A randomized phase II trial of fludarabine, cyclophosphamide and mitoxantrone (FCM) with or without rituximab in previously treated chronic lymphocytic leukaemia

Authors


Professor Peter Hillmen, Department of Haematology, Level 3, Bexley Wing, St. James’s University Hospital, Beckett Street, Leeds, LS9 7TF, UK. E-mail: peter.hillmen@nhs.net

Summary

Combination fludarabine (F), cyclophosphamide (C) and rituximab (R) is the standard front-line therapy in chronic lymphocytic leukaemia (CLL), but appropriate treatment of relapsed/refractory CLL is less clear. Combined FC and mitoxantrone (M) has been reported to be effective in a single arm study, and rituximab when added to chemotherapy in CLL is synergistic. A randomized, two-stage, Phase II trial of FCM and FCM-R was conducted in relapsed CLL. The primary endpoint was response rate 2 months after therapy, assessed according to the 2008 International Workshop CLL criteria. In addition, minimal residual disease (MRD) in the marrow was studied 2 months after therapy, with MRD negativity defined as <0·01% CLL cells. Fifty-two patients were entered, 26 in each arm. The overall response rates to FCM and FCM-R were 58% and 65% respectively. Combined complete response (CR) and CR with incomplete marrow recovery [CR(i)] was 15% (95% confidence interval [CI]:4–35%) for FCM and 42% (95%CI:23–63%) for FCM-R, with eight patients achieving MRD negativity (3 FCM; 5 FCM-R). The toxicity of both regimens was acceptable. In conclusion, the addition of rituximab to FCM improves the response rates in relapsed CLL, resulting in more complete remissions and without additional safety concerns. Efficacy and safety should be fully tested in a randomized Phase III trial.

Chronic lymphocytic leukaemia (CLL) is the most common adult leukaemia in Europe and North America. Whilst recently there have been significant developments in the therapy of CLL, the disease still remains incurable. Most patients will eventually become resistant to treatment. The initial treatment of CLL has now moved away from monotherapy to combination immunochemotherapy following evidence from large randomized trials (Hallek et al, 2010). FCR (fludarabine, cyclophosphamide, rituximab) is now the standard first-line treatment for patients with CLL who require therapy by conventional criteria and are able to tolerate fludarabine-based therapy, except for the small minority (approximately 7%) of patients whose CLL has deletion of chromosome 17p. Treatment for relapsed and refractory CLL is more difficult and is dependent on prior therapy and duration of remission. There is little published data to assess which therapies are most appropriate for second-line and subsequent treatments. Fludarabine combinations are often used in patients considered fit enough to tolerate such therapies, who have either not received fludarabine combinations previously or who responded with a relatively durable remission. The definition of a durable remission is dependent on initial therapy, and may be considered as over 6 months with fludarabine monotherapy, and over 2 years with fludarabine-based combinations, such as FCR. Alternatives for patients with refractory disease or with shorter remissions include alemtuzumab, ofatumumab, combination chemotherapy [e.g. CHOP (cyclophosphamide, doxorubicin, vincristine, prednisolone)] and possibly allogeneic stem cell transplantation.

Rituximab as a single agent is not very effective either as an initial treatment or in a relapsed/refractory setting(McLaughlin et al, 1998; Nguyen et al, 1999; Winkler et al, 1999; Foran et al, 2000; Huhn et al, 2001; Itäläet al, 2002; Hainsworth et al, 2003). At the time of designing this trial, Phase II data suggested improved rates of overall response (ORR), complete remission (CR), progression-free survival (PFS) and overall survival (OS) when rituximab is given in combination with FC in relapsed/refractory CLL (Wierda et al, 2005). Results from a randomized phase III trial (n = 552) comparing FC with FCR in relapsed CLL substantiated the efficacy of this combination(Robak et al, 2008). For FCR patients compared with FC, the trial showed a superior ORR (70% vs. 58% (P = 0·0034)), CR (24% vs. 13% (P = 0·0007)) and median PFS (30·6 vs. 20·6 months (P = 0·0002)) after a median follow-up of 25 months. Toxicities were comparable in both arms. Thus FCR therapy appears to be very effective in both untreated and previously treated patients with CLL.

