Adverse Metaphase Cytogenetics Can Be Overcome by Adding Bortezomib and Thalidomide to Fractionated Melphalan Transplants

With the incorporation of novel agents into autotransplant-supported high-dose melphalan trials, the outlook of patients with multiple myeloma has greatly improved. In our Total Therapy (TT) program, applying all multiple myeloma–active agents up front to reduce or even eliminate the development or survival of drug-resistant subclones, major advances have been linked to the addition of thalidomide in TT2 and bortezomib in TT3 (1, 2). However, when examined in the context of gene expression profiling (GEP) of CD138-purified plasma cells, such benefit was limited to the 85% of patients presenting with GEP70-based low-risk multiple myeloma (LR-MM) (3, 4). In recognition of the lack of progress in high-risk multiple myeloma (HR-MM), we decided in 2008 to assign low- and high-risk patients to separate protocols. We are now reporting on clinical outcomes of 289 patients with LR-MM enrolled in a randomized phase III trial of Total Therapy 4 (TT4), comparing a light arm (TT4-L) with a standard arm (TT4-S), with the goal of reducing toxicity while maintaining efficacy in TT4-L.

The protocol schema is portrayed in Supplementary Table 1. Briefly, eligible patients with GEP70-defined LR-MM were randomized between TT4-S and TT4-L. TT4 induction in both arms was similar to TT3b, except for the addition of melphalan 10 mg/m² test-dosing 48 hours following bortezomib 1.0 mg/m² test-dosing for the purpose of pharmacogenomic investigations (5, 6). TT4-L differed from TT4-S by reduction from two cycles to one cycle each of induction with M-VTD-PACE prior to and consolidation with dose-reduced VTD-PACE after tandem transplants. The transplant regimen in L-TT4 was altered from a single melphalan dose of 200 mg/m² (MEL200) in TT4-S to a fractionated 50 mg/m²/d for 4 days (MEL50 × 4) schedule with the aim to avoid peak melphalan dose levels and reduce mucosal toxicity. To compensate for the potentially reduced efficacy of this strategy, bortezomib, thalidomide, and dexamethasone (VTD) were added to the fractionated melphalan to exploit the observed synergism between melphalan and VTD (7, 8). Maintenance therapy was with bortezomib, lenalidomide, and dexamethasone in both arms for 3 years. A total of 289 patients were randomized (TT4-S, n = 145; TT4-L, n = 144) and stratified by ISS stage and the presence of metaphase cytogenetic abnormalities (CAs). Metaphase abnormalities were deemed to be present if at least two cells with the same karyotype were present. Protocol eligibility included age >18 years and ≤75 years and normal cardiopulmonary function; liver function tests could be up to twice normal, and creatinine levels up to 3mg/dL were acceptable. Patients had to have active and measurable multiple myeloma fulfilling CRAB criteria (9).All patients had to sign a written informed consent in keeping with institutional, national, and Helsinki Declaration guidelines. The protocol and its revisions had been approved by the Institutional Review Board, which received annual progress reports. As TT4 was supported with a grant from the NCI (Bethesda, MD), a Data Safety and Monitoring Board (DSMB) reviewed toxicities and efficacies at least annually. We also had independent data audits every 6 to 8 months to verify adherence to protocol stipulations, examine pharmacy records and informed consents, and verify recorded toxicities, especially causes of death (COD) and clinical efficacy in terms of complete response (CR) and CR duration (CRD), progression-free survival (PFS), and overall survival (OS).

Patient work-up was per protocol and was similar to our practice in TT3 (2). Special studies included GEP of plasma cells (GEP-PC) and whole bone marrow biopsies (GEP-BMBX), procured from both random iliac crest sites and from MRI-defined focal lesions under computer-assisted tomography guidance. Random bone marrow samples were submitted for metaphase cytogenetic analysis and interphase FISH for the detection of deletion 17p and amplification of 1q21. The protocol called for serial short-term and long-term studies of GEP and MRI as well as 18-fluoro-deoxyglucose PET. Pharmacogenomic studies called for repeat GEP sampling from random iliac crest bone marrow and MRI-defined focal lesion(s) 48 hours after test-dosing with bortezomib 1.0 mg/m² and again 48 hours after test-dosing with melphalan 10 mg/m² prior to starting the full VTD-PACE program. PET follow-up studies were scheduled on day 5 after commencing VTD-PACE and prior to first transplantation. These correlative studies will be the subject of a separate report. Most patients were treated in the outpatient setting and were checked daily by multiple myeloma–experienced nursing staff and weekly by their physicians. Hematopoietic progenitor collection was started when CD34 collection criteria were met with the goal to procure at least 2 × 10⁷ CD34 cells/kg body weight.

The primary clinical endpoint was reduction in toxicity in TT4-L versus TT4-S. Secondary endpoints included CR defined according to IMWG criteria, the duration of which was measured from its onset to progression or death (10). OS was calculated from registration until the date of death. PFS was similarly calculated, but also incorporated progressive disease as an event. Time to progression (TTP) was measured from registration until the date of progression or relapse, whereas time to relapse (TTR) focused on the subset of patients achieving CR. Death without prior progression or relapse was included as a competing risk in analyses of TTP and TTR. Postrelapse survival (PRS) was measured from the date of relapse or progression until death. For all time-to-event analyses, patients were censored at the date they were last known to be alive. COD included treatment-related mortality (TRM), myeloma-related mortality (MRM), and a third category capturing other or indeterminate causes (OIM), such as fatal car accident, stroke, and other causes that could not be attributed to treatment or disease. Toxicities were graded according to Version 3 of the NCI.

After accrual and randomization of 289 patients, the DSMB recommended closure of the TT4-L arm due to lack of reduction in toxicity vis-à-vis TT4-S. Two patients randomized to TT4-L were treated according to the standard arm after the DSMB recommended closure of TT4-L.These patients are accounted for in the 289 patients. Higher mortality at 1 year in patients ≥65 years on both arms (10.6% and 10.2% for TT4-L and TT4-S) prompted subsequent accrual of 74 such patients to Total Therapy 6 (TT6) designed for previously treated patients, because its less dose-intense therapies had resulted in TRM of 0% (11).Younger patients continue enrollment to TT4-S but only the 289 randomized patients are the subject of this report.

The dataset was created on January 16, 2015, on 289 patients randomized between TT4-S and TT4-L. In accordance with the intent-to-treat principle, all patients were analyzed according to their randomized arm. The Kaplan–Meier method was used to estimate the distributions of OS, PFS, and CR duration (12). Cumulative incidences for COD, TTP, and TTR were calculated using the method of Gooley and colleagues (13). Group comparisons for survival endpoints and cumulative incidence were performed using the log-rank test (14). Multivariate models of prognostic factors were carried out using Cox regression (15). P < 0.05 was considered statistically significant.

Patient characteristics are summarized in Table 1 and were not different between the two arms. The consort flow diagram (Supplementary Fig. S1) portrays the progression of patients through the various protocol steps and lists the off-study reasons. Grade ≥3 toxicities occurred with similar frequencies in the two arms of TT4 (Supplementary Table S2), with no significant differences observed between major toxicity categories. With a median follow-up of 4.5 years in both arms, 112 and 108 patients are alive at the time of analysis, while 100 and 92 are progression-free on TT4-S and TT4-L, respectively (Fig. 1). Two-year OS estimates are similar in both arms: 90% and 87% (Fig. 1A); the corresponding PFS estimates of 84% and 79% were also comparable (Fig. 1B). At the time of analysis, 90 and 92 had achieved CR status on TT4-S and TT4-L, respectively, for 2-year estimates of 59% and 61% (Fig. 1C); 2-year CRD estimates are 87% and 81% (P = 0.05; Fig. 1D). Median TTP was similar in both arms, with 2-year estimates of 8% and 11%, whereas there was a trend toward earlier TTR from CR on the TT4-L arm (12% vs. 8% at 2 years, P = 0.07; data not shown). COD were divided into MRM (P = 0.70), TRM (P = 0.61), and OIM (P = 0.25) and were not different between both arms (Supplementary Fig. S2A). PRS was similar in the two arms (Supplementary Fig. S2B).

As in previous TT trials, the presence of CA affected clinical outcomes (Fig. 2). Surprisingly, the presence of CA had opposite prognostic connotations in the two study arms. In TT4-S, CA showed a strong trend toward inferior OS (Fig. 2A), whereas the reverse applied to TT4-L (Fig. 2B). Nonsignificant trends in opposite directions were noted for PFS (Fig. 2C and D). CRD tended to be inferior in patients with CA-type multiple myeloma in TT4-S (Fig. 2E) with an opposite trend in TT4-L (Fig. 2F), although neither comparison was statistically significant. TTR was significantly steeper in the presence of CA in TT4-S (Fig. 2G) and did not affect this outcome variable in TT4-L (Fig. 2H).

Next, we examined the impact of baseline factors on outcomes (Tables 2 and 3). The univariate Cox analysis of baseline features associated with OS and PFS across TT4 arms is summarized in Table 2. OS and PFS were inferior in the case of older age ≥65 years; high β-2-microglobulin (B2M; both ≥3.5 and >5.5 mg/L), presence on PET-CT of more than three fluorodeoxyglucose avid focal lesions (FLs), low albumin levels (<3.5 g/dL) posed a significant hazard for OS but not for PFS, and the reverse applied to CRP elevation (≥8 mg/L). There was a trend for better OS and PFS for the CD1 and CD2 molecular subgroups (Table 2). In the presence of CA, in comparison with its absence (no CA), OS tended to be shorter in TT4-S (P = 0.08) and longer in TT4-L (P = 0.07). On multivariate analysis (Table 3), older age, B2M >5.5mg/L, and PET-FL >3 independently imparted inferior OS and PFS. Low albumin adversely affected OS and was of borderline significance for PFS. Patients with CA enjoyed superior OS when randomized to TT4-L with a strong trend apparent also for PFS. In the case of TT4-S, differences were in the opposite direction, although not significant. The different effect of CA on outcomes by treatment arm (interaction) was statistically significant for both OS and PFS (P = 0.0035 and 0.0491, respectively).

As we had previously reported on prognostic implication of imaging parameters, we sought to determine whether the adverse clinical impact of PET-FL >3 pertained to both TT4 arms (16). Regardless of treatment arm, OS and PFS were inferior for PET-FL >3 (Supplementary Fig. S3). The timing of onset of CR was not affected by PET-FL, while CRD was superior in case of PET-FL ≤3 for the TT4-S arm but not for TT4-L. A trend for more rapid TTP with PET-FL >3 was observed in TT4-S, whereas TTP was similar in TT4-L. TTR was significantly faster with PET-FL >3 only in case of TT4-S.

The observation of CA's favorable OS impact in TT4-L was unexpected and at variance with findings in all previous TT protocols (17). We therefore analyzed whether the presence or absence of metaphase CA could be linked to certain gene probes, which might explain the superior performance in TT4-L of fractionated melphalan with added bortezomib and thalidomide. A training set of patients with LR-MM enrolled in TT3a was chosen for this endeavor. Among 266 untreated patients with available baseline GEP studies, 90 (34%) exhibited CA. Among the test set of 164 patients with baseline GEP accrued to TT3b, 67 (41%) qualified as having CA. Using an FDR of 0.0001, 51 probes significantly distinguished patients with and without CA (Supplementary Table S3). Seven of the 51 genes function in DNA replication, recombination, and repair, five in nucleic acid metabolism, and four in RNA posttranslational modification and RNA damage and repair. Ingenuity Pathway Analysis (IPA) identified a network of eight interrelated genes that were overexpressed in the CA group, indicating that these multiple myeloma cells have a higher proliferative activity (Supplementary Fig. S4). We then examined clinical outcomes by the GEP51–CA prediction model in the two arms of TT4 (Fig. 3). In TT4-S, GEP51/no-CA had superior OS and PFS compared with GEP51/CA (Fig. 3A and B), which was not observed in TT4-L (Fig. 3C and D).

The results of this phase III TT4 trial failed to show that TT4-L was less toxic and was, in fact, inferior to TT4-S in terms of CRD with a trend for inferiority in case of TTR. One-year mortality was similar, 7.6% in TT4-L and 6.9% in TT4-S (Fisher exact test P = 0.75), but low in both arms in younger patients (TT4-S, 6.2%; TT4-L, 5.2%). Unanticipated were the opposite implications of the presence of CA on OS: adverse in TT4-S (P = 0.08) as in all previous TT trials and favorable in TT4-L (P = 0.07), with significant opposite prognostic implications in the two study arms. The other clinical endpoints were only marginally affected except for significantly faster TTR in case of CA in TT4-S. When analyzed in the context of all competing variables, TT4-L affected OS of patients with CA-type multiple myeloma favorably, with a strong trend apparent also for PFS. It seems reasonable to speculate that CA-type multiple myeloma derived benefit from conditioning with fractionated melphalan VTD, as T4-L employed one less cycle of induction and consolidation treatment. This combination has also been found to be synergistic when used to treat patients with VMP ineligible for stem cell transplantation in the VISTA study (18).

