16 Department of Nephrology and Intensive Care, Campus. 17 Department of Abdominal and Transplant Surgery,

Transcription

1 American Journal of Transplantation 2015; 15: Wiley Periodicals Inc. Brief Communication C Copyright 2015 The American Society of Transplantation and the American Society of Transplant Surgeons doi: /ajt International Prognostic Index, Type of Transplant and Response to Rituximab Are Key Parameters to Tailor Treatment in Adults With CD20-Positive B Cell PTLD: Clues From the PTLD-1 Trial R. U. Trappe 1,2, *, S. Choquet 3, D. Dierickx 4, P. Mollee 5, J. M. Zaucha 6, M. H. Dreyling 7, U. Dührsen 8, C. Tarella 9, O. Shpilberg 10, M. Sender 11, G. Salles 12, F. Morschhauser 13, A. Jaccard 14, T. Lamy 15, P. Reinke 16, R. Neuhaus 17, H. Lehmkuhl 18, H. A. Horst 1, M. Leithäuser 19, P. Schlattmann 20, I. Anagnostopoulos 21, M. Raphael 22, H. Riess 2, V. Leblond 3 and S. Oertel 23 for the German PTLD Study Group and the European PTLD Network 1 Department of Internal Medicine II: Hematology and Oncology, University Medical Centre Schleswig-Holstein, Campus Kiel, Kiel, Germany 2 Department of Hematology, Oncology and Tumor Immunology Charité Universitätsmedizin Berlin, Campus Virchow-Klinikum, Berlin, Germany 3 Département d Hématologie, Hopital Pitie-Salpêtriere, Université Pierre et Marie Curie, Paris, France 4 Department of Hematology, University Hospital Gasthuisberg Leuven, Leuven, Belgium 5 Department of Hematology, Pathology Queensland, Princess Alexandra Hospital, Brisbane, Australia 6 Department of Propedeutic Oncology, Poland and Polish Research Lymphoma Group, Medical University of Gdansk, Gdansk, Poland 7 Department of Internal Medicine III, University of Munich Campus Großhadern, Munich, Germany 8 Department of Hematology, Essen University Hospital, University of Duisburg-Essen, Essen, Germany 9 Department of Hematology, A.O. Mauriziano, University of Torino, Torino, Italy 10 Institute of Hematology, Rabin Medical Centre, Petach Tikva and Sackler Faculty of Medicine, Tel Aviv University, Tel Aviv, Israel 11 Department of Hematology, Sahlgrens Hospital, Gothenburg, Sweden 12 Hospices Civils de Lyon, Université de Lyon, Centre Hospitalier Lyon-Sud, Département d Hématologie, Pierre-Benite, France 13 Département d Hématologie, Centre Hospitalier Universitaire, Lille, France 14 Département d Hématologie, Centre Hospitalier Universitaire, Limoges, France 15 Département d Hématologie, Centre Hospitalier Universitaire, Rennes, France 16 Department of Nephrology and Intensive Care, Campus Virchow-Klinikum, Charité Universitätsmedizin Berlin, Berlin, Germany 17 Department of Abdominal and Transplant Surgery, Campus Virchow-Klinikum, Charité Universitätsmedizin Berlin, Berlin, Germany 18 Department of Cardiothoracic and Vascular Surgery, German Heart Institute Berlin, Berlin, Germany 19 Medical Department III Hematology, Oncology, Palliative Care, University of Rostock, Rostock, Germany 20 Department of Medical Statistics, Computer Sciences and Documentation, Jena University Hospital, Jena, Germany 21 Department of Pathology, Charité Universitätsmedizin Berlin, Campus Mitte, Germany 22 Service d Hématologie, Immunologie Biologiques et cytogénétique, Hôpital Bicêtre, Université Paris-Sud, Le Kremlin-Bicêtre, France 23 AbbVie Deutschland GmbH & Co. KG, Wiesbaden, Germany Corresponding author: Ralf Ulrich Trappe, r.trappe@diako-bremen.de Tailoring treatment by patient strata based on the risk of disease progression and treatment toxicity might improve outcomes of patients with posttransplant lymphoproliferative disorder (PTLD). We analysed the cohort of 70 patients treated in the international, multicenter phase II PTLD-1 trial (NCT ) to identify such factors. Of the previously published scoring systems in PTLD, the international prognostic index (IPI), the PTLD prognostic index and the Ghobrial score were predictive for overall survival. None of the scoring systems had a considerable effect on the risk for disease progression. Age and ECOG performance status were the baseline variables with the highest prognostic impact in the different scoring systems. Baseline variables not included in the scoring systems that had an impact on overall survival and disease progression were the type of transplant and the response to rituximab at interim staging. Thoracic organ transplant recipients who did not respond to rituximab monotherapy were at particularly high risk for death from disease progression with subsequent CHOP-based chemotherapy. Patients in complete remission after four courses of rituximab and patients in partial remission with low-risk IPI had 1091

