Department of Hematology and Medical Oncology, Emory University School of Medicine, Winship Cancer Institute of Emory University, Atlanta, Georgia
Corresponding author: Sagar Lonial, MD, Department of Hematology and Medical Oncology, Winship Cancer Institute of Emory University, 1365 Clifton Road NE, C4004, Atlanta, GA 30322; Fax: (404) 778-5530; email@example.com
Presented in abstract form at the 53rd Annual Meeting of the American Society of Hematology; December 10-13, 2011; San Diego, California.
The objective of this meta-analysis in patients with myeloma was to test the hypothesis that the addition of bortezomib to induction therapy not only improves the depth of response but also improves post-transplant progression-free survival (PFS) and overall survival (OS) outcomes.
Phase 3 trials that randomized newly diagnosed, transplant-eligible patients with myeloma to receive either a bortezomib-containing induction regimen (BCIR) or a nonbortezomib-containing induction regimen (NBCIR) were identified. Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines were adapted for data synthesis, and comprehensive meta-analysis software was used to report pooled data as hazard ratios or odds ratios under a random-effects model.
Four published phase 3 trials that included 2169 patients were analyzed. The postinduction and post-transplant pooled odds ratio for achieving a complete response/near complete response or a very good partial response or better and the overall response rate were higher with BCIR. The pooled hazard ratios for 3-year PFS and OS were 0.71 (95% confidence interval, 0.60-0.83; P < .00,001) and 0.79 (95% confidence interval, 0.66-0.96; P = .014), respectively, favoring BCIR. The odds of developing selected grade ≥3 toxicities (peripheral neuropathy and varicella-zoster virus reactivation) also were higher with BCIR.
The use of high-dose therapy (HDT) and autologous stem cell transplant (ASCT) was the first therapeutic approach that improved overall survival (OS) for patients with multiple myeloma.[1, 2] However, to date, no specific induction regimen has demonstrated an impact on post-transplant survival, suggesting that the use of high-dose melphalan was the major factor impacting disease control and, ultimately, OS. The choice of induction regimen had no impact, because most induction regimens were largely alkylator or steroid based and resulted in a very low overall response rate (ORR) and a low complete remission rate. It is established that the use of newer, more active agents as part of induction regimens, such as bortezomib,[3, 4] thalidomide,[5, 6]and lenalidomide,[4, 7-9] improve the ORR and the complete response (CR) rate before HDT and ASCT, but it is not clear from current phase 3 trials whether the impact of achieving a better response before and after ASCT has an impact on OS. This lack of a quantifiable impact of induction regimens on survival is likely a consequence of smaller sample sizes and multiple, confounding maintenance approaches within a single trial, ultimately rendering OS an uninterruptable endpoint in the context of individual randomized clinical trials.
In the recent past, 4 randomized trials[11-14] were performed that compared a bortezomib-containing induction regimen (BCIR) versus a standard, nonbortezomib-containing induction regimen (NBCIR). It was our hypothesis that a greater depth of response before and after HDT associated with the use of BCIR would improve the post-transplant endpoints of progression-free survival (PFS) and OS. To test this hypothesis, we performed a meta-analysis of these 4 large, randomized trials that, in aggregate, compared the impact of BCIR versus NBCIR on attaining a pretransplant CR/near complete response (nCR) or a “very good partial response or better” (≥VGPR) and on the ORR in transplant-eligible patients; and we evaluated the sequential impact of these improved responses on post-transplant PFS and OS. We also assessed the impact of BCIR and NBCIR on regimen-related grade ≥3 toxicities associated with the inclusion of bortezomib to better understand the full impact of BCIR from a risk and benefit perspective.
