Post-Transplant Lymphoproliferative Disorders Following Liver Transplantation: Incidence, Risk Factors and Survival

Authors


Abstract

This study investigates retrospectively the incidence, risk factors and mortality of post-transplant lymphoproliferative disorders (PTLD) in adult orthotopic liver transplant (OLT) recipients. Among 1206 OLT recipients at a single institution, 37 developed a PTLD. The incidence of PTLD was highest during the first 18 months and relatively constant thereafter with cumulative incidence of 1.1% at 18 months and 4.7% at 15 years. The risk of PTLD was approximately 10% to 15% of the risk of death without PTLD. During the first 4 years following OLT, PTLD were predominantly related to EBV, while afterward most PTLD were EBV negative. Significant risk factors for PTLD in OLT recipients were transplantation for acute fulminant hepatitis during the first 18 months following OLT (HR = 2.6, p = 0.007), and rejection therapy with high-dose steroids (HR = 4.5, p = 0.049) and OKT3 (HR = 3.9, p = 0.016) during the previous year. Therapy with high-dose steroids or OKT3 (HR = 3.6, p = 0.0071) were also significant risk factors for PTLD-associated mortality. OLT recipients remain at risk for PTLD years after transplantation. The strong association of PTLD with rejection therapy and the worse post-PTLD prognosis among recipients of rejection therapy indicate the need to balance the risk of immunosuppression against the risk of PTLD following rejection treatment.

Introduction

Immunodeficiency states, either congenital, iatrogenic or acquired predispose to the development of lymphoid neoplasms (1,2). These lymphoproliferative disorders differ from lymphomas seen in immunocompetent individuals as they are frequently extranodal, rarely correspond to low-grade B-cell lymphomas, have an aggressive course, are frequently associated with Epstein-Barr virus (EBV) and may respond to reduction or withdrawal of immunosuppression (3–5). Among iatrogenic immune deficiency states, post-transplant lymphoproliferative disorder (PTLD) is quite common (6). The risk of PTLD depends upon the organ transplanted and reported frequencies range between 1% and 10% (7–10). However, the incidence of PTLD depends upon the length of follow-up following transplantation, and is expected to be higher in studies with longer follow-up.

PTLD has been linked to the type and level of immunosuppression and EBV status (11–14). EBV infection is ubiquitous with little consequences in the normal population but in organ transplant recipients EBV is associated with a range of disorders from reactive polyclonal hyperplasia to monoclonal malignant lymphomas (15). Although earlier reports of PTLD were frequently reported to be EBV positive, more recent reports suggest an increasing frequency of EBV negative PTLD (16–18). These are viewed as distinct from the EBV positive PTLD because they tend to occur later after transplantation and have a worse prognosis (16–20). Studies are conflicting regarding the potential role of hepatitis C virus, alcoholic cirrhosis and older age as risk factors for PTLD (21–25). Therefore, besides immunosuppression and EBV, relatively little is known about the pre-transplant clinical factors associated with the risk of PTLD.

In this study, we examined the cumulative incidence of PTLD, both overall and EBV positive and EBV negative separately, adjusting for competing risk of death without PTLD. We also examined baseline characteristics including etiology of liver disease and immunosuppression agents as risk factors for PTLD. Further, we compared the EBV positive and EBV negative PTLD groups with respect to pathologic findings, clinical characteristics and survival.

Patients and Methods

This retrospective study was approved by the Institution Review Board. The study population included all consenting adult recipients of a first solitary orthotropic liver transplantation (OLT) between March 1, 1985 and December 31, 2004 at the Mayo Clinic, Rochester, Minnesota. Patients with multiple transplants such as liver/kidney or liver/heart transplants were excluded, but those who later received a retransplant or transplant of another organ at a later time, e.g. kidney transplantation as a result of calcinuerin inhibitor-related nephrotoxicity, were included. Follow-up for vital status and occurrence of PTLD lasted through December 31, 2004.

