Utility of Semiquantitative Polymerase Chain Reaction for Epstein-Barr Virus to Measure Virus Load in Pediatric Organ Transplant Recipients with and without Posttransplant Lymphoproliferative Disease

Abstract

We examined the utility of Epstein-Barr virus (EBV) load as a test for the presence of posttransplant lymphoproliferative disease (PTLD). A semiquantitative (SQ) EBV polymerase chain reaction (PCR) on peripheral blood mononuclear cells (PBMC) was used to determine virus load. We compared the values from pediatric patients, both with and without PTLD, with those from healthy pediatric and adult subjects. The virus loads for asymptomatic healthy subjects had a range of 0–1 log₁₀ cells/10⁶ PBMCs. Among transplant recipients (n = 135), the mean virus load (± standard deviation) at the time of diagnosis of PTLD was 3.1 ± 1.2 log₁₀ cells/10⁶ PBMCs versus a baseline value of 1.3 ± 1.4 log₁₀ cells/10⁶ PBMCs in children without PTLD (P < .0001). A cutoff of ⩾3 log₁₀ cells/10⁶ peripheral blood leukocytes resulted in the following values for use of virus load as a test for PTLD: sensitivity, 69%; specificity, 76%; positive predictive value, 28%; and negative predictive value, 95%. We conclude that determination of EBV load by use of SQ PCR is more useful in ruling out than in indicating the presence of PTLD.

Epstein-Barr virus (EBV)-related posttransplant lymphoproliferative disease (PTLD) affects 2%–5% of adult solid-organ transplant recipients [1, 2] and up to 14%–22% of pediatric solid-organ transplant recipients [3, 4]. Risk factors for PTLD include primary EBV infection, the type of organ that was transplanted, type and intensity of immunosuppression, and coexistent cytomegalovirus (CMV) infection [2–10]. Early diagnosis of this entity is important in maximizing the chances of a successful outcome.

The determination of EBV load by use of methods such as EBV semiquantitative (SQ) PCR to analyze peripheral blood mononuclear cells (PBMCs) has been suggested as a way of identifying patients who are at an increased risk of PTLD to facilitate early diagnosis. This approach is based on the premise that there are increasing levels of circulating EBV-infected transformed lymphocytes in the peripheral blood before the overt clinical manifestations of PTLD occur [11–15]. Patients have been shown to have high EBV loads in the peripheral blood at the time of diagnosis of PTLD [11].

The primary objective of this study was to examine the value of EBV load as a means of determining the PTLD status of solid-organ transplant recipients. We also examined the EBV load in the PBMCs in a population of pediatric solid-organ transplant recipients and compared this with the EBV load in PBMCs obtained from healthy EBV-seropositive patients.

Methods

Setting. The study was conducted at The Hospital for Sick Children, Toronto, Canada, which is a quaternary-care pediatric transplant center.

Study design and target group. This was a descriptive observational cohort study with a control group. All solid-organ transplant recipients whose EBV loads were determined from May 1996 through April 1998 were evaluated. The control subjects were healthy asymptomatic adults and children.

Diagnosis of PTLD. Diagnosis of PTLD was based on presence of clinical and radiographic evidence of PTLD, laboratory evidence of EBV infection, and histopathologic confirmation of disease.

Testing protocol for EBV load in PBMC. Since May 1996, we have routinely performed the EBV SQ PCR at our center. Patients who received transplants before 1996 were tested every 1–3 months. New transplant recipients were tested every 1–2 weeks for the first 6 weeks after transplantation, then every 4 weeks for the first 6 months after transplantation. Thereafter, the patients were tested every 1–3 months. Children who developed acute rejection, CMV disease, and clinical flare-ups of EBV disease were tested every 1–2 weeks for 6–8 weeks. Finally, EBV SQ PCR testing was part of the work-up for patients with suspected PTLD. In this report, “baseline” refers to the first virus load test that was performed during the posttransplant period.

