Salvage Regimens With Autologous Transplantation for Relapsed Large B-Cell Lymphoma in the Rituximab Era

Abstract

Purpose

Salvage chemotherapy followed by high-dose therapy and autologous stem-cell transplantation (ASCT) is the standard treatment for relapsed diffuse large B-cell lymphoma (DLBCL). Salvage regimens have never been compared; their efficacy in the rituximab era is unknown.

Patients and Methods

Patients with CD20⁺ DLBCL in first relapse or who were refractory after first-line therapy were randomly assigned to either rituximab, ifosfamide, etoposide, and carboplatin (R-ICE) or rituximab, dexamethasone, high-dose cytarabine, and cisplatin (R-DHAP). Responding patients received high-dose chemotherapy and ASCT.

Results

The median age of the 396 patients enrolled (R-ICE, n = 202; R-DHAP, n = 194) was 55 years. Similar response rates were observed after three cycles of R-ICE (63.5%; 95% CI, 56% to 70%) and R-DHAP (62.8%; 95 CI, 55% to 69%). Factors affecting response rates (P < .001) were refractory disease/relapse less than versus more than 12 months after diagnosis (46% v 88%, respectively), International Prognostic Index (IPI) of more than 1 versus 0 to 1 (52% v 71%, respectively), and prior rituximab treatment versus no prior rituximab (51% v 83%, respectively). There was no significant difference between R-ICE and R-DHAP for 3-year event-free survival (EFS) or overall survival. Three-year EFS was affected by prior rituximab treatment versus no rituximab (21% v 47%, respectively), relapse less than versus more than 12 months after diagnosis (20% v 45%, respectively), and IPI of 2 to 3 versus 0 to 1 (18% v 40%, respectively). In the Cox model, these parameters were significant (P < .001).

Conclusion

In patients who experience relapse more than 12 months after diagnosis, prior rituximab treatment does not affect EFS. Patients with early relapses after rituximab-containing first-line therapy have a poor prognosis, with no difference between the effects of R-ICE and R-DHAP.

INTRODUCTION

During the last decade, the addition of the anti-CD20 monoclonal antibody rituximab to various chemotherapies^1–3 has dramatically improved response rates in diffuse large B-cell lymphoma (DLBCL), with complete responses (CRs) in 75% to 80% of patients. The use of rituximab in first-line treatment improved 5-year event-free survival (EFS) from 29% to 47% in the initial study of patients between age 60 and 80 years⁴ and improved 3-year EFS from 59% to 79% in patients age 18 to 60 years;⁵ rituximab was also associated with improved overall survival (OS). Before the rituximab era, 5-year OS rate for relapsed DLBCL was 53% after high-dose chemotherapy with autologous stem-cell transplantation (ASCT).⁶ Various parameters greatly affect the results of ASCT, including chemotherapy sensitivity before ASCT,⁷ time from diagnosis to relapse of less than 12 months,⁸ and the presence of prognostic factors at relapse, as defined by the secondary age-adjusted International Prognostic Index (saaIPI).^9,10 The addition of rituximab to second-line chemotherapy followed by ASCT significantly improved progression-free survival (PFS) in patients not exposed to rituximab as part of their first-line treatment.¹¹

For patients who have experienced relapse, no comparative studies have thus far been performed to our knowledge to evaluate the efficacy of the different salvage regimens.¹² Therefore, we compared the effects of two established salvage regimens followed by ASCT, attempted to identify the parameters influencing the effectiveness of each regimen, and aimed to establish whether or not the widespread use of rituximab as part of first-line therapy affects the outcome of patients with relapsed DLBCL.⁶

The present Collaborative Trial in Relapsed Aggressive Lymphoma (CORAL) study was a collaborative effort by 12 countries worldwide. Patients with refractory or relapsed CD20⁺ DLBCL were randomly assigned to one of the following two widely used regimens that included rituximab: rituximab, ifosfamide, carboplatin, and etoposide (R-ICE)¹³ or rituximab, dexamethasone, high-dose cytarabine, and cisplatin (R-DHAP).¹⁴ In responding patients, peripheral progenitor cells were collected after chemotherapy and reinfused after a high-dose chemotherapy conditioning regimen. We also investigated the impact of post-transplantation rituximab administration. Here, we report the results of the comparison between these two salvage regimens and the factors affecting outcome.

PATIENTS AND METHODS

Patients

Eligible patients were age 18 to 65 years and had aggressive CD20⁺ B-cell non-Hodgkin's lymphoma, including DLBCL, and had experienced relapse or did not achieve CR with a standard anthracycline-based regimen composed of cyclophosphamide, doxorubicin, vincristine, and prednisone (CHOP). Before enrollment, CD20⁺ aggressive B-cell lymphoma was histologically confirmed in all patients. Patients eligible for inclusion had a performance status of 0 to 1. Exclusion criteria included CNS involvement, a history of HIV infection, post-transplantation lymphoproliferative disorders, and inadequate organ function. Patients were fully evaluated by examinations that included thoracic and abdominal computed tomography scans and bone marrow biopsy. saaIPI factor status was determined by the absence or presence of risk factors, poor performance status, elevated lactate dehydrogenase, and disseminated stage before salvage treatment.^9,10 The study was approved by the relevant institutional review boards or ethics committees, and all patients gave written informed consent.

The study was registered under Europen Union Drug Regulating Authorities Clinical Trials (EudraCT) No. 2004-002103-32 and ClinicalTrials.gov NCT 00137995. Four hundred patients were enrolled between July 2003 and September 2007 for part 1 of the study. On an intent-to-treat basis, 396 patients were randomly assigned (202 patients to the R-ICE arm and 194 patients to the R-DHAP arm), and 388 patients were actually treated (Fig 1). Patient characteristics are listed in Table 1. No significant differences between the two arms were observed. Histology was reviewed by local hematopathologists attached to the participating centers. In addition, an international central review was performed in 289 (73%) of 396 patients. Only 13 patients did not have DLBCL; three patients had grade 3 follicular lymphoma, six patients had grade 2 follicular lymphoma, two patients had T-cell lymphoma, and two patients had Hodgkin's lymphoma. Only four patients were CD20⁻, and CD20 status was not documented in 13 patients. All of the patients were included in an intent-to-treat analysis and received the protocol arm.

