Published ahead of print on August 28, 2008, doi:10.1164/rccm.200804-531OC
© 2008 American Thoracic Society doi: 10.1164/rccm.200804-531OC
Post-transplantation Lymphoproliferative DiseaseEpstein-Barr Virus DNA Levels, HLA-A3, and Survival1 The School of Medicine, 2 Department of Pathology and Laboratory Medicine, 3 Department of Microbiology and Immunology, and 4 Division of Pulmonary and Critical Care Medicine, University of North Carolina, Chapel Hill, North Carolina Correspondence and requests for reprints should be addressed to Robert Aris, M.D., CB# 7020, 4131 Bioinformatics Building, University of North Carolina at Chapel Hill, Chapel Hill, NC 27599. E-mail: aris{at}med.unc.edu
Rationale: Elevation in Epstein-Barr virus (EBV) circulating DNA has been proposed as a marker for development of post-transplant lymphoproliferative disease (PTLD), but few published data exist in the study of lung-transplant recipients. Objectives: To determine if elevated EBV DNA levels, in combination with other risk factors, were predictive of PTLD. Methods: We conducted a retrospective, single-center study examining all lung transplant recipients (n = 296) and EBV DNA levels (n = 612) using real-time TaqMan polymerase chain reaction. There were 13 cases of PTLD overall, of which 5 occurred in the era of EBV DNA monitoring. Measurements and Main Results: EBV DNA levels were distributed differently among seropositive and seronegative patients, with the latter having higher values (P < 0.0001). Among the cohort of pretransplantation seropositive patients, there was one diagnosed with PTLD. The EBV DNA level in this patient was elevated at the time of PTLD diagnosis (sensitivity = 100%, specificity = 100% for PTLD). Among the cohort of pretransplantation seronegative patients, there were four with a diagnosis of PTLD. In all four patients, the EBV DNA level was detectable (sensitivity = 100%, specificity = 24%), but in only two was it elevated (sensitivity = 50%, specificity = 22%). HLA-A3 expression in the recipient and/or donor conferred additional risk for PTLD among the seronegative patients (P = 0.026 to 0.003). No other PTLD risk factor was found. Conclusions: EBV DNA levels are a useful but imperfect predictor of PTLD in patients with lung transplants. Pretransplant EBV status affected the results of the assay and should be considered when interpreting test results. HLA-A3 was strongly linked to PTLD and may be a novel marker of PTLD risk.
Key Words: lung transplantation • EBV • DNA • HLA • lymphoma • PTLD
Post-transplantation lymphoproliferative disease (PTLD) is a well-documented and often fatal complication of organ transplantation (1, 2). Data from several major transplant centers suggest that lung transplant recipients are among those who are at the greatest risk for developing this condition, with a relative risk at least twofold higher than that observed after the transplantation of other organs (including kidney, heart, and liver) (3, 4). Although the exact pathogenesis of PTLD remains uncertain, it is generally accepted that infection with the Epstein-Barr virus (EBV) and high-dose immunosuppressive drug regimens are major risk factors (2, 5). Other factors suggested as plausible risk factors for PTLD after lung transplantation include the following: race, male sex, age, primary cytomegalovirus infection, underlying diagnosis (cystic fibrosis) and rejection frequency, and use of induction immunosuppressants (5, 6). The immunosuppressive agents required to prevent graft rejection after organ transplantation suppress both the de novo (naive) and, to a lesser extent, recall (memory) responses of EBV-specific cytotoxic T-cell activity. These events may prevent the immune system from mounting an effective response and allow uncontrolled viral replication and, ultimately, B cell transformation (5). Preexisting antigen-educated EBV-specific memory T cells lower the risk of PTLD in seropositive individuals. Seronegative patients have a 10- to 20-fold higher risk because the vast majority are inoculated with EBV from "passenger" lymphocytes in EBV-positive donor organs (which occur in approximately 90% of adult organ donors), and these patients lack memory T and adequate tumor surveillance functions to suppress viral replication and lymphomagenesis (5). In this situation, EBV may cause an infectious mononucleosis-like illness, but often is virtually silent with the first manifestation of infection being PTLD. Previous work in the PTLD arena, as it pertains to EBV DNA testing, has largely been confined to pediatric liver and kidney transplant recipients for whom PTLD is a particularly large problem (1, 7). Little research on EBV viral load testing has been performed after lung transplantation. Thus, we conducted this retrospective study with the hypothesis that elevated EBV loads predate tumor formation and are a reliable diagnostic marker of PTLD in lung transplant recipients. In addition, we assessed other factors (e.g., pretransplant EBV serostatus, levels of immunosuppression, antiviral therapy, and HLA types) as risk or ameliorative factors for PTLD. Some of this research has been published in abstract form (8).