Patient samples consisted of 3 mL of blood anticoagulated with EDTA. PBMCs were separated by means of centrifugation on a Ficoll-Hypaque cushion (Pharmacia). PBMCs were resuspended in PBS and counted on an automated cell counter (Coulter MAXM). Cells were then serially diluted to obtain 4 aliquots of 10⁶, 10⁵, 10⁴, and 10³ cells, respectively. Total DNA was extracted from each aliquot by use of the QIAamp blood kit (Qiagen), and the DNA was resuspended in 200 µL of double-distilled molecular-grade water. For the PCR analysis, 10 µL of the DNA solution was used.

PCR. PCR was performed using EBV-specific primers (5′-TTTGCCAGCCTCTACCCG-3′ and 5′-GCCAGCAGCTTCTTGATGG-3′). The primers bracket a 234-bp fragment of the EBV DNA polymerase gene. Reactions were performed in a total volume of 50 µL. The master mix contained in each tube was 5 µL of 10× Cetus buffer II (Perkin Elmer), 5 µL of 25-mM MgCl₂, 5 µL of dNTPs mix (each dNTP 2 mM), 5 µL of each primer stock solution (10 pmol/µL), 14.5 µL of molecular-grade water, and 0.5 µL of AmpliTaq Gold (Perkin Elmer), for a volume of 40 µL. To each 40-µL aliquot of master mix was added 10 µL of template DNA solution.

The PCR was performed in a Robocycler 40 (Stratagene) with the following cycling parameters: 10 min at 95°C, then 40 cycles consisting of 1 min at 95°C, 1 min at 64°C, and 1 min at 72°C; after completion of the last cycle, a final incubation at 72°C for 3 min was performed. Amplicons were detected by means of agarose gel electrophoresis and ethidium bromide staining. The sensitivity of the PCR was established with DNA extracted from serial dilutions of Namalwa cell line (which contained 2 EBV genome copies per cell). EBV could be reproducibly detected from the equivalent of 1 cell, and it is conservatively estimated that the PCR sensitivity is 1–10 genome copies. By use of this assay, no amplicons were obtained from DNA of EBV-negative cells or from herpes simplex virus type 1 DNA and CMV DNA templates.

SQ estimate of EBV-infected cell load. “EBV-infected cell load” was defined as the smallest aliquot of the cells to test positive for EBV by PCR (e.g., 10⁵), which is then converted as a proportion of the EBV-positive cells in 10⁶ cells; we expressed this as a range to account for the fact that only a fraction of the extracted DNA was used in the PCR (e.g., 10–100 cells/10⁶ PBMCs). The ranges that could be reported are therefore as follows: no EBV detected in 10⁶ PBMCs, 1–10 cells/10⁶ PBMCs, 10–100 cells/10⁶ PBMCs, 100–1000 cells/10⁶ PBMCs, and >1000 cells/10⁶ PBMCs, respectively. For the purposes of statistical analysis, we expressed these values by using an approximate log₁₀ scale based on the upper value of the ranges, as follows: 0, 1, 2, 3, and 4 log₁₀ cells/10⁶ PBMCs. (Values >1000 cells/10⁶ PBMCs were approximated to 4 log₁₀ cells/10⁶ PBMCs.)

Statistical analyses. Data management and analyses were facilitated by use of Epi Info software (Centers for Disease Control and Prevention). Mean values were compared by use of Student's t test; a nonparametric procedure was used to compare medians. Proportions were compared by use of the χ² test or Fisher's exact test, as appropriate. Data are presented as mean values ± SD.

Results

We evaluated 135 children, of whom 16 (11.9%) had PTLD. The descriptive characteristics of the study subjects are shown in table 1. The mean patient age at the time of transplantation was 72 ± 62 months. The majority of transplant recipients received liver transplants (72 [53.3%] of 135). Children who were EBV-seronegative recipients of transplants accounted for 64 (47.4%) of the 135 study subjects, whereas 51 (37.8%) of the subjects were EBV-seropositive before they underwent transplantation. EBV status was undetermined for 20 pediatric transplant recipients (14.8%). The majority of the patients (86 [63.7%] of 135) were CMV seronegative before they underwent transplantation.

Table 1

Descriptive characteristics of 135 pediatric transplant recipients evaluated for use of Epstein-Barr virus (EBV) load as a marker of posttransplant lymphoproliferative disease.