PTLD were classified according to the World Health Organization classification of tumors of lymphoid tissues (26). At the time of diagnosis with PTLD, all patients underwent staging work up to detecting systemic diseases, including CT scans (abdomen, chest and pelvis) and bone marrow aspiration and biopsy. Immunophenotyping for B-cell- and T-cell-associated antigens was performed using frozen section or paraffin immunoperoxidase staining or flow cytometry as previously described (27). Clonality was determined by demonstrating immunoglobulin light chain restriction using monoclonal antibodies to kappa and lambda light chains on fresh cell suspensions using flow cytometry, on frozen or paraffin-embedded tissue sections using immunoperoxidase staining or by detecting immunoglobulin or T-cell receptor gene rearrangements by Southern blot analysis on frozen tissues using probes for immunoglobulin heavy and light chain genes (μ-J, μ-C, κ-J and λ-C) and T-cell receptor beta and gamma chain genes (Jβ1, Jβ2, Jγ1.3 and Jγ2.3) (28). EBV genome was detected in the tumor cells by in situ hybridization using probes for Epstein-Barr virus encoded RNA (EBER) (27).

Statistical methods

Data were summarized using means ± standard deviation for numeric variables, and percents and counts for categorical variables. Incidence rate of PTLD and mortality (before PTLD) following OLT were derived jointly using the competing risk extensions of the Kaplan-Meier and Cox methods (29,30). Cox proportional hazards regression models were used to estimate the influence of potential risk factors on PTLD and survival following PTLD. Hazard ratios (HR) were computed univariately for each potential risk factor, and accompanied with 95% confidence intervals. A multivariable model was developed using a stepwise process, including terms with p ≤ 0.05. Survival following PTLD diagnosis was estimated by the Kaplan-Meier method.

Results

Patient characteristics

The study population comprised 1206 OLT recipients with a mean age of 59.7 years (standard deviation [SD] 10.7, range 18 to 71 years) and 671 (56%) were men (Table 1). The most common indications for OLT were chronic hepatitis (31%), primary sclerosing cholangitis (20%), primary biliary cirrhosis (14%), alcohol liver disease (10%) and acute fulminant hepatitis (4%). CMV serology was available in 1192 patients, and of whom, 219 (18%) were CMV donor positive/recipient negative (CMV mismatch). As EBV serology at the time of transplantation was generally not available for donor and recipient, this could not be assessed. FK506 was the primary immunosuppression agent in 676 (56%) patients followed by cyclosporine in 530 (44%) patients. Mycophenolate mofetil was the primary adjuvant immunosuppressive therapy in 420 (35%) patients, azathioprine in 758 (63%) and sirolimus in 28 (2%) patients. Rejection treatment included high-dose steroids in 259 (21%) patients, OKT3 in 101 (8%) patients and polyclonal antibodies in 74 (6%) patients.

Table 1. Characteristics of 1206 OLT recipients transplanted between March 1,1985 and December 31, 2004 at the Mayo Clinic, Rochester, Minnesota and followed-up through December 31, 2004
CharacteristicsN (%)*
  1. *For discrete data values represented as count (percent) and for numeric data values represented mean ± standard deviation.

  2. **Serology available in 1192 patients.

Age, mean (SD), years59.7 (±10.7)
Men671 (56%)
Length of follow-up, mean (SD), years8.6 (±5.3)
Etiology of liver disease
 Chronic hepatitis372 (31%)
 Primary sclerosing cholangitis240 (20%)
 Primary biliary cirrhosis164 (14%)
 Acute fulminant hepatitis49 (4%)
 Alcohol liver disease119 (10%)
 Other262 (22%)
CMV serology donor positive recipient negative**219 (18%)
Retransplantation133 (11%)
Immunosuppressive therapy during follow-up
 FK506676 (56%)
 Cyclosporine530 (44%)
 Mycophenolate mofetil420 (35%)
 Azathioprine758 (63%)
 Sirolimus28 (2%)
 Polyclonal antibodies as induction therapy56 (5%)
 Rejection therapy301 (25%)
  High-dose steroids301 (25%)
  OKT3101 (8%)
  Polyclonal antibodies21 (2%)

Incidence of PTLD and EBV status

Over a mean follow-up of 8.6 years (SD 5.3 years), 304 patients (25%) were deceased without having developed PTLD and 37 (3.1%) developed PTLD. PTLD was observed from 3 months post-transplant to 15½ years post-transplant. The incidence of PTLD was highest during the first 18 months post-OLT with cumulative incidences 0.5%, 0.9% and 1.1% at 6, 12 and 18 months, respectively (Figure 1). Thereafter, the rate of PTLD was relatively constant. The cumulative incidence was 2.1% at 5 years, 4.2% at 10 years, and 4.7% at 15 years with 676, 304 and 88 patients being followed beyond these time points. At the longest follow-up of 19 ¾ years the cumulative incidence reached 5.4%.

