Introduction

Induction therapy (IT) followed by high-dose melphalan (HDM) and autologous stem cell transplantation (ASCT) is the standard of care for younger multiple myeloma (MM) patients.1, 2, 3 ITs provide a crucial foundation for modern multi-drug-based myeloma treatment approaches, progressively inducing high rates of deep responses,4, 5, 6, 7, 8 thereby leading to prolonged progression-free survival (PFS) and overall survival (OS).3, 9, 10, 11, 12

Bortezomib/dexamethasone-based IT regimens are commonly combined with either cytotoxic agents, such as doxorubicin6, 7 or cyclophosphamide,5, 13, 14, 15, 16 or immunomodulatory drugs4, 17 (IMiDs, thalidomide and lenalidomide). The superiority of bortezomib-containing regimens to ITs without novel agents has been demonstrated in a number of phase III trials.4, 7, 8, 9, 13, 17 Until now, however, there has been no reported randomized phase III trial comparing a bortezomib/dexamethasone-based IT, with either doxorubicin or cyclophosphamide.

The open-label, randomized, multicenter phase III clinical trial MM5 of the German-speaking Myeloma Multicenter Group (GMMG, EudraCT No. 2010-019173-16) was designed to assess two independent primary end points:

1. Demonstration of non-inferiority of bortezomib/cyclophosphamide/dexamethasone (VCD) IT compared to bortezomib/doxorubicin/dexamethasone (PAd) IT with respect to response rates (very good partial response (VGPR) or better, ⩾VGPR).

2. Determination of the best of four treatment strategies with respect to PFS. The treatment strategies are defined by PAd or VCD IT followed by standard intensification therapy (HDM+ASCT), lenalidomide consolidation and maintenance treatment with either lenalidomide for 2 years or lenalidomide until complete response (CR) is achieved.

Herein we report the results of the first primary end point of the MM5 trial.

Patients and methods

Eligibility criteria

The key inclusion criteria were patients 18–70 years of age with newly diagnosed MM18 who require systemic chemotherapy based on ‘CRAB’ criteria;18 World Health Organization (WHO) performance status 0–2 or 3, if MM related; and measureable MM disease.19 Key exclusion criteria included: systemic light chain amyloidosis; peripheral neuropathy/neuropathic pain ⩾2° (National Cancer Institute Common Terminology Criteria for Adverse Events, NCI CTCAE, version 4.0). Patients with renal impairment (RI) or renal failure were not excluded from the study. Detailed inclusion/exclusion criteria are provided in the study protocol (see online Supplementary Material 1).

Study design and treatment

The MM5 trial is a prospective, open-label, randomized multicenter phase III clinical trial (EudraCT No. 2010-019173-16). A total of 31 transplant centers and 75 associated sites throughout Germany are participating in this trial. The study was initiated by the GMMG and approved by ethics committees in the University of Heidelberg and at all participating sites. The MM5 trial is conducted according to the European Clinical Trial Directive (2005) and the Declaration of Helsinki.

Five hundred and four patients were included in the trial between July 2010 and October 2012. At present the study is ongoing. Patients were equally randomized to each of the four treatment arms (A1, A2, B1 and B2; Figure 1) using block randomization, stratified by the International Staging System (ISS) stage.20 Treatment consisted of either three 4-week cycles of PAd (A1+B1) or three 3-week cycles of VCD (A2+B2). Thereafter, standard intensification according to local protocols (GMMG standard) was performed, including stem cell mobilization and leukapheresis followed by single HDM+ASCT or, for patients not achieving near CR (nCR) or better, tandem HDM+ASCT. Subsequently, consolidation therapy consisted of two cycles of lenalidomide (25 mg, days 1–21) followed by lenalidomide maintenance (for the first 3 months 10 mg/day continuously and thereafter 15 mg/day continuously) for either 2 years (A1+A2) or until CR (B1+B2).

Figure 1
figure 1

Flow chart and consort diagram of the GMMG-MM5 trial. (a) Flow chart: randomization was performed prior to start of induction therapy. Maintenance therapy consisted of either lenalidomide for 2 years or lenalidomide if no complete remission (CR) was achieved. (b) Consort diagram: 502 patients were randomized and received the hence displayed treatment. Abbreviations: ASCT, autologous stem cell transplantation; CAD, cyclophosphamide/doxorubicin/dexamethasone; G-CSF, granulocyte-colony stimulating factor (lenograstim); ITT, intention-to-treat population; MM, multiple myeloma; nCR, near complete remission; PAd, bortezomib/doxorubicin/dexamethasone; PP, per-protocol population; safety, safety population; VCD, bortezomib/cyclophosphamide/dexamethasone.

