Associations Between EBV Serostatus and Organ Transplant Type in PTLD Risk: An Analysis of the SRTR National Registry Data in the United States

Authors


Vikas R. Dharnidharka, vikasmd@ufl.edu

Abstract

In a prior multiorgan transplant database study, recipient Epstein–Barr virus (EBV) seronegativity was not associated with increased risk for posttransplant lymphoproliferative disorders (PTLD) in liver transplants (LTX), at variance with prior single center reports and with data from kidney and heart transplants (KTX and HTX). The Scientific Registry of Transplant Recipients (SRTR) in the United States is the only other registry with data on the required variables for comparison.Our study set comprised 112 756 KTX (580 PTLDs; 0.51%), 13 937 HTX (140 PTLDs; 1.0%) and 40 437 LTX (383 PTLDs; 0.95%) performed January 2003 onward. The unadjusted hazard ratio (HR) for PTLD if recipient EBV seronegative was 5.005 for KTX, 6.528 for HTX and 2.615 for LTX (p < 0.001 for all). In models adjusted for multiple covariates, the adjusted HR was 3.583 (p < 0.001) for KTX, 4.037 (p < 0.001) for HTX, 1.479 (p = 0.03) for LTX. Interaction models using EBV seropositive KTX as reference group showed significantly higher risk for all other EBV seronegative organ transplant groups and also for EBV seropositive LTX (AHR 2.053, p < 0.0001).Recipient EBV seronegativity is still significantly associated with risk for PTLD in LTX, though less so because of higher baseline risk in the EBV seropositive LTX group.

Abbreviations: 
AHR

adjusted hazard ratio

CTS

Collaborative Transplant Study

DD

deceased donor

EBV

Epstein–Barr virus

HTX

heart transplant

K–M

Kaplan–Meier

KTX

kidney transplant

LD

living donor

LTX

liver transplant

mTORi

mammalian target of rapamycin inhibitor

OPTN

organ procurement and transplant network

SRTR

scientific registry of transplant recipients

Introduction

Few complications of solid organ transplantation are as feared as posttransplant lymphoproliferative disorders (PTLDs). This potentially fatal complication is characterized by myriad presentations, can be difficult to diagnose and treatment is not always successful. Solid organ PTLDs are believed to be caused in many cases from Epstein–Barr virus (EBV) infection and transformation of recipient or donor B cells, with impaired host immune response from exogenous immunosuppression. Though not all PTLDs are associated with EBV, this virus clearly plays a central role in pathogenesis where the tumor is found to be EBV positive. Several studies of kidney transplantation, at single center level or from multicenter registries, have demonstrated 3- to 33-fold increased risk for PTLD if the renal transplant recipient is EBV seronegative at time of transplant (1–6). In such a setting, primary EBV infection often occurs through the allograft itself transporting latent EBV in renal tubular epithelium but with the naive host unable to mount a primary immune response and with no EBV-directed memory cells as a backup. Similarly, single center reports have shown high PTLD risk in cardiac or liver transplantation if recipient EBV seronegative (7–9). However, a recent multicenter registry report demonstrated no such significantly increased risk in liver transplantation, in contrast to kidney or heart transplantation and contrary to expectation (10). We, therefore, undertook this analysis of the SRTR registry database in the United States, the only other large transplant database with data on kidney transplants (KTXs), liver transplants (LTXs) and heart transplants (HTXs) and with EBV serostatus and PTLD data.

Methods

We analyzed data from the Scientific Registry of Transplant Recipients (SRTR) in the United States, which uses base Organ Procurement and Transplant Network data mandated on all solid organ transplants performed within the United States. Though the field existed before 2002 also, recipient EBV serostatus was made a required data field in 2002, so we analyzed transplants performed from January 1, 2003 onward till February 1, 2010, followed up until August 2, 2010, which also represents the recent era of immunosuppression. In the SRTR database, EBV status for donor and recipient at time of transplant can be reported by the centers as positive, negative, not done, unknown or missing. We determined time to PTLD by Kaplan–Meier (K–M) methods for each organ transplant (kidney, heart and liver), stratified by recipient EBV serostatus at time of transplant. We extracted data on recipient characteristics (age, sex and race), transplant characteristics (living or deceased donor), posttransplant immunosuppression at initial discharge (induction agent use, oral maintenance agent use) and acute rejection episode as reported by center at initial discharge. We then fitted Cox proportional hazard models and determined adjusted hazard ratios (HR) for PTLD in each organ system after confirmation of proportional hazards assumptions. Although unique primary diseases may also impart higher PTLD risk in each organ system, we did not include primary disease in our models because we wanted to compare identical variables for each organ system. In another model, an interaction variable was then created by interacting organ transplant by EBV status and a new Cox model was fitted that included all the covariates described above. All analyses were considered significant at p ≤ 0.05 and were carried out using SAS 9.2 (SAS Institute, Cary, NC, USA).