Figure 1.

Cumulative incidence of PTLD following OLT, adjusting for the competing risk of death.

Cumulative incidences of EBV-positive and EBV-negative PTLD are shown in Figure 1. Early PTLD were mostly EBV positive whereas, the later PTLD were mostly EBV negative. Of the 37 PTLD, 22 were EBV positive, 13 EBV negative while in 2 cases EBV in situ hybridization was not done. PTLD observed during the first 18 months following OLT were predominantly EBV positive (10 +, 2 −, 1 untyped), from 18 months to 4 years a mix of both EBV positive and EBV negative (3 +, 3 −), from 4 years to 8 years a mix (6 +, 2 −, 1 untyped) and after 8 years predominantly EBV negative (3 +, 6 −). By 15 years post-transplant, the cumulative incidences of both types were similar (Figure 1).

Figure 2 illustrates the cumulative incidence of (a) PTLD, (b) death without PTLD treated as a competing risk (i.e. mortality) and (c) death or PTLD. For time points beyond 1 year post-OLT, the cumulative incidence of PTLD appeared to be approximately 10% to 15% of that of mortality (i.e. for every 100 deaths there are roughly 10 to 15 PTLDs). Specifically, at 5, 10 and 15 years post-OLT, the observed cumulative incidence of PTLD was 2.1%, 4.2% and 4.7%, whereas the mortality (without PTLD) was 18% and 28% and 40%.

Figure 2.

Cumulative incidence of PTLD or death (without PTLD) following OLT treating PTLD and death as competing risks.

Risk factors for PTLD

Cox regression models were used to examine the univariate association of PTLD with various clinical and therapy characteristics (Table 2). Age, sex, CMV-mismatch and retransplantation were not significantly associated with the risk of PTLD. Among the various etiologies for liver disease, only acute fulminant hepatitis (AFH) was associated with a significantly higher risk of PTLD within 18 months post-transplant (HR 9.25, 95% confidence intervals [CI] (2.55, 33.61), p = 0.0067). This significant association was derived numerically from 3 PTLD in 49 AFH patients with an estimated cumulative incidence of 6.7% at 18 months as opposed to 10 events in 1157 non-AFH patients with an estimated cumulative incidence at 18 months of 0.9%. Of the 3 PTLDs in AFH patients, 1 was EBV+, 1 EBV− and in 1 case in situ hybridization was not done.

Table 2. Evaluation of risk factors for PTLD in 1206 OLT recipients transplanted between March 1,1985 and December 31, 2004 at the Mayo Clinic, Rochester, Minnesota
 Hazard ratio* (95% CI)p value
  1. *Hazard ratios and 95% confidence intervals (CI) from univariate Cox proportional hazards models.

Age, per 10 years increase0.83 (0.62, 1.12)0.23
Men (vs. women)1.45 (0.75, 2.81)0.27
Etiology of Liver Disease
 Chronic hepatitis1.23 (0.63, 2.42)0.55
 Primary sclerosing cholangitis/ biliary cirrhosis0.80 (0.41, 1.56)0.50
 Alcohol liver disease Acute fulminant hepatitis0.62 (0.15, 2.60)0.49
 Overall2.64 (0.81, 8.62)0.16
 First 18 months9.25 (2.55,33.61)0.0067
 Other etiologies0.91 (0.35, 2.35)0.84
CMV donor +/ recipient − serology0.89 (0.25, 3.13)0.86
Retransplantation
 For 18 months following retransplant1.03 (0.14, 7.74)0.98
 All time post-retransplant0.67 (0.16, 2.79)0.58
Years since January 1, 19960.95 (0.88, 1.02)0.15
Immunosuppressive therapy during follow-up
 FK506 vs. cyclosporine0.79 (0.39, 1.60)0.52
 Mycophenolate mofetil vs. azathioprine0.58 (0.20, 1.72)0.27
 OKT3 – ever1.83 (0.76, 4.40)0.21
  In previous month27.08 (3.45, 212.3)0.032
  In previous year4.51 (1.25, 16.28)0.048
 High-dose steroids–ever1.03 (0.50, 2.10)0.94
  In previous month17.78 (3.92, 80.64)0.0065
  In previous year3.37 (1.22, 9.35)0.03
 Polyclonal antibodies–ever0.72 (0.17, 2.99)0.63

We also examined temporal trends in incidence of PTLD by examining the risk according to years since January 1, 1996. There was no indication for a temporal trend (per calendar year increase HR 0.95 (0.88, 1.02)), suggesting that changes in immunosuppressive therapies in more recent years were unlikely to have had an effect on the risk of PTLD.

