT GGT ACA CAC TCC CCC GC), granzyme B (forward primer: AAT CCT AAG AAC TTC TCC AAT GAC ATC, reverse primer: GCA CAG CTC TGG TCC GCT, FAM probe: GGC CTT TCT CTC CAG CTG CAG TAG CA), and 18S rRNA (forward primer: ACA TCC AAG GAA GGC AGC AG, reverse primer: TTC GTC ACT ACC TCC CCG G, VIC probe: CGC GCA AAT TAC CCA CTC CCG A). Probes were ordered with the quencher TAMRA (Applied Biosystems). Primers were obtained from Genset Oligos (Paris, France).

The 5′ nuclease assay was performed using the ABI PRISM 7700 Sequence Detection System and TaqMan Universal Master Mix (Applied Biosystems), containing heat-activated polymerase, AmpErase uracil-N-glycosylase (UNG), and the passive reference dye ROX. Final concentrations of primers and probes were 300 and 200 nm, respectively. Cycling parameters were: 50 °C for 2 min, 95 °C for 10 min, and 45 cycles of 95 °C for 15 s and 60 °C for 60 s.

Gene expression was measured as increased fluorescence, corresponding to amplification and degradation of the probe. The cycle in which fluorescence exceeds the set detection threshold is termed the ‘threshold cycle’ (CT). Larger quantities of initial specific template result in threshold-exceeding fluorescence during earlier cycles (lower CT). CT values for the endogenous control (18S rRNA), perforin, and granzyme B genes were obtained for each sample. 18S rRNA was chosen as a control, since inter-sample variability is typically lower for rRNA than that observed with other housekeeping genes (14,15). In order to correct for differences between samples, 18S rRNA CT values were used to normalize target gene expression for each patient, referred to as the perforin or granzyme B ‘Delta CT’ (ΔCT) value:

image

‘Target’ samples were compared relative to a patient-specific postoperative or pre-anti-rejection therapy ‘calibrator’, considered as 100% (16). In short, normalized ΔCT values for calibrator and target samples are substituted into the following equation:

image

and relative expression is calculated according to the formula:

image

For prediction of rejection, the first post-transplant sample received for each patient (median day 3) was chosen for calibration. We elected to use a postoperative sample, since it reflects individual patient differences in both immune system activity and response to immunosuppression. For analysis of the effect of anti-rejection therapy, the second-last sample taken prior to therapy served as the calibrator (median day 7). Receiver operating characteristic (ROC) curves and nonparametric Mann–Whitney and Wilcoxon tests were used for statistical analysis.

Results

Gene expression in rejector and non-rejector groups

Samples from patients with acute rejection (n = 17) were compared to those from patients without rejection (n = 50) (Table 1). Only samples taken before anti-rejection therapy were used in this comparison. Recipients with acute rejection had significantly higher levels of perforin transcript on postoperative days 5–7, 8–10, 11–13, 17–19, 20–22, and 26–29 as compared to patients without rejection (Figure 1A). The difference in expression was also significant for granzyme B on days 5–7, 8–10, 11–13, 17–19, 20–22, and 26–29 (Figure 1B). Since 14 non-rejector group patients were recipients of living donor organs (Table 1) and all rejector group patients were recipients of cadaver organs, an additional analysis limited to cadaver organ recipients was made. Similar results were obtained (perforin; days 5–7: p = 0.012, 8–10: p = 0.006, 11–13: p = 0.008, 17–19: p = 0.017, 20–22: p = 0.014, 26–29: p = 0.027, granzyme B; days 5–7: p = 0.033, 8–10: p = 0.019, 11–13: p = 0.008, 17–19: p = 0.017, 20–22: p = 0.017, 26–29: p = 0.037) (data not shown).

Figure 1.

Relative expression of perforin (A) and granzyme B (B) gene expression in rejector and non-rejector clinical groups. Gene expression was determined relative to the postoperative calibrator, taken on days 1–4, which was set to 100%. Values shown correspond to individual patient samples taken at each day interval. Sample numbers (indicated above each data cluster) lower than the patient group maximum are the result of patient release, exclusion of patients for whom anti-rejection therapy has been initiated, and missing samples. (A) Clinically diagnosed rejectors had significantly higher perforin expression on days 5–7 (p = 0.004), 8–10 (p = 0.006), 11–13 (p = 0.008), 17–19 (p = 0.014), 20–22 (p = 0.018), and 26–29 (p = 0.027), as compared to patients without rejection. (B) The difference in expression between the two groups was also significant for granzyme B as observed on days 5–7 (p = 0.01), 8–10 (p = 0.01), 11–13 (p = 0.01), 17–19 (p = 0.01), 20–22 (p = 0.027), 26–29 (p = 0.036).

ROC curves were used to analyze the diagnostic potential of the assay. Although higher levels of gene expression were evident in samples from recipients with rejection as early as 5–7 days post transplantation, assay sensitivity and specificity was highest during the first 2 weeks on days 8–10. Perforin expression in day 8–10 samples exceeding 114% of the calibrator value was indicative of rejection (sensitivity 82%, specificity 90%), whereas samples with granzyme B expression above 92% of the calibrator value were predictive of rejection (sensitivity 72%, specificity 87%) (Table 2). Using ROC-curve determined thresholds, 82% (14/17) of clinically rejecting patients could be categorized as rejectors based on increased expression in one or both of the genes during the first month post transplantation. In the case of the earliest clinical diagnosis on day 6, rejection was indicated using gene expression measurements in the first post-calibrator sample collected on the same day. The other 13 patients could be categorized as rejectors a median of 11 days before clinical diagnosis (range 2–30), regardless of whether increased perforin or granzyme B expression was used as the diagnostic criterion. Whereas 10 of the 17 patients with rejection were not clinically diagnosed until after 2 weeks post transplantation, 7 of these 10 patients had levels of perforin and granzyme B expression above the ROC-defined thresholds on days 5–7. The serial measurements show that up-regulation of gene expression was not restricted to the immediate period preceding clinical rejection diagnosis, but instead was consistently high, beginning with the day 5–7 sample and increasing in those patients not receiving anti-rejection therapy (Figure 1A,B).

Table 2. : Diagnostic value of perforin and granzyme B gene expression measurements
GenePost-op
days
AUCp-valueSamplesThresholdaSPSENPVPPV
  • a

    Values given are gene expression percentages relative to the postoperative (postop) calibrator gene expression set to 100%.

  • AUC: Area under the curve, SP: specificity, SE: sensitivity, NPV: negative predictive value, PPV: positive predictive value.

  • Values for threshold, SE, SP, NPV, and PPV are only shown for those tests with p < 0.05.

Perforin5–70.7450.004n = 56111%65%75%87%46%
 8–100.7850.006n = 41114%90%82%93%75%
 11–130.7650.008n = 50139%85%73%92%57%
 14–160.6170.279n = 47
 17–190.7850.014n = 40105%78%75%93%46%
 20–220.8190.018n = 29336%100%67%92%100%
 23–250.7540.104n = 18
 26–290.8550.027n = 16235%100%80%92%100%
Granzyme B5–70.7220.010n = 56133%68%69%84%46%
 8–100.7530.014n = 4192%87%73%90%67%
 11–130.7580.010n = 50101%74%82%94%47%
 14–160.6460.176n = 47
 17–190.7930.011n = 4081%66%88%95%39%
 20–220.7970.027n = 29237%100%67%92%100%
 23–250.7850.068n = 18
 26–290.8360.036n = 16183%100%80%92%100%

For combined analysis of the diagnostic value of perforin and granzyme B, ROC-curve defined thresholds were applied to score samples from days 5–7, 8–10, 11–13, and 17–19 as positive or negative in each gene. Eighty-eight per cent (215/244) of these samples were assigned scores of high or low expression for both genes (‘double-positive’ or ‘double-negative’), 5% (11/244) of the samples had high perforin and low granzyme B expression, and 7% (18/244) of the samples had high granzyme B and low perforin expression.

Assay specificity and sensitivity was recalculated, for intervals from Table 1 with significant differences between the two analyzed groups, using double-positivity or double-negativity as diagnostic criteria. When double-positivity was used as the diagnostic criterion for rejection, specificity improved from 65–68% to 73% on days 5–7, from 87–90% to 97% on days 8–10, and from 74–85% to 90% on days 11–13 (Table 3). When double-negativity was used to determine non-rejecting patients, specificity increased from 65–68% to 81% on days 5–7, and from 66–78% to 88% on days 17–19 (Table 3). For days 8–10 and 11–13, specificity was higher in comparison to using measurements of granzyme B expression but lower in comparison to using perforin as the diagnostic criterion (Tables 2 and 3).

Table 3. : Diagnostic value of simultaneous perforin and granzyme B gene expression
Perforina Granzyme BaPost-op daysbp-valuecSamplesSPSE
  • a

    Positive or negative scoring was determined from relative gene expression either above (+) or below (–) the gene and time-interval specific threshold values listed in Table 2.

  • b

    Days 20–22 and 26–29 were excluded since all samples were either double-positive or double-negative, thus sensitivity and specificity would remain the same as indicated in Table 1.

  • c

    Determined using Fisher's exact test.

  • SP: specificity, SE: sensitivity.

++5–7  0.030n = 5673%63%
  8–10< 0.001n = 4197%73%
  11–13< 0.001n = 5090%73%
  17–19  0.005n = 4081%75%
5–7  0.007n = 5681%60%
  8–10  0.001n = 4182%80%
  11–13  0.004n = 5082%72%
  17–19  0.017n = 4088%63%

In contrast to granzyme B and perforin determinations, serial fas-ligand gene expression, measured early in the study, demonstrated a highly fluctuating ‘zig-zag’ character, from which no stable curves from individual patients in either group could be obtained (data not shown). As a result, measurements in this gene were discontinued after testing in 46 patients.

Gene expression in patients receiving anti-rejection therapy

The effect of anti-rejection therapy on gene expression was investigated in 10 recipients in whom two samples both preceding and subsequent to therapy initiation were available within the first month. Eight of these patients showed increasing perforin expression in the two samples preceding anti-rejection therapy, which then decreased after therapy initiation in all patients. Gene expression remained low in the next sample obtained a median of 7 days after therapy initiation, and a further decrease in expression was observed in five patients (Figure 2A). Granzyme B expression increased in 7 of the 10 patients in the samples preceding therapy. A significant decrease in the expression of this gene was observed in the sample subsequent to therapy initiation (9/10 patients), and expression remained low throughout the patient group in the next sample, as compared to pretreatment levels (Figure 2B).

Figure 2.

Effect of anti-rejection therapy on perforin (A) and granzyme B (B) gene expression. Gene expression was determined relative to the ‘−2’ sample (calibrator), which was the second-last consecutive sample taken before therapy initiation, and set to 100%. Individual patient values are shown for 10 rejectors with two samples preceding and two samples subsequent to anti-rejection therapy initiation during the first month post transplantation. The ‘−1’ sample corresponds to the last sample preceding anti-rejection therapy in each patient. The calibrator ‘−2’ sample was collected a median of 4 ± 0.6 days prior to this time point. ‘ + 1’ and ‘ + 2’ samples were collected a median of 3 ± 0.6 and 7 ± 0.5 days after the ‘− 1’ sample, respectively. Group medians for gene expression are shown as horizontal bars. (A) Perforin expression tended to increase slightly during the course of rejection and prior to initiation of anti-rejection therapy (median increase 32%, p = 0.017), followed by decreases in the next two consecutive samples relative to the ‘−2’ sample (median decreases 47% and 67%, respectively, p = 0.005 for both). (B) Granzyme B expression increased during the course of rejection in 7 of 10 patients in the samples prior to therapy initiation (median increase 36%), but this change was not statistically significant. Significant decreases after therapy initiation were apparent in both consecutive samples (median decreases 54% and 64%, p = 0.017 and p = 0.005, respectively).

Discussion

In the present study, we analyzed the diagnostic usefulness of perforin and granzyme B gene expression measurements in the peripheral blood for prediction of acute rejection. In contrast to previous studies measuring gene expression in samples collected at single time-points, we present the first full analysis of serial perforin and granzyme B gene expression in the peripheral blood during the first month post transplantation. Patients were identified as rejectors a median of 11 days before clinical diagnosis, based on increased gene expression on days 5–7 with assay specificity of 65% and 68% and sensitivity of 75% and 69% for perforin and granzyme B, respectively. But specificity values were much improved for both genes on days 8–10, corresponding to fewer false positives (3–4 vs. 13–14 for granzyme B and perforin, respectively).

Perforin and granzyme B are important cytotoxic genes effecting target cell lysis by both CTLs and natural killer (NK) cells. CTLs were shown to play an important role in tissue destruction during rejection (11,12), whereas the role of NK cells in solid organ transplantation has not been thoroughly investigated. Since, in our samples, NK cells only contributed to 6 ± 6.9% of the total lymphocyte count (data not shown), CTLs were probably the major source of perforin and granzyme B expression. The relevance of perforin and granzyme B as molecular correlates of rejection is further supported by recent studies which have demonstrated significantly higher levels of granzyme B gene expression in the peripheral blood at the time of biopsy (14) and of both perforin and granzyme B in the urine of rejecting patients, during the first 9 days and after the first month post transplantation using competitive PCR (10).

There are various factors which can potentially lower the specificity of an assay based on lymphocyte gene expression. Low cyclosporine bioavailability is associated with increased gene expression in lymphocytes post transplantation, but increased expression of perforin and granzyme B was observed in patients with rejection despite adequate cyclosporine trough levels. Clinical infection poses a further problem in the later post-transplantation period, but prediction of rejection during the first 10 days may, in the majority of cases, precede infection typically associated with long-term immunosuppression, such as CMV. None of our patients had clinical CMV infection during the study period. Furthermore, histological signs of rejection, observed in surveillance biopsies in patients without clinical signs of rejection (17), combined with the fact that we did not have surveillance biopsy results for non-rejecting patients in our study, suggest that some samples in our study considered as false-positive may originate from patients with ‘subclinical’ rejection. Rush et al. conducted a randomized prospective study to demonstrate that improved graft function and survival could be achieved in patients with surveillance biopsies and treatment of subclinical rejection (18). Gene transcription in surveillance biopsy infiltrate was phenotypically similar to that of infiltrate from clinical rejection biopsies with both perforin and granzyme B expression up-regulated, but at a slightly lower level in subclinical as compared to clinical biopsy infiltrate (19).

A maximum assay sensitivity of 88% suggests that CTLs, the major producers of perforin and granzyme B in the peripheral blood, are important players in the immune reaction against the allograft but are not solely responsible for rejection. This is not surprising since it has previously been demonstrated that rejection can occur in the absence of CTLs. Cosimi et al. found that elevated ratios of helper T cells (CD4+) to CTLs (CD8+) were indicative of rejection (20) and Morton et al. demonstrated that CD4+ cells are both necessary and sufficient to cause rejection in heart and skin allografts with MHC-class I mismatches (21). Humoral alloreactivity, which has been related to inferior graft outcome (22), offers an alternative or supporting mechanism of rejection initiation that cannot be detected with this assay.

Combined up-regulation of gene expression as a diagnostic criterion for rejection generally increased specificity at the expense of sensitivity in comparison to measurements for a single gene alone. Although ‘double-positive’ assay specificity was quite high on days 8–10 (97%), no cumulative increase in both specificity and sensitivity in comparison to the measurement of perforin expression alone is achieved. The additional expense associated with measuring granzyme B expression may thus not be warranted, as perforin is clearly the stronger predictor of rejection in the early post-transplantation period.

Negative predictive values for our assay were quite high (84–94%) during the first 2 weeks, but positive predictive values were lower, ranging from 46 to 75% (Table 2). These results are strongly related to patient group size, since fewer patients in the rejector group decreases the maximum number of true-positive results. Positive predictive value and consequently clinical value could be maximized through application of these assays for monitoring in recipient groups with a higher incidence of rejection, such as pediatric recipients and adult recipients of cadaver organs.

Elevated levels of cytotoxicity must be considered disadvantageous in the critically important early post-transplantation period. The question remains as to whether this warrants the use of increased immunosuppression or pre-emptive anti-rejection therapy. The main advantage of our assay, as compared with conventional post-transplantation monitoring techniques, is the potential for earlier intervention by the physician, prior to substantial organ damage. Since acute rejection is a major risk factor for later chronic rejection, earlier intervention may result in prolonged graft function and reduction in other costs such as dialysis and retransplantation. Considering the lack of uniformity in transplant center procedures, the actual implementation of this system in the clinic depends on independent center judgments on the value associated with earlier intervention. This decision is likely to be made more easily in the future through expansion of research into this area such as further testing in randomized studies.

Acknowledgments

This study was supported in part by a grant of Forschungsschwerpunkt Transplantation Heidelberg.

Ancillary