We addressed the effect of post-transplant lymphoproliferative disorder (PTLD) treatment with rituximab monotherapy or CHOP-based chemotherapy (± rituximab) after upfront immunosuppression reduction (IR) on renal graft function in a longitudinal analysis of 58 renal transplant recipients with PTLD and 610 renal transplant controls. Changes in the estimated glomerular filtration rate over time were calculated from a total of 6933 creatinine measurements over a period of >1 year using a linear mixed model with random and fixed effects. Renal graft function significantly improved with treatment of PTLD, especially in the chemotherapy subgroup. Patients treated with IR+chemotherapy ± rituximab had a noninferior graft function compared with untreated controls suggesting that the negative impact of IR on the renal graft function can be fully compensated by the immunosuppressive effect of CHOP. The immunosuppressive effect of single agent rituximab may partially compensate the negative impact of IR on the graft function. Thus, it is possible to reduce immunosuppression when using chemotherapy to treat PTLD.
Post-transplantation lymphoproliferative disorder (PTLD), a spectrum of lymphatic diseases associated with the use of potent immunosuppressive drugs following transplantation, represents one of the most common neoplastic diseases following solid organ transplantation (1). Patients with PTLD commonly present with stage III or IV disease, of varying histologies, frequently involving extranodal sites (2–6). Organ dysfunctions, especially renal or liver function impairment, are also a common feature at presentation (7). Initial therapy in PTLD is immunosuppression reduction (IR). Dose reductions of immunosuppressive drugs can result in complete remission (CR) (8,9), but response rates are low (10). Additionally, IR is known to be associated with a high risk of graft rejection and graft loss, and a rejection rate of 37% has recently been reported in a first prospective trial systematically evaluating IR in PTLD (10). Subsequent therapies in patients failing to respond significantly to upfront IR are rituximab monotherapy in CD20-positive B-cell PTLD (11–14), CHOP-like chemotherapy (15–19) and a combination of rituximab and CHOP, either synchronous or in sequence. Both options are highly effective in the treatment of PTLD after an initial failure of upfront IR. Overall response rates of about 40–60% for single agent rituximab (11–14) and up to 90% for sequential treatment with four courses of rituximab followed by four cycles of CHOP have been reported (20). While none of the three prospective clinical trials of rituximab monotherapy after failure of IR reported an increased risk of graft failure (11–13); a systematic evaluation of graft function is lacking. Published data on CHOP-based chemotherapy in PTLD are usually retrospective and focus on treatment efficacy rather than on graft survival. However, an increased incidence of graft failure after chemotherapy has not been reported so far and many clinicians would agree that graft rejection occurs less frequently after IR + chemotherapy than after IR alone. To assess in more detail the effect of subsequent therapies on graft function after upfront IR we performed a retrospective study in 58 renal transplant recipients requiring first-line treatment with rituximab monotherapy or CHOP-based chemotherapy (± rituximab) after failing to respond to IR for newly diagnosed PTLD. Renal transplant function was monitored for a period of >1 year after PTLD diagnosis and compared with that of 610 renal transplant controls matched for age, sex, time since transplantation and transplant function at start of follow-up.
Patient data collection
This study was conducted using pooled data of adult renal transplant recipients from two prospective, international, multicenter trials of first-line treatment of PTLD in patients not responding to IR (20,21), and single patient data from first-line therapy in adult renal transplant recipients prospectively reported to the German PTLD registry (PTLD D2006–2012). Regular reporting on renal function is included in both trials and in the registry at start of therapy and at least at 4 weeks, 6 months and 12 months after completing therapy. Additional data on patient characteristics were retrieved from databases of the different transplant centers. These were cause of end stage renal disease (ESRD), number of previous transplantations, donor age, donor type (living/deceased), number of HLA mismatches, cold-ischemia-time (CIT), initial transplant function after transplantation, type of immunosuppression and trough levels before and after diagnosis of PTLD.
Diagnosis of PTLD
PTLD diagnosis was based on an examination of histological material, obtained by open biopsy or core needle biopsy. Routinely, diagnostic tissue samples were subsequently reviewed at a central institution by a single expert pathologist and were classified morphologically according to the WHO classification (2004). In order to detect lymphoproliferative sites, all patients with diagnosed PTLD underwent a physical examination, blood tests (white blood cell [WBC] count, biochemical tests, liver tests and lactate dehydrogenase [LDH] assay, upper limit 248 U/L), bone marrow biopsy and computed tomography (CT) scans of the neck, chest, abdomen and pelvis.
