A Phase II, Randomized, Double-Blind, Placebo-Controlled Study of Siltuximab (Anti-IL-6 mAb) and Bortezomib versus Bortezomib Alone in Patients with Relapsed or Refractory Multiple Myeloma

Abstract

We compared the safety and efficacy of siltuximab (S), an anti-interleukin-6 chimeric monoclonal antibody, plus bortezomib (B) with placebo (plc)+B in patients with relapsed/refractory multiple myeloma in a randomized phase II study. Siltuximab was given by 6 mg/kg IV every 2 weeks. On progression, B was discontinued and high-dose dexamethasone could be added to S/plc. Response and progression-free survival (PFS) were analyzed pre-dexamethasone by EBMT criteria. For the 281 randomized patients, median PFS for S+B and plc+B was 8.0 and 7.6 months (HR 0.869, p=0.345), overall response rate was 55% vs. 47% (p=0.213), complete response rate was 11% vs. 7%, and median overall survival (OS) was 30.8 vs. 36.8 months (HR 1.353, p=0.103). Sustained suppression of C-reactive protein, a marker reflective of inhibition of interleukin-6 activity, was seen with S+B. Siltuximab did not affect B pharmacokinetics. S/plc discontinuation (75% vs. 66%), grade ≥3 neutropenia (49% vs. 29%), thrombocytopenia (48% vs. 34%), and all-grade infections (62% vs. 49%) occurred more frequently with S+B. The addition of siltuximab to bortezomib did not appear to improve PFS or OS despite a numerical increase in response rate in patients with relapsed or refractory multiple myeloma.

Introduction

Studies have long shown that the pleiotropic cytokine interleukin (IL)-6 plays an important role in the pathogenesis of multiple myeloma (MM), with proliferative and anti-apoptotic effects in neoplastic plasma cells [1]. Elevated serum IL-6 levels are associated with poor prognosis and short survival in advanced MM [2]. IL-6 promotes myeloma cell survival via phosphorylation of signal transducers and activators of transcription (STAT)-3 and up-regulation of anti-apoptotic molecules, such as myeloid cell leukemia-1 (Mcl-1) and c-Myc [3–5]. IL-6 also induces vascular endothelial growth factor expression in myeloma cells [6–8], contributing to the enhanced angiogenesis seen in myeloma.

Although the recent development of the proteasome inhibitor bortezomib has improved survival in patients with MM [9, 10], its efficacy is limited by a number of resistance mechanisms. One of the most important is the heat shock protein (HSP) and stress response pathways which, through members such as HSP-70 and mitogen-activated protein kinase (MAPK) phosphatase, oppose the pro-apoptotic activities of bortezomib [11, 12].

IL-6 inhibition was hypothesized to enhance the activity of bortezomib by interfering with the induction of the HSP response and Mcl-1. IL-6 activates STAT-1 [13], which in turn interacts with heat shock transcription factor (HSF)-1 to facilitate transcription of HSP-70 and HSP-90 [14]. Preclinical studies demonstrated that the addition of siltuximab (formerly CNTO 328), a chimeric (human-murine), anti-IL-6 monoclonal antibody, to bortezomib had an additive effect in inducing apoptosis in IL-6-dependent and IL-6-independent MM cell lines [11]. Treatment with siltuximab reduced bortezomib-induced HSP-70 and potently attenuated bortezomib-mediated increases in Mcl-1 by inhibiting IL-6−mediated downstream signaling pathways via STAT-1, STAT-3, and p44/42 MAPK phosphorylation.

A large, open-label, dose-finding phase I study of single-agent siltuximab was conducted in 67 patients with B-cell non-Hodgkin’s lymphoma, Castleman’s disease (CD), or relapsed MM. Siltuximab could be given up to 12 mg/kg once every 2 or 3 weeks without dose-limiting toxicity and over prolonged dosing with no evidence of cumulative toxicity [15]. The most frequently reported possibly drug-related adverse events (AEs) were transient, and reversible thrombocytopenia, neutropenia, hypertriglyceridemia, leukopenia, hypercholesterolemia, and anemia. Twelve of 36 evaluable CD patients showed radiologic response, most of whom were treated at 12 mg/kg. Two of 13 MM patients achieved complete response, and 1 MM patient had prolonged disease stabilization.

The purpose of this current study was to evaluate the safety and efficacy of the combination of siltuximab (S) and bortezomib (B) in patients with relapsed or refractory MM.

Methods

Patients

Patients were at least 18 years old and had a confirmed diagnosis of MM with measurable secretory disease (ie, serum M-protein ≥1 g/dL or urine M-protein ≥200 mg/24 hours). Patients must have had 1 to 3 prior lines of therapy, relapsed or refractory disease, and had undergone or were unsuitable for autologous hematopoietic stem cell transplantation (SCT). Other eligibility criteria included an Eastern Cooperative Oncology Group performance status (ECOG-PS) score of ≤2 and adequate organ function. Key exclusion criteria were prior bortezomib use, allogeneic transplantation, and chemotherapy washout of <30 days.