The addition of mitoxantrone to fludarabine-based therapy has been found to result in high response rates in a variety of indolent lymphoproliferative disorders, including follicular lymphoma (Wilder et al, 2002) and mantle cell lymphoma (Zinzani et al, 1999). The combination of fludarabine, cyclophosphamide and mitoxantrone (FCM) has previously been reported in two trials of 60 patients and 29 patients who had relapsed/refractory CLL (Bosch et al, 2002;Hendry et al, 2004). The ORRs were 78% and 79% respectively, with 30(50%) patients and 9(32%) patients achieving a CR. Interestingly, 10 of the 30 patients in CR had an eradication of detectable minimal residual disease (MRD) by a sensitive multi-parameter flow cytometric test, and these patients had a significantly prolonged survival compared to the other patients in this series. Therefore, it appears likely that the combination of mitoxantrone with FC will yield the highest remission rates of chemotherapy-based regimens. The combination therapy FCM has subsequently been reported in 69 previously untreated patients with CLL (Bosch et al, 2008), with a high ORR of 90%. A CR was achieved in 64% of patients, with 26% of the total achieving an MRD negative CR. The addition of rituximab to FCM (FCM-R) appears to increase response rates even further, as the same group reported treatment with FCM-R in 72 previously untreated patients resulting in an ORR of 93% and a CR rate of 82%, of which 46% achieved an MRD negative CR (Bosch et al, 2009).

The observation that some patients achieved the eradication of detectable MRD following either FCR or FCM-R is of considerable interest. The current literature suggests that patients who respond to salvage therapy and do not have detectable CLL by extremely sensitive techniques have a significantly prolonged survival (Rawstron et al, 2001; Kwok et al, 2009). Therefore, one of the endpoints of this study was to assess whether FCM-R eradicates detectable CLL, and whether it is reasonable to consider FCM-R in future Phase III Trials powered to demonstrate an improvement in OS.

Patients and methods

Trial design

The trial was a multi-centre, randomized, controlled, open, two-stage, parallel group, Phase II trial assessing FCM and FCM-R for patients with CLL requiring therapy, who had received at least one prior therapy. Patients were randomized to receive fludarabine with cyclophosphamide plus mitoxantrone alone (FCM) or with simultaneous rituximab (FCM-R) on a 1:1 basis. The objective of the trial was to assess the efficacy and safety of both regimens in previously treated CLL patients. The primary endpoint was ORR, defined as complete (CR) or partial remission (PR) by the International Workshop on CLL (IWCLL) Criteria (Hallek et al, 2008). Secondary endpoints were the proportion of patients with undetectable MRD, OS, PFS and toxicity.

The sample size was estimated for a single arm trial to inform the number of patients required for the experimental FCM-R arm. The sample size was doubled in order to include a randomized concurrent control arm, FCM, to protect against a possible patient selection bias, and give internal and external validity of the results (Van Glabbeke et al, 2002). Based on estimating a 50% ORR in the FCM-R arm with 90% power at a 10% level of precision, using Gehan’s two-stage approach (Machin et al, 1997), four patients in the first stage and then up to 21 patients in the second stage of the trial were required. A maximum of 56 patients across the two arms was planned (allowing for dropouts). Randomization was by stratified permuted blocks, stratifying for previous treatment with fludarabine (refractory or not refractory/naïve).

An independent Data Monitoring and Ethics Committee (DMEC) was established. Stopping rules were designed to stop the trial if treatment-related mortality (TRM) was unacceptably high, the observed response rate was unacceptably low or the response rate was so high it would be desirable to stop the trial and proceed to a Phase III study. The DMEC monitored treatment-related deaths as they occurred. The safety stopping rule was based on an unacceptable TRM rate in the FCM-R arm of 10% (Goldman & Hannan, 2001). The trial would be stopped if the second treatment-related death occurred in the first two patients, if the third occurred in the first 14 patients or if the fourth occurred in the full 25 patients. The DMEC would have considered stopping the trial for efficacy reasons after the first stage of the trial if the FCM-R treatment arm failed to achieve any overall (complete or partial) responses, or if all four patients responded. As the primary endpoint of response was not measured until around 9 months into the trial, the DMEC used an earlier indication of likely response, change in absolute lymphocyte count from baseline, to monitor the stopping rule in a timely manner.