2 Trappe et al a low risk of disease progression. We speculate that chemotherapy might not be necessary in this patient cohort. Abbreviations: CHOP, chemotherapy with cyclophosphamide, doxorubicin, vincristine and prednisone; CR, complete remission; EBV, Epstein Barr virus; ECOG, performance status Eastern Oncology Group performance status; HIV, human immunodeficiency virus; HR, hazard ratio; IPI, international prognostic index; IV, intravenous; LDH, serum lactate dehydrogenase activity; OS, overall survival; PD, progressive disease; PFS, progression free survival; PO, peroral; PR, partial remission; PTLD, posttransplant lymphoproliferative disorder; SD, stable disease Received 24 December 2013, revised 13 October 2014 and accepted for publication 19 October 2014 Introduction Posttransplant lymphoproliferative disorder (PTLD), a spectrum of lymphatic diseases associated with the use of potent immunosuppressive drugs after transplantation, ranges from polyclonal early lesions associated with primary Epstein Barr virus (EBV) infection to monomorphic lymphoma (1). The PTLD-1 trial, the largest prospective phase II trial in the field so far, has demonstrated the efficacy and safety of sequential therapy (rituximab followed by CHOP chemotherapy) with an overall response rate of 90% and 6.6 years median overall survival (2). Due to the risk of treatment-related complications, particularly infections in chronically immunosuppressed transplant recipients, tailoring treatment to patients according to their risk profile could potentially improve outcomes even further. PTLD are clinically different from lymphoma in the immunocompetent due to their higher incidence (3), their frequent association with Epstein Barr virus and their good response to immunotherapy and chemotherapy (2,4). Furthermore, extranodal disease and advanced stage, two of the risk factors included in the international prognostic index (IPI), (5) are very common in PTLD (2). In an attempt to take these differences from lymphoma in the general population into account, a steadily growing number of prognostic indices have been put forward in PTLD by groups in Australia, Europe and the US (6 12). The differences in the selection criteria of the underlying cohorts and the changes in treatment over time are reflected in different results. Due to the rarity of PTLD, the majority of these studies have been limited by either their retrospective or single-institution design (6,7,9,10,12), or non-standardized treatment (6,7,10 12). Furthermore, while some included PTLD both after solid organ transplantation (SOT) and after hematopoietic stem cell transplantation (12), others included solid organ transplant recipients (6 8,10) or only kidney transplant recipients (9,11). Despite different results and numerous scoring systems, there were some trends across cohorts: poor ECOG performance status was repeatedly identified as a prognostic factor for overall survival in univariate (6 10,12) and multivariable analysis (6 8); age (8,11,12) and elevated LDH (8,9,11) were similarly prognostic in multivariable analyses in three publications each. Extranodal disease (7,10,12) and CNS involvement (6,10,11) were significant predictors of overall survival in three studies in univariate analysis. Regarding prognostic scoring systems, the international prognostic index (IPI) (5) was a significant predictor of overall survival in four cohorts (7 9,12), whereas the Leblond prognostic index (6,9) and the PTLD prognostic index (8,9) were significant in two cohorts. Heart and lung transplant recipients have an increased risk of PTLD (13), and poor overall survival has been described in patients with PTLD after lung transplantation in particular (14,15). In addition, early response during treatment has been identified as a parameter for treatment stratification and is associated with time to progression and overall survival (2). Methods Patients The current analysis is based on the published dataset of the PTLD-1 trial (NCT , N ¼ 70) (2). Data from the German and French rituximab monotherapy trials (N ¼ 60) (8) were used for comparison. Treatment and patient characteristics are summarized in Table 1. Data analysis Exploratory univariate analyses were performed by applying Cox-regression models for time-to-event outcomes and x 2 tests to categorical variables. Multivariable analyses were performed using Cox-regression models (likelihood ratio test, backward elimination). The performance of the models was accessed by calibration and discrimination. Discrimination denotes a model s ability to separate patients with events from those without events and was quantified using Somer s D for censored data. Somers D has a range from 1 toþ1, with 1 for a perfect negative correlation, 0 for no correlation and þ1 for a perfect positive correlation. Calibration refers to the extent of bias. Ideally, observed and predicted survival at a certain time would be identical. Here, we used the difference between the observed and the predicted 2-year survival probability based on a flexible hazard regression approach. Internal model validation was performed using bootstrapping. The procedure allows to report mean calibration errors together with corresponding 0.9 quantiles (16,17). The level of significance was set at p < Statistical tests were performed using IBM SPSS 21 and SAS (version 9.2, SAS Institute GmbH, Heidelberg, Germany). Model validation was performed with R (18) using the R-package rms (19). Response to treatment and disease progression were classified according to the World Health Organization criteria. Overall response was complete or partial remission. Time to progression was time from start of treatment to disease progression after interim staging with censoring of patients who died before restaging. Overall survival was time from start of treatment to death from any cause. Treatment-related mortality was any death due to treatmentrelated complications American Journal of Transplantation 2015; 15:

3 Prognostic Factors in the PTLD-1 Trial Table 1: Type of treatment and baseline characteristics PTLD-1 trial (2) Key inclusion criteria: Solid organ transplant recipients with CD20-positive PTLD unresponsive to a reduction of immunosuppression Key exclusion criteria: CNS involvement, HIV infection, severe organ dysfunction not related to PTLD, ECOG >2 Treatment: Four weekly courses of 375 mg/m 2 rituximab IV followed by 4 weeks without treatment and four cycles of CHOP chemotherapy (cyclophosphamide 750 mg/m 2 IV day (d) 1, doxorubicin 50 mg/m 2 IV d1, vincristine 1.4 mg/m 2 IV d1, and prednisone 50 mg/m 2 PO d1-5) at 3-week-intervals starting on day 50. In case of disease progression with rituximab monotherapy, patients proceeded to chemotherapy immediately. G-CSF-support was mandatory and antibiotic prophylaxis (cotrimoxazole and ciprofloxacin) was recommended. Primary endpoint: Treatment efficacy Median follow up: 5.1 years Patient characteristics: Number of patients: 70 Median age: 53.3 years Type of transplant: Kidney 41% Liver 23% Heart 20% Lung 6% Heart and lung 3% Kidney and pancreas 6% Bone marrow 1% Time to PTLD (1 year posttransplant): 76% Monomorphic histology: 96% EBV-association: 44% Stage of disease (Ann Arbor stage III or IV): 74% Extranodal involvement: 81% Graft involvement: 16% Elevated LDH: 75% ECOG performance status : ECOG 0 44% ECOG 1 26% ECOG 2 19% ECOG 3 7% ECOG 4 1% Combined data from the German and French rituximab monotherapy trials (8) Key inclusion criteria: Solid organ transplant recipients with CD20-positive PTLD unresponsive to a reduction of immunosuppression Key exclusion criteria: CNS involvement, HIV infection, severe organ dysfunction not related to PTLD Treatment: Four weekly courses of single agent rituximab (375 mg/m 2, IV). Rituximab treatment was stopped if the lymphoma progressed, if the patient declined to continue, or if it was considered necessary to do so because of concomitant illness or adverse events. Primary endpoint: Treatment efficacy Median follow up: 1.2 years Patient characteristics: Number of patients: 60 Median age: 49.7 years Type of transplant: Kidney 37% Liver 18% Heart 27% Lung 13% Heart and lung 5% Time to PTLD (1 year post transplant): 70% Monomorphic Histology: 68% EBV-association: 53% Stage of disease (Ann Arbor stage III or IV): 63% Extranodal involvement: n.d. Graft involvement: n.d. Elevated LDH: 55% ECOG performance status: ECOG 0 28% ECOG 1 37% ECOG 2 20% ECOG 3 15% ECOG 4 0% unknown in 3%, n.d., not determined. American Journal of Transplantation 2015; 15:

4 Trappe et al Results Relevance of previously published prognostic indices in the PTLD-1 trial A number of prognostic indices have been suggested in PTLD. We evaluated their prognostic value in the PTLD-1 trial cohort (Table 2, Figure 1, supplemental Figure S1). While the Leblond score and the Hourigan score delineate three prognostic groups, the Ghobrial score and the PTLD prognostic index delineate four prognostic groups. The international prognostic index delineates up to six prognostic groups but might be broken down for simplification. The Evens score was not included in this analysis due to lack of data in two-third of patients, while the high proportion of patients with monomorphic disease in the PTLD-1 trial (67/70, 96%) limited the value of the Ghobrial score. International prognostic index (IPI) and PTLD prognostic index The IPI is based on the baseline parameters age, stage of disease, ECOG performance status, involvement of extranodal sites and LDH, while the PTLD prognostic index is a variation of the IPI restricted to the baseline factors age, ECOG performance status and LDH. Both the IPI (p < 0.001) and the PTLD prognostic index (p ¼ 0.006) reached significance for overall survival. However, low-risk patients according to IPI did worse than low- or highintermediate-risk patients. Breaking the IPI score down into just low- (0,1,2 points) or high-risk (3,4,5 points) (7) resulted in two groups of almost similar size. While the significant effect on overall survival was retained in this more simplicistic form of IPI (p ¼ 0.032), none of the indices had a significant effect on time to progression. However, all indices correlated with treatment-related