Patients provided written, informed consent. The study protocol was approved by the institutional review board or ethics committee for each site and was conducted in accordance with the Declaration of Helsinki.

Study Design

The first part of this study was an open-label, single-group, run-in to evaluate the safety of S+B. In the second part of this study, patients were randomized 1:1 to S+B or placebo (plc)+B, stratified by International Staging System (ISS) stage [16], prior autologous SCT (yes/no), and number of prior lines of therapy (1 or >1). During the randomized treatment phase, patients received blinded treatment in 42-day cycles for ≤4 cycles: with S 6 mg/kg or plc infused over 2 hours every 2 weeks (q2w) and B 1.3 mg/m² injected intravenously twice weekly (days 1, 4, 8, 11, 22, 25, 29, 32). Patients with stable disease (SD) or better could continue to receive maintenance therapy: S 6 mg/kg or plc q2w and weekly B (days 1, 8, 15, 22 every 5 weeks) until progressive disease (PD). For the first 113 patients enrolled, those with PD on B treatment were permitted, at the investigator’s discretion, to continue on assigned treatment with the addition of low-dose dexamethasone (dex; 20 mg on the day of and day after each B administration). To align the study protocol with clinical practice, it was thereafter amended to require B discontinuation at the time of PD or intolerable B-related toxicity (ie, after 2 B dose reductions), with optional high-dose dex salvage (on days 1−4, 9−12, 17−20 for four 28-day cycles, then on days 1−4 for subsequent cycles) while blinded S/plc was continued. Treatment-related toxicities were managed by protocol-specified dose delays (up to 14 days) for S and dose modifications for B and dex, and ultimately treatment discontinuation.

The focus of this report is the pre-dex portion of the blinded randomized treatment phase. Results from the open-label, safety run-in and the post-dex salvage phase are summarized separately.

Assessment of Efficacy

The primary endpoint in the randomized treatment phase was progression-free survival (PFS) in all randomized patients (the ITT population), according to European Bone Marrow Transplant registry group (EBMT) criteria [17] implemented by a validated computer algorithm and modified to include near complete response (nCR). Responses were evaluated for patients who had measurable disease at baseline, ≥1 study-agent administration, and ≥1 post-baseline disease assessment pre-dex (the response-evaluable population). In the randomized treatment phase, major secondary efficacy endpoints were overall response rate (ORR; complete response [CR] + partial response [PR] before dex/study withdrawal) confirmed ≥6 weeks, 1-year overall survival (OS) rate, CR rate, and OS. Other efficacy endpoints were time to response, time to progression (TTP), and response duration. Response was assessed q6w using laboratory disease assessments evaluated by a central laboratory.

Assessment of Safety, Pharmacokinetics, Immunogenicity, Pharmacodynamics, and Patient-Reported Outcomes

AEs were recorded through 30 days after the last study-agent dose according to the National Cancer Institute Common Terminology Criteria for Adverse Events (NCI-CTCAE) v3.0. Safety assessments included routine laboratory tests, physical examination, vital signs, electrocardiography, and recording of concomitant medications. All patients who received ≥1 dose of study drug were included in the safety analysis.

Blood samples were collected from all patients in the safety run-in and a subset of patients in the randomized treatment phase during cycle 1 to evaluate plasma B concentrations using an assay based on a validated liquid chromatography-mass spectrometry/mass spectrometry method. The B pharmacokinetic analysis included all patients who received ≥1 full dose of B and had B pharmacokinetic samples obtained, at a minimum, on cycle 1 days 4, 8, and 11. Blood samples were collected immediately following S dosing and before B administration in cycles 1, 2, and 3 from all patients in the safety run-in, and a subset of patients in the randomized treatment phase to evaluate serum S concentrations using a validated assay based on an electrochemiluminescent immunoassay format on the Meso Scale Discovery (MSD) platform. The S pharmacokinetic analysis included all patients who received ≥1 full dose of S and had ≥1 post-dose S pharmacokinetic sample.

Blood samples to evaluate immunogenicity were collected from all patients before the first S administration, at treatment discontinuation, and every 3 months (up to 3 times) after the last S dose, and when an infusion reaction occurred.