Among various immunosuppressive agents, only OKT3 and high-dose steroid therapy were significantly associated with the risk of PTLD during the following 1 month and 12 months, but not indefinitely beyond the 12-month period (Table 2). Therapy with polyclonal antibodies was not associated with the risk of PTLD at any time during follow-up. Fitting a multivariable model, AFH for the first 18 months after OLT (HR 8.13, (2.11, 29.75), p = 0.010) and high-dose steroids in the previous 12 months (HR 3.08, (1.20, 8.62), p = 0.046) were significant risk factors or PTLD, while OKT3 in the previous 12 months entered into the model after these factors (HR 1.78, (0.36, 8.87), p = 0.48) was not significant.

PTLD characteristics and post-PTLD survival

Of the 34 PTLD cases where cell lineage was determined 32 (94%) were of B-cell type and 2 (5%) were of T-cell type. The PTLD morphology was monomorphic in 29/34 (85%), polymorphic in 4/34 (12%) and early lesion (plasma cell hyperplasia [PCH]) in 1/34 (3%) of the cases. Among the monomorphic PTLD, 22 (77%) were diffuse large B-cell lymphoma-like (DLBCL), 2 (7%) were marginal zone B-cell [mucosa-associated lymphoid tissue] lymphoma-like (MZBCL) in the stomach and parotid gland, 2 were plasmacytoma-like (PL) and there was 1 each (3%) of Burkitt lymphoma-like (BL), peripheral T-cell lymphoma-like (PTCL) and T-cell anaplastic large cell lymphoma-like (T-ALCL). Both T-cell PTLD had initial presentation in extranodal sites. The PTCL was EBV positive, whereas the ALCL was not. The ALCL was CD30+ but lacked expression of p80 indicating no abnormality of the ALK gene locus. Among the 2 polymorphic PTLD for which determinations were made both were polyclonal. Monoclonality was observed in all 17 monomorphic PTLD for which conclusive results were obtained by demonstrating immunoglobulin light chain restriction (frozen section immunoperoxidase [3], paraffin immunoperoxidase [8] or flow cytometry [4]) or by clonal immunoglobulin (2) gene rearrangements. Six monomorphic PTLD had indeterminate results for clonality by immunohistochemistry and 6 were not tested. Polyclonality was detected in both polymorphic PTLD on which testing was performed and in the PCH. Overall, monoclonality was identified in 17/26 (55%), polyclonality in 3/26 (12%) and the type of clonality was indeterminate in 6/26 (23%) of the cases. Further PTLD characteristics are described in Table 3.

Table 3. Clinical and histological comparison of EBV-positive and EBV-negative PTLD
 EBV negative (N = 13)EBV positive (N = 22)EBV not typed (N = 2)p value
  1. For discrete data values represented as count (percent) and for numeric data values represented mean ± standard deviation (sample size). p values compare EBV+ to EBV− only, Fisher exact for discrete variables and t-test for continuous variables.