mortality, which was significantly increased in higher risk groups. Unfortunately, none of the indices performed well in terms of discrimination looking at their discriminatory power using Somer s D. The IPI broken down to just low and high showed a Somer s D score of The six-parameter IPI score showed a score of and the PTLD prognostic index showed a score of Using bootstrap methods for internal validation, all indices were remarkably stable. Looking at their calibration at 2 years, all indices were biased. While the PTLD prognostic index showed a rather large bias (mean error ¼ 0.15), the IPI score broken down into just low- and high-risk showed a relatively low bias (mean error ¼ 0.07). Leblond score, Hourigan score and Ghobrial score The Leblond score and the Hourigan score did not reach significance for overall survival, time to progression, treatment-related mortality or overall response rate. The Ghobrial score, which includes the baseline risk factors ECOG performance status >2, monomorphic disease and graft involvement, reached significance for overall survival (p < 0.001) and had an effect on response to treatment. Table 2: Predictive value of prognostic indices in the PTLD-1 trial Somer s Overall survival Somer s D index D corrected- Calibration Error (mean, 0.9 HR 95% CI p 1 D p 2 D Quantile) Prognostic index N Groups Distribution /13/11/25/10/ , 0.28 IPI (5) one point each for age >60, advanced stage, >2 extranodal sites, ECOG 2 and elevated LDH IPI / , 0.11 PTLD prognostic index (8) one point each for age >60, /31/20/ < , 0.22 ECOG 2 and elevated LDH Ghobrial score (7) one point each for ECOG >2, /50/12/ < < , 0.32 monomorphic disease and graft involvement Hourigan score (9) one point each for elevated LDH /23/ , 0.21 and B-symptoms Leblond score (6): one point each for ECOG 2 and /42/ , 0.23 involved sites > 1 N, number of patients with available data, Groups total number of groups in a given scoring system, Distribution number of patients within the different groups, HR, hazard ratio; CI, confidence interval; p 1, significance based on the likelihood ratio test; Somer s D measure of discrimination: for the Cox model D is the rank correlation between predicted log hazard and time to event. Somers D has a range from 1toþ1, with 1 for a perfect negative correlation, 0 for no correlation and þ1 for a perfect positive correlation, Somer s D corrected is based on bootstrap internal validation as well as the calibration error. p 2, significance based log-rank statistics; IPI, international prognostic index; ECOG, eastern cooperative oncology group performance status; LDH, serum lactate dehydrogenase; PTLDPI, PTLD prognostic index; and CNS, central nervous system American Journal of Transplantation 2015; 15:

5 Prognostic Factors in the PTLD-1 Trial Figure 1: Kaplan Meier estimates for overall survival in the PTLD-1 trial: (A) stratified by IPI low versus high risk, (B) stratified by Ghobrial score and (C) stratified by the PTLD prognostic index. The score represented by each line and the number of patients (N) included in the subgroup is indicated. The composition of the scoring systems is given in Table I. P significance is based on log-rank statistics. While the impact of ECOG performance status on response to sequential treatment has been described before in the PTLD-1 trial cohort (2), this result might be a reflection of ECOG status, taking into account that 96% of patients had monomorphic disease and only 16% had graft involvement. American Journal of Transplantation 2015; 15: Impact of individual baseline parameters To identify the prognostically relevant baseline components of the different scoring systems, we performed univariate analyses for ECOG performance status, age, Ann Arbor stage, LDH, presence of extranodal disease, graft involvement and monomorphic disease. In addition, we analysed the impact of thoracic organ transplantation and response to rituximab at interim staging. Only ECOG (p ¼ 0.002, HR 1.800) and age (p ¼ 0.001, HR 1.049/year) as continuous variables as well as ECOG >2 (p < 0.001, HR ) and age >60 years (p ¼ 0.004, HR 2.663), response to rituximab at interim staging (CR vs. PR vs. SD vs. PD, p ¼ 0.010, HR 0.647) and thoracic organ transplantation (p ¼ 0.039, HR 2.084) correlated with overall survival. Only ECOG >2 (p¼ 0.001, HR ) and response to rituximab at interim staging (CR/PR vs. 1095