Blood samples were also collected on day 1 of each cycle from all patients for serum biomarker analysis of baseline IL-6 and percent change from baseline in C-reactive protein (CRP). Low- and high-molecular weight IL-6 complexes were measured using a newly developed, single-plex, panoptic IL-6 assay based on the MSD platform with a lower limit of quantification (LLOQ) of 9.77 pg/mL. Low-molecular weight IL-6 was also measured using the standard MSD Human IL-6 Ultra-Sensitive Kit. IL-6 was measured only at baseline because accurate quantification of IL-6 in post-treatment samples is not currently possible, as siltuximab-neutralized antibody-IL-6 complexes distort current immunological-based IL-6 quantification methods. CRP concentration was analyzed at a central laboratory (Covance) using a high-sensitivity CRP assay with a LLOQ of 0.20 mg/L. In the randomized treatment phase, patient-reported outcomes were assessed using the European Organization for Research on the Treatment of Cancer Quality of Life Questionnaire-Core 30 (EORTC QLQ-C30) [18], the Functional Assessment of Chronic Illness Therapy – Fatigue scale (FACIT-Fatigue) [19], and the pain intensity scale from the brief pain inventory (BPI) [20].

Statistical analyses

The sample size for the open-label, safety run-in was not based on hypothesis testing. An independent data monitoring committee (IDMC) reviewed safety results and decided that enrollment should begin for the randomized treatment phase. This phase enrolled approximately 270 patients (135/group) using dynamic randomization. The data cutoff for the primary analysis was to be after ≥193 PFS events had occurred and all ongoing patients had been treated for ≥6 months. Assuming 50% improvement in median PFS of the combination group over the control group, the study had at least 80% power to achieve statistical significance at a two-sided level of 0.05, under exponential distribution for PFS. Sensitivity analyses of PFS using Cox proportional hazards regression were performed according to pre-specified subgroups (age, gender, race, ECOG-PS, β₂-microglobulin level, ISS stage, prior SCT, number of prior therapies, BPI score, FACIT-Fatigue category and score). An interim futility analysis was performed by the IDMC approximately 2 months after 104 patients (39% of sample size) were randomized. The IDMC also monitored unblinded safety data in the randomized phase. Descriptive statistics were used to summarize data. For time-to-event variables, Kaplan-Meier estimates and 95% confidence intervals (CIs) were calculated. Analysis of variance was used to compare all continuous variables. The Cochran-Mantel-Haenszel test was used to compare discrete variables. Log-rank testing stratified by β2-microglobulin (< or ≥3.5 mg/L) and number of prior lines of therapy (1 or >1) was used for time-to-event variables. All tests were at a 2-sided α of 0.05.

Results

From November 2006 through July 2009, a total of 307 patients with relapsed or refractory MM were enrolled at 88 sites in 18 countries (Belgium, Brazil, Bulgaria, Canada, Czech Republic, France, Germany, Greece, Hungary, Netherlands, Poland, Portugal, Romania, Russia, Slovakia, Spain, United Kingdom, United States).

Safety Run-In

Twenty-one patients received S+B for a median of 4 cycles. The median duration of treatment with S and B was 142 and 144 days, respectively, and the majority of patients discontinued treatment due to AEs (29%) or PD (24%; Supplementary Table I). Demographics and disease characteristics are summarized in Supplementary Table II. At the interim analysis, the safety of S+B was deemed acceptable by the IDMC, and enrollment was initiated for the randomized treatment phase. AEs are summarized in Supplementary Tables III and IV.

Pre-Dexamethasone Randomized Treatment Phase

A total of 286 patients were randomized (ITT population: 142 S+B. 144 plc+B), and 281 were treated (safety population: 142 S+B, 139 plc+B; Fig 1). Five patients were not treated due to unmet entry criteria (n=3), AE (n=1), or withdrawal of consent (n=1). Three patients randomized to plc+B incorrectly received treatment with S+B at some time point during treatment; these patients were included in the plc+B group for efficacy analysis but in the S+B group for safety. Siltuximab/plc was discontinued in 106 (75%) S+B and 91 (66%) plc+B patients, mainly due to PD (20% vs. 22%), AE (18% vs. 16%), or CR (7% vs. 3%). Bortezomib was discontinued in 134 (94%) S+B and 125 (90%) plc+B patients, mainly due to PD (32% vs. 42%) or AE (28% vs. 21%). The majority of S/B discontinuations occurred during the first 4 cycles.

Fig 1.

Patient disposition in the randomized treatment phase pre- and post-dexamethasone. ITT, intention-to-treat.

Baseline characteristics of patients in the randomized treatment phase are presented in Table I. The median age was 64 years in S+B and 61 years in plc+B. The median duration since diagnosis was 2.7 years (range 0.2, 24) in S+B and 2.4 years (range 0.2, 14) in plc+B. Similar proportions of patients in S+B and plc+B had ISS stage III (26% vs. 25%) disease at randomization. The proportions of patients receiving 1 (49% vs. 53%), 2 (33% each), or 3 (18% vs. 14%) lines of prior therapy or prior SCT (36% vs. 38%) were comparable between groups. Common prior anti-myeloma therapies were corticosteroids (98%), alkylating agents (88%), anthracycline (46%), and thalidomide/lenalidomide (41%). In S+B and plc+B, 58% versus 51% had relapsed and refractory disease, defined as PD while on treatment (24% vs. 19%), SD on treatment (15% vs. 17%), or PD within 60 days (19% vs. 15%) of the last therapy. Baseline characteristics were similar between S+B and plc+B except for immunoglobulin A subtype (27% vs. 20%), ECOG-PS score of 2 (13% vs. 19%), age ≥65 years (48% vs. 40%), and FACIT-Fatigue score ≥30 (71% vs. 61%).