Fig 1.

CONSORT diagram of distribution of patients according to arm resulting from the first random assignment. CRF, case report forms; R-ICE, rituximab, ifosfamide, carboplatin, etoposide; R-DHAP, rituximab, dexamethasone, high-dose cytarabine, cisplatin; BEAM, carmustine, etoposide, cytarabine, melphalan; ASCT, autologous stem-cell transplantation.

Table 1.

Baseline Patient Demographics and Clinical Characteristics (intent to treat)

Demographic or Clinical Characteristic	No. of Patients		P
	R-ICE (n = 202)	R-DHAP (n = 194)
Age, years
Median	54	55
Range	19-65	19-65	NS
Sex
Male	125	118
Female	77	76	NS
Ann Arbor stage
I-II	81	66
III-IV	119	121	NS
Extranodal site > 1	55	64	NS
Bone marrow involvement	17	19	NS
Elevated LDH	104	94	NS
saaIPI at relapse
0-1	119	107
2-3	75	74	NS
Time to relapse after diagnosis, months	89	87	NS
< 12*	112	103
≥ 12	122	122	NS
Prior rituximab treatment
Prior first-line CHOP-like chemotherapy	171	167	NS
Intensified CHOP	28	23

Abbreviations: R-ICE, rituximab, ifosfamide, carboplatin, and etoposide; R-DHAP, rituximab, dexamethasone, high-dose cytarabine, and cisplatin; NS, not significant; LDH, lactate dehydrogenase; saaIPI, secondary age-adjusted international prognostic index at relapse; CHOP, cyclophosphamide, doxorubicin, vincristine, and prednisone.

^*Including patients not achieving complete response after first-line treatment.

Study Design and Treatment

This study was a phase III multicenter randomized trial designed to compare the efficacy of R-ICE and R-DHAP in patients with previously treated DLBCL followed by ASCT with or without rituximab maintenance therapy (Fig 2). There were two random assignments, the first for salvage therapy and the second for maintenance treatment. The efficacy of the two salvage regimens is the subject of this report.

Fig 2.

Treatment protocol. R1, first random assignment; R-DHAP, rituximab, dexamethasone, high-dose cytarabine, cisplatin; R-ICE, rituximab, ifosfamide, carboplatin, etoposide; PBPC, peripheral-blood progenitor cells; CR, complete response; PR, partial response; PD, progressive disease; SD, stable disease; BEAM, carmustine, etoposide, cytarabine, melphalan; ASCT, autologous stem-cell transplantation; R2, second random assignment.

Patients were stratified according to participating country, prior rituximab treatment, and relapse occurring less than or more than 12 months after diagnosis. Every 3 weeks, patients were given three cycles of chemotherapy, followed by ASCT. In both regimens, rituximab (375 mg/m²) was administered before chemotherapy, and in the first course, additional rituximab was given on day −1. The R-ICE¹³ regimen consisted of etoposide (100 mg/m² per day) on days 1 through 3, ifosfamide (5,000 mg/m²) infused continuously for 24 hours on day 2 with mesna; and carboplatin (area under the curve = 5; maximum dose, 800 mg) on day 2. The R-DHAP regimen¹⁴ consisted of cisplatin (100 mg/m²) on day 1 via continuous 24-hour infusion, followed on day 2 by cytarabine (2 g/m²) in a 3-hour infusion repeated after 12 hours, and dexamethasone (40 mg/d) for 4 consecutive days. Granulocyte colony-stimulating factor was administered after R-ICE and, depending on site policy, with R-DHAP, but always after the third cycle until the end of leukaphereses.

Leukaphereses were performed after the third or second course of salvage therapy to obtain a target of 2,000,000 CD34⁺ hematopoietic stem cells per kilogram for cryopreservation. In case of inadequate peripheral stem-cell collection after the third course, patients were considered to be experiencing treatment failure and withdrawn from the study.

Assessment of Response and Follow-Up

Response was assessed by conventional diagnostic methods, including computed tomography scans, after the third chemotherapy course. Bone marrow biopsies were only repeated if abnormal before treatment.

Response was assessed using the International Working Group criteria.¹⁵ CR was defined by the disappearance of all documented disease; unconfirmed CR (CRu) was used when a residual mass was present without evidence of active disease. Partial response (PR) was defined as a 50% reduction of measurable disease. The mobilization response rate was defined as the objective CR and PR rates associated with the target mobilization of the peripheral stem cells (2,000,000 CD34⁺ hematopoietic stem cells/kg). Response was evaluated 3 months after transplantation. Follow-up procedures included a physical examination every 3 months for the first year and every 6 months thereafter for 2 years and a complete evaluation at the end of the first year or earlier if necessary.

ASCT

Patients who achieved a CR or PR after the third cycle of salvage treatment were given carmustine, etoposide, cytarabine, and melphalan (BEAM) high-dose chemotherapy. The BEAM regimen included carmustine (300 mg/m²) on day −6, etoposide (200 mg/m²), cytarbine (200 mg/m²) on days −5 to −2, and melphalan (140 mg/m²) on day −1. Peripheral-blood stem cells were reinfused on day 0, at least 24 hours after completion of BEAM.

Radiotherapy after transplantation was not allowed and was considered to be an event. Supportive treatments were given according to standard use in each center.

Statistical Analysis

The primary end point was the mobilization-adjusted response rate after three cycles of chemotherapy. A higher favorable response rate was expected for R-ICE than for R-DHAP, with fewer failed stem-cell collections. To detect a difference of 15% in the mobilization-adjusted response rate between R-ICE, for which this rate was 60% (75% response minus 15% mobilization failure), and R-DHAP, with a corresponding rate of 45% (65% response minus 20% mobilization failure) with a power of 82% and a 5% significance level, 400 patients had to be randomly assigned to the two chemotherapy arms. This allowed the second random assignment of 240 patients, with an expected dropout rate of 40% (Appendix, online only).