Patient Population A retrospective, single-center study was performed, under the auspices of our institutional review board, using an electronic database. All patients (n = 296) who underwent lung transplantation from 1990 to 2008 were included. The follow-up for all patients exceeded 1 year. Of 296 patients, only 125 had EBV DNA level measurements, because this test first became available in 1999. The surgical and medical management has been previously described (3, 9).
EBV DNA Testing
EBV Serology
Other Risk Factors for PTLD
PTLD Diagnosis
Clinical Management of PTLD Patients
Statistical Analysis
Study Population Patient characteristics are shown in Table 1. All 296 patients were assessed for PTLD risk factors, but only 125 patients who had 612 different EBV DNA levels measured were included in the analysis to determine the value of viral load testing as it pertained to PTLD risk. These 125 patients had similar characteristics to the total population of patients (n = 296) who underwent lung transplantation (Table 1). Overall, 13 of 296 patients (4.4% incidence) developed PTLD, and 5 of 125 patients with viral load data developed PTLD. Twelve PTLD diagnoses were confirmed histologically and one clinically. The histology and molecular characteristics of the most recent seven cases are in Table 2; the first six cases have been described previously (3). Twelve cases occurred in patients entering transplant seronegative for EBV and one case occurred in a seropositive patient. The median time from transplant to PTLD diagnosis was 95 days (mean ± SD = 243 ± 342 d) with 11 of 13 cases occurring in the first year, one at 1.2 years and one at 4.2 years (the late case was a Burkitt's lymphoma). Tumor locations included the lungs (12 of 13), tonsils (1), abdominal wall (1), mesenteric lymph nodes (1), liver (1), small intestine (1), and sigmoid colon (1).
Predictive Value of EBV DNA Levels The sensitivities and specificities of the different EBV DNA levels are shown in Table 3. Overall, true positive levels consistent with active infection yielded a specificity of 30% and a sensitivity of 60% for PTLD. As might be expected, better sensitivity and worse specificity was obtained if any value in the indeterminate range was considered positive. A true negative DNA level indicated 0 risk for PTLD in our study (100% negative predictive value). However, when DNA testing was viewed in the context of EBV serology, its predictive value appeared to improve (Table 1). The rationale for considering serology when interpreting the EBV DNA levels was that the frequency distribution of EBV DNA levels was significantly different between the seropositive and seronegative patients (rightwardly skewed toward higher viral loads in the seronegatives), making it likely that a fundamental biological difference was responsible for these differences (Figure 1; P < 0.001). Most of the false positive DNA levels for PTLD in the seronegative cohort probably represented primary EBV infection, lessening the ability of this assay to predict PTLD among seronegative patients. Taken together, these data suggest that a viral load greater than the normal range (i.e., indeterminate and positive values combined) should be used by clinicians to evaluate seronegative patients for PTLD risk to avoid missing any PTLD cases (i.e., improve sensitivity), whereas true positive viral load levels may be the most clinically useful to assess PTLD risk among patients who were seropositive. The pattern and number of positive and/or indeterminate viral loads did not help discriminate between PTLD and no PTLD, but the retrospective nature of this study probably compromised our ability to detect important trends in this regard.
Clinical Usefulness of EBV DNA Levels in the Treatment of PTLD In patients who developed PTLD, which was stage IV in each case, EBV DNA levels fell immediately and usually were negative within 2 weeks of reducing immunosuppression and starting antiviral therapy (see Figure E1 in the online supplement). In contrast, the size of the tumor took a mean (±SD) of 40 (±24) days to begin decreasing in size and a mean (±SD) of 131 (±40) days to resolve by imaging studies.
HLA-A3 and Seronegativity as Risk Factors for PTLD
Other Risk Factors for PTLD No other independent risk factor for PTLD, including types or levels of immunosuppressants, antiviral therapy, sex, age. or diagnostic group, was found. Several patients developed PTLD while taking antiviral agents (including acyclovir, valganciclovir, foscarnet, intravenous ganciclovir), indicating that this intervention alone was not protective.