Table I.

Baseline characteristics in the pre-dexamethasone randomized treatment phase

	Siltuximab + Bortezomib	Placebo + Bortezomib
Patients treated
Patients randomized	142	144
Age (years)	64 [36, 82]	61 [37, 81]
Sex
Male	72 (51)	85 (59)
Female	70 (49)	59 (41)
Race
Caucasian	126 (89)	130 (90)
Black	6 (4)	6 (4)
Asian	2 (1)	2 (1)
Other	8 (6)	6 (4)
ECOG performance status score
0	36/141 (26)	44 (31)
1	86/141 (61)	72 (50)
2	18/141 (13)	28 (19)
≥3	1/141 (1)	0
Time from diagnosis (years) Immunoglobulin isotype*	2.7 [0.2, 24]	2.4 [0.2, 14]
Immunoglobin G	92/142 (65)	102/143 (71)
Immunoglobulin A	38/142 (27)	29/143 (20)
Light chain	10/142 (7)	10/143 (7)
ISS stage at randomization
I	48/141 (34)	44/143 (31)
II	57/141 (40)	63/143 (44)
III	36/141 (26)	36/143 (25)
Relapsed and refractory disease
PD on last prior therapy	34 (24)	27 (19)
SD on last prior therapy	21 (15)	24 (17)
PD within 60 days of last prior therapy	27 (19)	22 (15)
Prior lines of therapy
1	70 (49)	76 (53)
2	47 (33)	48 (33)
3	25 (18)	20 (14)
Prior treatment regimens
Autologous stem cell transplantation	51 (36)	54 (38)
Corticosteroids	138 (97)	143 (99)
Alkylating agents	120 (85)	131 (91)
Anthracycline	64 (45)	68 (47)
Thalidomide/lenalidomide	57 (40)	60 (42)
CRP serum concentration (mg/L)	2.88 [0.10, 161]	3.54 [0.10, 299]

^*In siltuximab + bortezomib, 1 patient had immunoglobulin D and 1 patient was not detected. In placebo + bortezomib, 2 patients had immunoglobulin D. Data presented as n (%) or median [range]. CRP; C-reactive protein; ECOG, Eastern Cooperative Oncology Group; ISS, International Staging System; PD, progressive disease; SD, stable disease.

Efficacy

The primary efficacy endpoint analysis in the ITT population was performed after 192 PFS events were observed in the pre-dex randomized treatment phase. Median PFS was 8.0 months in S+B and 7.6 months in plc+B (HR 0.869 [95% CI: 0.650, 1.162], p=0.345; Fig 2A). Forty-nine and 45 patients were censored, respectively, in S+B and plc+B groups due to initiation of dex (69% each), withdrawal of consent (14% vs. 20%), or clinical cutoff (16% vs. 11%). No significant differences in PFS were found between treatment groups according to pre-specified subgroup sensitivity analyses.

Fig 2.

Kaplan-Meier estimates of (A) progression-free survival (pre-dexamethasone) and (B) overall survival in the randomized treatment phase.

After 24.5 months of follow-up, the median OS in the S+B and plc+B groups was 30.8 versus 36.8 months (p=0.103; Fig 2B). Sixty-seven versus 53 deaths were observed, the majority during the follow-up period. The 1-year OS rate with S+B and plc+B was 81% versus 85% (p=0.380), and the 2-year OS rate was 57% versus 68% (p=0.060). From an ad-hoc analysis, an asymmetry in the use of rescue treatment was apparent between the S+B and plc+B groups with protocol-specified high-dose dex (32 vs. 46 patients) and with SCT (7 vs. 16 patients). An ad-hoc analysis of OS in patients who did not receive subsequent per-protocol high-dose dex salvage or SCT found no decrease in OS with S and numerically longer median OS for S+B (n=106) versus plc+B (n=85) (34.4 vs. 24.9 months, p=0.631).

One hundred thirty-one patients in S+B and 137 patients in plc+B comprised the tumor response-evaluable population. ORR was 55% versus 47% (p=0.213), CR was 11% versus 7% (p=0.351). Time to CR was shorter with S+B than plc+B (4.1 vs. 5.1 months). Near CRs without bone marrow confirmation were 4% versus 2%, and nCRs with positive immunofixation were 7% versus 5%. PR was seen in 44% versus 39%, and minimal response was seen in 6% versus 10%.