Administration of an alternative treatment was considered as an event. EFS was defined as the time from the start of treatment to progression, relapse, new treatment, or death (irrespective of cause), whichever event occurred first. PFS was defined as the time from study entry until disease progression or death. OS was defined as the time from the start of treatment to death.

The Kaplan-Meier method was used to estimate EFS, PFS, and OS, and 95% CIs were calculated.¹⁶ Cox regression analysis was used to calculate the hazard ratio between the two arms.¹⁷ All reported P values are two-sided, and P < .05 was considered significant. All analyses were carried out with SAS 9.1.3 software (SAS Institute, Cary, NC).

The study was designed by the Steering Committee of CORAL. The same investigator (C.G.) checked the data for medical coherence, analyzed and interpreted the data, and was the principal writer of this article (Appendix).

RESULTS

Response to Treatment

At diagnosis, 62% of the patients had been treated with a CHOP-like regimen with rituximab. Before inclusion, after first-line treatment, 65% of patients had achieved a first CR, 20% had achieved a PR, 4% had stable disease, and 11% had progressive disease.

After salvage chemotherapy but before transplantation, the overall response rate, including CR, CRu, and PR, was 63.5% (95% CI, 56.8% to 70.7%) in the R-ICE arm and 62.8% (95% CI, 55.6% to 69.7%) in the R-DHAP arm (Table 2). The factors significantly affecting the overall response rate in the univariate analysis (P < .001) were refractory disease/relapse less than 12 months after diagnosis, secondary IPI of 2 to 3, and prior rituximab treatment, but not the treatment arm (Table 3). In total, 206 patients received BEAM and ASCT per protocol, and five more patients had stable disease. The main reason for premature withdrawal from the study was disease progression (Fig 1). Three months after transplantation and random assignment, 132 (73%) of 181 evaluable patients had CR or CRu, 24 (13%) had PR, one had stable disease, and 17 (9%) had progressive disease.

Table 2.

Response After Induction Treatment (including death) for All Patients

Response	R-ICE (n = 197)		R-DHAP (n = 191)
	No. of Patients	%	No. of Patients	%
Complete response	48	24	53	28
Unconfirmed complete response	24	12	22	12
Partial response	53	27	45	24
Stable disease	23	12	22	12
Progressive disease	38	19	35	18
Death	6	3	10	5
Premature withdrawal, not evaluated	4	2	4	2
Autologous transplantation
Median CD34+ cells collected, million/kg	4.5		4.9
Collection failure < 2,000,000 CD34+ cells	20	10	15	8
Mobilization-adjusted response	103	52.3	104	54.5
Consolidation with BEAM performed per protocol	101	51	105	55

Abbreviations: R-ICE, rituximab, ifosfamide, carboplatin, and etoposide; R-DHAP, rituximab, dexamethasone, cytarabine, and cisplatin; BEAM, carmustine, etoposide, cytarabine, and melphalan.

Table 3.

Response Rate and Survival According to Prognostic Factors

Factor	Total No. of Patients	Response CR/CRu/PR			3-Year Event-Free Survival		3-Year Overall Survival
		No. of Patients	%	P	%	P	%	P
All patients	398	246	63		31		50
CR/CRu		148	38		51		70
Prior rituximab
No	147	122	83	< .001	47	< .001	66	< .01
Yes	244	124	51		21		40
Relapse, > 12 months	160	140	88	< .001	45	< .001	64
Refractory, < 12 months	228	106	46		20		39	< .001
saaIPI
< 2	224	160	71	< .001	40		62
> 1	146	76	52		18	< .001	32	< .001

Abbreviations: CR, complete response; CRu, unconfirmed complete response; PR, partial response; saaIPI, secondary age-adjusted International Prognostic Index.

Survival

After a median follow-up time of 27 months, the 3-year EFS rate was 31% (95% CI, 26% to 36%) and was not significantly different between the R-ICE and R-DHAP arms (26% and 35%, respectively; P = .6). Three-year PFS was 37% (95% CI, 31% to 42%), and again, the R-ICE and R-DHAP arms were not significantly different (31% and 42%, respectively; P = .4). Three-year OS (Figs 3A and 3B) was 49% (95% CI, 43% to 55%), with no difference between the R-ICE and R-DHAP arms (47% and 51%, respectively; P = .4). For patients who underwent ASCT, 3-year PFS was 53% (Fig 4A). There was no difference between the numbers of patients who achieved CR and PR just before ASCT (Fig 4B).

Fig 3.

(A) Overall survival according to the first random assignment (intent to treat). (B) Progression-free survival according to treatment arm. (C) Event-free survival (EFS) according to prior rituximab treatment and relapse less than 12 months after diagnosis. (D) EFS according to prior rituximab treatment and relapse more than 12 months after diagnosis. R-ICE, rituximab, ifosfamide, carboplatin, etoposide; R-DHAP, rituximab, dexamethasone, high-dose cytarabine, cisplatin.

Fig 4.

(A) Progression-free survival (PFS) of patients undergoing autologous stem-cell transplantation (intent to treat; n = 206). (B) PFS according to response after salvage regimen (including death) for all patients: complete response (CR) plus unconfirmed complete response (CRu; n = 147) and partial response (PR; n = 98).

Three-year EFS, PFS, and OS were affected by prior rituximab treatment, early relapse, and saaIPI (Table 3). In the Cox model, all of these parameters remained significant (P < .001) for EFS, PFS, and OS; prior rituximab treatment was significant at a lower level (P = .01). The treatment arm was not significant.