Full size image

The PAd IT consisted of bortezomib 1.3 mg/m2 on days 1, 4, 8 and 11; doxorubicin 9 mg/m2 intravenously (i.v.) on days 1–4; and oral (p.o.) dexamethasone 20 mg on days 1–4, 9–12 and 17–20 (240 mg/cycle, repeated every 28 days). VCD consisted of bortezomib 1.3 mg/m2 on days 1, 4, 8 and 11; cyclophosphamide 900 mg/m2 i.v. on day 1; and p.o. dexamethasone 40 mg on days 1–2, 4–5, 8–9 and 11–12 (320 mg/cycle, repeated every 21 days; see Figure 2). Antibiotic (cotrimoxazole) and antiviral (acyclovir) prophylaxis was mandatory throughout IT. Intravenous bisphosphonate administration was recommended every 4 weeks. The route of administration of bortezomib was changed from i.v. to subcutaneous (s.c.) in all study arms following a protocol amendment in February 2012 after the enrollment of 314 patients. Thereafter, equal numbers of patients in the VCD and PAd groups received s.c. bortezomib. These results are based on the dataset as of July 2013. At that time, all patients had completed IT and had either proceeded under maintenance/consolidation treatment or were in follow-up.

Figure 2
figure 2

Induction treatments. Schematic view of induction therapy regimens within the GMMG-MM5 trial.

Full size image

Response assessments and end points

The response was assessed by a local investigator and verified by a medical monitor according to Durie et al.19 (International Myeloma Working Group criteria). In addition, the European Society for Blood and Marrow Transplantation criteria21 were implemented to assess minimal response. nCR was defined as the absence of serum and urine M-protein on standard electrophoresis and/or standard 24-h urinary measurement with a positive or missing immunofixation status in the serum and/or urine. For the confirmation of CR, negative serum and urine immunofixation as well as cytological bone marrow assessment were required. Enriched CD138+ plasma cells from bone marrow aspirates were analyzed according to the standardized interphase fluorescence in-situ hybridization (iFISH) procedure.22

Statistical design and analysis

As mentioned earlier, the MM5 trial is designed to address two independent primary objectives.

To guarantee a family-wise error rate of 5.0%, each primary objective is tested with the two-sided alpha level of 2.5%. The first primary objective was tested in a group-sequential way with a significance level split into 0.1% for the interim and 2.4% for the final analysis. The interim analysis was based on 75 patients in each IT with the possibility to stop for futility if a single arm showed <30% response.

In the final analysis, the non-inferiority analysis was performed for the intention-to-treat (ITT) and the per-protocol (PP) population with a non-inferiority margin of 10% for the difference in ⩾VGPR rates. Therefore two-sided confidence intervals were calculated by using the Newcombe's hybrid score interval.23 The test of Farrington and Manning24 was used to test the one-sided null hypothesis of the non-inferiority of VCD to PAd. The two-sided significance level for this final analysis was set to 2.4%, the one-sided level accordingly to 1.2%. To demonstrate non-inferiority of VCD, non-inferiority for both ITT and PP populations needs to be confirmed. Adverse events (AEs) are summarized per patient and Fisher’s exact test is used to compare AE frequencies and response rates. Unless indicated otherwise, numbers from the ITT population are listed. All analyses were carried out using R versions 2.14.0 and 2.15.3 (www.r-project.org).

Results

Patients and adherence to treatment

Of the 504 patients randomized to the four treatment arms (Figure 1a), two were excluded from ITT due to a violation of inclusion criteria. One of these patients has still been included in the safety population. One ITT patient was randomized to PAd but received VCD instead and was therefore excluded from PP but included in the ITT and the safety populations. In total, 251 patients were randomized to the PAd and VCD arms, respectively. In both arms, 249 patients were evaluable for the safety analysis. Nineteen (PAd) and 12 patients (VCD) were excluded from the PP analysis. ITT patients with less than three IT cycles due to progressive disease (PD) were included in the PP assessment. A total of 233 patients in the PAd group (92.5%) and 240 patients in the VCD group (95.2%) completed all three planned IT cycles and subsequent response assessment. After IT, eight patients in each treatment arm (PAd and VCD) left the study due to high-risk disease and received allogeneic transplantation. The consort diagram is displayed in Figure 1b.