There were no statistically significant differences between the S+B and plc+B groups in the other secondary efficacy endpoints. Time to response (CR or PR) was 2.6 with S+B versus 4.0 months with plc+B (HR 1.298 [95% CI 0.923, 1.827], p=0.133). TTP (EBMT) was 9.2 versus 9.0 months (HR 0.849 [95% CI 0.632, 1.155], p=0.296) for the S+B and plc+B groups. Duration of response was 10.1 with S+B versus 8.7 months with plc+B (HR 0.784 [95% CI 0.511, 1.204], p=0.266). Serum M-protein response was evaluable for patients in the ITT population with measureable serum M-protein at baseline: in 131 and 133 patients in S+B and plc+B, respectively, and 24% versus 19% achieved 100% reduction while 58% versus 56% achieved ≥50% reduction. Urine M-protein response was evaluated in 10 patients in S+B and plc+B with non-measurable (<1 g/dL) serum M-protein at baseline; among them, 7 versus 2 patients achieved 100% reduction. The median time to subsequent treatment for MM was 10 versus 9.2 months (HR 0.872 [95% CI 0.665, 1.142], p=0.318) for the S+B and plc+B groups. No significant differences were observed between treatment groups for EORTC QLQ-C30, FACIT-Fatigue, or BPI.

According to ad-hoc analyses, differences were found between the 40% of patients enrolled in North America/Western Europe (NA/WE) and 60% of patients enrolled in the rest of the world (ROW; Supplementary Table V). A statistically significant difference in median PFS was observed between S+B and plc+B in NA/WE (9.7 vs. 5.6 months, p=0.010) but not in ROW (7.7 vs. 8.4 months). This discordant regional treatment effect is interpreted as heavily influenced by the longer PFS on plc+B in the ROW region compared with the PFS on plc+B in the NA/WE region. This may be related to the longer bortezomib treatment duration (158 days and 106 days for ROW and NA/WE, respectively) and the higher cumulative bortezomib dose (36.5 mg/m² and 21.1 mg/m², respectively) in the plc+B group in the ROW region. The prolonged bortezomib exposure probably masked the treatment effect that siltuximab was adding to the combination. Median OS was not different by treatment group in NA/WE but was worse with S+B in ROW, possibly related to asymmetric use of high-dose dex salvage in S+B versus plc+B (NA/WE: 16% vs. 29%, ROW: 28% vs. 33%), SCT (NA/WE: 5% vs. 10%, ROW: 5% vs. 12%) and subsequent lenalidomide (NA/WE: 56% vs. 57%, ROW: 13% vs. 12%).

Safety

In the pre-dex randomized treatment phase, the median duration of S treatment for S+B was 155 days, with median cumulative dose of 59.9 (range 6, 357) mg/kg. The median duration of B for S+B and plc+B was 135.5 and 155 days, respectively, with median cumulative dose of 28.9 (range 3, 129) mg/m² and 29.4 (range 1, 172) mg/m². Key safety events and common all-grade AEs are summarized in Table II. Almost all patients had ≥1 AE with S+B (99%) and plc+B (98%), including 95% and 91% patients with AEs reasonably related to B and 69% and 64% patients with AEs reasonably related to S. Common hematologic AEs were neutropenia (59% vs. 36%), thrombocytopenia (57% vs. 45%), anemia (31% vs. 29%), and leukopenia (25% vs. 10%). More patients in S+B than plc+B received colony-stimulating factors (38% vs. 28%), although use of anti-anemic preparations (ie, erythropoietins) was less with S+B (14% vs. 19%). Transfusion rates were similar (26% vs. 24%), with fewer S+B than plc+B patients receiving transfusions due to anemia (78% vs. 88%). The higher incidence of grade ≥3 AEs associated with S+B was primarily due to neutropenia (49% vs. 29%) and thrombocytopenia (48% vs. 34%); however, grade ≥3 bleeding events were infrequent (2% each), and only 1 patient per group had a grade 5 bleeding event. Infections were more frequent with S+B than plc+B (62% vs. 49%), although grade ≥3 infections were similar (16% vs. 14%) and grade ≥3 febrile neutropenia was infrequent (0.7% vs. 1.4%). Common infections included nasopharyngitis (11% vs. 9%), upper respiratory tract infection (11% vs. 7%), bronchitis (10% vs. 5%), urinary tract infection (8% vs. 4%), and pneumonia (each 6%). Other frequently reported AEs in S+B and plc+B were peripheral sensory neuropathy (49% vs. 51%, although grade ≥3 events were low: 12% vs. 10%), diarrhea (36% vs. 35%), fatigue (each 27%), decreased appetite (23% vs. 18%), neuralgia (22% vs. 23%), constipation (20% vs. 15%), and vomiting (20% vs. 19%). Hypercholesterolemia (1% vs. 4%) and hypertriglyceridemia (3% each) were low and similar between the S+B and plc+B groups.