6 Trappe et al SD/PD, p ¼ 0.022, HR 0.380) correlated with time to progression. In the PTLD-1 trial, 9/70 patients (13%) suffered treatmentrelated mortality including 7/66 patients (11%) who died CHOP-associated. Of these seven patients, two had stable disease at interim staging and five had progressive disease after rituximab monotherapy. From the parameters associated with overall survival, only ECOG >2 was associated with treatment-related mortality (p ¼ 0.002, 3/6 vs. 5/62), while age >60 (p ¼ 0.311, 4/21 vs. 5/49), response to rituximab at interim staging (p ¼ 0.218, 0/14 vs. 3/28 vs. 2/12 vs. 4/16) and thoracic organ transplantation (p ¼ 0.652, (2/20, 7/50) were not. However, patient numbers were small and CHOP-associated mortality was analysed together with treatmentrelated mortality from rituximab treatment and immunosuppression reduction. Multivariable analysis We have previously published multivariable Cox-regression analyses of factors affecting overall survival and time to progression in the PTLD-1 trial (2). Age, sex, time-to-ptld, EBV association, advanced stage, extranodal disease, serum LDH, ECOG 2 and violation of exclusion criteria were analysed. To test whether the variables overall response to rituximab at interim staging and thoracic organ transplantation remain prognostic when analysed in combination with these parameters, we have included them in these previously described models (2). In order to identify risk factors for treatment stratification, continuous variables were replaced by dichotomized variables using previously characterised cut-off values (age > 60 instead of age and late PTLD [>1 year after transplantation] instead of time-to-ptld) and violation of exclusion criteria (due to ECOG >2 in 6/7 cases) was left out. Backward likelihood-ratio Cox regression analyses confirmed the variables thoracic organ transplantation and overall response to rituximab (Table 3). In total, three parameters remained significant for overall survival in the PTLD-1 trial cohort in multivariable analysis: age >60 (p ¼ 0.001, HR 4.423), thoracic organ transplantation (p < 0.001, HR 7.827) and overall response to rituximab at interim staging (p ¼ 0.017, HR 0.322). Cox-regression analysis for time to progression identified the two parameters: overall response to rituximab at interim staging (p ¼ 0.008, HR 0.213) and thoracic transplantation (p ¼ 0.075, HR 2.983). Patients who might need less chemotherapy In 2007, the European PTLD Network introduced riskstratification according to response to rituximab into the PTLD-1 trial: Patients in complete remission after four courses of rituximab monotherapy were considered low-risk and received four further courses of rituximab monotherapy at 3-week intervals. Patients who did not achieve a complete remission (PR, SD or PD) were considered high-risk and continued with subsequent chemotherapy (R-CHOP). First interim results have been encouraging (20). To test whether additional stratification by IPI might identify further patients who might not need chemotherapy, survival data from the German and French rituximab monotherapy trials were analysed by response to rituximab and IPI (0,1,2 versus 3,4,5) (Figure 2A). One-year progression free survival for patients in complete remission after four courses of rituximab was 77% and overall survival was 91%. For patients in partial remission after four courses of rituximab, Table 3: Multivariable Cox-regression analyses for overall survival and time to progression in the PTLD-1 trial Risk factor p HR 95% CI Cox regression model n ¼ 62, Step 6, p ¼ Overall survival Thoracic organ transplantation < Age > Overall response to rituximab Late PTLD Advanced Stage Cox regression model n ¼ 60, Step 9, p ¼ Time to progression Overall response to rituximab Thoracic organ transplantation Factors included in these analyses were: age > 60, sex, late PTLD (x year after transplantation), EBV association, advanced stage, extranodal disease, elevated serum LDH, ECOG 2, thoracic organ transplantation and overall response to rituximab (CR/PR versus SD/PD) HR, hazard ratio; CI, confidence interval Overall significance of the Cox-regression models is assessed based on the likelihood ratio test. This p-value is based on the Wald test. The discriminatory power of the models assessed by Somer s D is 0.47 for the overall survival model (optimism corrected: 0.33) and 0.41 for the time to progression model (optimism corrected: 0.18). The mean calibration error of the overall survival model is 0.14 (90%-quantile: 0.18). The mean calibration error of the time to progression model is 0.2 (90%-quantile: 0.31). Bootstrap estimates of calibration accuracy for 2-year survival for the prognostic models using adaptive linear spline hazard regression are shown in supplemental Figure S American Journal of Transplantation 2015; 15:

7 Prognostic Factors in the PTLD-1 Trial 1-year progression free survival was 100% for patients with low-risk IPI criteria (0,1 or 2), but only 25% for patients with high-risk IPI criteria (3 5), p ¼ (Figure 2A). One-year overall survival was 100% and 78%, respectively. In contrast to the PTLD-1 trial cohort, patients in the rituximab monotherapy trials had subsequent chemotherapy only in case of disease progression and not all progressing patients had received chemotherapy. To test whether subsequent chemotherapy with CHOP started at day 50 per protocol improved the outcome of patients in partial remission after rituximab monotherapy with high-risk IPI criteria (IPI: 3,4,5), response and IPI stratification was applied to the PTLD-1 trial cohort (Figure 2B). For patients in partial remission after four courses of rituximab, 1-year progression free survival was 100% for patients with low-risk IPI criteria (0,1 or 2) and 71% for patients with high-risk IPI criteria (3 5) (Figure 2B). One-year overall survival was 100% and 82%, respectively. Thus, chemotherapy consolidation with CHOP starting at day 50 improved 1-year PFS in patients in partial remission after rituximab monotheray with high-risk IPI criteria, but did not improve outcome of patients in partial remission after four courses of rituximab monotheray with low-risk IPI criteria. Patients in partial remission with low-risk IPI criteria therefore might not need chemotherapy in 1st-line treatment of PTLD. Patients who might need more chemotherapy To identify patients at high risk for death from disease progression in the PTLD-1 cohort, we analysed the prognostic scores and applicable risk factors from multivariable analysis, namely age >60 and thoracic organ transplantation and their effect on overall survival and time to progression to sequential therapy in the high-risk group of rituximab non-responders (SD and PD cohort). This analysis was an intention-to-treat analysis of note, three of the 28 patients never received chemotherapy (one due to treatment-related mortality before start of therapy, two because they were considered not fit enough for chemotherapy). Thoracic organ transplantation was a significant risk factor for overall survival (p ¼ 0.034, HR 3.252) and time to progression (p < 0.001, HR 8.941) in univariate analysis, whereas no CHOP-associated death was observed in this subgroup (0/4). Non-response to rituximab monotherapy in heart- and lung-transplant recipients thus predicts CHOP refractory disease (Figure 3A). One-year overall survival was 25% in this patient cohort (Figure 3B). Figure 2: Kaplan Meier estimates for progression-free survival stratified by response to rituximab and IPI (low-risk vs. high-risk) in (A) the combined data set of the German and French rituximab monotherapy trials (8) and (B) the PTLD-1 trial cohort. While patients in partial remission after single agent rituximab with low-risk IPI criteria had a very favorable outcome, patients in partial remission with high-risk IPI criteria had a high risk for disease progression (PFS at 1 year: 25%). Most of these patients had second-line treatment including rituximab and/or CHOP. However, it is important to notice that overall survival in the rituximab monotherapy trials was considerably worse than OS after sequential treatment (2.4 years vs. 6.4 years) (2). In the PTLD-1 trial cohort where CHOP was given per protocol at day 50 to all patients (B), patients with high-risk IPI criteria who reached a PR with four courses of single-agent rituximab still did worse than those with low-risk IPI criteria, but 1-year progression-free survival was improved to 71%. IPI: international prognostic index, PR: partial remission after four courses of rituximab monotherapy, CR: complete remission after four courses of rituximab monotherapy, N: number of patients included in respective groups. American Journal of Transplantation 2015; 15:

8 Trappe et al Figure 3: Kaplan Meier estimates for progression free (A) and overall survival (B) stratified by response to rituximab at interim staging and type of transplant (heart or lung versus all others) in the PTLD-1 trial cohort. While renal and liver transplant recipients not responding to four courses of single-agent rituximab had a relatively good outcome, heart and lung transplant recipients not responding to upfront singleagent rituximab had a high risk for death from disease progression (OS at 1 year: 0%). SD: stable disease at interim staging, i.e. after four courses of rituximab monotheray, PD: progressive disease at interim staging, N: number of patients included in respective groups. Discussion This analysis of the international, prospective, multicenter PTLD-1 trial confirms the prognostic role of IPI, Ghobrial score and PTLD prognostic index for overall survival in PTLD treated with sequential immunochemotherapy with rituximab and CHOP. Neither the IPI, the Ghobrial score nor the PTLD prognostic index were associated with time to progression, but a high IPI and a high PTLD prognostic index were associated with an increased treatment-related mortality. Age and ECOG performance status were the most important baseline parameters driving prognosis of the indices and ECOG >2 correlated with time to progression and treatment-related mortality. Thoracic organ transplantation and response to rituximab were identified as important prognostic factors for both, time to progression and overall survival, by multivariable analysis. When the IPI is broken down into just low (<3) and high (3), it s prognostic power is almost retained. The PTLD prognostic index arguably performed best in the prediction