Baseline characteristics are listed in Table 1. Patients in the VCD arm were slightly younger and had a more favorable distribution of WHO performance status scores.

Table 1 Baseline patient and disease characteristics
Full size table

Trial medication

The mean delivered doses of bortezomib did not differ between treatment groups in all the three cycles (mean values (mg); cycles I/II/III: PAd: 9.6/9.6/9.4 versus VCD: 9.6/9.5/9.9).

The proportions of patients who received a scheduled or delayed full dexamethasone dose in the PAd and VCD arms were similar (% receiving full dexamethasone dose; cycles I/II/III: PAd: 81.5/80.2/85.9 versus VCD: 81.5/82.2/84.8).

A full doxorubicin dose was applied to 94.3/88.0/91.9% of patients in cycles I/II/III, respectively, in the PAd group. A full cyclophosphamide dose was administered to 89.9/82.6/86.4% of patients in cycles I/II/III, respectively, in the VCD group.

Primary end point

In the final analysis of the primary objective, non-inferiority of VCD compared to PAd was established with the lower confidence limit for the difference in ⩾VGPR rates exceeding the non-inferiority margin of −10.0% in ITT (VCD-PAd: 2.8% (−6.8%; 12.3%), P=0.001) and PP analyses (VCD-PAd: 1.4% (−8.6%; 11.4%), P=0.005).

Actual ⩾VGPR rates were 34.3/36.9% (PAd) and 37.0/38.3% (VCD) for ITT and PP, respectively (Figure 3).

Figure 3
figure 3

Response rates after three cycles of PAd or VCD induction therapy. ITT population consisted of 502 patients; PP population consisted of 473 patients. Graphs display percentage of patients achieving the indicated response in PAd or VCD induction therapy arm for ITT population (a) and PP population (b). Abbreviations: CR, complete remission; ITT, intention-to-treat population; missing, no response assessment performed; MR, minimal response; nCR, near complete remission; PAd, bortezomib/doxorubicin/dexamethasone; PD, progressive disease; PP, per-protocol population; PR, partial response; SD, stable disease; VCD, bortezomib/cyclophosphamide/dexamethasone; VGPR, very good partial response.

Full size image

Response rates

Exploratory analyses of response rates after IT are shown in Table 2. Rates of PR or better (⩾PR, defined as overall response rate, ORR) did not differ in PAd 72.1% versus VCD 78.1%, P=0.15.

Table 2 Response rates after induction therapy
Full size table

PD occurred in 4.8 and 0.4% of patients within PAd and VCD arms (P=0.003). Among patients presenting with at least one adverse cytogenetic aberration (del17p, t(4;14), gain 1q21>2 copies), PD rates were 6.2 versus 0.0% (P=0.01, PAd versus VCD). Among patients with RI (serum creatinine ⩾2 mg/dl), PD rates were 10.8 versus 0.0% (P=0.05, PAd versus VCD).

The response rates for the PAd and VCD groups based on ISS staging are shown in Supplementary Table 1. The PD rates for patients with ISS stage III were significantly higher in the PAd group (9.7 versus 1.3%, P=0.03).

Safety and toxicity

Safety data are shown in Table 3. Application of VCD led to a significant higher proportion of leukocytopenia and/or neutropenia (CTCAE ⩾3°, VCD 35.2% versus PAd 11.3%, P=0.001).

Table 3 AE, SAE and deaths during PAd and VCD induction therapies
Full size table

Neuropathy (CTCAE ⩾2°) was observed more frequently in the PAd arm than in the VCD arm (14.9 versus 7.6%, P=0.03).

The number of patients with at least one serious AE (SAE) during IT was significantly higher in the PAd group (32.7 versus 24.0%, P=0.04). SAEs due to thromboembolic events (including pulmonary embolism) were significantly more frequent in the PAd than in VCD arm (2.8 versus 0.4%, P=0.04).