Outcomes for PTLD
The primary goal of this research was to investigate the value of elevated EBV loads and other risk factors as predictors of PTLD. We found that EBV loads were a valuable but imperfect measure of PTLD risk. Using our definition of positive viral load test, EBV DNA levels had a specificity of 30% and a sensitivity of 60% for PTLD. The sensitivity and specificity data of the EBV viral load assay, used in a "real world" fashion as occurred in our study, were not very impressive. Of course, the sensitivity and specificity of the assay may have been better if a specific algorithm for targeting high-risk patients was used. Although a negative viral load reassuringly ruled out PTLD in our cohort (100% negative predictive value [NPV]), it should be noted that the small sample of PTLD cases in our study may have compromised our ability to categorically assign no risk to those with negative loads. Importantly, the EBV load assay performed differently between seropositive and seronegative patients. At any post-transplant time point, seronegative patients were significantly (P < 0.001) more likely to have positive and indeterminate load values than seropositive patients. In all likelihood, this probably reflected primary EBV infection. For this reason, our data suggest that EBV DNA threshold values for clinically meaningful events may be different for seropositive and seronegative patients. Our data suggest that viral loads in the indeterminate area are less likely to be associated with clinically relevant events (i.e., PTLD development) among seropositive patients (44 patients with indeterminate loads had no PTLD) compared with seronegative patients (5 patients had indeterminate loads and 2 developed PTLD). For both groups, a negative or normal DNA level was not associated with PTLD. To the best of our knowledge, this is the first study using EBV viral load testing to predict PTLD after a lung transplant. To date there have been few published reports in this area and most have not focused on using load testing to predict PTLD risk but rather looked at the use of load testing as a measure of systemic immunosuppression. Thus, with a different focus, these previous studies were hampered by the use of small numbers of patients (<75) and even smaller numbers of PTLD cases. Stevens and colleagues found that four of six patients with PTLD had positive levels (>2,000 EBV DNA copies/ml in blood) and 78% of all samples tested before the diagnosis were greater than this value as well; whereas only 4 of 117 (3.4%) samples from two of eight patients free of PTLD were positive (15). However, no sensitivity and specificity data were reported. Bakker and colleagues found greater than 10,000 EBV DNA copies/ml in the blood of 26 of 75 (35%) lung transplant recipients but only 1 developed PTLD (specificity = 4%) (16). Ahya and colleagues found that 41% of 32 recipients of lung transplant developed a positive viral load but none developed PTLD (17). It should be emphasized that there is no nationally or internationally accepted standard for EBV viral load testing, and many different assays exist using different denominators for expressing load values and different thresholds for positivity. Despite the consensus that PTLD patients have a higher EBV load than non-PTLD transplant recipients, based largely on data outside the lung transplant world (18), it is not clear what threshold values are predictive for PTLD. Many different threshold values have been reported, all with different sensitivity (60–100%) and specificity (71–100%) (19–23). These differences may be due to the patient populations, the types of transplants performed, the immunosuppressive regimens (including induction versus no induction) used, and the blood compartments in which EBV DNA was measured (24). The point needs to be made that the sensitivity and specificity values from these previous studies are more auspicious than those found in our study and other cohort analyses published more recently (15–17). When analyzing other risk factors for PTLD, we only found one, but it probably is the most provocative and potentially most important finding in our study. As a risk factor for PTLD, HLA-A3 may be as important as elevated EBV viral loads. HLA-A3 was strongly associated with the development of PTLD among the seronegative patients. We previously showed that seronegative patients were at an approximately 15-fold higher risk for developing PTLD (incidence 33%) after lung transplant compared with seropositives (incidence 1.8%) (3) and our current study, as well as those from others (25), confirmed that finding. In our previous study, we were unable to identify which seronegative patients were at highest risk (3). Our current HLA-A3 results help quantify risk within the seronegative cohort. The PTLD risk to a seronegative recipient with HLA-A3 was 58% compared with 22% in recipients without HLA-A3 (P = 0.026). Even more interesting, when neither the recipient nor the donor expressed HLA-A3, the PTLD risk was only 7% compared with 55% when either the donor or recipient expressed A3 (P = 0.003). To date, we have not seen reports on the role of particular HLA alleles in the pathogenesis of PTLD. Nonetheless, the role of HLA proteins and EBV infection has been investigated at a basic level, and there is a growing literature to indicate that variations in major histocompatibility complex class I proteins (i.e., HLA-A and HLA-B), which are responsible for presentation