Table II.

Key safety events in the pre-dexamethasone randomized treatment phase

	Siltuximab + Bortezomib	Placebo + Bortezomib
Patients treated	142	139
Patients with
AEs	140 (99)	136 (98)
AEs of grade ≥3	129 (91)	103 (74)
SAEs	41 (29)	33 (24)
SAEs of grade ≥3	38 (27)	29 (21)
AEs leading to discontinuation of siltuximab	34 (24)	27 (19)
AEs leading to discontinuation of bortezomib	47 (33)	33 (24)
AEs causing temporary dose interruption of siltuximab	86 (61)	62 (45)
AEs causing temporary dose interruption or modification of bortezomib	118 (83)	101 (73)
AEs leading to death	11 (8)	7 (5)
Patients with common adverse eventsa	140 (99)	136 (98)
Neutropenia	84 (59)	50 (36)
Thrombocytopenia	81 (57)	63 (45)
Peripheral sensory neuropathy	69 (49)	71 (51)
Diarrhea	51 (36)	48 (35)
Anemia	44 (31)	40 (29)
Fatigue	38 (27)	37 (27)
Nausea	38 (27)	40 (29)
Leukopenia	35 (25)	14 (10)
Decreased appetite	32 (23)	25 (18)
Neuralgia	31 (22)	32 (23)
Constipation	29 (20)	21 (15)
Vomiting	28 (20)	27 (19)
Pain in extremity	24 (17)	13 (9)
Arthralgia	22 (15)	13 (9)
Asthenia	21 (15)	26 (19)
Back pain	21 (15)	24 (17)
Pyrexia	9 (6)	25 (18)

Data presented as n (%) unless otherwise noted. AEs, adverse events; SAEs, serious adverse events.

^aReported by ≥15% of patients in either treatment group.

SAEs occurred in 29% compared with 24% of patients in the S+B and plc+B groups. AEs leading to discontinuation of S/plc were reported in 24% and 19% of patients, with peripheral sensory neuropathy (5% each) and thrombocytopenia (2% each) being the only events reported in ≥3 patients. AEs leading to discontinuation of B occurred in 33% and 24% of patients, with peripheral sensory neuropathy (9% vs. 10%), neuralgia (5% vs. 3%), and thrombocytopenia (3% vs. 2%) being the only events reported in ≥3 patients. Infusion reactions reasonably related to S/plc occurred in 5% of patients per group, and all were low-grade except 1 case of grade 3 diarrhea with S+B. Eighteen (11 S+B, 7 plc+B) patients died during the pre-dex treatment phase, including 4 patients in S+B and 2 patients in plc+B considered reasonably related to study treatment.

Slight increases in alanine aminotransferase and aspartate aminotransferase were similarly observed on day 1 of cycle 2 in the S+B and plc+B groups; however, mean values remained within normal range (<35 IU/L) throughout the randomized treatment phase. There was a steady increase from baseline in triglycerides that was similarly observed with S+B (48/142 patients) and plc+B (45/139 patients). Analysis of median neutrophil, platelet, and hemoglobin nadirs by cycle showed no evidence of cumulative toxicity for either treatment group. Median neutrophil and platelet nadirs were lower in the S+B than the plc+B group. There was little change from baseline in hemoglobin values, and the median nadirs were generally similar between groups.

An ad-hoc analysis of regional differences comparing NA/WE to ROW showed an overall lower incidence of AEs in ROW for both treatment groups. However, AEs leading to death were notably higher in the ROW. In the S+B group, 14% of AEs resulted in death in the ROW compared with 2% in NA/WE. The incidence of AEs leading to death was similar in the plc+B group (6% in ROW compared with 4% in NA/WE).

Post-Dexamethasone Salvage Phase

Seventy-eight (27%) of the 286 randomized patients stayed on study after PD or intolerable B-related toxicity and received dex salvage. Common reasons for initiating dex for patients in the S+dex and plc+dex groups were PD (76% vs. 78%) and toxicity (18% vs. 16%). The median treatment duration for S+dex was 84.5 (range 1, 309) days, with median cumulative S dose of 38.7 (range 0, 134) mg/kg; median treatment duration for plc+dex was 155 (range 1, 603) days. Seventy-six patients (33 S+dex, 43 plc+dex) discontinued S/plc due to PD (70% vs. 62%), withdrawal of consent (3% vs. 11%), or AE (9% vs. 7%). Five patients in S+B received combination S+B+dex; 11 patients in the plc+B group received combination plc+B+dex, with 3 patients per group continuing treatment after database cutoff.

The post-dex ORR (EBMT) for evaluable patients was 4/24 (17%) in S+dex and 9/36 (25%) in plc+dex (Supplementary Table VI). One patient per group had a CR.