Age of patient49.7 ± 11.1 (13)44.9 ± 12.4 (22)37.0 ± 7.1 (2)0.26
Male gender indicator9 (69%)13 (59%)0 (0%)0.72
Etiology of liver disease 0.97
 Acute fulminant hepatitis1 (8%)1 (5%)1 (50%) 
 Alcohol liver disease0 (0%)2 (9%)0 (0%) 
 Chronic hepatitis5 (39%)7 (32%)1 (50%) 
 Other2 (15%)3 (14%)0 (0%) 
 PBC/PSC5 (39%)9 (41%)0 (0%) 
CMV serology at discharge2 (17%)5 (26%)1 (50%)0.68
CMV donor +/ recipient −2 (15%)3 (14%)1 (50%)>0.99
Years since OLT of PTLD6.2 ± 3.94 (13)3.9 ± 3.97 (22)3.0 ± 2.14 (2)0.099
PTLD within 1 year of OLT2 (15%)8 (36%)0 (0%)0.26
Immunosuppressive therapy during follow-up
 FK506 vs. cyclosporine8 (67%)9 (48%)1 (50%)0.46
 Mycophenolate mofetil vs. azathioprine2 (15%)2 (9%)0 (0%)0.62
 High-dose steroids4 (31%)7 (32%)1 (50%)>0.99
 OKT32 (15%)4 (18%)0 (0%)>0.99
PTLD Characteristics
 Monoclonal8 (61%)8 (80%)1 (50%)0.48
 Monomorphic12 (100%)16 (88%)1 (50%)0.50
 Nodal7 (54%)10 (45%)0 (0%)0.73
 Extranodal10 (77%)13 (59%)2 (100%)0.46
 Lymphoma in multiple organs3 (23%)6 (29%)2 (100%)>0.99
 Lymphoma in grafted organ3 (23%)7 (32%)1 (50%)0.71
 IPF1.8 ± 1.40 (12)1.7 ± 1.12 (22)2.5 ± 0.71 (2)0.81
 IPF >= 34 (33%)6 (27%)1 (50%)0.72
 Stage 3.1 ± 1.20 (10) 3.5 ± 1.07 (19)4.0 ± 0.00 (2)0.40
 Performance score 1.2 ± 1.03 (12) 1.5 ± 1.25 (21)4.0 ± . (1)0.41
 LDH198 ± 129 (9)266 ± 192 (9)   . ± . (0)0.40
 LDH >= 2502 (22%)3 (33%)0 (0%)>0.99
 Anti-CD209 (100%)9 (69%)1 (100%)0.11
 Rituxan3 (30%)3 (14%)0 (0%)0.36

Following diagnosis, 24 of the 37 patients (65%) underwent a trial of significant decrease in immunosuppression, i.e. removal or reduction by ½ of one or more agent. For patients in whom other treatments were indicated, and in patients with stable disease or progression despite reduction of immunosuppression other therapies were administered. These included (singly or jointly) surgical resection in 13 (35%), chemotherapy in 11 (30%) (CHOP or BACOP), antibody therapy (anti-CD20 (Rituximab)) in 5 (14%), radiation in 4 (11%), high-dose steroids in 2 (5%) and Acyclovir indefinitely in 3 patients (8%).

Of the 37 patients with PTLD, 20 (54%) were deceased and 17 (46%) were still alive through December 31, 2004. Six deaths (30%) were due directly to PTLD, 9 (45%) were due to other causes but with PTLD, 4 (20%) were due to other causes without PTLD, and the cause of death was unknown in 1 patient (5%). Survival post-PTLD diagnosis was very similar for the EBV-positive and EBV-negative PTLD (see Figure 3). The overall survival at 1 year post-PTLD was 56.5% (±SE 8.6%), and at 5 years 40.8% (±SE 9.2%). Considering the variables of Table 3, significant univariate risk factors for post-PTLD mortality included rejection therapy with high-dose steroids (HR 3.00, (1.23, 7.28), p = 0.0160), as well as the well-known clinical risk factors of mortality in non-transplant lymphoma patients of International Prognostic Factors (IPF, HR 1.95, (1.28, 2.97), p = 0.0015), Stage (HR 2.11, (1.29, 2.59), p = 0.013), Performance Score (HR 1.83, (1.29, 2.59), p = 0.0011), lactate deydrogenase (LDH) levels (per 100 U/L unit increase HR 2.97 (1.46, 6.06), p = 0.0005), LDH > 250 U/L (HR 4.84, (1.18, 19.95), p = 0.036). The risk factors polymorphic (HR 3.01, (0.84, 8.5), p = 0.084) and IPF ≥ 3 (HR 2.52, (0.97, 6.30), p = 0.058) were marginally statistically significant. No other factors such as age (per 10 years HR 0.82 (0.57, 1.18), p = 0.29), male (HR 0.59, (0.25, 1.43), p = 0.25), etiology of chronic hepatitis (HR 2.10, (0.85, 5.18), p = 0.11), primary sclerosing cholangitis or primary biliary cirrhosis (HR 0.58, (0.22, 1.52), p = 0.26), alcohol liver disease (HR 0.49, (0.06, 4.02), p = 0.47), acute fulminant hepatitis (HR = 1.74 (0.39, 7.70), p = 0.49), other etiology (HR 0.68, (0.16, 2.95), p = 0.86), CMV positive serology at discharge (HR 2.14 (0.82, 5.58), p = 0.13), CMV mismatch (HR 0.89, (0.25, 3.13), p = 0.86), EBV-positive PTLD (HR 2.04, (0.66, 6.27), p = 0.19), time of PTLD since OLT (HR 0.96 per year, (0.86, 1.08), p = 0.51), PTLD within 1 year of OLT (HR 0.86, (0.30, 2.46), p = 0.78), immunsuppressive therapy during follow-up of FK506 vs. cyclosporine (HR 1.27, (0.49, 3.31), p = 0.63), mycophenolate mofetil vs. azathioprine (HR 1.26 (0.28, 5.62), p = 0.76), OKT3 (HR 2.05, (0.77, 5.44), p = 0.17), PTLD characteristic of monoclonal (HR 0.67, (0.14, 3.12), p = 0.62), nodal (HR 1.93, (0.76, 4.89), p = 0.44), extranodal (1.36, (0.52, 3.56), p = 0.52), in multiple organs (HR 2.06, (0.79, 5.37), p = 0.15), in grafted organ (HR 2.09, (0.86, 5.10), p = 0.11), anti-CD20 positive (HR 0.39, (0.09, 1.60), p = 0.22), patient received rituxan (HR 1.20, (0.27, 5.44), p = 0.81) were significant at the 0.10 level.