of EBV antigens to CD8 cytotoxic T cells, affect EBV replication and host response. The fundamental underlying hypothesis being tested is that different HLA class I proteins are less or more efficient antigen presenters and that the host response to contain viral replication is altered by these differences, ultimately affecting lymphomagenesis. To date, more than 25 different HLA proteins have been shown to participate in presentation of EBV antigens. With regard to Hodgkin lymphoma, Diepstra and colleagues were the first to identify alleles of two microsatellite markers (D6S265 and D6S510) in germ-line DNA from 200 patients that were significantly associated with tumor formation (26). Interestingly, in haplotype prediction studies, both the D6S510 and D6S265 microsatellite markers were shown to have strong linkage disequilibrium with the HLA-A locus (27). Most recently, Niens and colleagues genotyped exons 2 and 3, encoding the peptide-binding groove of HLA-A, for 32 single nucleotide polymorphisms in 70 patients with EBV+ Hodgkin lymphoma, 31 patients with EBV-negative Hodgkin lymphomas, and 59 control participants and found that HLA-A1 was significantly overrepresented and HLA-A2 was significantly underrepresented in patients with EBV-positive Hodgkin lymphomas (28). Other studies of HLA, EBV, and nasopharyngeal carcinoma (another EBV-associated cancer) have found that tumor formation was linked to the HLA-A locus (29). With regard to infectious mononucleosis, McAulay and colleagues demonstrated that the class I alleles, HLA-A1 and HLA-A3, were more frequently found in symptomatic patients compared with silent seroconverters after EBV infection (30). Taken together, these studies further support an important role of HLA in antigen presentation of EBV proteins. Although we have not tested this theory, some similarities and some differences exist between our results and those cited above. However, it should be noted that PTLD is not a Hodgkin lymphoma and that both recipient and donor cells are capable of antigen presentation in the post-transplantation environment, unlike in the studies cited above. One of our most clinically important results was that survival with PTLD was not different than survival for our entire cohort of lung-transplanted patients. This finding is in contrast to some recent reports of short median survival times (approximately 4 mo) and high mortality for PTLD (31, 32). Although the optimal therapy for PTLD remains unknown, most experts advocate reducing immunosuppression as the first step. It should be noted that EBV viral load fell faster in response to PTLD treatment (which may be valuable to the clinician) than did the imaging studies show changes in the size of the PTLD. There are several shortcomings of this research. The most important one is the retrospective nature of the study. To help lessen selection bias, we chose to include all patients and every viral load measured. Nonetheless, a prospective study would have been a stronger design. In addition, PTLD is a relatively rare complication (with an incidence of 2–20%) after lung transplant and, thus, we only had 13 cases of PTLD with only 5 occurring during the time of EBV load monitoring. Therefore, our calculated sensitivities and specificities for various viral load cutoffs for PTLD need to be viewed in light of our small sample size. Also, we blended data from two different viral load tests. However, we used the recommended thresholds for positive, indeterminate, and negative for blending purposes and did not attempt to generate arbitrary values in a post hoc fashion to better fit the data. Last, our HLA A3 results need further molecular work to determine which of three A3 alleles and several subtypes is linked to susceptibility. In conclusion, this study shows that positive EBV DNA levels have a fair to good sensitivity but relatively poor specificity for PTLD. On the basis of our results, it may be important to consider the EBV serostatus of the patient before transplantation when interpreting the EBV DNA levels. Last, HLA A3 expression was significantly associated with PTLD risk and may represent a novel marker of PTLD risk, but this finding requires confirmation from larger studies.
The authors thank our funders, the National Institute of Diabetes, Digestive and Kidney Diseases, the Cystic Fibrosis Foundation, and the Clinical and Translational Science Award (CTSA) from the National Institutes of Health. They also thank all the patients who underwent lung transplantation at the University of North Carolina; the surgeons and physicians who performed the operations and cared for the patients; our coordinators, Becky Cicale and Ken Davis, without whom this program could not function; and W. June Brickey for help with data analysis.
Supported by the National Institute of Diabetes and Digestive and Kidney Diseases and the Cystic Fibrosis Foundation. This article has an online supplement, which is accessible from this issue's table of contents at www.atsjournals.org Originally Published in Press as DOI: 10.1164/rccm.200804-531OC on August 28, 2008 Conflict of Interest Statement: None of the authors has a financial relationship with a commercial entity that has an interest in the subject of this manuscript. Received in original form April 8, 2008; accepted in final form August 28, 2008
This article has been cited by other articles:
|
|||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
TOP
ABSTRACT
AT A GLANCE COMMENTARY