The overall incidence of AEs post-dex was similar for the S+dex (91%) and plc+dex groups (87%; Supplementary Table VII). The incidence of SAEs post-dex was similar (24% each).

Pharmacokinetics, Immunogenicity, and Pharmacodynamics for the Overall Study

The pharmacokinetic parameter estimates for the first dose Cmax (92.97±78.00 µg/mL) and the seventh dose Cmax (156.56±54.622 µg/mL) of S in the randomized treatment phase were similar to the values previously reported in renal cell carcinoma patients using the same dose and schedule [21]; therefore, combination therapy with B did not appear to affect S pharmacokinetics. Accounting for inter-subject variability, plasma B concentrations were similar when dosed in combination with S and plc, eg, on cycle 1 day 32, the mean ± standard deviation was 59.0±69.1 ng/mL with plc+B and 56.2±75.4 ng/mL with S+B. Therefore, S did not appear to affect the B pharmacokinetic profile. Antibodies to S were not detected in the 118 patients with appropriate samples for evaluation.

Effective CRP suppression (>85% median percent change) was observed as early as cycle 2 day 1 with S+B treatment compared to a 6.5% decrease in the plc+B group at the same time point (Fig 3). CRP concentrations were suppressed to or below LLOQ (0.2 mg/L) with S+B throughout the treatment period in both phases of the study, whereas CRP concentrations remained variable and above LLOQ with plc+B. Exploratory analysis of baseline CRP concentrations did not show an association with ORR per EBMT criteria. Post-treatment CRP decrease was observed irrespective of clinical response.

Fig 3.

Mean (+ standard deviation) serum C-reactive protein (CRP) concentrations in the overall study. C, cycle; D, day.

Of the 272 patients tested for IL-6, 65 (24%) had evaluable baseline IL-6 concentrations above the LLOQ (9.77 pg/mL) of the panoptic IL-6 assay, with 33 of 65 patients showing >5-fold concentrations of IL-6 using the panoptic IL-6 assay compared to the standard MSD IL-6 assay. However, baseline circulating serum IL-6 concentrations, as determined by both assays, were not predictive of clinical response per EBMT criteria.

Discussion

In this randomized, controlled study of patients with relapsed or refractory MM who had received 1 to 3 prior therapies not containing bortezomib, the median PFS by EBMT criteria before dexamethasone salvage was not significantly improved when siltuximab was added to bortezomib (8.0 months) compared with bortezomib alone (7.6 months). Although combination treatment led to faster time to response and numerically longer duration of response than bortezomib alone, the response with bortezomib alone improved with prolonged treatment. Notably, the median PFS with single-agent bortezomib in this study is longer than in the APEX [9] and DOXIL-MMY3001 [22] studies, because patients in our study received bortezomib until PD, while those studies had a definite treatment cutoff, suggesting that longer bortezomib treatment had an enhanced effect. This effect was corroborated in our study by the regional differences in bortezomib exposure, showing that a higher cumulative bortezomib dose led to a greater PFS benefit without a remarkable increase in adverse effects or toxicities.

Although most efficacy endpoints were numerically better with combination therapy, a trend toward worse OS was observed with combination therapy. After 24.5 months of follow-up, median OS was 30.8 months with S+B and 36.8 months with plc+B. Although the median OS in the combination group was comparable to the survival rate observed in historical studies conducted with bortezomib in this population, the median OS of the bortezomib alone group was better than those observed in the APEX study (29.8 months) [9] and in the study of subcutaneous versus intravenous bortezomib administration (~28.7 months) [23]. In our study, there was no OS difference between treatment groups in NA/WE. The OS disadvantage noted for the ROW for combination treatment may be explained in light of the asymmetric use of dexamethasone rescue and subsequent SCT. Patients in the ROW also received less lenalidomide as subsequent anticancer therapy compared with NA/WE.

Despite preclinical studies suggesting a synergistic interaction between bortezomib and siltuximab in IL-6−dependent models, the combination of S+B was not superior to single-agent bortezomib in this relapsed or refractory population. IL-6 has been shown to induce expression of CRP, a surrogate marker of IL-6 bioactivity [21], by activating transcription factors STAT3 and CCAAT-enhancer-binding protein (C/EBP) β. Activation of NF-κB may also enhance the effect of STAT3 and C/EBPβ, and blockade of NF-κB by bortezomib has been shown to down-regulate IL-6−triggered cascades [24]. Preliminary pharmacokinetic/pharmacodynamic modeling from an open-label study showed that siltuximab given at 11 mg/kg q3w would decrease CRP to below 1 mg/L in multicentric CD patients [25]. While S+B suppressed systemic CRP concentrations to below the level of detection, single-agent bortezomib also led to a less pronounced CRP decrease (Fig 3), which suggests that siltuximab and bortezomib may have had overlapping mechanisms. Systemic CRP suppression was not associated with clinical response. This may explain to some degree why the combination of these two agents did not show a significant additive effect.