Figure 3.

Kaplan-Meier curves showing the survival following diagnosis of PTLD as a function of the lymphoma EBV status. The curve for all EBV types includes individuals whose lymphomas were not typed and the survival curve is greater than 0 beyond 144 months. The difference in survival between the EBV+ and EBV− groups is not significant (p = 0.19).

Discussion

Previous reports of PTLD following OLT in adults have reported frequencies in the 2% to 3.5% range (10). Using competing risk survival analysis, we observed a cumulative incidence of 5.4% at 15 ½ years post-transplant. Detailed assessment of the cumulative incidence curves for PTLD in OLT recipients indicated that the incidence of PTLD is highest during the 18 months following OLT, whereas the incidence is relatively constant beyond the 18-month period. The wide range of reported frequencies of PTLD in earlier reports (10) stems in part from differences in length of follow-up, and the inadequacy of reporting proportions, as opposed to accounting for length of follow-up and the competing risk of death in estimating PTLD incidence. If we were to put the risk of PTLD into perspective by comparing with all cause mortality, the risk of PTLD appears to be approximately 10% to 15% of the risk of death without PTLD. If the risk of PTLD in relation to all cause mortality should remain constant beyond 15 years then, the incidence of PTLD following OLT could exceed the 5.4% incidence observed in our study.

Interestingly, the usual Kaplan-Meier method for estimating incidence of PTLD yields an overestimate of the true incidence whereas the naïve proportion is typically an underestimate (29). However, the competing risk approach, which simultaneously estimates the two ‘competing’ outcomes of PTLD and death without (before) PTLD yields an accurate estimate of the probability of PTLD as a function of time since transplant. Heuristically, the competing risk method provides estimates of the fraction of patients with PTLD and of the fraction of patients dead without PTLD that one would expect as a function of time if one could follow all patients indefinitely. In contrast, by treating death as a censoring mechanism, the Kaplan-Meier procedure estimates the probability of a patient developing a PTLD conditional upon the patient not dying until after developing a PTLD. Though more involved in its calculation, the competing risk cumulative incidence estimates are then more directly interpretable than those of the Kaplan-Meier procedure.

