Complete remission sustained 3 years from treatment initiation is a powerful surrogate for extended survival in multiple myeloma




Complete response (CR) has been considered a necessary although not sufficient early clinical endpoint for extended survival in multiple myeloma.


By using Total Therapy 2 (TT2) clinical outcome data in 668 patients, whether sustained CR (SUS-CR) was potentially a superior surrogate for survival than attaining CR status per se was evaluated.


Compared with not achieving CR (NON-CR) and especially achieving and subsequently losing CR status (LOS-CR) within a 3-year landmark from treatment initiation, SUS-CR was associated with highly superior postlandmark survival (P < .0001). These results were validated in 231 untreated patients enrolled in the predecessor trial, TT1 (hazard ratio [HR] = 0.54, P = .013) and in 1103 previously treated patients on other transplant protocols (HR = 0.49; P < .001).


In all 3 trial settings the survival benefit of SUS-CR was independent of metaphase abnormalities as a dominant adverse parameter. Given its bleak prognosis despite high CR rates, SUS-CR should be evaluated as a primary trial endpoint in high-risk myeloma. Cancer 2008. © 2008 American Cancer Society.

The efficacy of therapy for multiple myeloma has long been judged by the frequency with which complete remission (CR) can be induced, as prolongation of overall survival (OS) appeared tied to this early treatment endpoint in high-dose melphalan-based autotransplants.1–5 However, other data suggest that long-term survival beyond 10 years can be achieved in the absence of CR,6 whereas CR induction appeared only critical in the setting of high-risk myeloma.7 Furthermore, despite similar initial CR rates, survival varies widely as a result of differences in baseline characteristics, especially cellular and molecular genetics.8–10 In addition, in Total Therapy 2 (TT2), survival and CR duration were not prolonged among patients randomized to thalidomide, although CR frequency and event-free survival were superior in comparison with the control arm.11 Because in the era of new agent investigations much attention has been placed on the CR rate as a primary trial objective, we examined whether a minimum duration of CR might be a better surrogate for survival than its initial frequency.


Study Populations

TT2 evaluated the role of thalidomide in a randomized trial design.11 With a median follow-up of 5 years, median survival was similar in both study arms and exceeds 8 years. For the purpose of this analysis, all 668 patients initially enrolled were considered. Survival was examined as a function of CR not obtained (NON-CR), obtained and sustained (SUS-CR), or obtained and lost (LOS-CR) at increasing time landmarks of 18 months, 24 months, 30 months, and 36 months from protocol enrollment. The results were subsequently validated in 2 other patient cohorts, 231 newly diagnosed patients treated on predecessor protocol Total Therapy 1 (TT1) and 1103 previously treated patients receiving other protocol-based tandem transplants (non-TT).

Clinical Endpoints of Study

Response and relapse definitions used criteria similar to those reported by Blade et al.12 CR required the absence of M-protein in serum and urine on immunofixation analysis as well as normal bone marrow aspirate and biopsy by microscopy and flow cytometry (absence of aneuploidy and monoclonal cytoplasmic light chain); in addition, metaphase cytogenetic abnormalities had to be absent. CR criteria had to be documented on at least 2 subsequent occasions at least 2 months apart. Relapse from CR implied the reappearance of M-protein in serum or urine on immunofixation, reappearance of monoclonal bone marrow plasmacytosis or of cytogenetic abnormalities, the development of extramedullary disease or of new bone lesions recognized on at least annually performed magnetic resonance imaging (MRI) or skeletal surveys. Nonsecretory relapses in the absence of M-protein recurrence also constituted relapse, such as new focal lesions on MRI or positron emission tomography (PET) scan, reemergence of bone marrow monoclonal plasmacytosis, or extramedullary disease often associated with elevated levels of lactate dehydrogenase (LDH).

Statistical Methods

The Kaplan-Meier Method was used in estimating OS and CR duration,13 in the context of baseline laboratory parameters that included gene expression profiling (GEP)-based risk models.9 Survival was defined from the date of registration or from various landmarks until death from any cause; survivors were censored at the time of last contact. CR duration was measured from the onset of CR until relapse or death from any cause. The results of different trials were compared using the log-rank test. Multivariate models of prognostic factors were performed using Cox regression.14


We examined increasing time intervals from initiation of TT2 to identify an optimal split in postlandmark survival according to the 3 response categories (SUS-CR, NON-CR, LOS-CR) (Fig. 1. As is apparent from a comparison of 4 landmarks (18 months, 24 months, 30 months, 36 months), the 36-month landmark provided the greatest separation in subsequent survival among the 3 response groups: the 4-year postlandmark survival estimates were 83% for 256 patients with SUS-CR versus 59% among 211 in the NON-CR group versus 19% among 39 in the LOS-CR category (all P < .0001) (Fig. 1d). The 36-month landmark also separated postlandmark survival in other protocols (Fig. 2). Thus, in TT1 the median postlandmark survival was 10.6 years for SUS-CR, 6.0 years for NON-CR, and 2.9 years for LOS-CR (Fig. 2a); in non-TT protocols the median postlandmark survival durations were longest in case of SUS-CR (6.4 years) and progressively shortened for NON-CR (3.9 years) and LOS-CR (1.8 years) (Fig. 2b). When examined for its independent impact on postlandmark survival in the context of standard prognostic factors, 3-year SUS-CR was a favorable variable countering the adverse implications of cytogenetic abnormalities in all 3 trial settings (Table 1). Survival was also shorter in the case of LDH elevation and IgA isotype in TT1, whereas creatinine levels affected survival adversely in TT2 and advanced age suggested poor outcome in non-TT protocols.