Among patients for whom specific treatment or prophylaxis information was complete, 36 (30%) of 120 recipients received ganciclovir prophylaxis or treatment before monitoring by EBV SQ PCR, and 61 (47.7%) of 128 patients received acyclovir prophylaxis or treatment. The use of immunoglobulin preparations was as follows: standard immunoglobulin for 9 (6.7%) of 135 patients, CMV high-titer immunoglobulin for 51 (38.3%) of 133 patients, and CytoGam (MedImmune) for 7 (5.2%) of 134 patients. Details on the use of these agents were unavailable for a proportion of patients who received organ transplants at external centers several years prior to the study period.

The healthy control subjects consisted of 46 children and 16 adults. The mean patient ages were 129.8 ± 6.9 months for children and 36.9 ± 2.7 years for adults. These control subjects were asymptomatic, with the only recent illness being an upper respiratory tract infection in 1 (6.3%) of the 16 adults. The mean age of the 16 patients who had PTLD was 63.9 ± 59.5 months. This disease was diagnosed at a mean time of 3.4 ± 2.6 years after the patients underwent transplantation. Four (25%) of 16 patients with PTLD received their diagnoses within the first year after they underwent transplantation, whereas 7 (44%) of 16 had PTLD diagnosed within 2 years after they underwent transplantation.

The mean virus loads as defined previously among healthy children and adults were 0.11 ± 0.1 log₁₀ cells/10⁶ PBMCs and 0.5 ± 0.13 log₁₀ cells/10⁶ PBMCs, respectively (P = .01). This is consistent with a greater likelihood for children to be EBV seronegative. In this regard, serologic testing of a subset (n = 15) of these children revealed a rate of seronegativity of 53.3% (8 of 15 children); none of the seronegative children had detectable EBV DNA in the PBMCs.

The mean baseline virus load for transplant recipients with PTLD was 3.1 ± 1.3 log₁₀ cells/10⁶ PBMCs, compared with 1.3 ± 1.4 log₁₀ cells/10⁶ PBMCs for transplant recipients without PTLD (P < .0001). The corresponding median values (ranges) were 4 (0–4) log₁₀ cells/10⁶ PBMCs and 1 (0–4) log₁₀ cells/10⁶ PBMCs, respectively (P < .0001).Table 2 shows the virus loads for transplant recipients both with and without PTLD. As indicated above, baseline virus load values were significantly higher among patients with PTLD than they were among patients without PTLD. In addition, we compared the virus loads at the time of diagnosis of PTLD with the baseline loads for patients without PTLD. This allowed an appreciation of the extent to which virus load increased among patients with PTLD, compared with the baseline values for patients without PTLD. The differences were statistically significant: 3.1 ± 1.2 mean log₁₀ cells/10⁶ PBMCs at the time of diagnosis of PTLD versus 1.4 ± 1.5 mean log₁₀ cells/10⁶ PBMCs at baseline in patients without PTLD (P < .0001). The median values (ranges) were 3.5 (0–4) log₁₀ cells/10⁶ PBMCs and 1 (0–4) log₁₀ cells/10⁶ PBMCs, respectively (P < .0001). The peak values recorded in each group within 12 months after baseline were also compared. This enabled us to examine the extent to which differences in virus load between patients with and patients without PTLD were sustained. Patients with PTLD attained greater virus loads than did those without PTLD (mean, 3.4 ± 0.5 log₁₀ cells/10⁶ PBMCs vs. 1.8 ± 1.5 log₁₀ cells/10⁶ PBMCs [P < .0001]; median [range], 3 [1–4] log₁₀ cells/10⁶ PBMCs vs. 2 [0–4] log₁₀ cells/10⁶ PBMCs, respectively [P < .0001]).

Table 2

Epstein-Barr virus (EBV) semiquantitative PCR results among patients with and without posttransplant lymphoproliferative disease (PTLD).

Table 3 shows virus loads at baseline in relation to recipient serostatus. Although patients with PTLD had higher virus loads, the differences between PTLD and non-PTLD patients were more pronounced among those patients who were EBV- or CMV-seronegative transplant recipients (P = .0004 and P = .0005, respectively).

Table 3

Epstein-Barr virus (EBV) semiquantitative PCR test results at baseline in relation to the transplant recipients' serostatus.

The results of univariate analyses indicated no statistically significant association between virus load and the following clinical and treatment variables: acute rejection of transplant, CMV disease, previous ganciclovir prophylaxis or treatment, immunoglobulin prophylaxis, cyclosporine use, and tacrolimus use (P > .05 for all). However, patients who received acyclovir prophylaxis had higher baseline virus loads than did those who did not receive acyclovir prophylaxis, with mean values of 1.9 ± 1.7 log₁₀ cells/10⁶ PBMCs and 1.2 ± 1.3 log₁₀ cells/10⁶ PBMCs, respectively (P = .02). This association was not present when the analysis was done by selecting CMV- or EBV-seropositive recipients of transplants (P = .07 and P = 0.79), but it was present when the data were analyzed by selecting CMV- or EBV-seronegative recipients of transplants (P = .002 and P = .021, respectively).

We examined the characteristics of the EBV SQ PCR assay when used as a diagnostic test for the presence of PTLD. Table 4 shows characteristics for 3 different cutoff values that were used: ⩾1, ⩾2 and ⩾3 log₁₀ cells/10⁶ PBMCs. These results show that the “test” has a high negative predictive value but a low positive predictive value. The highest positive predictive value (28%) corresponds to a cutoff of ⩾3 log₁₀ cells/10⁶ PBMCs (negative predictive value, 95%).

Table 4

Epstein-Barr virus semiquantitative PCR test characteristics at the time of diagnosis of posttransplant lymphoproliferative disease.

Discussion

Given the generally poor prognosis of patients with patients with PTLD, early detection is important to maximize the chances of a successful outcome. EBV load is regarded as a useful indicator of the changes in viral dynamics that precede the development of PTLD. This test can also detect the increased viral replication associated with acute (nonmalignant) EBV disease. By use of this PCR assay, it is possible to detect EBV DNA in the peripheral blood lymphocytes of healthy seropositive patients. Such patients are believed to have ⩽1 infected lymphocyte per 10⁶ peripheral blood lymphocytes [16].

In our cohort, a large proportion of the subjects were EBV seronegative before they underwent transplantation. Children who were EBV seronegative before they received a transplant and who developed PTLD tended to have higher EBV loads detected by SQ PCR than did children who had PTLD but who were EBV seropositive before they underwent transplantation. This may be related to a better ability of patients in the latter group to exert some control of EBV replication due to cytotoxic T lymphocyte (CTL) activity [1, 17–20]. The fact that transplant recipients without PTLD had virus loads that were greater than the mean value for healthy children and adults is also consistent with a relative inability of these patients to control the lymphoproliferative aspects of EBV replication when compared with individuals who possess good CTL surveillance. High EBV loads, although not necessarily indicative of PTLD, suggest an environment in which EBV replication and proliferation of transformed lymphocytes are enhanced.

These results indicate that patients with PTLD had higher EBV SQ PCR values at baseline and at the time of diagnosis of PTLD than did patients without PTLD. This would suggest that the test is of value for evaluation of patients with suspected PTLD. However, the low positive predictive value is of concern. This, coupled with the high negative predictive value, would suggest that the test is useful in ruling out PTLD but has limited usefulness in indicating the presence of PTLD. Additional investigation is warranted to improve the test or to develop a composite test in which virus loads are combined with other laboratory or clinical parameters.

Higher virus loads were shown to be more likely in children who had received previous acyclovir prophylaxis or treatment; however, this was not independent of the transplant recipient's EBV or CMV serostatus prior to transplantation. In this context, this effect was only significant among CMV- or EBV-seronegative transplant recipients. It is tempting to speculate that acyclovir use may not have been effective in limiting virus load in patients who received this agent before the determination of their baseline virus load. However, it is possible that acyclovir use was merely a marker of high-risk patients (i.e., donor-seropositive and recipient-seronegative EBV or CMV status) who were identified as being in need of antiviral prophylaxis. This is consistent with the finding that there was a trend toward higher virus loads in patients who were treated with immune globulin. Although no association was found between ganciclovir use and higher virus load, more subjects had been treated with acyclovir and with immune globulin than had been treated with ganciclovir, which may have enhanced the ability to detect differences.