When patients were analyzed according to early relapse and prior rituximab treatment, there was no difference in PFS, EFS, or OS for patients with relapse more than 12 months after diagnosis (Figs 3C and 3D). Early relapse and prior rituximab treatment (n = 187) defined a population with a poor response rate to the standard treatment; thus, their 3-year PFS was only 23%. However, for responding patients who underwent ASCT (n = 68), 3-year PFS was 39%, compared with 14% for patients who did not receive transplantation (n = 119; P < .001; Appendix Fig A1, online only). At the time of our analysis, 92 deaths (47%) had occurred in the R-ICE arm, and 82 deaths (43%) had occurred in the R-DHAP arm, mainly as a result of lymphoma.

Relapse and Progression

Progression or relapse was experienced by 104 patients in the R-ICE arm and 97 patients in the R-DHAP arm, mostly at the initial site and by half of patients during the treatment period. Various treatments were administered, including radiotherapy and chemotherapy, with or without transplantation (32 autotransplantations and 14 allografts; Appendix Tables A1 to A3, online only). A second CR was experienced by 32 of 176 patients. In all, 48 patients, 24 in each treatment arm, reported an event as a result of a new treatment after progression.

Adverse Events

The median time between salvage cycles was 22 days for both arms for the 230 patients who completed three cycles. Grade 3 to 4 hematologic toxicities were more severe in the R-DHAP arm than the R-ICE arm, and more patients required at least one platelet transfusion during the induction phase (57% in R-DHAP arm v 35% in R-ICE arm). In all, 90 serious adverse events occurred in 58 patients in the R-ICE arm, and 120 serious events occurred in 68 patients in the R-DHAP arm.

In both arms, the most common serious adverse events were infections, with a similar rate of infection as a result of neutropenia (16%) in both arms. Grade 3 to 4 nonhematologic toxicities were more severe in the R-DHAP arm and included grade 4 renal toxicity in 11 patients (Appendix Tables A4 and A5, online only). Patients who underwent BEAM followed by ASCT experienced the usual patterns of hematologic and nonhematologic toxicity, and three toxic deaths occurred.

DISCUSSION

In DLBCL, two populations are candidates for salvage treatment followed by high-dose chemotherapy and ASCT—patients who experience a relapse after achieving CR and those who do not achieve CR but are still responding to treatment. From the PARMA data,⁶ patients experiencing early relapses less than 12 months after diagnosis have the same poor prognosis as incomplete responders. Such patients constituted 57% of all patients in the present study. Because this study was performed between 2003 and 2007, not all of the patients had access to rituximab as first-line treatment. This fact enabled us to prospectively enroll patients who did and did not have prior rituximab treatment (62% and 36%, respectively).

Because no randomized comparison of any salvage regimens had ever been previously reported, it was not clear which regimen was preferable for treatment of relapsed DLBCL.¹² The R-ICE regimen was chosen because we assumed that rituximab would improve its results, as suggested by the Memorial Sloan-Kettering Cancer Center.¹³ Because DHAP has been widely used all over the world and was the salvage regimen of the PARMA study, it was used here as comparator.^5,12 Both regimens were supplemented with rituximab, which has been shown to improve treatment results of patients with relapsed DLBCL^11–13 not previously treated with rituximab.

The present results show a similar response rate of 63% for the two regimens, with a CR rate of only 38%, even after adjustment for mobilization failure. Furthermore, similar prospective mobilization failure rates of 10% were observed after both regimens. Only 50% of patients were able to undergo ASCT. Toxicities were similar, but there were more platelets and renal toxicity in the R-DHAP arm. An important finding was that several independent factors significantly affected response rates after salvage therapy, including saaIPI score, early relapse less than 12 months after diagnosis, and prior rituximab treatment. The same independent factors were found for OS, EFS, and PFS. R-ICE and R-DHAP gave similar results for all conceivable situations, thus demonstrating that it will be difficult to improve therapy without new drugs.

In this study, it was possible to identify a population with late relapse who benefited from the introduction of rituximab into their salvage regimen and exhibited an 80% response rate and a 3-year EFS ranging from 40% to 50%. Here, the standard treatment with ASCT reproduced the PARMA results.⁶ However, there was a group of patients with a poor prognosis whose prior rituximab treatment was predictive, in cases of early relapse, of a response rate of 50% and 3-year EFS of only 20%. For these patients, the results of standard therapy should be improved, and new approaches are needed.

At the time of this analysis, there were not enough events (85 of 140 events) to determine the impact of rituximab administered as post-transplantation maintenance therapy. For patients who underwent transplantation, 3-year PFS was 53% (Fig 4).

Our results seem less favorable than those reported in a nonrandomized study¹³ with R-ICE and in a study using high-dose rituximab before and after transplantation.¹⁸ In the randomized CORAL study, the three courses of R-ICE were separated by a 3-week interval instead of 2 weeks, which may have helped to lower the CR rate. However, the patients in the present study differed from those in both of the previously cited studies because they had not had previous rituximab treatment and their response was evaluated by functional imaging.¹³ We believe, however, that our results are more representative of the general population with relapsed DLBCL than those reported by single institutions with limited numbers of patients and no random assignment. When we looked at the initial prognostic parameters before failure/relapse according to prior rituximab treatment, patients who had received rituximab had more adverse factors, a finding likely to prove representative of the patients we will have to treat in the future.¹⁹ Consequently, new drugs designed to increase the response rate of salvage regimens and new approaches,²⁰ including allogeneic transplantation, should be explored.^21,22 In the era of antibody chemotherapy, novel targeted therapy resulting from better understanding of the biology of DLBCL, including studies of patient tumor specimens, will play a key role in these respects.