Results

The study set comprised 112 756 KTXs (580 PTLDs; 0.51%), 13 937 HTX (140 PTLDs; 1.0%) and 40 437 LTX(383 PTLDs; 0.95%). Table 1 lists the detailed demographic parameters of the study group. On the basis of recipient EBV serostatus alone, the unadjusted HR for PTLD, if recipient EBV seronegative was 5.005 (p < 0.001) for kidney, 6.528 (p < 0.001) for heart and 2.615 (p < 0.001) for liver. In Cox proportional hazard models adjusted for multiple covariates, the adjusted HR dropped slightly to 3.583 (p < 0.001) for kidney (Table 2), 4.037 (p < 0.001) for heart (Table 3), 1.479 (p = 0.03) for liver (Table 4), remaining significant for all organs.

Table 1.  Demographic data of the study group
 KidneyHeartLiver
  1. 1SRTR does not report an AR at initial discharge field for heart transplants.

Transplant (n)112 75613 93740 437
PTLD n (%)580 (0.51%)140 (1.0%)383 (0.95%)
Recipient race Caucasian n (%)62 330 (69.51%)9553 (78.81%)28 745 (87.77%)
Living donor n (%)44 085 (39.10%)1 (0.01%)1842 (4.56%)
Recipient gender male n (%)68 307 (60.58%)10 194 (73.14%)26 609 (65.8%)
Recipient EBV serostatus
 Positive n (%)65 281 (57.90%)8698 (62.41%)21 845 (54.03%)
 Negative n (%)12 701 (11.26%)2051 (14.72%)5447 (13.47%)
 Missing n (%)1550 (1.37%)113 (0.81%)900 (2.23%)
 Not done n (%)29 158 (25.96%)2487 (17.84%)10 024 (24.79%)
 Unknown n (%)4066 (3.61%)588 (4.22%)2218 (5.49%)
Donor/recipient EBV seromatch
 Negative/negative n (%)1256 (1.1%)153 (1.1%)325 (0.8%)
 Positive/negative n (%)7003 (6.21%)1056 (7.58%)2545 (6.29%)
 Negative/positive n (%)3179 (2.82%)401 (2.88%)766 (1.89%)
 Positive/positive n (%)39 067 (34.65%)4065 (29.17%)9861 (24.39%)
 Unknown n (%)62 251 (55.21%)8262 (59.8%)26 940 (66.62%)
 Recipient age <18 years n (%)5704 (5.06%)2084 (14.95%)3631 (8.98%)
At initial discharge
 Any induction agent used n (%)81 216 (73.03%)7304 (52.4%)9409 (23.26%)
 Tacrolimus used n (%)86 946 (79.7%)7809 (58.9%)34 420 (87.5%)
 Mycophenolate used n (%)93 884 (86.06%)11 517 (86.90%)25 749 (65.52%)
 Steroids used n (%)99 195 (90.93%)11 999 (90.62%)35 286 (89.7%)
 mTORi used n (%)5216 (4.78%)337 (2.54%)467 (1.18%)
 Acute rejection n (%)863 (0.77%)N/A1230 (0.57%)
Table 2.  Cox proportional hazard model results for risk factors for PTLD development in kidney transplant recipients
Variables in modelReference groupAHR95% LCL95% UCLp-Value
Recipient race whiteBlack2.0841.6112.696<0.0001
Recipient age <18Age ≥ 181.7031.2712.280.0004
Recipient EBV serostatus
 NegativePositive3.5832.5585.017<0.0001
 MissingPositive1.3180.72.4810.393
Not donePositive1.4251.0651.9070.017
 UnknownPositive3.4551.9426.147<0.0001
 Donor type—deceasedLiving1.8331.4752.277<0.0001
Induction group
 AlemtuzumabNone1.6741.1362.4670.0092
 T-cell depletingNone1.5251.1811.970.0012
 IL-2 R AbNone1.4121.0761.8510.0127
 Azathioprine useMycophenolate0.7970.3281.9350.6157
 No antimetabolite usedMycophenolate0.7640.5421.0770.1241
 Cyclosporine usedTacrolimus0.7590.580.9930.044
 No CNI usedTacrolimus1.0570.7291.5330.7705
 Steroids not usedYes1.0730.7471.5420.7031
 Sirolimus/everolimus not usedYes0.8090.5651.1580.2462
Donor/recipient EBV
 Positive/negativeNegative/negative2.7471.3275.6870.0065
 Negative/positiveNegative/negative2.2020.8355.8090.1108
 Positive/positiveNegative/negative1.470.6613.2720.3449
 UnknownNegative/negative0.9860.4642.0970.9711
 Acute rejection at initial dischargeNo4.8822.310.359<0.0001