MATERIALS AND METHODS
Search Methodology and Data Sources
We conducted an electronic literature search of PubMed (MEDLINE), the Cochrane Library, and the EMBASE database using the keywords “myeloma,” “bortezomib,” “induction therapy,” and “transplant-eligible,” restricting the search to exclusively identify randomized controlled trials (RCTs) that took place from May 2003 (when the US Food and Drug Administration approved bortezomib for use in patients with relapsed myeloma who had received at least 1 prior therapy) to August 2012. In addition, we performed a manual search of the annual meeting conference proceedings of American Society of Hematology (ASH), the American Society of Clinical Oncology (ASCO), and the European Hematology Association (EHA). Prospective trial registers (available at: www.clinicaltrials.gov) were explored for relevant, ongoing trials. Our search strategy for study selection is outlined in Figure 1.
Selection Criteria and Data Extraction
We followed the systematic methodology for trial identification and data synthesis described in the Cochrane Collaboration for Systematic Reviews and Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines for reporting the results of the meta-analysis. We included only phase 3 RCTs that randomized younger, transplant-eligible patients with myeloma to receive BCIR versus NBCIR therapies before transplant. For this meta-analysis, studies were considered if they reported the clinical outcomes of response rates (CR, ≥VGPR, ORR postinduction, ORR post-ASCT), post-transplant time to events (PFS and OS), and grade ≥3 toxicities. The appropriate outcome data were extracted from the published reports of the RCTs using previously described methods.[16, 17] The pertinent abstracts from the potential studies identified by the literature search were manually reviewed for eligibility. The full text of any abstract that appeared to be eligible was carefully examined for inclusion after a meticulous assessment of the methodological quality of the trial. In total, 69 RCTs were screened, as illustrated in Figure 1. Among the 12 eligible RCTs that used BCIR before transplant, 6 trials that did not have an NBCIR comparator arm to assess the impact of bortezomib as a part of the induction regimen were excluded. Among the remaining RCTs, 2 were excluded, because 1 study reported subgroup data and another reported updated survival from a second randomization. Any disagreements were resolved by consensus among the authors. Data were extracted and independently confirmed from the selected articles using standardized case report forms. The name of the group performing the specific study has been used to identify each trial in this analysis. In the Program for Study and Therapeutics of Malignant Hemopathies (PETHEMA) trial, the 2 BCIR arms (bortezomib, thalidomide, and dexamethasone [VTD] and vincristine, carmustine [BCNU], melphalan, cyclophosphamide, prednisone/vincristine, BCNU, doxorubicin, dexamethasone/bortezomib [VBMCP/VBAD/B]) were combined, and aggregate results were compared with results from the NBCIR arm (thalidomide and dexamethasone [TD]).
Statistical Analysis and Outcome Measures
The current meta-analysis was performed using the Comprehensive Meta-Analysis software package (version 2.2.057; Biostat Inc., Englewood, NJ). A random model meta-analysis was performed for the primary analysis to calculate the relative risk of BCIR with the comparator arm. The outcome data were pooled and are reported as hazard ratios (HRs) or odds ratios (ORs) with corresponding 95% confidence intervals (CIs). The primary outcomes of interest were response rates (CR, ≥VGPR, and ORR) after induction and the same outcomes after ASCT. Secondary outcomes of interest were PFS and OS from study entry and grade ≥3 toxicities. Pooled outcomes under the random-effects model meta-analysis are displayed as forest plots. An HR <1.00 for PFS or OS signifies an advantage for BCIR, with lower rates of death and/or disease progression. The relative risk, defined as an OR >1.00 for CR, ≥VGPR, and ORR, suggests the superior efficacy of BCIR. An OR <1 for grade ≥3 toxicities implies lesser toxicity with BCIR. To improve the efficiency and reliability of the survival endpoints in the current analysis, we used previously described methods for extracting estimates. The Hemato-Oncology Foundation for Adults in the Netherlands/German-Speaking Myeloma Multicenter Group (HOVON-GMMG) study was the only trial that reported an OS difference. To prove that no single trial influenced the aggregate results of the meta-analysis, a “1 study excluded” analysis also was performed.
Consistency of Effect Sizes
The consistency of results (effect sizes) among studies was investigated by means of 2 heterogeneity tests: the chi-square–based Cochran Q test and the I2 statistic, which was determined from the forest plot statistics. We considered that heterogeneity was present when the P value of the Cochran Q test was < .1 and the I2 statistic was >50% (P ≥ .1 and I2 < 0.25 reflected low-level heterogeneity). In addition, to further assess for publication bias, we used Begg and Egger funnel plots.
Four RCTs that included a total of 2169 patients and were available as published articles were included in the final analysis. All trials had similar inclusion criteria for newly diagnosed, transplant-eligible patients. All studies reported the clinical outcomes of postinduction response rates postinduction and post-ASCT (CR, ≥VGPR, and ORR), survival outcomes (PFS and OS), and grade ≥3 toxicities. Three trials (the French Myeloma Intergroup [IFM] trial, the GIMEMA trial, and PETHEMA/Spanish Myeloma Group [PETHEMA/GEM] trial) were prospectively designed to assess the rates of CR/nCR postinduction as the primary endpoint. The primary endpoint of the fourth trial, the HOVON-GMMG trial, was PFS post-ASCT.
Patient Characteristics of Included Trials
Summary characteristics of the patients enrolled in the 4 RCTs are provided in Table 1. The median sample size for the RCTs was 478 patients, with a median of 250 patients in the BCIR arm and 240 patients in the NBCIR arm. The median age of all included patients was 57 years in both arms. The BCIR and NBCIR arms were well balanced, as expected from RCTs. The percentages of men and women were 56% and 44%, respectively, in the BCIR arm versus 55% and 45%, respectively, in the NBCIR arm; the percentages of patients who had the paraprotein isotypes immunoglobulin G, immunoglobulin A, and immunoglobulin free light chain in the BCIR arm versus the NBCIR arm were 65% versus 61%, 20% versus 22%, and 16% versus 14%, respectively; the percentages of patients with International Staging System stage I disease, stage II disease, and stage III disease in the BCIR arm versus the NBCIR arm were similar between the arms at 43% versus 45%, 36% versus 38%, and 21% versus 22%, respectively. All BCIR studies used the conventional bortezomib intravenous dosing of 1.3 mg/m2 administered on days 1, 4, 8, and 11 every 3 weeks.
Table 1. Summary of Patient Characteristics, Response Rates, and Toxicities Across All Trialsa
Abbreviations: ASCT, autologous stem cell transplant; BCIR, bortezomib-containing induction regimen; CR, complete response; del, deletion; DVT, deep vein thrombosis; EBMT, European Group for Blood and Marrow Transplantation; FISH, fluorescence in situ hybridization; GIMEMA, Italian Group for Hematologic Diseases in Adults; HOVON/GMMG, Hemato-Oncology Foundation for Adults in the Netherlands/German-Speaking Myeloma Multicenter Group; IFM, French Myeloma Intergroup; IgA, immunoglobulin A, IgG, immunoglobulin G; ISS, International Staging System; mo, months; NBCIR, nonbortezomib-containing induction regimen; nCR, near complete response; ORR, overall response rate; OS, overall survival; PETHEMA, Program for Study and Therapeutics of Malignant Hemopathies; PFS, progression-free survival; PN, peripheral neuropathy; t, translocation; TTP, time to progression; VGPR, very good partial response; VZV, varicella-zoster virus; y, years.
All BCIR studies used bortezomib 1.3 mg/m2 on days 1, 4, 8, and 11. Cytogenetic abnormalities identified by FISH were available for 218 patients who received combined bortezomib, thalidomide, and dexamethasone and for 223 patients who received thalidomide plus dexamethasone in the GIMEMA trial; FISH results were available for 330 patients in the PETHEMA trial.
ORR postinduction, CR/nCR post-ASCT, safety, and toxicity
CR/nCR post-ASCT; TTP, PFS, OS, and safety
PFS, OS, and safety
OS, safety, and toxicity
No. of Patients (%)
Total no. of patients
Age: Median [Range]. y
Free light chains
Cytogenetic abnormalities by FISH: del 17p, t(4;14), t(14;16)
Postinduction responses by EBMT criteria/≥VGPR by International Uniform criteria
Post-ASCT responses by EBMT criteria; ≥VGPR by International Uniform criteria
68% at 3 y
56% at 3 y
81.4% at 3 y
77.4% at 3 y
86% at 3 y
84% at 3 y
72% at 4 y
65% at 4 y
61% at 5 y
55% at 5 y
Grade ≥3 toxicities
Efficacy Assessments Before and After Transplant
The difference in ORR favored BCIR, as demonstrated in Figure 2. The pooled CR/nCR rate, ≥VGPR rate, and ORR postinduction in the BCIR arm were 23%, 46%, and 83%, respectively (4 trials; n = 1148), compared with 10%, 22%, and 65%, respectively, in the NBCIR arm (4 trials; n = 1021). Similar pooled CR/nCR rates post-ASCT were 45% versus 25% comparing the BCIR arm versus the NBCIR arm, respectively. After ASCT, the ≥VGPR rate for BCIR versus NBCIR was 65% versus 44%, respectively, and the ORR was 87% versus 79%, respectively (3 trials; n = 889 patients in the BCIR arm and n = 894 patients in NBCIR arm; specifics were unavailable for the PETHEMA/GEM trial). Heterogeneity was present for the ORR post-ASCT because of a single study (IFM) that reported a lower ORR (P = .022; I2 = 73.93) (Table 2). Hence, the significance of BCIR in the resultant ORR post-ASCT may have been underestimated. When the IFM trial was excluded from the analysis, the OR for ORR increased from 2.0 to 2.4 (95% CI, 1.75-3.32).
Table 2. Treatment Regimens and Schema for Individual Studies
Abbreviations: ASCT, autologous stem cell transplant; BCIR, bortezomib-containing induction regimen; CAD, cyclophosphamide, doxorubicin, and dexamethasone; Cy, cyclophosphamide; DCEP/none, randomized to receive dexamethasone, cyclophosphamide, etoposide, and cisplatin or none; Dm, dexamethasone maintenance; GCSF, granulocyte colony-stimulating factor; GIMEMA, Italian Group for Hematologic Diseases in Adults; HOVON/GMMG, Hemato-Oncology Foundation for Adults in the Netherlands/German-Speaking Myeloma Multicenter Group; IFM, French Myeloma Intergroup; IFNα2bm/Tm/VTm, randomized to receive either interferon α-2b maintenance versus thalidomide versus thalidomide plus bortezomib; Mel200, melphalan 200 mg/m2; NBCIR, nonbortezomib-containing induction regimen; PAD, bortezomib, doxorubicin, and dexamethasone; PETHEMA, Program for Study and Therapeutics of Malignant Hemopathies; Rm/placebo: randomized to receive either lenalidomide maintenance or placebo according to French Myeloma Intergroup study 2005-02; TD × 2, thalidomide and dexamethsone consolidation; Tm, thalidomide maintenance; VAD, vincristine, doxorubicin, and dexamethasone; VBAD/B, vincristine, carmustine, doxorubicin, dexamethasone/bortezomib consolidation; VBMCP, vincristine, carmustine, melphalan, cyclophosphamide, and prednisone; VD, bortezomib and dexamethasone; Vm, bortezomib maintenance; VTD × 2, bortezomib, thalidomide, and dexamethasone consolidation.
In the PETHEMA/GEM trial, the 2 BCIR arms were combined, and aggregate results are presented.
For each individual trial, the PFS favored BCIR; and, in aggregate, the pooled PFS favored BCIR (HR, 0.71; 95% CI, 0.60-0.83; P = .00,001). When OS was evaluated, only 1 of the 4 RCTs demonstrated improved OS; however, in aggregate, OS favored BCIR (HR, 0.79; 95% CI, 0.66-0.95; P < .014) (Fig. 3). There was no significant heterogeneity in the HRs of individual trials (PFS: P = .725; I2 = 0.00; OS: P = .941; I2 = 0.00) (Table 3).
Table 3. Tests of Heterogeneity
Abbreviations: ASCT, autologous stem cell transplant; CR, complete response; DF, degrees of freedom; DVT, deep vein thrombosis; I2, statistic for heterogeneity; ORR, overall response rate; OS, overall survival; PFS, progression-free survival; PN, peripheral neuropathy; Q, statistic derived from the chi-square–based Cochran Q test; VGPR, very good partial response; VZV, varicella-zoster virus.
Grade ≥3 toxicities
The Begg and Egger funnel plots for all endpoints demonstrated a symmetrical distribution, suggesting an absence of publication bias (Fig. 3).[20, 21] We performed sensitivity analyses with 1 study removed for both the primary and secondary endpoints, and those analyses indicated that no individual study influenced the pooled results of the RCTs. The results excluded the chance that any single trial was over-weighted with respect to its effect on the outcome overall.
In assessing grade ≥3 adverse events, the OR for peripheral neuropathy (PN) and varicella-zoster virus (VZV) reactivation was higher for BCIR therapies (PN: OR, 3.03; 95% CI, 1.97-4.32; P = .000; VZV: OR, 4.807; 95% CI, 1.88-12.30; P = .001). The risk of grade ≥3 thrombocytopenia with BCIR was nonsignificant in the meta-analysis. When the PETHEMA/GEM trial was excluded from the analysis, the OR for thrombocytopenia was 2.16 (95% CI, 0.93-5.05; P = .07), trending toward significance. It is known that bortezomib induces transient and reversible thrombocytopenia because of temporary arrest in megakaryopoeisis. The risk of grade ≥3 infections, anemia, and neutropenia was similar between the BCIR and NBCIR arms (Fig. 4).
A hallmark of the use of bortezomib and lenalidomide in myeloma therapy has been higher response rates in the induction setting. However, the impact of these novel agents on post-transplant outcomes has been less clear. Although the short-term endpoints of ORR and depth of response are clearly important, demonstrating an improvement in the more objective endpoint of OS has remained challenging because of the numerous postinduction variations in therapy that confound survival analysis. However, the aggregation of data in the current meta-analysis demonstrates that BCIRs not only collectively improve pretransplant ORR, ≥VGPR, and CR rates but also improve the post-ASCT endpoints of PFS and OS. The current meta-analysis provides an opportunity to view these 4 trials in aggregate rather than as 4 separate BCIRs and to directly compare them with “older,” conventional NBCIRs. The demonstration of superiority for BCIR is even more striking in the context that each trial used different high-dose (single vs tandem) and post-transplant (maintenance and consolidation) strategies, which could have confounded the interpretation had this been a smaller analysis. To eliminate the possibility that the final result was because of an over-influence of any individual study, each individual trial was removed, and the remaining 3 trials were analyzed for the same endpoints. When we used this single-trial exclusion method, there was no change in the final conclusion of the analysis (Fig. 5). This supports our hypothesis that the benefit of BCIR in PFS and OS was not influenced by any single trial or maintenance strategy.
Our current approach has potential limitations common to all meta-analyses, and these include the inclusion of trials with different methodologies, different study designs, inconsistent endpoints, and different duration of follow-up. Given these differences among the 4 large randomized trials, some degree of statistical heterogeneity was anticipated, yet this was only noted with regard to the post-transplant ORR. An explanation for this low rate of heterogeneity is the relative uniformity of the target group of younger patients that was common to each trial. Although each trial used a different post-transplant strategy (Table 2), what was common to all of them was the use of bortezomib as part of the induction phase of therapy. Furthermore, the Begg and Egger funnel plots for all primary and secondary endpoints and the analysis specifically designed to evaluate for possible publication bias further supported the validity and robustness of the analyses and the absence of publication bias. One other limitation of our meta-analysis was the limited number of studies (4 RCTs) that fit the inclusion criteria, which precluded us from pursuing any further subgroup analysis (eg, comparing BCIR vs either combined vincristine, doxorubicin, and dexamethasone or TD).
Although BCIR is able to confer a survival benefit, some noteworthy adverse events are observed with this approach and should be used to balance the risk/benefit ratio when assessing an individual patient or treatment strategy. It is well known that the incidence of PN is increased with bortezomib-based induction therapy, and 6% to 11% of patients suffered from grade 3 or 4 PN among those who received BCIRs. Most patients who were enrolled on the 4 included trials had received treatment before data were available evaluating the use of either weekly or subcutaneous bortezomib dosing, thus making it difficult to interpret the incidence and severity of adverse events in the context of more recent administration standards. It remains clear that appropriate and timely evaluations before initiating each cycle of therapy with prompt dose reductions of bortezomib upon the development of early signs of PN can modulate these adverse effects, but they can be limiting despite the use of careful monitoring and newer administration strategies. In addition, the incidence of VZV reactivation is associated with bortezomib-based treatment and is now managed through the use of antiviral prophylaxis for all patients who receive bortezomib. Thrombocytopenia is also a known effect of BCIR, although, in the current analysis, this was only observed to any significant degree in 1 of the 4 clinical trials (PETHEMA) and may have be a consequence of combining bortezomib with alkylator-based induction therapy.
Although the scope of this meta-analysis applies only to BCIR versus NBCIR, we could make the case that any regimen that induces a high ORR and a high depth of response would yield similar results. Thus, induction regimens that contain lenalidomide or the newer second-generation proteasome inhibitors (carfilzomib or MLN9708), which have demonstrated the ability to induce a high ORR and a high CR rate, probably would produce a similar improvement in PFS and OS post-transplant compared with older, conventional treatments. Thus, the current analysis supports the use of newer strategies directed toward an improved pretransplant response with the goal of improving post-transplant PFS and OS.
In conclusion, the current data support the use of BCIR for patients who are eligible for transplant, demonstrating improved ORR before and after ASCT as well as improved PFS and OS compared with NBCIR. Although bortezomib-based induction regimens are used extensively in the United States and Europe, their use in other parts of the world is more limited. Our analysis supports their use over conventional induction regimens to improve OS for younger, transplant-eligible patients. Future studies with second-generation proteasome inhibitors or combinations with better tolerated immunomodulatory drugs may further augment the risk/benefit ratio by reducing the incidence and severity of treatment-related adverse events associated with BCIR.
Dr. Flowers was supported by grant R21 CA158686-01 from the National Cancer Institute.
CONFLICT OF INTEREST DISCLOSURES
Dr. Kaufman has received compensation as a consultant to Millennium Pharmaceuticals, Inc.; Onyx Pharmaceuticals, Inc.; Novartis AG; and Celgene Corporation. Dr. Flowers has received compensation as a consultant to Millennium Pharmaceuticals, Inc.; Genentech, Inc.; and OptumRx; he has received grants from Celgene Corporation, Janssen Pharmaceutica, Millennium Pharmaceuticals, Inc./The Takeda Oncology Company; and Spectrum Pharmaceuticals, Inc.; and he has received compensation for the development of educational presentations from Educational Concepts, Inc. Dr. Gleason has received compensation as a consultant to Celgene Corporation. Dr. Boise has received compensation as a consultant to Onyx Pharmaceuticals, Inc. Dr. Lonial has received compensation as a consultant to Onyx Pharmaceuticals, Inc.; Millennium Pharmaceuticals, Inc.; Celgene Corporation; Novartis AG; Bristol-Myers Squibb Company; Sanofi SA; and Johnson & Johnson.