Figure 1.

Searching for a landmark in Total Therapy 2 (TT2) that optimally separates postlandmark survival among 3 response categories (achieved and sustained CR status [SUS-CR], never achieved CR status [NON-CR], attained and lost CR status [LOS-CR]). a: 18-month landmark. b: 24-month landmark. c: 30-month landmark. d: 36-month landmark.

Figure 2.

Applying the 36-month landmark to other treatments. a: Overall survival from a 36-month landmark in Total Therapy 1 (TT1) according to 3 response subgroups. b: Overall survival from a 36-month landmark in other trials (non-TT) according to 3 response subgroups.

Table 1. Multivariate Analyses of Baseline Features and Sustained Complete Remission After 36-Month Landmark Survival in 3 Treatment Settings
n/N (%)HR (95% CI)Pn/N (%)HR (95% CI)Pn/N (%)HR (95% CI)P
  1. TT2 indicates total therapy 2; TT1 total therapy 1; Non-TT, no total therapy; n/N (%); n indicates number with factor; N, number with valid data for factor; HR, hazard ratio; CI, confidence interval, LDH, lactate dehydrogenase; SUS-CR, sustained complete remission.

Creatinine ≥ 2.0 mg/dL23/256 (9)4.90 (2.40, 10.04)<.001
Cytogenetic abnormalities, CA72/256 (28)2.18 (1.19, 4.00).01229/154 (19)1.82 (1.12, 2.95).016128/464 (28)1.49 (1.12, 1.97).005
LDH ≥ 190 U/L45/154 (29)1.67 (1.10, 2.55).017
IgA myeloma25/154 (16)2.06 (1.24, 3.41).005
Age ≥ 65 y87/464 (19)1.60 (1.17, 2.19).003
SUS-CR133/256 (52)0.31 (0.16, 0.61)<.00142/154 (27)0.54 (0.33, 0.88).013187/464 (40)0.49 (0.37, 0.65)<.001

We also examined whether the 3 response categories were associated with standard prognostic variables. As depicted in Table 2, LOS-CR was prominently correlated with several adverse baseline features, including GEP-defined high-risk, elevations of beta-2 microglobulin (B2M) and LDH, as well as IgA isotype and low albumin concentration. However, despite these associations, SUS-CR was an independent favorable outcome variable for all 3 protocol settings (Table 1).

Table 2. Univariate Analysis of Baseline Features Associated With 3 Response Subgroups in Total Therapy 2 (TT2) Patients: Achieved and Sustained CR Status, Never Achieved CR Status, Attained and Lost CR Status
FactorResponse subgroupP
  • Nonsignificant variables were gender, race, age, Hb, creatinine, CRP, and cytogenetic abnormalities.

  • LOS-CR indicates lost CR status; CR, complete remission; NON-CR, never achieved CR status; SUS-CR, achieved and sustained CR status; B2M, beta-2 microglobulin; LDH, lactate dehydrogenase.

  • *

    Chi-square test.

  • Fisher exact test.

B2M ≥ 3.5 mg/L20/39 (51%)69/211 (33%)79/256 (31%)0.0130.67
Albumin < 3.5 g/dL12/39 (31%)32/209 (15%)35/255 (14%)0.0070.94
LDH ≥ 190 U/L19/39 (49%)41/210 (20%)78/255 (31%)0.0020.007
IgA myeloma15/39 (38%)37/211 (18%)63/256 (25%)0.0150.064
70 gene-defined high-risk4/19 (21%)3/111 (3%)11/138 (8%)0.0290.073


The historical pursuit of extending survival in myeloma by increasing the frequency of CR has paid off in the case of melphalan-based therapy, for which a profound dose-response relationship was observed that also translated into marked survival prolongation. Thus, raising CR frequency from less than 5% with standard melphalan-prednisone toward 50% with high-dose melphalan-based transplants was associated with an extension of median survival from less than 3 years to beyond 5 years.4, 5 For the reasons cited earlier, however, it is now apparent that CR per se is not a guarantor for long survival.7, 11 In the current era of an unprecedented discovery rate of novel agents displaying anti-myeloma activity when used alone and especially in combination with each other and with melphalan,15–20 CR has become a popular trial endpoint in phase 2 studies. We recently reported that, among the pretreatment determinants evaluated in TT3, the only parameters positively associated with CR frequency were those recognized as imparting high risk for overall survival (elevated serum levels of LDH and free light chains, IgA isotype).21 Thus, the clinical trial results have to be interpreted in the proper context of critical prognostic factors.

Here we demonstrate that SUS-CR is a new independent favorable posttreatment variable associated with prolonged survival not only in TT2 but also in 2 other transplant trial settings. The worst outcome was observed among the LOS-CR group of patients who also shared baseline characteristics least favorable for OS, whereas the NON-CR category had an intermediate prognosis. The underlying mechanism likely relates to greater therapeutic sensitivity in highly proliferative tumors that, upon failure to achieve profound tumor cytoreduction and possibly cure, regrow rapidly and, in the process, acquire additional mutations that underlie their ultimate multiagent resistance and patient death.7 We propose that SUS-CR be evaluated as a potentially more robust survival surrogate endpoint that permits, within a timeframe of 3 years, rapid trial throughput, especially toward identifying more effective therapies for high-risk myeloma.