It is also possible that IL-6 was insufficiently suppressed in the bone marrow microenvironment given the possibility of paracrine, autocrine, and intracrine IL-6 production by tumor cells [26–28]. Although IL-6 in the local environment is a well-known growth factor for tumor plasma cells [29], only 24% of our patients had detectable serum IL-6 concentrations before study entry, and no consistent association was observed between systemic IL-6 levels and the clinical stage or phase of disease. Treatment with S+B resulted in sustained suppression of CRP, a marker for indirect measurement of in vivo IL-6 neutralization, but systemic IL-6 concentrations do not necessarily reflect IL-6 concentrations in the tumor niche, or the tumor cell utilization of IL-6, which are more likely to impact response to treatment. Additionally, while our study was in progress, pharmacokinetic/pharmacodynamic modeling showed that the 6 mg/kg q2w regimen may be suboptimal [21] and a clear dose-response with optimal efficacy was observed at 12 mg/kg q3w in a B-cell lymphoproliferative disorder population without dose-related toxicity [15].

Even if complete IL-6 neutralization was achieved in our study, simply blocking one cytokine in the bone marrow microenvironment may not have had a significant effect due to compensatory increases in other myeloma growth factors, such as insulin-like growth factor 1 [29], or other mechanisms activated in later disease. A concurrent study in relapsed or refractory MM patients who had received ≥2 therapies including bortezomib showed no clinical activity with single-agent siltuximab and only modest response with siltuximab-dexamethasone [30]. Despite providing earlier responses and higher CR/nCR rates, siltuximab plus bortezomib-melphalan-prednisone in newly diagnosed, transplant-ineligible MM did not translate into a PFS benefit compared with standard bortezomib-melphalan-prednisone in a randomized phase 2 study [31]. Although we believe that IL-6 remains a good target in MM based on a strong rationale from the literature [32], the role of IL-6 may be more prominent in earlier disease. A study of siltuximab monotherapy in high-risk smoldering MM is ongoing.

Combination treatment did not appear to affect the pharmacokinetic profile of either agent. The AE profile before dexamethasone salvage was mainly driven by bortezomib and was consistent with the APEX study [9]. Adding siltuximab to bortezomib compared with bortezomib alone did not change the type of AEs but increased the frequency of neutropenia (59% vs. 36%), thrombocytopenia (57% vs. 45%), and gastrointestinal events (60% vs. 53%). However, the higher incidence of AEs in the blood and lymphatic system did not translate into an increased incidence of high-grade infections (16% vs. 14%) or bleeding (2% each). The incidence of peripheral neuropathy was similar (49% vs. 51%), suggesting that IL-6 is not involved in bortezomib-induced peripheral neuropathy, or that blocking IL-6 alone is insufficient to alleviate this type of neuropathy. The incidence of AEs leading to bortezomib discontinuation was higher in the combination therapy group (33% vs. 24%), which prevented longer bortezomib dosing in the S+B group and explains in part the higher cumulative bortezomib dose in the single-agent bortezomib group.

In conclusion, this randomized, phase 2 study does not provide evidence that the addition of siltuximab to bortezomib improved outcomes in patients with relapsed or refractory MM who have received 1 to 3 prior therapies not containing bortezomib. Baseline IL-6 levels and systemic suppression of CRP were not associated with IL-6-blockade−induced clinical response. Neutropenia and thrombocytopenia were more frequent and severe when siltuximab was administered in combination with bortezomib. The results of this study do not support further investigation of siltuximab in the treatment of advanced MM.

This manuscript presents original, final results of a phase II study of siltuximab. Preliminary results from this study have been presented as the following abstracts:

• Orlowski RZ, Gercheva L, Williams C, et al. Phase II, randomized, double blind, placebo-controlled study comparing siltuximab plus bortezomib versus bortezomib alone in pts with relapsed/refractory multiple myeloma. In: Proceedings from the American Society of Clinical Oncology; June 1–5, 2012; Chicago, IL. Abstract 8018.

• Rossi J-F, Manges RF, Sutherland HJ, et al. Preliminary results of CNTO 328, an anti-interleukin-6 monoclonal antibody, in combination with bortezomib in the treatment of relapsed or refractory multiple myeloma. Paper presented at: 50th Annual Meeting of the American Society of Hematology; December 6–9, 2008; San Francisco, CA. Abstract 867.

• Manges RF, Sutherland HJ, Jagannath S, et al. Preliminary results of CNTO 328, an anti-interleukin (IL)-6 monoclonal antibody (mAb), in combination with bortezomib in the treatment of relapsed or refractory multiple myeloma. Poster presented at: 49th Annual Meeting of the American Society of Hematology; December 8–11, 2007; Atlanta, GA. Abstract 1183.