Appendix

The following centers and principal investigators included patients in the study: Australia (n = 37): J. TROTMAN, Concord Repatriation General Hospital, Ph. CAMPBELL, Geelong Hospital, I. LEWIS, Royal Adelaide Hospital, R. LOWENTHAL, Royal Hobart Hospital, R. HERRMANN, Royal Perth Hospital, D. MA, St Vincent's Hospital, Sydney, D. GILL, P. MARLTON, Princess Alexandra Hospital – Woodville, G. HILL, Royal Brisbane And Women's Hospital – Herston, J. GIBSON, Royal Prince Alfred Hospital – Camperdown, K.E. FAY, Royal North Shore - St Leonards – NSW, C.L. SMITH, Austin Hospital – Heidelberg, A.P. GRIGG, Royal Melbourne Hospital - Parkville – Victoria, G. CULL, Sir Charles Gardiner Hospital - Nedlands – WA. New Zealand (n = 13): P.J. BROWETT, Auckland Hospital, Christchurch Hospital, C.S. KARAPETIS, Ch. MUSUKA, Dunedin Hospital, G. FORGESON, Palmerston North Hospital. Switzerland (n = 17): W. MINGRONE, Kantonsspital Aarau AG, C. BERETTA, Fmh Onkologie-Hamatologie – Rheinfelden, D.C. BETICHER, Inst Fur Medizinische Onkol Der Univ – Bern, M. GHIELMINI, Ospedale Civico – Lugano, C. HELG, Hug Geneve – A. LOHRI, Geneve, Kantonsspital – Basel, C. CASPAR, Kantonsspital – Baden. Sweden (n = 13): B. MALMER, Umea University Hospital, H. HAGBERG, Akademiska Sjukhuset – Uppsala. United Kingdom (n = 28): D.W. MILLIGAN, Birmingham Heartlands Hospital, Ch. POCOCK, Kent and Canterbury Hospital, M. JOYNER, Royal Devon and Exeter Hospital, A. PETTITT, Royal Liverpool University Hospital, D. LINCH, University College London Hospitals, S. MONTOTO, St Bartholomew's Hospital – London, J. RADFORD, Christie Hospital – Manchester, T. MAUGHAN, Velindre Hospital – Cardiff, A. KRUGER, Royal Cornwall Hospital – Truro, Ch. HATTON, John Radcliffe Hospital – Oxford, J. NEILSON, Russells Hall Hospital – Dudley, R. PETTENGELL, St Georges Hospital – London, S.A.J. RULE, Derriford Hospital – Plymouth, M. MACHETA, Blackpool Victoria Hospital – Blackpool. Ireland (n = 2): H. ENRIGHT, AMNCH – Dublin, E. VANDENBERGHE, St James's Hospital – Dublin. Czech Republic (n = 32): I. VASOVA, FN Brno, P. ZAK, FN Hradec Kralove, T. KOZAK, FN Kralovske Vinohrady, M. TRNENY, VFN Praha 2 - Charles University Général Hospital, T. KOZAK, FN Kralovske Vinohrady – Praha. Israel (n = 11): H. ROSENBAUM, Rambam – Haifa, O. BAIREY, Rabin Medical Center - Beilinson Hospital – Petah Tikva, A. AVIGDOR, Sheba Medical Center – Tel Hashomer, D. BEN YEHUDA, Hadassah Medical Center – Jerusalem. United States (n = 6): C. MOSKOWITZ, A. ZELENETZ, Memorial Sloan-Kettering Cancer Center – New York City. France (n = 110): A. THYSS, Centre Antoine Lacassagne Nice, H. TILLY, Centre Henri Becquerel Rouen, Ch. ALLARD, Centre Hospitalier Meaux, M. JANVIER, Centre René Huguenin Saint Cloud, M. BLANC, CH Chambéry, B. CHRISTIAN, CH Metz, F. MORSCHHAUSER, CHU de Lille, O. TOURNILHAC, CHU Clermont-Ferrand, O. CASASNOVAS, CHU Dijon, J.C. EISENMANN, CHU Mulhouse, C. RECHIER, CHU Toulouse, B. COIFFIER, CHU Lyon Sud, A. DELMER, CHU Reims, B. AUDHUY, Hôpital Pasteur Colmar, C. FERME, Institut Gustave Roussy Villejuif, K. BOUABDALLAH CHU Pessac, D. DECAUDIN, Institut Curie – Paris, C. GISSELBRECHT, CHU Saint Louis – Paris, N. MILPIED, CHU – Nantes, T. De Revel, Hôpital d'Instruction des Armées Percy – Clamart, A. DELMER, CHU Hôtel Dieu – Paris, A.M. PENY, Centre Francois Baclesse – Caen, C. SEBBAN, Centre Léon Bérard – Lyon, R. BOUABDALLAH, Institut Paoli Calmette – Marseille, J. GABARRE, Hôpital de la Pitié Salpétrière – Paris, M. MACRO, CHU Clémenceau – Caen, P. FENAUX, CHU Avicenne, C. HAIOUN, CHU – Créteil. Belgium (n = 28): A. VAN HOOF, A.Z. Sint Jan AV, B. DE PRIJCK, CHR de la Citadelle, A. TRIFFET, CHU Charleroi-Vésale, G. FILLET, CHU de Liège, M. ANDRE, Grand Hopital de Charleroi, H. DEMUYNCK, Heilig Hart Ziekenhuis, F. OFFNER, Universitair Ziekenhuis Gent, A. BOSLY, Université Catholique de Louvain Mont Godinne, A. KENTOS, E. VAN DEN NESTE, Université Catholique de Louvain Saint Luc – Bruxelles, D. BRON, Institut Jules Bordet – Bruxelles. Germany (n = 103): O. SEZER, Charite Berlin Mitte, R. MUCK, Diakonissenkrankenhaus Stuttgart, L. BALLEISEN, Evangelisches Krankenhaus Hamm, LINK, Kaiserslautern, G. SCHLIMOK, Klinikum Augsburg, E.G. HIDDEMANN, Klinikum Grosshadern Der Lmu Munchen, H. BODENSTEIN, Klinikum Minden, B. METZNER, Klinikum Oldenburg, FISCHER, Stadt Klinikum Karlsruhe, C.U. DUHRSEN, Univ Klinikum Essen, G. FINKE, Univ Klinikum Freiburg, H. PRALLE, Univ Klinikum Giessen, G. HESS, Univ Klinikum Mainz, H. DOHNER, Univ Klinikum Ulm, Innere Medizin III, Kneba, Universitatsklinikum Kiel, T. WAGNER, Universitatsklinikum Lubeck, D. PEEST, University Hospital Hannover, LIERSCH, Universtatsklinikum Munster, THOMSSEN, Klinikum Bremn Mitte – Bremen, M. PFREUNDSCHUH, Universitatskliniken Des Saarlandes – Homburg, H. EIMERMACHER, St-Johanne – Hagen, N. SCHMITZ, Asklepios Klinik St. GEORG – Hamburg.

Statistics

Statistical analysis was planned and performed as follows.

Descriptive statistics.

Patient characteristics were compared between the two treatment arms using the Pearson χ² or Fisher's exact test. Study end points were complete response and partial response rate, event-free survival (EFS), progression-free survival, and overall survival. Patients without progression or relapse who were still alive were censored at the date of last contact.

Quantitative variables were summarized in tables displaying sample size, mean, standard deviation, median, and range; quartiles were presented when considered relevant. Qualitative variables were described in terms of frequencies of each response category, and frequencies were converted into percentages of the number of patients or adverse events examined depending on the statistical unit under investigation.

Censored data were presented as Kaplan-Meier plots of time to first event and summary tables of Kaplan-Meier estimates for criterion rates at fixed time points with 95% CIs. The median time to event was calculated (if reached) with 95% CIs. Estimates of the treatment effect were expressed as hazard ratios based on Cox regression with 95% CIs.

Statistical inference.

Statistical tests were two-sided and performed using a 5% level of significance. When relevant, 95% CIs were also presented. Survival end points were analyzed using the log-rank test (unstratified), the Cox model for corresponding hazard ratios, P values of treatment effects, multivariate models with 95% CIs, and P values based on the likelihood ratio test in unadjusted and adjusted analysis.

The number and proportion of responders and nonresponders in each treatment group, together with the two-sided 95% Pearson-Clopper CIs, were presented, as well as the difference between the proportions, the two-sided 95% asymptotic CI, and the P value of a χ² test. All statistical analyses were carried out using SAS 9.1.3 software (SAS Institute, Cary, NC).

Determination of Sample Size

Part I: induction.

The primary end point was mobilization-adjusted response rate after three cycles of chemotherapy. It was expected that to detect a difference in mobilization-adjusted response rate of 15% between rituximab, ifosfamide, carboplatin, and etoposide (R-ICE) 60% (75% response rate and 15% mobilization failure) and rituximab, dexamethasone, high-dose cytarabine, and cisplatin (R-DHAP) 45% (65% response rate and 20% mobilization failure) with 82% power at the 5% significance level, 400 patients should be randomly assigned between the two chemotherapy arms. Initially, 400 patients were randomly assigned 1:1 to either R-ICE or R-DHAP.

It was expected that 40% of these patients would either not achieve complete response or partial response or would drop out before autologous stem-cell transplantation (ASCT). It was expected that there would be 240 patients (400 × 60%) available immediately before ASCT for the second random assignment (1:1) into the maintenance or rituximab arms. The first safety analysis on 100 patients (reviewed by the Data and Safety Monitoring Committee on November 14, 2005) and first interim analysis on 200 patients (April 18, 2007) showed that the dropout rate was 50%. As a result, to keep the planned power with 240 patients for the maintenance or rituximab arms, we increased the initial sample size from 400 to 480 (240 on each arm). The enrollment was completed in June 2008.

Part II: maintenance.

The primary end point of EFS was used to assess sample size. If, after transplantation, we wished to detect a change in the 2-year EFS rate of 15% in favor of the rituximab arm (65%) versus no maintenance (50%), 240 transplantation patients randomly assigned 1:1 between the two treatment groups recruited over 3 years and observed for a minimum of 2 years would provide 80% power to detect the expected difference at the overall 5% (two-sided) significance level.

Interim Analysis

An interim analysis of the two parts, response rate and EFS efficacy parameters, was planned after 200 patients, necessitating an adjustment of the nominal significance (α level) for the final analysis to maintain the overall global significance level. The O'Brien-Fleming adjustment will be used to partition the α level with α = .003 at the first interim for response and α = .05 at the final analysis.

An interim analysis of the primary efficacy parameter was planned after the inclusion of 200 patients, leading to 100 patients randomly assigned to the maintenance treatment. This necessitates an adjustment of the nominal significance (α level) for the final analysis to maintain the overall global significance level. The O'Brien-Fleming adjustment will be used to partition the α level with α = 8.10⁻⁵ (40 events) at the first interim and α = .05 at the final analysis. The expected number of events during the 5 years is 140 to 145.

Efficacy Evaluation

Eligible patients for analysis.

Only patients who turned out not to have been eligible for the study were excluded from analysis, whereas eligible patients who were not treated according to protocol were analyzed according to treatment arm. The data were analyzed as of February 2008.

The following five populations of patients were identified. Induction full analysis set (following the intent-to-treat [ITT] principle) refers to all randomly assigned patients regardless of whether they received study treatment or not (396 patients analyzed according to the therapy they were randomly assigned to receive; 202 in the R-ICE arm and 194 in the R-DHAP arm). Induction ITT population refers to patients receiving at least one injection of the study treatment regardless of the quantity injected (388 patients analyzed according to the therapy they were randomly assigned to receive; 197 in the R-ICE arm and 191 in the R-DHAP arm). Induction safety population refers to patients receiving at least one injection of the study treatment (388 patients analyzed according to the therapy they actually received; 197 in the R-ICE arm and 191 in the R-DHAP arm). Maintenance ITT population refers to all patients formally randomly assigned in the second part of the study (197 patients analyzed according to the therapy they were randomly assigned to receive; 102 in the rituximab arm and 95 in the observation arm). Maintenance safety population refers to all patients formally randomly assigned in the second part of the study who received at least one dose of rituximab or have only been observed and have at least one maintenance follow-up assessment (388 patients analyzed according to the therapy they actually received; ie, a patient will be included in the rituximab arm if he or she received at least one dose of rituximab during any maintenance visit; otherwise, the patient will be included in the observation arm; thus, 97 in the rituximab arm and 94 in the observation arm). Because all patients received randomly assigned induction treatment, induction ITT and safety populations are equivalent.

Data Monitoring and Regulatory Aspects

Investigators.

This was an intergroup study. The participating groups are Groupe d'Etude des Lymphomes de l'Adulte (GELA) from France and Belgium, German High-Grade Non-Hodgkin Lymphoma Study Group (DSHNHL) from Germany, National Cancer Research Institute (NCRI) from the United Kingdom, Australasian Leukemia Lymphoma Group (ALLG) from Australia and New Zealand, Swiss Group for Clinical Cancer Research (SAKK) from Switzerland, centers from Sweden and Ireland, Memorial Sloan-Kettering Cancer Center (MSKCC) from the United States, Czech Lymphoma Study Group (CLSG) from the Czech Republic, and Israel Society of Hematology (ISH) from Israel.

Participating centers were determined by each lymphoma group, and participation was restricted to transplantation centers. The local organization of care within the network of the group was authorized as long as Good Clinical Practice procedures could be followed. Before any inclusion, each center must have received an ethical committee approval for this study and government authorization according to procedures in each country. To be declared as a participating center, the respective principal investigator must have sent his curriculum vitae to the international coordinator.

Sponsor: GELA Recherche Clinique (GELARC) Collaborative Trial in Relapsed Aggressive Lymphoma (CORAL).

This was an intergroup collaborative study organized by a steering committee including the principal investigator of each study group. The steering committee will be represented by an intergroup protocol coordinator to organize the study. Therefore, the study shall be conducted under the sponsorship of the collaborative groups as mentioned and specified in the protocol (GELA, NCRI, DSHNHL, ALLG, Sweden, Ireland, and US centers, hereinafter referred to as CORAL Collaborative Group). The principal investigator of each individual collaborative group was responsible for answering all clinical questions concerning eligibility, treatment, and evaluation of the patients and for study coordination within his group (eg, administrative procedures, ethics committee approval, serious adverse event reporting to local authorities) in collaboration with a local investigator. The principal investigator of the group was in communication with the coordinating center (GELARC) and the protocol coordinator. In case of absence, another coordinator should be designated by the group.

All participating countries had to sign an agreement with the sponsor GELARC describing duties, flow of data, and responsibilities. The data were analyzed and centralized at the GELARC data center in Lyon, France.

The steering committee will be responsible for running the study with the protocol coordinator. It will give its scientific advice during the study and in the elaboration of data reports (eg, manuscripts, speakers, ancillary research).

Study coordination center: GELARC.

Although each group was responsible for the organization within its center, general coordination on time was necessary, as well as centralized data management. GELARC structure will act as the coordination center. GELARC is located in France (Paris for random assignment and part of the data management, and Lyon Sud for another part of the data management and statistical analysis). The data were collected by the principal investigator at each participating center, checked for accuracy in each country by the coordinator and research assistants of the lymphoma group, and sent to GELARC. Queries were sent to each country coordinator, and data entry was performed after resolution.

The roles of GELARC were as follows: random assignment procedure; distribution and collection of case report forms in collaboration with study group principal investigator; data entry and validation; elaboration and mailing of queries; reporting of serious adverse events (see chart); coordination of response review; coordination of histologic review; coordination of monitoring procedures for each group; relation with investigators and newsletter; transmission of the data to the group on a regular basis; statistical analysis and report; and any demand from the steering committee.

Fig A1.

(A) Progression-free survival (PFS) according to prior exposure to rituximab (intent to treat [ITT]). (B) PFS according to time to failure from diagnosis (ITT). (C) PFS of induction ITT population according to autologous stem-cell transplantation (ASCT) or no ASCT in patients with prior rituximab and failure from diagnosis less than 12 months. NA, not available.

Table A1.

Patients With Progression/Relapse (induction ITT)

Progression/Relapse	Arm A (R-ICE)		Arm B (R-DHAP)
	No. of Patients	%	No. of Patients	%
Progression/relapse No. 1
Yes	104	53	97	51
No	93	47	94	49
Progression/relapse No. 2
Yes	11	6	14	7
No	186	94	177	93
Progression/relapse No. 3
Yes	4	2	5	3
No	193	98	186	97
Progression/relapse No. 4
Yes	2	1	0	0
No	195	99	191	100
Progression/relapse No. 5
Yes	2	1	0	0
No	195	99	191	100
Total	197	100	191	100

NOTE. One hundred four patients (53%) in the R-ICE arm and 97 patients (51%) in the R-DHAP arm presented with a first progression/relapse.

Abbreviations: ITT, intent to treat; R-ICE, rituximab, ifosfamide, carboplatin, and etoposide; R-DHAP, rituximab, dexamethasone, cytarabine, and cisplatin.

Table A2.

Period of First Progression/Relapse (induction ITT)

Period of Progression/Relapse	Arm A (R-ICE)		Arm B (R-DHAP)
	No. of Patients	%	No. of Patients	%
Treatment period	52	50	53	55
Follow-up period	52	50	42	43
Total	104	100	95	98

Abbreviations: ITT, intent to treat; R-ICE, rituximab, ifosfamide, carboplatin, and etoposide; R-DHAP, rituximab, dexamethasone, cytarabine, and cisplatin.

Table A3.

Type of Treatment for First Progression/Relapse (induction ITT)

Treatment	Arm A (R-ICE)		Arm B (R-DHAP)
	No. of Patients	%	No. of Patients	%
Chemotherapy
Not available	1	1	0	0
Yes	74	77	64	80
No	21	22	16	20
Radiotherapy
Not available	3	3	1	1
Yes	31	32	27	34
No	62	65	52	65
Immunotherapy
Not available	3	3	1	1
Yes	29	30	24	30
No	64	67	55	69
Transplantation
Not available	3	3	0	0
Yes	22	23	22	28
No	71	74	58	73
Other treatment
Not available	2	2	1	1
Yes	11	11	16	20
No	83	86	63	79
Total	96	100	80	100

NOTE. A total of 53 patients in the R-ICE arm and 36 patients in the R-DHAP arm experienced a new progression later on.

Abbreviations: ITT, intent to treat; R-ICE, rituximab, ifosfamide, carboplatin, and etoposide; R-DHAP, rituximab, dexamethasone, cytarabine, and cisplatin.

Table A4.

Main Toxicity of Salvage Regimens

Toxicity	R-ICE		R-DHAP
	No. of Patients	%	No. of Patients	%
Hemoglobin, g/dL
Day 10
Median	10.6		11.1
Range	6-15		7-16
Day 14
Median	10.3		10.1
Range	5-15		7-15
WBC, cells/μL
Day 10
Median	3.8		6
Range	0-30		0-86
Day 14
Median	6.3		6
Range	0-71		0-68
Platelets, platelets/μL
Day 10
Median	190		101
Range	9-1,088		2-2,940
Day 14
Median	58		47
Range	2-616		1-2,360
Transfusion		35		57
Patients who received 3 cycles	169	86	161	84
Infection with neutropenia grade 3 or 4	33	17	31	16
Infection without neutropenia grade 3 or 4	11	6	15	8
Renal grade 3 or 4	2	1	11	6
Toxic death	1		3

NOTE. Toxicities were assessed according to the National Cancer Institute Common Toxicity Criteria (version 3.0).

Abbreviations: R-ICE, rituximab, ifosfamide, carboplatin, and etoposide; R-DHAP, rituximab, dexamethasone, cytarabine, and cisplatin.

Table A5.

Time Intervals for Hematologic Recovery After Transplantation With BEAM and Peripheral Stem Cells

Parameter	Actual Arm of Treatment
	Arm A (R-ICE)	Arm B (R-DHAP)
Neutrophils > 1 × 109/L, days after transplantation
No. of patients	95	97
Mean	23.1	16.0
Standard deviation	57.97	18.65
Median	11.0	12.0
Minimum	0	9
Maximum	424	174
Neutrophils > 0.5 × 109/L, days after transplantation
No. of patients	94	96
Mean	15.6	14.3
Standard deviation	37.83	17.92
Median	11.0	11.0
Minimum	0	3
Maximum	375	174
Platelets > 20 × 109/L, days after transplantation
No. of patients	97	100
Mean	16.0	18.2
Standard deviation	37.24	39.66
Median	12.0	12.0
Minimum	0	1
Maximum	375	401

Abbreviations: BEAM, carmustine, etoposide, cytarabine, and melphalan; R-ICE, rituximab, ifosfamide, carboplatin, and etoposide; R-DHAP, rituximab, dexamethasone, cytarabine, and cisplatin.

AUTHORS' DISCLOSURES OF POTENTIAL CONFLICTS OF INTEREST

Although all authors completed the disclosure declaration, the following author(s) indicated a financial or other interest that is relevant to the subject matter under consideration in this article. Certain relationships marked with a “U” are those for which no compensation was received; those relationships marked with a “C” were compensated. For a detailed description of the disclosure categories, or for more information about ASCO's conflict of interest policy, please refer to the Author Disclosure Declaration and the Disclosures of Potential Conflicts of Interest section in Information for Contributors.

Employment or Leadership Position: None Consultant or Advisory Role: Christian Gisselbrecht, Roche (U); David C. Linch, Roche (C) Stock Ownership: None Honoraria: Bertram Glass, Roche Pharma AG; David C. Linch, Roche; Marek Trneny, Roche; Ofer Shpilberg, Roche; Norbert Schmitz, Roche Research Funding: Christian Gisselbrecht, Roche, Baxter, Chugai Pharmaceutical; Bertram Glass, Roche Pharma AG; David C. Linch, Roche; Marek Trneny, Roche; Andre Bosly, Roche; Craig H. Moskowitz, Genentech; Norbert Schmitz, Roche Expert Testimony: None Other Remuneration: None

AUTHOR CONTRIBUTIONS

Conception and design: Christian Gisselbrecht, Nicolas Mounier, Devinder Singh Gill, David C. Linch, Marek Trneny, Andre Bosly, Hans Hagberg, David Ma, Craig H. Moskowitz, Norbert Schmitz

Administrative support: Christian Gisselbrecht, David C. Linch, Marek Trneny, Andre Bosly, Nicolas Ketterer, Craig H. Moskowitz, Norbert Schmitz

Provision of study materials or patients: Christian Gisselbrecht, Bertram Glass, Nicolas Mounier, Devinder Singh Gill, David C. Linch, Marek Trneny, Andre Bosly, Nicolas Ketterer, Ofer Shpilberg, Hans Hagberg, David Ma, Craig H. Moskowitz, Norbert Schmitz

Collection and assembly of data: Christian Gisselbrecht, Nicolas Mounier, Devinder Singh Gill, Marek Trneny, Andre Bosly, Nicolas Ketterer, Ofer Shpilberg, David Ma, Craig H. Moskowitz, Norbert Schmitz

Data analysis and interpretation: Christian Gisselbrecht, Bertram Glass, Nicolas Mounier, Ofer Shpilberg, Norbert Schmitz

Manuscript writing: Christian Gisselbrecht, Bertram Glass, Nicolas Mounier, David Ma, Norbert Schmitz

Final approval of manuscript: Christian Gisselbrecht, Bertram Glass, Nicolas Mounier, Devinder Singh Gill, David C. Linch, Marek Trneny, Andre Bosly, Nicolas Ketterer, Ofer Shpilberg, Hans Hagberg, David Ma, Josette Brière, Craig H. Moskowitz, Norbert Schmitz