The trial was approved by all relevant institutional ethical committees and regulatory review bodies, and was conducted in accordance with the Declaration of Helsinki and Good Clinical Practice.

Patients

Eligible patients had CLL requiring therapy, previous treatment with at least one chemotherapeutic regimen, a World Health Organization (WHO) performance status (PS) of 0, 1, or 2 and a life expectancy of at least 12 weeks. Patients were required to be able to give written informed consent. Patients could not have received previous treatment with fludarabine (or other purine analogues) combined with cyclophosphamide and mitoxantrone (FCM), or previous treatment with rituximab, either alone or in combination with chemotherapy, or have a past history of anaphylaxis following exposure to rat or mouse-derived complementarity-determining region (CDR)-grafted humanized monoclonal antibodies.

Response assessment

Response to treatment was assessed according to the 2008 IWCLL Guidelines (Hallek et al, 2008). MRD was assessed by highly sensitive multi-parameter flow cytometry with a level of detection to below 1 CLL cell in 10 000 leucocytes (Rawstron et al, 2007).

Treatment

Treatment with FCM and FCM-R was repeated every 28 d for a total of six cycles. Fludarabine was administered orally at 24 mg/m2/d on days 1–5. Cyclophosphamide was administered orally at 150 mg/m2/d on days 1–5. Mitoxantrone was administered intravenously at 6 mg/m2 on day 1. In addition, rituximab was administered intravenously on day 1 of each cycle in the FCM-R group. The dose of rituximab was originally 375 mg/m2 for all cycles, but the protocol was amended to increase the dose of rituximab to 500 mg/m2 for cycles 2–6, which reflected an update to the Summary of Product Characteristics. Three patients were treated prior to this amendment and received all cycles at 375 mg/m2. Patients unable to tolerate oral fludarabine and cyclophosphamide were permitted to receive the equivalent intravenous dosing of fludarabine (25 mg/m2/d for 3 d) and cyclophosphamide (250 mg/m2/d for 3 d).

Statistical methods

The ORR in each arm, defined as CR or PR by the National Cancer Institute criteria (Cheson et al, 1996), and by assessment of MRD, are summarized with 95% confidence intervals (CIs). The response criteria were updated in 2008 by the IWCLL (Hallek et al, 2008b) to include a new response category: clinical CR with a morphologically normal marrow but persistent cytopenias [CR(i)], i.e. platelet count <100 × 109/l and/or neutrophil count <1·5 × 109/l, and responses assessed by these criteria are also summarized. PFS and OS for all patients are summarized for each arm, and described using Kaplan–Meier curves. No formal statistical comparisons are presented for the endpoints as the trial was not designed to have the power to test for a clinically relevant difference between the arms. Toxicity was assessed using adverse event rates and laboratory changes, NCI toxicity grades and TRM rates. Planned subgroup analyses summarized the primary endpoint data for the clinical subgroups: patients previously treated with fludarabine; fludarabine refractory patients; all patients resistant to prior treatments; and all patients who had responded to previous treatment. In addition, the primary endpoint data was summarized for the following mutational subgroups: good risk (mutated IGHV genes excluding IGHV3-21), standard risk (unmutated IGHV or IGHV3-21 usage and/or 11q del and/or β2M > 4) and poor risk (>10% 17p del) disease, and also for the group of patients with 11q deletion independent of other prognostic markers.

Results

Patient Characteristics

Fifty-two patients were randomized between July 2005 and January 2007 (26 FCM, 26 FCM-R) from 11 UK institutions with local ethical and management approval. Written informed consent was received from all patients. The Consolidated Standards of Reporting Trials (CONSORT) diagram (Moher et al, 2001) (Fig 1) shows the flow of patients throughout the trial. As there were no safety concerns, and for practical reasons, there was no break in recruitment between Stages I and II of the trial.

Figure 1.

 CONSORT (Consolidated Standards of Reporting Trials) Diagram to show the flow of patients throughout the trial.

Baseline characteristics of the patients are displayed in Table I. The median age was 68 years (range 32–79) with 79% men. 23/37 had a β2-microglobulin (β2m) >4 mg/l. The median number of prior therapies was two (range 1–6), 33(63%) had prior fludarabine and 11 (21%) were refractory to, or relapsed <6 months after, fludarabine. 26/45 had unmutated IGHV genes (11/22 FCM, 15/23 FCM-R); 11 patients had deletion of 11q (six FCM, five FCM-R); and one patient had >20% 17p deleted cells (FCM-R).

Table I.   Baseline characteristics of patients by treatment group.
 FCM-R (%)FCM (%)Total (%)
  1. WHO, World Health Organization; NCI, National Cancer Institute; FISH, fluorescence In Situ hybridization.

  2. *Good = not 17p del, IGHV mutated (<98% homology), and no IGHV3-21 usage; Standard = not 17p del, IGHV unmutated (≥98% homology) or IGHV 3-21 usage, 11q del, or β2M > 4; poor = >10% 17p del.

Total26 (100·0)26 (100·0)52 (100·0)
Previous treatment with fludarabine
 Refractory5 (19·2)6 (23·1)11 (21·2)
 Not refractory13 (50·0)9 (34·6)22 (42·3)
 Naïve8 (30·8)11 (42·3)19 (36·5)
Gender
 Male22 (84·6)19 (73·1)41 (78·8)
 Female4 (15·4)7 (26·9)11 (21·2)
Age, years (as randomized)
 Median66·068·067·5
 Range(44, 79)(32, 79)(32, 79)
 ≤6511 (42·3)11 (42·3)22 (42·3)
 >6515 (57·7)15 (57·7)30 (57·7)
WHO performance status
 021 (80·8)12 (46·2)33 (63·5)
 15 (19·2)14 (53·8)19 (36·5)
Disease stage (BINETS criteria)
 A2 (7·7)0 (0·0)2 (3·8)
 A (progressive)2 (7·7)5 (19·2)7 (13·5)
 B11 (42·3)4 (15·4)15 (28·8)
 C10 (38·5)16 (61·5)26 (50·0)
Number of previous lines of treatment received for CLL
 17 (13·5)9 (17·3)16 (30·8)
 212 (23·1)11 (21·2)23 (44·2)
 34 (7·7)2 (3·8)6 (11·5)
 42 (3·8)3 (5·8)5 (9·6)
 50 (0·0)1 (1·9)1 (1·9)
 61 (1·9)0 (0·0)1 (1·9)
NCI response to most recent treatment received for CLL
 CR7 (26·9)3 (11·5)10 (19·2)
 PR10 (38·5)8 (30·8)18 (34·6)
 SD4 (15·4)6 (23·1)10 (19·2)
 PD1 (3·8)4 (15·4)5 (9·6)
Beta-2 microglobulin (β2M) (mg/l)
 <46 (23·1)8 (30·8)14 (26·9)
 ≥4 or <812 (46·2)7 (26·9)19 (36·5)
 ≥82 (7·7)2 (7·7)4 (7·7)
Risk stratification*
 Good1 (3·8)7 (26·9)8 (15·4)
 Standard22 (84·6)17 (65·4)39 (75·0)
 Poor1 (3·8)0 (0·0)1 (1·9)
IGHV mutation
 Mutated7 (26·9)11 (42·3)18 (34·6)
 Unmutated15 (57·7)11 (42·3)26 (50·0)
 Clonal1 (3·8)0 (0·0)1 (1·9)
FISH summary
 11q normal19 (73·1)19 (73·1)38 (73·1)
 11q deleted5 (19·2)6 (23·1)11 (21·2)
 17p deleted1 (3·8)0 (0·0)1 (1·9)

The arms were mainly well balanced at baseline, although more patients in the FCM-R arm had a WHO performance status of 0 rather than 1 [FCM 12(46%), FCM-R 21(81%)], and more patients in the FCM-R arm had responded to their most recent treatment with at least a PR [FCM 11(42%), FCM-R 17(65%)]. This degree of imbalance is not unexpected by chance for factors that are not stratified at randomization and with small patient numbers.

Treatment

All 52 patients received at least one dose of study drug as per protocol. Seven patients were randomized to the study before the protocol amendment to increase the dose of rituximab, three of whom were randomized to the rituximab arm.

Of the 52 patients entered into the trial, 26(50%) patients did not receive the specified six cycles of treatment [14(54%) FCM, 12(46%) FCM-R]. 36(69%) received four or more cycles of therapy with no difference between FCM and FCM-R (18/26 in each arm). Reasons for early discontinuation for patients in the FCM-R arm were: toxicity (7), death (1), patient choice (1), chest infection (2), unknown (1).

Efficacy

The stage I analysis population consisted of the first four patients randomized to the higher dose of rituximab (after protocol amendment implementation). Three out of the four patients achieved a PR or greater with FCM-R (two CRs and one PR), and the other had stable disease (SD). Based on the study design, this number of responders would have meant that only three patients per arm were required for stage II of the trial. However, as toxicity was an important secondary end-point in deciding whether to study the FCM-R combination in a large randomized Phase III trial, the DMEC recommended continuing recruitment so a more informative safety profile for the treatments could be established.

The efficacy endpoints of response at 2 months post-treatment on all patients randomized to both stages of the trial are summarized in Table II. ORRs were 58% (95% CI: 40%, 77%) and 65% (95%CI: 44%, 83%) in the FCM and FCM-R arms respectively. CRs were achieved by 2(8%) patients randomized to FCM, and 4(15%) patients randomized to FCM-R. Progressive disease (PD) was observed in 3(12%) patients randomized to FCM and 2(8%) to FCM-R. Six patients had missing responses due to death (4), withdrawal (1) or a missing sample (1).

Table II.   Response and minimal residual disease (MRD) summaries by treatment group.
Trial armAchieved at least PR (%)Did not achieve PR (%)95% CI for percentage achieving at least a PR*
  1. CR, complete remission; CR(i), CR with incomplete bone marrow recovery; PR, partial remission; SD, stable disease; PD, progressive disease; NCI, National Cancer Institute; IWCLL, international workshop on chronic lymphocytic leukaemia; 95% CI, 95% confidence interval.

  2. *Confidence intervals are calculated using the exact method.

Total32 (61·5)14 (26·9)[47·0, 74·7]
FCM-R17 (65·4)7 (26·9)[44·3, 82·8]
FCM15 (57·7)7 (26·9)[36·9, 76·6]
 NCI criteria 
CR (%)PR (%)SD (%)PD (%) 
Total6 (11·5)26 (50·0)9 (17·3)5 (9·6) 
FCM-R4 (15·4)13 (50·0)5 (19·2)2 (7·7) 
FCM2 (7·7)13 (50·0)4 (15·4)3 (11·5) 
 IWCLL criteria
CR (%)CR(i) (%)PR (%)SD (%)PD (%) 
Total6 (11·5)9 (17·3)17 (32·7)9 (17·3)5 (9·6) 
FCM-R4 (15·4)7 (26·9)6 (23·1)5 (19·2)2 (7·7) 
FCM2 (7·7)2 (7·7)11 (42·3)4 (15·4)3 (11·5) 
 Minimal residual disease (MRD)
MRD negative (%)MRD positive (%)95% CI for percentage achieving MRD Negativity*
Total8 (15·4)37 (71·2)[6·9, 28·1]
FCM-R5 (19·2)19 (73·1)[6·6, 39·4]
FCM3 (11·5)18 (69·2)[2·4, 30·2]

In addition, according to the 2008 IWCLL criteria, a further 2(8%) patients randomized to FCM and 7(27%) patients randomized to FCM-R attained a CR(i). To date, 3(20%) of the 15 patients who achieved a CR or CR(i) have died, compared to 21(68%) of the 31 who did not. Similarly, 5/15(33%) patients who achieved a CR or CR(i) have progressed, compared to 23/26(88%) of those with PR or SD.

An MRD-negative response was seen in 3(12%) and 5(19%) of patients receiving FCM or FCM-R respectively. To date, 1(13%) of the eight MRD-negative patients has died, but seven remain in remission after a median follow-up of 29 months (range 24–46). In contrast, 23/37(62%) and 31/37(84%) of the MRD-positive patients have died and progressed respectively.

At the time of reporting, the median follow-up for survivors was 38 months (range 24–46). 29 patients have died and 38 have progressed or died in total. Kaplan–Meier OS and PFS curves are presented in Figs 2 and 3 respectively.

Figure 2.

 Kaplan–Meier curve demonstrating the overall survival from randomization by treatment group, after 38 months median follow-up time for survivors. 29 of the 52 patients have died in total.

Figure 3.

 Kaplan–Meier curve demonstrating the progression-free survival from randomization by treatment group, after 38 months median follow-up time for survivors. 38 of the 52 patients have progressed or died in total.

The ORRs in all planned clinical subgroups (range 57–64%) were of a similar magnitude to the ORR in all trial patients (62%). Of the 39 patients with standard risk disease, the ORR was 69% (95% CI: 52%, 83%), with no difference between the arms. Of the 11 patients with deletion of chromosome 11q, 73% (95% CI: 39%, 94%) achieved at least a PR; 100% (5/5) of those randomized to FCM-R (1 = CR, 3 = CR(i), 1 = PR), and 50% (3/6) of those randomized to FCM (1 = CR, 2 = PR, 1 = SD, 1 = PD, 1 = died).

Safety

Twenty-nine patients who were taking part in the trial have died: 13 of the patients received FCM, and 16 received FCM-R. The primary cause of death for four patients was infection due to treatment, one patient received FCM, and three received FCM-R. None of the other deaths were thought to be treatment-related. There were no treatment-related deaths in the stage I analysis population. The stopping rules for TRM in the FCM-R arm were not triggered since there were no treatment-related deaths in the first two patients, one in the first 14(7%) patients and three in all 26(12%) patients.

Forty-one Serious Adverse Events (SAEs) were reported from 27(52%) patients (Table III). Nineteen of the events were from 13(50%) patients who received FCM, and 22 were from 14(54%) patients who received FCM-R. Thirty-one SAEs were suspected to be related to protocol treatment; 19(61%) in patients who received FCM-R and 12(39%) in patients who received FCM. Five of the SAEs were suspected to be related to rituximab, alone or in combination.

Table III.   Serious adverse events by treatment group: suspected relationships and body systems.
 FCM-RFCM N (%)Total N (%)
  1. F, fludarabine; C, cyclophosphomide; M, mitoxantrone; R, rituximab.

Relationship to experimental treatment
 Suspected, unexpected1 (4·5)0 (0·0)1 (2·4)
 Suspected, expected18 (81·8)12 (63·2)30 (73·2)
 Not suspected3 (13·6)7 (36·8)10 (24·4)
 Suspected: fludarabine3 (13·6)3 (15·8)6 (14·6)
 Suspected: rituximab2 (9·1)0 (0·0)2 (4·9)
 Suspected: F, C, M and R3 (13·6)0 (0·0)3 (7·3)
 Suspected: F, C, and M10 (45·5)8 (42·1)18 (43·9)
 Suspected: F and C0 (0·0)1 (5·3)1 (2·4)
 Suspected: other combination of products1 (4·5)0 (0·0)1 (2·4)
 Not suspected: CLL related1 (4·5)3 (15·8)4 (9·8)
 Not suspected: uncertain0 (0·0)2 (10·5)2 (4·9)
 Not suspected: other reason2 (9·1)2 (10·5)4 (9·8)
 Total22 (100)19 (100)41 (100)
Suspected SAEs by body system
 Inflammatory and immunosuppressive disorders2 (10·5)0 (0·0)2 (6·5)
 Blood and lymphatic system disorders (haematological disorders)12 (63·2)3 (25·0)15 (48·4)
 Gastrointestinal disorders0 (0·0)3 (25·0)3 (9·7)
 Renal and urinary disorders1 (5·3)1 (8·3)2 (6·5)
 Respiratory, thoracic and mediastinal disorders4 (21·1)4 (33·3)8 (25·8)
 Skin and subcutaneous disorders0 (0·0)1 (8·3)1 (3·2)
 Total19 (100)12 (100)31 (100)

One Suspected, Unexpected, Serious Adverse Reaction (SUSAR) was reported for a patient who received FCM-R. The patient suffered persistent neutropenia that was deemed to be related to fludarabine, but the length of time that the neutropenia persisted after the end of treatment was unexpected.

Two-hundred and seventy-six adverse events (AEs) were reported from 24 patients who received FCM-R, and 211 were reported from 23 patients who received FCM (Table IV). Of these, 69 events from 28 patients (14 from each trial arm) were grade 3 or 4 neutropenia.

Table IV.   Adverse events by treatment group: CTC grades and causalities.
 FCM-R (%)FCM (%)Total (%)
  1. CTCAE, common toxicity criteria for adverse events; NA, not applicable.

CTCAE grade
 00 (0·0)2 (0·9)2 (0·4)
 1127 (46·0)91 (43·1)218 (44·8)
 278 (28·3)58 (27·5)136 (27·9)
 343 (15·6)32 (15·2)75 (15·4)
 422 (8·0)28 (13·3)50 (10·3)
 NA1 (0·4)0 (0·0)1 (0·2)
 Missing5 (1·8)0 (0·0)5 (1·0)
 Total276 (100)211 (100)487 (100)
Causality
 Almost certainly related to the trial drugs117 (42·4)100 (47·4)217 (44·6)
 Probably related to the trial drugs70 (25·4)51 (24·2)121 (24·8)
 Possible related to the trial drugs40 (14·5)15 (7·1)55 (11·3)
 Unlikely to be related to the trial drugs13 (4·7)18 (8·5)31 (6·4)
 Unrelated to the trial drugs31 (11·2)23 (10·9)54 (11·1)
 Missing5 (1·8)4 (1·9)9 (1·8)
 Total276 (100)211 (100)487 (100)

The reported adverse event and death information suggests that the safety profile of FCM-R is similar to that of FCM.

Discussion

The data from randomized phase III trials now suggest that addition of rituximab to fludarabine and cyclophosphamide is beneficial in the treatment of relapsed/refractory CLL (Robak et al, 2008). Non-randomized trials have shown that a combination chemotherapy with fludarabine, cyclophosphamide and mitoxantrone (FCM) is very effective in this setting (Wierda et al, 2005). However, to date there is no randomized trial reporting the addition of rituximab to this combination. This trial has demonstrated that the ORR to both FCM and FCM-R in patients with previously treated CLL are high, with 58% and 65% of patients responding respectively. Although the trial was not designed to have power to formally test for a difference between the arms, it is intriguing that comparison with the concurrent randomized control arm indicated higher CR rates, particularly including those with incomplete marrow recovery, which were 27%(95% CI: 3·4%, 50·4%) higher for FCM-R (42%) compared to FCM (15%). In addition, the toxicity of both regimens was acceptable. The FCM arm in this trial demonstrated lower ORR and CR rates compared to the previous trials, which may be due to the differences between patient groups, since around 70% of patients entered into the trial had two or more prior treatments and around 60% had previous fludarabine. In addition, the median age in our trial, 68 years, was older than other studies of fludarabine-based combination therapy in CLL (Catovsky et al, 2007). This reinforces the need for the concurrent control arm. This trial provides strong evidence that the addition of rituximab adds to the response rates in previously treated CLL and that these regimens can be safely used. Our data suggests that both the addition of mitoxantrone to chemotherapy as well as the addition of rituximab to fludarabine-based combinations should be tested in larger studies.

Acknowledgements

The authors would like to thanks all patients and hospital staff who contributed to this study. In addition, they acknowledge the invaluable support provided by the independent Data Monitoring and Ethics Committee (DMEC) and to Roche Pharmaceuticals for providing rituximab for the trial as well as an unrestricted grant to support the running of the trial. The study was included in the National Cancer Research Network Trials Portfolio.

Ancillary