As was the case in previous studies, the number of patients in our study who had PTLD was relatively small. This limited the extent to which multivariable analyses could be performed to examine the various factors that were associated with PTLD and with elevated virus loads.

A limitation of our log transformation procedure should be noted. Patients who had a virus load of >1000 cells/10⁶ PBMCs were regarded as having a load of 4 log₁₀ cells/10⁶ PBMCs. However, it is possible that some patients may have had virus loads that approached 4.5 log₁₀ cells/10⁶ PBMCs or more. Given the general trend toward higher virus loads that was seen among patients in the PTLD group, our truncation of the upper limit of the virus load measurements would be more likely to affect the group with PTLD and, therefore, to result in an underestimation of the differences in virus load between patients with PTLD and patients without PTLD.

Although we had a prospective surveillance protocol in place, in reality, the virus loads were determined at varying intervals after the patients underwent transplantation. Therefore, it was not possible to ascertain in some patients with low virus loads whether the levels had peaked before the determination of baseline values. It is possible that some patients with PTLD who had low virus loads may actually have had higher values before the determination of baseline values and values at the time of diagnosis of PTLD. The effect of this bias would be to falsely decrease the negative predictive value of the test. This would strengthen our results, given that we documented negative predictive values of >90%.

Our study did not evaluate the actions taken by clinicians in response to baseline virus load measurements. Modifications to immunosuppression could affect virus load values recorded after the initial baseline measurement. This clearly would not affect the baseline values, but it might affect the peak virus load values obtained after the initial baseline measurement. Alterations in immunosuppression that occurred before the development of PTLD may also have affected the EBV load at the time of diagnosis of PTLD. The effect of this bias would also be favorable to our results. In this regard, reductions in immunosuppression and any subsequent incremental improvements in CTL activity would tend to decrease, rather than increase, the virus load. Therefore, if this bias was operational in our study, the tendency would be for lower virus load values to be seen in patients at the time of diagnosis of PTLD.

Applications of the EBV SQ PCR test of virus load include use as an entry criterion in clinical trials on EBV prophylaxis [21] and use as a surrogate outcome marker in trials aimed at the prevention of PTLD. The test may be helpful in following patients with PTLD who have undergone treatment. However, it is relatively expensive, and the frequency and number of tests that need to be performed are the subject of much discussion. Qualitative EBV PCR may be used for surveillance of EBV-negative recipients of transplanted organs from EBV-positive donors until infection occurs.

The reporting of the laboratory results for EBV load should be standardized. Studies should be done to compare different techniques that use a given set of samples. This will enable clinicians to interpret better the results obtained by means of different techniques. In this context, it is necessary to compare methods that assess EBV genomic load with other methods that assess total infected cell burden. Measurement of the latter assesses latently infected cells as well as cells that harbor replicating virus. It is likely that the measurement of latently infected cells is more accurately correlated with PTLD. Quantitative assessments of virus genomic load done by measuring the number of genomic copies may reflect either the proliferation of EBV-infected B lymphocytes or an increase in the number of EBV genomes in each lymphocyte [11]. The issue is further complicated by the fact that different clones of EBV transformed cells may contain different numbers of genomic copies.

In summary, this study examined the utility of EBV SQ PCR testing in determining the virus load in pediatric solid-organ transplant recipients both with and without PTLD. Further studies are warranted to improve the ability of EBV load tests to discriminate between patients with PTLD and those without PTLD and in identifying patients at risk of this disease.

Acknowledgments

We thank Grant Johnson, laboratory technologist, The Hospital for Sick Children, and Kimberley Eddy, research assistant.

Author notes

Presented in part: 37th annual meeting of the Infectious Diseases Society of America, Philadelphia, 18–21 November 1999 (abstract 553).

Financial support: Research Institute, The Hospital for Sick Children, Toronto.