Earlier studies reported that the frequency of EBV positive PTLD is markedly higher than that of EBV negative PTLD (31). However, more recent studies report an increasing frequency of EBV negative PTLD (11,16–19). Our findings indicate that early PTLD are predominantly EBV positive whereas the late PTLD are EBV negative. The cumulative incidences of the two types of PTLD at 10 years and beyond are similar. Therefore, it is likely that the recent reports of increasing frequency of EBV negative PTLD result from the greater number of OLT recipients followed-up beyond 10 years or more. The recently observed increase in EBV negative PTLD is not likely a reflection of a changing disease pattern (16,17) but rather, a consequence of the increase in length of follow-up for the liver transplant recipients. The number or EBV negative PTLD are likely to increase in the future. Furthermore, EBV negative PTLD were previously reported to be more aggressive and associated with higher mortality than EBV positive PTLD. We found no indication for this to be the case. Both groups had about 40% mortality in the first year subsequent to diagnosis and had about a 50% survival at 5 years. Our results here are consistent with earlier reports from our institution considering all solid organ transplant programs where survival following diagnosis of PTLD in patients diagnosed within 1 year of transplant and diagnoses later were similar (32).

Treatment for rejection with high-dose steroids and OKT3 was associated with subsequent development of PTLD, but treatment with polyclonal antibodies was not. Interestingly, the risk of PTLD subsequent to OKT3 or steroid therapy was elevated for the 1-year period following treatment, but the effect disappeared thereafter. Similar observations as regards to the type, length and intensity of immunosuppressive drug therapies as risk factors for PTLD have been reported extensively (33–35). Yet, it is still unclear whether the relationship is causal or reflects is confounding by indication and disease severity (36). Lack of consistency of risk estimates across various immunosuppressive regimens in different studies, and the lack of temporal trends despite significant therapeutic changes over time (i.e. ecological evidence) suggest that the risk of PTLD could be, at least partially, related to the whole transplantation process and the immunosuppression, and not a specific immunosuppressive regimen per se (11). Therefore, high-dose steroids and OKT3 may be markers of immunosuppressive state as well as causal risk factors for PTLD. Unlike in kidney and cardiac transplants, rejection in liver transplantation is sometimes suggested not to impact patient survival or risk of graft loss (37). Given our findings that treatment for rejection does put the patients at higher risk for PTLD and an associated higher risk of mortality, it is important to balance the risk of greater levels of prophylactic immunosuppression against the risk of PTLD following rejection treatment. Further research using more specific study designs (38) is needed to elucidate the potential role of individual immunosuppressive agents.

Several previous studies have suggested that the type of disease that had led the patient to transplantation may be important in the risk of PTLD development (39,40). In OLT recipients, a preexisting autoimmune hepatitis, primary biliary cirrhosis was suggested to increase the PTLD risk (40). Although the number of patients was relatively low, we found a significant association between AFH as an indication for OLT and development of PTLD during the first 18 months following OLT. Jain et al. (23) also reported a higher occurrence of PTLD following OLT for AFH in 5.6% of patients as compared to 2.9% for adults overall, but their AFH group also included children. An impaired immune system as a result of the virulent nature of AFH with delayed recovery of the immune system remains a possible explanation. This finding warrants further investigation.

Expectedly EBV pre-transplant serostatus and in particular EBV donor positive and recipient negative (EBV mismatch) status is a significant risk factor for EBV positive PTLD (41). A limitation of this study is the unavailability of EBV serology on all but very recent transplants making a proper assessment impractical. Future work should consider the impact of EBV mismatch on PTLD risk. Further, given the possibly reduced risk of early onset PTLD incurred with use of antiviral therapies distinction in PTLD risk should also be made between patients treated with antiviral therapies and those not treated (42).

In conclusion, earlier reports of a lower incidence of PTLD late after OLT are likely to be due to fewer patients being followed for a prolonged period of time. With increased survival and longer follow-up of OLT recipients, the incidence of PTLD and especially, the number of EBV negative PTLD relative to the total number of PTLD, are likely to increase in the future. Significant risk factors for PTLD are treatment for rejection with high-dose steroids and OKT3 indicating the need to balance the risk of greater levels of prophylactic immunosuppression against the possible risk of PTLD following rejection treatment. Also, patients transplanted for AFH may be at increased risk for PTLD within the first 18 months post-transplantation. Upon diagnosis PTLD patients are at increased risk of mortality especially during the first year following diagnosis. Survival probability at 5 years was approximately 50%. The most important predictors of survival upon diagnosis of PTLD were the usual predictors of survival in lymphoma patients, though previous rejection treatment may also be a risk factor. Given the current advances in gene expression profiling and proteomics, these methods show promise for providing stronger predictive accuracy and identifying treatments to patient-specific disease (43).

Acknowledgment

T. M. Habermann received honoraria from Genentech, Inc.

Ancillary