We previously reported that bortezomib and VTD could be added to fractionated melphalan in doses up to 250 mg/m² in an advanced patient population with acceptable toxicity and high response rates (7, 8). Others have also explored the use of bortezomib and melphalan as a conditioning regimen prior to transplantation and found that both agents can be safely combined and possibly produce increased response rates (19–22).The current article is the first randomized study comparing single-dose melphalan 200 mg/m² with a fractionated MELVTD schedule. The results of TT4 indicate that CA-type myeloma benefits from MELVTD, whereas non-CA multiple myeloma fares better with application of a single high-dose melphalan 200 mg/m². Prior attempts to improve conditioning with melphalan have not been successful and included the addition of other drugs, skeletal targeting radioactive antibodies, or radiotherapy (23–28).

Mono- or polyubiquitination of DNA repair enzymes, such as H2AX, BRCA1, and FANCD2, is required for recruitment to sites of DNA double-stranded breaks and is essential to the DNA damage response (29). Bortezomib reduces the availability of nuclear ubiquitin and may thereby impair homologous recombination–mediated DNA repair. It has been suggested that bortezomib renders myeloma cells more vulnerable to alkylator-mediated DNA damage by essentially inducing a BRCAness type state (30). CA-type multiple myeloma is likely to harbor more genomic chaos and may be more dependent on efficient DNA repair, explaining the sensitivity to fractionated MEL-VTD conditioning. Availability of GEP led us to examine whether CA-linked gene probes could be identified. Indeed, 51 genes were identified that distinguished no-CA from CA subgroups. When examined for its clinical relevance, the GEP51 model was prognostic in TT4-S, so that GEP51/CA prediction was associated with inferior OS and PFS. In contrast, such Kaplan–Meier plots were superimposable between these two groups in TT4-L. IPA suggests that CA-type multiple myeloma has overexpression of genes critical to proliferation. The ability to examine metaphase cytogenetics inherently implies that the myeloma cells are able to survive and proliferate in vitro without the support of the myeloma microenvironment. Increased ability to proliferate may be a favorable evolutionary trait, but on the other hand, stresses requirements for DNA repair to prevent fatal mutations and cell death. In this context, one could speculate that exposure to DNA-damaging alkylators, such as melphalan in the setting of bortezomib-induced reduced DNA repair could lead to fatal “mitotic catastrophes.”

This is the first clinical trial that suggests that tailoring the transplant conditioning regimen to myeloma biology may further improve outcome. However, the study was not designed to specifically study the impact of conditioning on outcome in the context of metaphase CA. These preliminary findings therefore require confirmation in future phase III studies.

M. Zangari is a consultant/advisory board member for Celgene, Millennium, and Novartis. G. Morgan is a consultant/advisory board member for Bristol-Myers Squibb, Celgene, Janssen, Onyx, and Takeda. No potential conflicts of interest were disclosed by the other authors.

Conception and design: S. Waheed, J. Crowley, B. Barlogie, F. van Rhee

Development of methodology: M. Cottler-Fox, J. Crowley, G. Morgan, B. Barlogie, F. van Rhee

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): E. Tian, S. Waheed, R. Khan, X. Papanikolaou, M. Grazziutti, N. Petty, D. Steward, S. Panozzo, B. Barlogie, F. van Rhee

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): A. Mitchell, J. Epstein, S. Yaccoby, E. Tian, X. Papanikolaou, A. Hoering, J. Crowley, J. Sawyer, G. Morgan, B. Barlogie, F. van Rhee

Writing, review, and/or revision of the manuscript: Y.S. Jethava, A. Mitchell, J. Epstein, M. Zangari, S. Yaccoby, A. Hoering, J. Crowley, G. Morgan, B. Barlogie, F. van Rhee

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): Y.S. Jethava, N. Petty, D. Steward, S. Panozzo, C. Bailey, J. Crowley, G. Morgan, F. van Rhee

Study supervision: N. Petty, G. Morgan, B. Barlogie, F. van Rhee

Other (was involved in recruiting patients to the protocol as well): S. Waheed

Other (supervision of cell collection and processing and infusion for transplant): M. Cottler-Fox

Total Therapy 4 was supported by the NCI, NIH Program Project grant CA55819 (to G. Morgan).