The PT-LPD-1 study, instigated in 1999, was a prospective phase II trial evaluating efficacy and safety of single agent rituximab in PTLD. Patients for whom IR had provided no benefit received four once-weekly doses of rituximab (375 mg/m2) (11). In 2003, we commenced the multicenter PTLD-1 phase II trial using sequential treatment comprising four courses of rituximab followed by four courses of standard CHOP chemotherapy as a first-line treatment in PTLD unresponsive to IR (20). This trial had a major amendment in 2006 introducing risk-stratified sequential treatment (RSST). Under the RSST-approach, patients with a CR after four courses of rituximab monotherapy were regarded as ‘low risk’ while patients with a partial response (PR), stable disease (SD) or progressive disease (PD) were regarded as ‘intermediate/high risk’. Low-risk patients were treated with four further courses of rituximab monotherapy (once every 3 weeks). High-risk patients subsequently received four courses of R-CHOP. Finally, the PTLD D2006–2012 trial is a prospective national registry for rare PTLD subtypes and relapsed PTLD using different treatment protocols for distinct subtypes. Patients included in this analysis either had CD20-negative B-cell PTLD, T-cell PTLD, Hodgkin or Hodgkin-like PTLD or plasmocytic PTLD. Treatment protocols included CHOP (cyclophosphamide, hydroxydaunorubicine, oncovin, prednisolone), ABVD (adriamycin, bleomycin, vinblastine, dacarbazine) and VAD (vincristin, adriamycin, dexamethasone). The responsible local ethical committees approved all trials and all patients gave written informed consent according to the Declaration of Helsinki.
Evaluation of renal function
Renal transplant function was assessed by creatinine analysis. Creatinine values were assessed at diagnosis of PTLD (prior to start of treatment), 4 weeks after completing treatment (2–8 months after diagnosis of PTLD) and 6 months after completing treatment (8–14 months after diagnosis of PTLD). Estimated glomerular filtration rate (eGFR) was determined by the four-variable modification of diet in renal disease (MDRD) study equation. Stage 1 chronic kidney disease (CKD) was defined as eGFR >90 mL/min/1.73 m2. Stage 2 CKD was eGFR 60–89 mL/min/1.73 m2; Stage 3 CKD was eGFR 30–59 mL/min/1.73 m2; stage 4 CKD was eGFR 15–29 mL/min/1.73 m2; stage 5 CKD was eGFR <15 mL/min/1.73 m2 or permanent renal replacement therapy.
In total, 58 renal transplant recipients with PTLD were included: 9 from the PT-LPD-1 trial, 38 from the PTLD-1 trial and 11 from the PTLD registry. The mean patient age was 49.0 years at diagnosis of PTLD (range 18.3–79.8 years). The male to female ratio was 17: 41. Further details on patient characteristics are given in Table 1. Immunosuppression was reduced in all patients immediately after diagnosis of PTLD. 18/58 patients (31.0%) subsequently received monotherapy rituximab, 40/58 (69.0%) patients received CHOP or CHOP-like chemotherapy, 31 of these 40 patients also received rituximab (i.e. R-CHOP or sequential R-CHOP). The complete remission rate in this patient cohort was 69%. Treatment related mortality was observed in 12% of patients and was due to septic complications after chemotherapy. 8/58 patients (13.8%) died from PTLD and 2/58 (3.4%) from other reasons (1 myocardial infarction, 1 unknown).
Table 1. Baseline characteristics of patients and controls
Mean time from transplantation to PTLD / study entry (months)
88.0 ± 67.9
77.3 ± 44.6
Early PTLD (within the 1st year after transplantation)
Late PTLD (later than 1 year after transplantation)
Histology of PTLD
DLBCL, Burkitt or Burkitt-like PTLD
Hodgkin or Hodgkin-like PTLD
Other B-cell PTLD
Stage of disease
Limited (stage I or II)
Advanced (stage III or IV)
Renal graft involvement
Mean patient age at study entry (years)
49.0 ± 14.6
50.7 ± 11.2
Cause of ESRD
Metabolic dysfunction other than diabetes mellitus
Cystic renal degeneration
Others or unknown
Mean donor age (years)1
37.7 ± 17.7
39.5 ± 15.0
Living donor transplantation (yes/no)2
Combined transplantation (pancreas/kidney,
2nd, 3rd, 4th renal transplantation
2.1 ± 1.9
2.2 ± 1.7
1.6 ± 1.7
1.8 ± 1.6
Mean CIT (hours)4
14 ± 8
14 ± 7
Initial transplant function (yes/no)5
Mean CNI plasma levels (prior to diagnosis of PTLD)6
Cyclosporin A (ng/ml)
134 ± 37
117 ± 60
8.8 ± 3.9
6.9 ± 2.0
Mean eGFR at start of treatment (mL/min/1.73 m2)
48.4 ± 20.5
48.7 ± 18.9
Renal graft loss during the study period
From all renal transplant patients (n = 1212) consecutively admitted or referred from other German transplant units to the outpatient unit at the Charité University Hospital (Campus Virchow-Klinikum) in Berlin between January 1, 1998 and July 30, 2008, a frequency-matched control group was established. 610 patients were selected by automated, computerized process. Matching was performed for sex, age, time since transplantation and transplant function at start of follow-up. Subsequently, patient characteristics and creatinine levels over time were retrieved from the database. Control patients’ creatinine levels were measured at each visit and the number of visits differed depending on the clinical situation. Stable kidney patients were usually followed up in 3-month intervals. Renal function data for the first 15 days after transplantation were removed as well as creatinine data after day 460. This aligned the control group follow-up with the PTLD cohort, in which the earliest PTLD diagnosis was made at postoperative day 15 and the longest follow-up was until postoperative day 460. This finally resulted in 6784 single data points representing the time course of renal graft function over the study period of 1.3 years. For patients who recommenced hemodialysis, creatinine values were removed in both groups at the start of dialysis. The incidence of graft failure was 3.9% in the control group and 5.2% in the PTLD cohort (p = 0.722).
Frequencies of patient characteristics were estimated from the observed data and categorical variables were compared by the χ2 test with the Fisher exact test, where available. Continuous variables of patient characteristics were compared with controls using the t-test for independent samples with a previous test for equality of variances. When normality for the continuous variables could not be assumed using the corresponding normality test, we used the Mann–Whitney nonparametric test. Linear mixed effect models (short regression analysis) (22) were performed to calculate statistical significance to assess treatment differences over time. Treatment was included as a covariate together with an interaction term for time and treatment. Linear mixed effect models offer the advantage of allowing the investigation of variability between patients (heterogeneity) and simultaneously adjusting for correlation within subjects. In our analysis we allowed random effects for intercepts and for the effect of time. This implies that the change over time may vary between patients. As several potentially confounding factors might impact on the renal graft function, model reduction based on likelihood-ratio tests was performed in order to find the model that provided the best fit to our observational data with the smallest set of variables. Model regression coefficients are reported together with standard error estimate (SE). The level of significance in all cases was set at p < 0.05. Statistical tests were run on SPSS (version 16.0, SPSS Inc., Chicago, IL), SAS (version 9.2, SAS Institute GmbH, Heidelberg, Germany) and R (version 2.8, R Development Core Team).
Renal transplantation for end stage renal disease was either due to autoimmune disorders (26/58), diabetes (4/58), cystic renal degeneration (6/58) or metabolic dysfunction (1/58). In 21/58 patients the reason for end stage renal disease was unknown. Seven patients had a combined renal transplantation (4 kidney + pancreas, 3 kidney + heart), while 51 received a single kidney transplant. Seven patients had more than one renal transplantation. Additional patient characteristics are summarized in Table 1.
Reduction of immunosuppression and type of subsequent therapy
At diagnosis of PTLD, 30/58 patients were undergoing intensive immunosuppressive triple therapy. Of these, 24 were receiving calcineurin-inhibitor (CNI) + mycophenolate mofetil (MMF) or azathioprin (AZA) + steroid, two were receiving mTOR-inhibitors + MMF + steroid and four were receiving mTOR inhibitor + CNI + steroid. Of the 17/58 patients undergoing dual immunosuppressive therapy, 9 received CNI + MMF/AZA and 8 received CNI + steroid. 6/58 patients did not have CNI or mTOR inhibitors but had immunosuppression with MMF/AZA +steroid. 5/58 patients had single agent immunosupression with CNIs or mTOR-inhibitors.
After diagnosis of PTLD, immunosuppression was reduced in 57/58 patients. CNIs were stopped in 14 patients. Five patients had major dose reductions of CNIs (i.e. a reduction of 50–75%), 8 had minor dose reductions (i.e. a reduction of 25–49%). CNI plasma levels before and after IR were evaluable in a subset of 18 of 34 patients. In these patients, mean tacrolimus and mean cyclosporin A plasma levels had been reduced by 35 and 30%, respectively. 20/40 patients discontinued MMF/AZA, 3/40 had reductions. Previous treatment with mTOR inhibitors was usually maintained or increased while comedication was reduced or stopped. Details on immunosuppression and IR are given in Table 2.
Table 2. Reduction of immunosuppression (IR) after diagnosis of PTLD
Number of patients at diagnosis of PTLD
Number of patients after IR
1In patients receiving CNIs before and after IR.
Single agent immunosuppression
Mean cyclosporin plasma levels1
Mean tacrolimus plasma levels1
Subsequent therapy after failing to respond to IR was single agent rituximab in 18 patients and CHOP or CHOP-like chemotherapy ± rituximab in 40 patients. The mean number of single agent rituximab doses received was 5.5 (range: 2–8). In patients treated with chemotherapy ± rituximab, the mean numbers of cycles of chemotherapy received were 4.4 (range: 1–8) and 3.6 (range: 0–9) respectively. 9/40 patients in this group did not receive rituximab.
Renal graft function before and after treatment of PTLD
After treatment of PTLD 12/58 patients died within one year of diagnosis. Three of the 46 PTLD survivors recommenced hemodialysis. Of these, one patient who had been treated with rituximab monotherapy suffered from a hemolytic uremic syndrome. He had received a simultaneous pancreas–kidney transplantation and developed DLBCL-type PTLD confined to the grafted pancreas. While reduction of immunosuppression and single agent rituximab were not successful, explantation of the allografts resulted in CR (23). Two further patients with a pretherapeutic eGFR of 10.6 mL/min/1.73 m2 and 15.7 mL/min/1.73 m2 required hemodialysis shortly after chemotherapeutic treatment of PTLD. In contrast to these cases, 10/18 patients who received single agent rituximab and 25/40 patients who received chemotherapy ± rituximab had improved renal function. A further 6/18 patients who received rituximab and 12/40 who received chemotherapy ± rituximab showed no relevant changes in renal graft function. After completing PTLD therapy the 55 surviving patients had the following distribution of renal graft function (change from baseline): 4 CKD 1 (+2), 15 CKD 2 (–3), 29 CKD 3 (+2), 5 CKD 4 (–3), 2 CKD 5 (–1). There was a significant improvement in mean eGFR values 4 weeks after treatment compared with values taken before treatment (paired t-test: N = 53, mean change in eGFR: +7.2 mL/min/1.73 m2, p = 0.0017). Subgroup analyses for the different treatment groups, however, showed a significant improvement of renal graft function only in patients who had been treated with chemotherapy (paired t-test: N = 39, mean change in eGFR before versus 4 weeks after therapy: +9.7 ml/min/1.73 m2, p = 0.023).
Changes in renal graft function of patients and controls
To assess treatment differences in the course of time and in comparison to controls, a linear mixed effect model was applied. Model reduction identified a model including time, donation type, treatment and time-dependent treatment effects as fixed effect parameters and random effects to provide the best fit to the observational data with the smallest set of variables (p < 0.0001, Table 3). The subsequent solution for fixed effects identified (i) a significant deterioration of renal graft function with time (–2.01 ± 0.91 ml/min/1.73 m2/year, p = 0.0278), (ii) a significant improvement of renal graft function with time for patients who had received chemotherapy (+10.84 ± 4.21 ml/min/1.73 m2/year, p = 0.0100) and (iii) a statistically nonsignificant benefit with time for patients who had received rituximab monotherapy (p = 0.7276). Figure 1 summarizes the individual eGFR values of both treatment groups together with their local means. As the question of validity of the comparator group is crucial, particular potential imbalances were addressed. More specifically: (i) as subjects of living donor transplantation were more frequent in the PTLD cohort than in the control group, we added the interaction term for time and living donor transplantation to the mixed linear regression model shown in Table 3 resulting in an extended test model. This model failed, however, to identify any significant effect on the rate of change in graft function over time, by donation type (p = 0.1503). Meanwhile, time-dependent effects of chemotherapy stayed significant. Thus, the higher proportion of living donor transplantation in the PTLD cohort does not independently impact on treatment effects. (ii) Because patient death is more frequent in the PTLD cohort, we analyzed the graft function after treatment of PTLD in the subgroup of PTLD survivors only. This identified a significant increase of mean eGFR values with treatment of PTLD as already described for the total PTLD cohort (p = 0.0019). The effects were similar with an improvement of +4.93 and +7.69 ml/min/1.73 m2, respectively. Thus, there is no bias by a more frequent patient death in the patient group. All other parameters were well balanced.
Table 3. Linear mixed effect model calculation of the renal transplant function in follow-up
Fixed effects (eGFR ∼ time + treatment + subject of living donor transplantation + treatment × time)
Treatment 2 (chemotherapy ± rituximab) × time (years)
This is the first study to analyze renal transplant function over time in the context of IR followed by either rituximab monotherapy or CHOP-based chemotherapy. Currently, there is little or no standardization of either how IR should best be realized or of the most suitable endpoints to the intervention. This means immunosuppression is usually combined with other treatments, and although response rates to IR are low (10) it is difficult to eliminate a possible immune benefit effect. Hematotoxicity, leading to infectious complications and significant treatment-related mortality is the most important therapeutic limitation of CHOP. Anti-CD20 monotherapies are therefore the preferred treatment modality in PTLD when patients fail to respond significantly to upfront IR, although they seem to be less effective in high-risk patients (14). Optimizing IR in the context of subsequent therapies therefore might be crucial to improve efficacy and to reduce toxicity of subsequent treatment lines. Here, we have shown that IR followed by either single agent rituximab or CHOP-based chemotherapy ± rituximab is not associated with an increased risk of renal graft impairment compared with IR alone. In contrast, renal graft function slightly, but significantly improved following treatment of the lymphoma. Compared with renal transplant controls showing stable graft function, PTLD patients receiving chemotherapy had a noninferior graft function at follow-up. This suggests that the negative impact of IR on graft function can be compensated by the immunosuppressive effect of CHOP, suppressing both cell-mediated and humoral immunity through its effects on T and B cells (24). Conversely, patients with PTLD may be overimmunosuppressed. Whether the compensatory immunosuppressive effect of CHOP is potent enough to offer the possibility of stopping all immunosuppressive therapy during the time of chemotherapeutic treatment remains unclear, although these data encourage a more intensive IR when patients receive CHOP in order to reduce treatment-related mortality. For patients treated with IR + single agent rituximab noninferiority to the control group could not be proven, but follow-up graft function was not significantly worse than for controls. The immunosuppressive effect of rituximab, which acts mainly on B cells, may only be strong enough to partially compensate for IR. Peripheral B cell depletion strategies, however, have been successfully employed in the treatment of systemic autoimmune diseases, and recent studies have suggested that transient B cell depletion may act on non-B cell populations (25). In conclusion, IR + single agent rituximab and IR+CHOP-based chemotherapy ± rituximab are highly effective treatment modalities when patients fail to respond to IR alone. Chemotherapy ± rituximab and to a lesser extent rituximab monotherapy stabilize the graft function as both have immunosuppressant activity. IR therefore is a viable strategy in the context of subsequent treatment with rituximab and CHOP.
This project has been supported by grants from AMGEN, CLS Behring, Mundipharma GmbH and Roche Pharma AG to RT. The DPTLDSG is a member of the German Competence Network Malignant Lymphomas (KML). Financial support for scientific and educational presentations has been received from CLS Behring and Roche Pharma AG by RT. All other authors declare that no individual support has been received and that there is no relationship with any company in the context of this study.