Table 3.  Cox proportional hazard model results for risk factors for PTLD development in heart transplant recipients
Variables in modelReference groupAHR95% LCL95% UCLp-Value
Recipient race whiteBlack1.2680.7632.1080.3597
Recipient age < 18Age ≥ 183.8642.4766.03<0.0001
Recipient EBV serostatus
 NegativePositive4.0372.3716.874<.0001
 MissingPositive1.3120.3914.4040.6606
 Not donePositive1.6960.9612.9950.0684
 UnknownPositive000.9926
Induction group
 AlemtuzumabNone000.9758
 T-cell depletingNone0.8270.4951.3820.469
 IL-2 R AbNone1.1080.6651.8460.6945
 Azathioprine useMycophenolate0.4230.1631.0940.0758
 No antimetabolite usedMycophenolate0.6560.3121.380.2661
 Cyclosporine usedTacrolimus0.4850.3110.7580.0015
 No CNI usedTacrolimus0.9260.3892.2030.8613
 Steroids not usedYes1.7810.7094.4750.2198
 Sirolimus/everolimus not usedYes0.5390.2621.1120.0944
Donor/recipient EBV
 Positive/negativeNegative/negative1.1960.2635.4370.8163
 Negative/positiveNegative/negative0.9560.08211.150.9713
 Positive/positiveNegative/negative1.0420.1995.4510.9616
 UnknownNegative/negative0.5910.1342.6020.4868
Table 4.  Results of Cox proportional hazards models for PTLD development in liver transplants
Variables in modelReference groupAHR95% CL p-Value
Recipient race whiteBlack1.2840.8871.8570.185
Recipient age < 18Age ≥ 183.2652.4244.398<0.0001
Recipient EBV serostatus
 NegativePositive1.4791.0232.1370.0375
 MissingPositive0.6140.2821.3330.2173
 Not donePositive1.350.9561.9050.0882
 UnknownPositive1.9650.9424.0990.0716
 Donor type—deceasedLiving0.510.3420.7610.001
Induction group
 AlemtuzumabNone1.3420.5453.3040.5222
 T-cell depletingNone1.1980.7961.8020.3868
 IL-2 R AbNone1.0820.771.520.6511
 Azathioprine useMycophenolate1.5290.663.540.3214
 No antimetabolite usedMycophenolate1.2941.0121.6540.0397
 Cyclosporine usedTacrolimus0.7260.411.2870.2731
 No CNI usedTacrolimus0.3130.0990.9890.0478
 Steroids not usedYes0.6370.3821.0620.0837
 Sirolimus/Everolimus not usedYes1.4570.5933.580.4119
Donor/Recipient EBV
 Positive/negativeNegative/negative1.5430.4675.1010.4768
 Negative/positiveNegative/negative0.8350.1624.2960.8289
 Positive/positiveNegative/negative1.4650.4334.9570.5393
 UnknownNegative/negative0.3060.09410.0499
 Acute rejection at initial dischargeNo4.7761.18119.3150.0283

Other significant covariates that imparted higher risk, additional to EBV status, in the kidney model included recipient Caucasian race or pediatric age, deceased donor source kidney, use of T-cell depleting agents or tacrolimus (over cyclosporine). Other significant covariates imparting higher risk in the heart model included recipient age <18 at transplant and the use of tacrolimus (over cyclosporine). Other significant covariates imparting higher risk in the liver model included living donor source liver, pediatric recipient age group and whether an acute rejection episode was observed upon discharge. Because LTX recipients may have misleading recipient EBV positive results from maternal antibody at very young ages, we fitted separate Cox proportional hazard models excluding 1057 recipients under 1 year age. The risk for PTLD if recipient was EBV seronegative (vs. seropositive) increased very slightly, (adjusted hazard ratio [AHR] value 1.681, 95% CI 1.142, 2.473; p = 0.0084), remaining well below the AHR values for EBV seronegative recipient PTLD risk in heart and kidney recipients.

The time to PTLD K–M estimates by recipient EBV serostatus are shown in Figure 1(A) (kidney), Figure 1(B) (heart) and Figure 1(C) (liver). In each case, out to 1500 days posttransplant, recipient EBV seronegative recipients had the highest PTLD risk, EBV seropositive recipients had the lowest risk and the categories missing, not done and unknown expectedly fell somewhere in between. The interaction model with EBV seropositive KTX as reference group in the presence of previously described covariates showed significantly higher risk for each organ type if EBV seronegative, but also for LTX even with EBV seropositive (Table 5 and Figure 2).

Figure 1.

Panel A: Incidence of PTLD-free survival (y-axis) over days posttransplant (x-axis) in kidney transplants, by recipient EBV serostatus. Panel B: Incidence of PTLD-free survival (y-axis) over days posttransplant (x-axis) in heart transplants, by recipient EBV serostatus. Panel C: Incidence of PTLD-free survival (y-axis) over days posttransplant (x-axis) in liver transplants, by recipient EBV serostatus.

Table 5.  Results of Cox proportional hazards models for PTLD-free survival (higher hazard ratio = higher risk for PTLD), with interaction terms for organ type by recipient EBV serostatus
PTLD risk by organ type and recipient EBV status (reference group EBV seropositive recipient, kidney transplant)ReferenceAHR95% LCL95% UCLp-Value
Recipient EBV seronegative, kidney transplantRecipient positive /kidney3.7992.8235.115<0.0001
Recipient EBV seronegative, heart transplantRecipient positive/kidney5.6163.7728.361<0.0001
Recipient EBV seropositive, heart transplantRecipient positive/kidney1.2640.8681.8420.2222
Recipient EBV seronegative, liver transplantRecipient positive/kidney3.4972.4624.966<0.0001
Recipient EBV seropositive, liver transplantRecipient positive/kidney2.0531.5622.699<0.0001
Recipient race—CaucasianBlack1.4151.1361.7620.0019
Donor type—deceasedLiving2.0621.6272.613<0.0001
Recipient age < 18Age ≥ 182.5082.0553.061<0.0001
Induction group
 AlemtuzumabNone1.3170.9051.9170.1501
 T-cell depletingNone1.2330.9951.5280.0558
 IL-2 R AbNone1.2080.9791.490.0779
 Azathioprine useMycophenolate0.6570.3471.2440.1974
 No antimetabolite usedMycophenolate1.1120.9011.3720.3235
 Cyclosporine usedTacrolimus0.7080.550.9110.0073
 No CNI usedTacrolimus0.9570.6611.3860.817
 Steroids not usedYes0.8850.6381.2270.4629
 Sirolimus/everolimus not usedYes0.930.681.2710.6478
Donor/recipient EBV
 Positive/negativeNegative/negative2.1081.1893.7380.0107
 Negative positiveNegative/negative1.8250.8363.9820.131
 Positive/positiveNegative/negative1.5980.8622.9630.1369
 UnknownNegative/negative0.7180.3991.2930.2694
Figure 2.

Adjusted hazard ratios for liver (LI), heart (HR) and kidney (KI) transplants, using EBV seropositive recipient kidney transplant as reference group.