of outcome in the PTLD-1 cohort. However, given the rarity of PTLD with only few cases per year even at specialized hematological centers, applying the high-risk criteria according to IPI (IPI 3) due to the common risk factors extranodal disease and advanced stage has practical advantages over the use of disease-specific prognostic indices. It can delineate a population with a particular risk for treatment-related mortality (p ¼ 0.009, 8/36 vs. 0/27) that might benefit from more stringent supportive care. Applying IPI low-risk criteria to patients in partial remission after four courses of rituximab monotherapy can delineate a patient cohort with a relatively low risk of disease progression where chemotherapy consolidation might not be necessary similar to patients in complete remission after rituximab monotherapy. In contrast, patients in partial remission after rituximab monotherapy with high-risk IPI criteria (3) have a poor prognosis with rituximab monotherapy and CHOP consolidation improved outcome (2,8). The results of our multivariable analyses underlined the key role of response to rituximab and thoracic organ transplantation for overall survival and time to progression. Thoracic organ transplant recipients not responding to rituximab monotherapy had a particularly high risk of disease progression with subsequent CHOP, while their risk of treatment-related mortality was relatively low. More intensive treatment than CHOP therefore might improve outcome in this patient cohort. However, increasing intensity of chemotherapy runs risk for treatment-related mortality and should be done with caution. Despite utilizing the largest prospective trial cohort in PTLD to date, our analysis is limited by sample size due to the rarity of the disease. This limitation needs carefully taken into account when interpreting results from subgroup and multivariable analyses. In total, only 6% of patients in this trial were lung and only 20% were heart transplant recipients. Another 3% of patients had a combined heart 1098 American Journal of Transplantation 2015; 15:

9 Prognostic Factors in the PTLD-1 Trial and lung transplant. While the risk of PTLD, the type of immunosuppression and the response to rituximab monotherapy is different after heart and lung transplantation, response to sequential immunochemotherapy with rituximab and CHOP is similar (2). To minimize the limitations of multiple testing, we employed a step-wise approach to identify clinically defined high-risk subgroups. However, while some of our results confirm previous findings, especially the newly identified risk factor thoracic organ transplantation needs to be confirmed in additional cohorts. In addition, supportive treatment and immunosuppression of transplant recipients and PTLD patients will change over time, potentially modifying the relevance of some of the risk factors. Because treatment stratification using clinically defined risk groups requires rigorous prospective testing before parameters can be incorporated into clinical decision making, a protocol prospectively testing the described findings has been suggested (NCT ). Authors Contributions SO, RUT and SC are the principal investigators, coordinated the research and take primary responsibility for the paper. SC, PM, UD, MS, GS, FM, AJ, TL, PR, RN, HL, HAH, ML, HR, VL, SO and RUT recruited the patients. SC and RUT collected the data. SC, DD, PM, JMZ, MHD, UD, CT, OS, MS, GS, PR, HR, VL, RUT and SO analysed and interpreted the data. PS provided statistical advice. IA and MR served as reference pathologists. RUT and SO wrote the paper. All authors had full access to the final version of the manuscript and agreed to publication. Disclosures American Journal of Transplantation 2015; 15: The authors of this manuscript have conflicts of interest to disclose as described by the American Journal of Transplantation. SC received payment for lectures, research grants and consultancy fees from Roche. PM received payment for consultancy and travel support from Celgene. JMZ received payment for lecture and travel support from Amgen, Roche and Celgene. CT received honoraria from Roche. OS received grant support from Roche and Janssen. GS received payment for consultancy, honoraria and/or advisory board membership from Roche and/or Genentech and is an advisory board member for Celgene, Janssen-Cilag, Genzyme, Calistoga/Gilead and Mundipharma. FM received payment for consultancy, honoraria and/or advisory board membership from Roche and/or Genentech; he is advisory board member for Celgene, Spectrum and Mundipharma. VL is on the advisory board for Janssen-Cilag and Pharmacyclics, received travel support from Roche and Honoraria from Janssen-Cilag and Mundipharma. RUT received payment for lectures and consultancy from CSL Behring, Mundipharma and/or Roche, grant support from AMGEN, CSL Behring, Mundipharma and Roche and travel support from AMGEN, CLS Behring and Roche. SO planned and initiated the PTLD-1 trial in 2003 while employed at Charité university hospital Berlin, and he has been an employee at Hoffmann-La Roche Ltd. from 2005 to 2013 and owns Roche equity. All other authors declare no conflicts of interest. References 1. Swerdlow SH, Webber SA, Chadburn A, Ferry JA. Post-transplant lymphoproliferative disorders. In: Swerdlow SH, Campo E, Harris NL, Jaffe ES, Pileri SA, Stein H, et al, eds. WHO classification of tumours of haematopoetic and lymphoid tissues. 