During IT, six patients in the PAd group (2.4%) and one patient in the VCD group (0.4%) died (Table 3). In the PAd group, three patients died due to infections (sepsis, pneumonia and infection of an internal spinal fixation device), one due to pulmonary embolism, one because of suspected pulmonary embolism and one due to cardiac failure and concurrent PD. One patient in the VCD group died due to an infection (pneumonia).

Stem cell mobilization and collection

Stem cell mobilization and collection were effective and feasible in both arms: 86.9% patients (PAd group) and 88.4% (VCD group) collected at least one transplant (>2.0 × 106 CD34+ cells per kg bodyweight; Table 4).

Table 4 Stem cell collection after induction therapy
Full size table

Discussion

Since the introduction of thalidomide in 1999 ushered in the era of novel agents in MM,25 treatment options and outcomes have markedly improved.26 New challenges, such as long-term disease control, have been defined.27 The depth and duration of response are crucial steps to achieve long-term responses.10, 28, 29, 30

Bortezomib/dexamethasone in combination with either IMiDs,4, 17 alkylating agents5, 13, 14, 15, 31 (cyclophosphamide) or anthracyclines6, 7 (doxorubicin) is frequently used as IT. Decades of clinical experience and lower costs support the use of classical cytotoxic agents.

In line with our results, using an identical VCD regimen and number of cycles, Einsele et al.,16 reported an ORR of 84%. A phase II trial in the Mayo Clinic used an intensified (1500 mg/m2 per cycle cyclophosphamide) and prolonged cyclophosphamide/bortezomib/dexamethasone/VCD treatment regimen that, as would be expected, resulted in a higher ORR/⩾VGPR rate (88/61%)13, 14, 32 when compared to our study (ORR/⩾VGPR rate 78/37%). Kumar et al.5 assessed two VCD regimens with 1000 mg/m2/cycle (VCD) and 1500 mg/m2/cycle (VCD-mod) cyclophosphamide. The efficacy of the VCD-mod regimen (ORR/⩾VGPR rate 82/41%) was comparable to that of our study, which was achieved with a single i.v. administration of 900 mg/m2 cyclophosphamide. However, these previous studies included rather small numbers of patients, HDM/ASCT was not necessarily part of the first-line treatment and IT or maintenance treatment could be applied continuously.

Whether p.o. or i.v. cyclophosphamide should be applied remains difficult to determine. The applied cyclophosphamide doses and intensities vary between the investigated VCD regimen and no randomized studies exist. However, the split p.o. application of cyclophosphamide5, 14 might increase efficacy through a steady distribution and increased total cyclophosphamide dose in comparison to the high blood levels achieved with a single i.v. injection. Further studies are required to answer these questions.

Combinations of an IMiD and a proteasome inhibitor (PI), namely bortezomib/dexamethasone with either thalidomide (VTD) or lenalidomide (RVD33, 34 or VRD5) and carfilzomib/lenalidomide/dexamethasone (CRD) are progressively used as IT.

Three phase III trials investigating VTD IT (three to six cycles), reported higher ORR (85, 93 and 88%) and ⩾VGPR rates (60, 62 and 49%)4, 17, 35 compared to both IT administered in the present study (Table 5). No prospective trials have compared VCD versus VTD IT. The addition of cyclophosphamide to VTD, however, did not improve response rates but increased toxicity.36

Table 5 Comparison of different induction therapy regimen used in newly diagnosed multiple myeloma
Full size table

RVD/VRD has been investigated in three phase I/II trials with an ORR/⩾VGPR rate of 73/32%,5 75/11%33 and 93/58%34 after three to four cycles (Table 5). These response rates are, except for one study, lower compared to the present VCD regimen. However, with increasing cycle numbers, including upfront HDM/ASCT and/or consolidation/maintenance therapies, RVD/VRD regimens achieved an ORR between 85 and 100% and ⩾VGPR rates between 51 and 87%.5, 33, 34 Similarly, CRD yielded an ORR/⩾VGPR rate of 98/58% after a median of 12 cycles.37

Of note, many of these studies permitted the continuation of IT or consolidation/maintenance treatment beyond three to six IT cycles without the need for frontline or later HDM/ASCT.5, 17, 33, 34, 37 It is therefore difficult to compare the results and response rates of these studies to our present trial. In addition, PFS and OS results from our trial remain to be seen.

It should be noted that despite the high efficacy of VTD, VRD and CRD, there are no data available on whether the administration of two novel agents in first-line therapy has an adverse effect on the second-line/third-line PFS and whether OS is improved in comparison to IT/consolidation therapies containing either an PI, PI/alkylator or an IMiD (and spare an IMiD or PI for relapse). This is supported by the EVOLUTION trial, which demonstrates neither PFS nor OS differences between VRD and VCD/VCD-mod-treated cohorts.5 Moreover, recent data from Reeder et al.38 demonstrate an excellent 5-year OS of 70% for VCD-treated patients. Furthermore, two phase III trials investigating VTD versus bortezomib/dexamethasone or a bortezomib/dexamethasone/alkylator-containing regimen found no PFS/OS and no OS differences.4, 35 Prospective trials with a long-term follow-up and comparable settings (for example, comparable IT/consolidation cycle number and rate of upfront HDM/ASCT) are therefore needed to determine the role of PI/alkylator versus PI/IMiD induction/consolidation therapies.

Likewise, studies evaluating VRD and CRD included relatively small patient numbers. Results from randomized phase III trials are pending, but these regimens are cost intensive in comparison to a PI/alkylator-containing regimen such as VCD.

The lower ORR/⩾VGPR rates achieved with PAd in our present study (72/34%) when compared to PAd used in our previous HOVON65/HD4 study7 (78/42%), may be due to the reduced dose of dexamethasone (MM5: 240 mg/cycle and HOVON65/HD4: 480 mg/cycle).

The ORR and ⩾VGPR rates observed in our current trial and perhaps PFS might be increased upon application of more cycles. Therefore, we recommend, according to recent guidelines,1, 2 at least four cycles of VCD as pretransplantation therapy.

The clinical value of the not significantly different ORR between the VCD and PAd IT slightly favoring the VCD IT cannot be determined yet. Different cycle lengths and bortezomib dose intensities as well as the different dexamethasone doses (VCD 320 mg/cycle and PAd 240 mg/cycle; Figure 2) might influence comparability of efficacies in our current trial. Furthermore, an earlier response assessment due to a shorter cycle length in the VCD group might underestimate response in comparison to the PAd group, since the half-life time of monoclonal immunoglobulins can be up to 21 days.

We observed a significantly higher PD rate in the PAd group (PAd: 12 patients/4.8% versus VCD: 1 patient/0.4%; P=0.003). In the HOVON65/GMMG-HD4 trial,7 the PD rate during the PAd IT was 1.0%. Rates of PD in trials assessing VCD IT were 0.0%5 and 2.3%.16 In the current trial, 91.7% of the patients with PD in the PAd group had either RI at diagnosis or adverse cytogenetics. In the VCD group, there was no PD observed in these subgroups. Among patients with RI in the PAd group, 24.3% had either PD (10.8%) or stable disease (13.5%), whereas only 5.1% had stable disease and none had PD in the VCD group. Gain of 1q21, displaying an adverse prognostic factor,22 was present in all patients with adverse cytogenetics in the PAd group incurring PD (data not shown).

Together with the improved ORR, our results provide support for the use of VCD as an IT in patients with initial RI and adverse cytogenetics. However, follow-up data of this trial should be awaited to further interpret the impact of the two different IT for patients with adverse prognostic factors.

The rates of leukocytopenia/neutropenia were significantly higher in the VCD group compared to the PAd group (35.2 versus 11.3%, P<0.001). The scheduled total dose of dexamethasone per cycle was higher in the VCD than the PAd arm (320 versus 240 mg; Figure 2). This did, however, not translate into higher infection rates or infection-related SAEs (Table 3). In this study, we aimed at reducing the infection rate (CTCAE⩾2°) of 49% as seen with the use of PAd in the HOVON65/GMMG-HD4 trial7 by lowering the total dexamethasone dose. As intended, the rate fell to 24.6% in the PAd group.

Cyclophosphamide is the main cause for neutropenia observed with VCD. The rates of leukocytopenia/neutropenia in those who received the VCD regimen were comparable to some other studies.5, 16 The study from Reeder et al.14 using a cycle length of four weeks and p.o. cyclophosphamide observed lower rates of neutropenia (13%). Perhaps, splitting of cyclophosphamide doses can reduce the depth of leukocytopenia/neutropenia. Therefore, a weekly application of cyclophosphamide p.o. appears favorable and safe.5, 14

Neuropathy occurred more frequently in the PAd when compared to the VCD group (14.9 versus 8.4%, P=0.03). The applied doses of bortezomib did not differ between the two arms.

Similarly, low neuropathy rates as observed in our trial were reported by other trials investigating VCD (Table 5).5, 14, 16 Reeder et al.32 demonstrated that neuropathy rates can even be decreased without a loss of efficacy by a weekly (versus biweekly) application of bortezomib.

In the VCD group, the higher dose intensity of dexamethasone (VCD 106 mg/week versus PAd 60 mg/week; Figure 2) and cyclophosphamide, both immunosuppressive agents, might have conferred protection against the inflammatory component of bortezomib-induced peripheral neuropathy.39, 40 The shorter scheduled bortezomib treatment-free time in the VCD arm (VCD 10 days versus PAd 17 days; Figure 2) appears not to have offset this effect, though a reduced dose intensity of bortezomib is reported to result in lower neuropathy rates.32 Perhaps, the addition of doxorubicine to bortezomib might further enhance the neurotoxicity of PAd compared to VCD.

In the prospective, randomized MMY-3012 trial, Moreau et al. demonstrated non-inferiority of s.c. versus i.v. bortezomib regarding efficacy and PFS in relapsed/refractory MM. They also observed fewer AE and neurotoxicity/bortezomib-induced peripheral neuropathy in the s.c. group.41 Based on these results, we changed the route of administration of bortezomib in the MM5 trial from i.v. to s.c. The number of s.c. treated patients was balanced between the PAd and VCD groups.

In a recent subgroup analysis of the MM5 trial, comparing i.v. versus s.c. application of bortezomib during IT, we demonstrated that ORR was not affected by i.v. or s.c. application of bortezomib. Further, AEs were more frequently observed in i.v. treated patients as well as neuropathy rates in the third cycle of IT.42 Together with limited data from other centers incorporating s.c. bortezomib in VCD or VTD regimen,43 our data suggest that efficacy of s.c. bortezomib is retained in newly diagnosed MM. To date, no comparison between s.c. and i.v. bortezomib in large clinical trials of newly diagnosed MM have been published.

It should be noted that profound changes in the NCI CTCAE classification of neuropathy from version 3.0 (2006, applied in the HOVON65/GMMG-HD4 trial7) to version 4.0 (2010, applied in the MM5 trial) affect comparability of neuropathy grading. However, neuropathy rate in the PAd group decreased in comparison to the PAd in the HOVON65/GMMG-HD4 trial (24%)7 and is similar to a study using bortezomib/liposomal doxorubicin/dexamethasone (Table 5).6 The increased familiarity of study site attending physicians with emerging bortezomib-induced peripheral neuropathy may have contributed to the low neuropathy rates in the VCD group and the decreased rates in PAd (MM5) versus PAd (HOVON65/GMMG-HD4) group as observed previously.44

The increased number of thromboembolic AEs and SAEs in the PAd group demonstrates the well-described pro-thrombotic activity of doxorubicin.45, 46

Cardiotoxic events were rarely observed with either IT regimen and completely resolved in all but two cases.

The higher total SAE numbers might have resulted from the prolonged duration of treatment in the PAd group. In addition, the VCD regimen is easier to deliver than the PAd regimen, with only a single infusion per cycle.

The favorable toxicity profile of VCD would support the use of more than three cycles, for example, four to six, to further improve pretransplantation response rates and PFS.5, 14, 36

In summary, the MM5 trial demonstrates that VCD IT is not inferior to PAd IT in terms of achieving ⩾VGPR rates. Additional analyses revealed a trend toward higher ORR. PD rates in the VCD group were lower, especially in patients with gain of 1q21 and RI at diagnosis. Despite higher rates of leukocytopenia/neutropenia, which did not translate into increased infection rates, VCD has a favorable toxicity profile in terms of the total numbers of SAEs during IT, neuropathy rates and thromboembolic events. VCD is therefore an established standard IT in newly diagnosed transplant-eligible MM and is recommended to be preferred over PAd.