Discussion

The pathogenesis of posttransplant lymphoproliferative disorder is still not fully known. In most cases, this condition seems to arise secondary to EBV-induced transformation of infected B cells, without the proliferation control otherwise exercised by now immunosuppression-impaired CD8 T cells. Such conditions allow the transformed cells to escape from the growth cycle controls. Though a minority of PTLD cases are non-B cell or non-EBV-induced, in most cases this pathogenesis paradigm applies. In situations where EBV plays a role, a primary infection of the immunosuppressed host, with no prior memory cells or existing antibody, imparts an even more significant risk for PTLD development than a reactivation of preexisting latent virus in the host. Thus, recipient EBV seronegativity has long been known as the risk factor of highest magnitude (11,12). Such a risk has generally been found across organ types, such that similar higher risk if recipient EBV seronegative is noticed in KTXs (2,13), HTX (7,9) or LTXs (14,15). Pediatric recipients are often EBV seronegative, so that impact of EBV seronegativity may be confounded if many pediatric recipients are included in the study group. Nevertheless, even in study groups restricted to adults, recipient EBV seronegativity adds a significant risk in KTXs (1–6) and HTX (7–9). Notably, in liver transplantation, most of the reports are either single center (15,16) or, if multicenter, emanate from pediatric groups (17). In the reports with largely pediatric LTX data, recipient EBV seronegativity was universally cited as a major risk factor (14,17–19). In contrast, in the series with adult LTX data, results on EBV serostatus were sparse or all patients were EBV seropositive at time of transplant (16,20). A review of PTLD after liver transplantation by Leblond and Choquet (21) found and cited only two exclusively adult recipient PTLD studies (22,23), none with recipient EBV serostatus data. Results of tumor EBV positivity were conflicting, ranging from 10–38% (24,25) to 79–87% (26,27). In these studies, tumor EBV status was unknown in a significant proportion. In a single center study of 40 EBV seronegative adults, the PTLD rate was 33% (28). Kremers et al. found that most PTLD tumors within the first 4 years post-LTX in adult recipients were EBV positive, whereas most PTLD tumors after 4 years posttransplant were EBV negative (15).

More recently, the Collaborative Transplant Study (CTS) found that recipient EBV seronegativity imparted a significantly higher risk of non-Hodgkin lymphoma (NHL), a PTLD subtype, in KTX or HTX recipients, but not in LTX recipients (10). The CTS is a large transplant registry with extensive data on NHL across different organ transplants. The lack of significance of recipient EBV seronegativity in LTXs was described by the authors as contrary to expectation and surprising. They also found and reported a higher PTLD rate in the EBV seropositive LTX recipients. We have duplicated this latter finding but found an additional higher risk in the EBV seronegative LTX recipient, as the majority of the literature suggests. The blunting of relative risk in EBV seronegative recipients in our study can be attributed in large part to the higher baseline risk in the EBV seropositive recipient. Several possible reasons may be possible for the partial difference in results between our study and the CTS study. The CTS records data on NHL whereas the SRTR records data on PTLD, though in the past, risk factor studies from these two databases have obtained largely parallel results. This study includes a much larger transplant sample than was present in the CTS study for all three organs. Notably, the CTS study only reported on approximately 23 000 transplants from 1995 to 2007, though the total database comprises more than 400 000 transplants since 1966, suggesting that EBV serostatus data was available in approximately only 20% of their subjects. Our study is restricted to a more recent time period (January 2003 onward) and percentage of known EBV serostatus data was 76.74% for heart, 68.73% for kidney and 67.07% for liver. Because we used a more recent time period, the CTS analysis has longer duration of follow-up, which may also account for some of the differences. Indeed time to PTLD was longer for EBV+ recipients versus EBV– recipient in our analysis.

In summary, our study shows that recipient EBV seropositivity remained an important risk factor in proportional hazards models adjusted for multiple covariates, whether the allograft received was a kidney, heart or liver. The magnitude of risk was highest for heart, followed by kidney and then by liver. EBV seropositive LTX recipients also have an elevated risk, the reason for which is not known, but prior studies speculate that the larger lymphoid mass of the liver may favor EBV reactivation and progression to PTLD (10).

Acknowledgment

This study was presented at the American Transplant Congress in May 2011.

Disclosure

The authors of this manuscript have conflicts of interest to disclose as described by the American Journal of Transplantation.

This work was partially funded through the endowment of the Eminent Scholar Chair in Nephrology and Hypertension funded by the Central Florida Kidney Center, Inc. The authors do not have any direct conflicts of interest with regards to the data presented in this manuscript.

VRD is a consultant to Bristol–Myers–Squibb and on the Advisory board of Chimerix for work unrelated to this study. HUMK was an ad hoc consultant for Bristol–Myers–Squibb, Novartis, Pfizer and Roche. As of August 8, 2011, HUMK is an employee of Astellas, Inc. As of April 4, 2011, KL is an employee of Chronic Disease Research Group (CDRG). VRD has received grant support from Genzyme for work unrelated to this study.

No part of this manuscript was prepared by a commercial organization.

Ancillary