4th ed. Lyon: International Agency for Research on Cancer, 2008, pp Trappe R, Oertel S, Leblond V, et al. Sequential treatment with rituximab followed by CHOP chemotherapy in adult B-cell posttransplant lymphoproliferative disorder (PTLD): The prospective international multicentre phase 2 PTLD-1 trial. Lancet Oncol 2012; 13: Quinlan SC, Morton LM, Pfeiffer RM, et al. Increased risk for lymphoid and myeloid neoplasms in elderly solid-organ transplant recipients. Cancer Epidemiol Biomarkers Prev 2010; 19: Zimmermann H, Reinke P, Neuhaus R, et al. Burkitt posttransplantation lymphoma in adult solid organ transplant recipients: Sequential immunochemotherapy with rituximab (R) followed by cyclophosphamide, doxorubicin, vincristine, and prednisone (CHOP) or R-CHOP is safe and effective in an analysis of 8 patients. Cancer 2012; 118: The International Non-Hodgkin s Lymphoma Prognostic Factors Project. A predictive model for aggressive non-hodgkin s lymphoma. N Engl J Med 1993; 329: Leblond V, Dhedin N, Mamzer Bruneel MF, et al. Identification of prognostic factors in 61 patients with posttransplantation lymphoproliferative disorders. J Clin Oncol 2001; 19: Ghobrial IM, Habermann TM, Maurer MJ, et al. Prognostic analysis for survival in adult solid organ transplant recipients with posttransplantation lymphoproliferative disorders. J Clin Oncol 2005; 23: Choquet S, Oertel S, LeBlond V, et al. Rituximab in the management of post-transplantation lymphoproliferative disorder after solid organ transplantation: Proceed with caution. Ann Hematol 2007; 86: Hourigan MJ, Doecke J, Mollee PN, et al. A new prognosticator for post-transplant lymphoproliferative disorders after renal transplantation. Br J Haematol 2008; 141: Evens AM, David KA, Helenowski I, et al. Multicenter analysis of 80 solid organ transplantation recipients with post-transplantation lymphoproliferative disease: Outcomes and prognostic factors in the modern era. J Clin Oncol 2010; 28: Caillard S, Lamy FX, Quelen C, et al. Epidemiology of posttransplant lymphoproliferative disorders in adult kidney and kidney pancreas recipients: Report of the French registry and analysis of subgroups of lymphomas. Am J Transplant 2012; 12: Dierickx D, Tousseyn T, Sagaert X, et al. Single-center analysis of biopsy-confirmed posttransplant lymphoproliferative disorder: Incidence, clinico-pathological characteristics and prognostic factors. Leuk Lymphoma 2013; 54:

10 Trappe et al 13. Engels EA, Pfeiffer RM, Fraumeni JF, et al. Spectrum of cancer risk among U.S. solid organ transplant recipients: The transplant cancer match study. JAMA 2011; 306: Zimmermann H, Choquet S, Dierickx D, et al. Early and late posttransplant lymphoproliferative disorder after lung transplantation 34 cases from the European PTLD network. Transplantation 2013; 96: e18-e Muchtar E, Kramer MR, Vidal L, et al. Posttransplantation lymphoproliferative disorder in lung transplant recipients: A 15-year single institution experience. Transplantation 2013; 96: Harrell FE Jr, Lee KL, Mark DB. Multivariable prognostic models: Issues in developing models, evaluating assumptions and adequacy, and measuring and reducing errors. Stat Med 1996; 15: Harrell FE. Regression modeling strategies. Springer; Heidelberg: R Core Team. 2014, R: A language and environment for statistical computing. R Foundation for Statistical Computing; Vienna, Austria: URL Harrell Frank E Jr, 2014; rms: Regression Modeling Strategies. R package version Trappe RU, Dierickx D, Reinke P, et al. Interim analysis of the largest prospective trial to date in adult CD20-positive posttransplant lymphoproliferative disorder (PTLD): Introducing riskstratified sequential treatment (RSST) [abstract]. J Clin Oncol 2012; 30 (15_suppl) Supporting Information Additional Supporting Information may be found in the online version of this article. Figure S1: Bootstrap estimates of calibration accuracy for 2-year survival of established prognostic indices in the PTLD-1 trial cohort using adaptive linear spline hazard regression. Black lines show observed survival, blue lines optimism corrected survival and grey lines indicate ideal agreement. Figure S2: Bootstrap estimates of calibration accuracy for 2-year survival for the prognostic models evaluated in Table 3 using adaptive linear spline hazard regression. Black lines show observed survival, blue lines optimism corrected survival and grey lines indicate ideal agreement American Journal of Transplantation 2015; 15: