|
Advertisement |
|
|
||
 |
|
|||||
|
|||||
|
 |
||||||
|
||||||
|
||||||
|
Originally published as JCO Early Release 10.1200/JCO.2004.06.144 on March 22 2004 © 2004 American Society of Clinical Oncology. Benefit and Timing of Second Transplantations in Multiple Myeloma: Clinical Findings and Methodological Limitations in a European Group for Blood and Marrow Transplantation Registry StudyFrom the Haematology Department, Belfast City Hospital, Belfast, Northern Ireland; University Medical Centre, Leiden, the Netherlands; Department of Medicine, Karolinska Institutet, Huddinge University Hospital, Stockholm, Sweden; Division of Haematology and Oncology, University of Leipzig, Leipzig, Germany; on behalf of the Chronic Leukaemia Working Party Myeloma Subcommittee of the European Blood and Marrow Transplantation Group Address reprint requests to T.C.M. Morris, MD, Department of Haematology (C floor), Belfast City Hospital, Lisburn Rd, Belfast, BT9 7AB Northern Ireland; e-mail: curly.morris{at}bll.n-i.nhs.uk
PURPOSE: To use European Group for Blood and Marrow Transplantation registry data to assess the benefit and optimal timing of a double-autologous transplantation strategy for patients with myeloma. PATIENTS AND METHODS: 7,452 transplantation patients described as being either in a multiple graft program ("planned") or not, were analyzed on an intention-to-treat basis. Subsequent multivariate analyses concentrated on the real occurrence of second transplantation, survival, relapse, and transplant-related mortality. RESULTS: Although the transplantation rate in the planned group failed to reach 60%, the median survival from transplantation is 60 months for the planned, compared with 51 months for the remainder group. While the hazard ratio of the planned group is 0.89 (95% CI, 0.79 to 1.00; P =.05) before approximately 70 months, this "effect" is reversed after 70 months, with the hazard ratio estimated as 3.01 (95% CI, 1.07 to 8.46; P = .04). A time-dependent multivariate Cox analysis shows that, taking patients without a second transplantation as a reference group, those receiving a second transplantation in first remission (ie, before relapse) show an increased probability of transplant-related mortality, especially if the transplantation is performed more than 12 months after the first, and the reduction of the risk of relapse is less than when the transplantation is performed earlier. Performing a second transplantation after relapse does not seem to prolong survival, though a second transplantation before relapse is associated with a higher probability of mortality. CONCLUSION: To improve survival of tandem autologous transplantation in multiple myeloma, the second transplantation should preferably be performed before relapse and within 6 to 12 months of the first transplantation.
There is now evidence from various sources that autologous stem-cell transplantation prolongs survival in patients with myeloma,1-5 though the final data from prospective randomized studies addressing the value of tandem transplantations are awaited. Second transplantations given after relapse following a single transplantation have been used as a salvage therapy in many centers,6-8 but to date, no studies have examined if it is necessary to perform two transplantations within a limited period, or if advantage can be obtained by performing a second transplantation only after the patient has relapsed. Inspection of the European Group for Blood and Marrow Transplantation (EBMT) myeloma registry shows that a significant number of patients have received second transplantations, though the period between transplantations varies substantially and sometimes includes a relapse after the first transplantation. We have considered various statistical approaches for analyzing these observational data with respect to benefit obtained from the second transplantation and the importance of its timing.
To address the question "is performing a second transplantation after relapse as good as tandem transplantation up front?" we need to have information about the therapeutic strategy chosen at the time of the first transplantation. Any other method based on the exclusion and/or on the grouping of patients using an outcome measure or the evolution of the disease and its treatment (eg, transplantation, relapse) would result in a biased analysis. For example, a direct comparison of patients receiving one transplantation against those receiving two transplantations is not methodologically (statistically) valid, since in order to receive a double transplantation, patients must survive for a sufficient period to receive the second transplantation. The EBMT Med B form for autotransplantation allows indication by the transplantation center if a patient is part of a "planned" multiple transplantation program. We initially adopted the working hypothesis that a code for an "unplanned" double transplantation might act as a surrogate for transplantations that were to be performed only after relapse, and therefore allow a comparison of the double transplantation "up front" strategy against postrelapse retransplantation (in effect "planned" v "wait-and-see"). We therefore considered 7,452 patients from 362 EBMT centers who have had their first autologous peripheral blood stem-cell transplantation for myeloma between January 1993 and March 2002, and for whom information on the intention to treat was reported on Med B forms. Patients with amyloid and other conditions were excluded. Of this group, 2,655 patients were identified as being in a double transplantation program; the remaining 4,797 patients were considered the "unplanned" group. This grouping is not based on whether patients actually had a second transplantation or not (since this would bias the analysis severely). Patients receiving an allogeneic transplantation as the second transplantation were also excluded. As the grouping "planned" versus "unplanned" does not reflect the situation in a randomized trial, the analysis has been adjusted for the main known risk factors9 including age, type of myeloma, beta 2 microglobulin (B2m) at diagnosis, remission status at transplantation, use of total-body irradiation (TBI; Table 1). To evaluate the quality and consistency of the coded information regarding the therapeutic strategy, we have reviewed the relationship between the "planned/unplanned" classification and occurrence of relapse and transplantation, and time to second transplantation in the two groups of "planned" and "unplanned." The approach above can be seen as an "intention-to-treat" analysis.
In a separate analysis, we ignored the information on "planned" versus "unplanned," and described instead the data in terms of hazard ratios (HRs) for relapse, transplantation-related mortality (TRM), and overall survival (OS) with respect to the occurrence of second transplantation and the timing of this intervention. We have therefore studied relapse and TRM, comparing at each point in time patients who at that time had not (yet) received a second transplantation stem-cell transplantation (SCT) and patients who had (already) received a second transplantation SCT, taking into account their time to second transplantation. We have further analyzed the survival of relapsed patients, taking into account the effect of having already had a second transplantation at the time of relapse, the time from first transplantation to relapse, and the effect of transplantation after relapse. These analyses are carried out estimating Cox models with time-dependent covariates on the proper data sets, adjusting for the main risk factors.
A series of risk factors known from the literature were considered as adjustment variables in the multivariate analyses; factors were eliminated from the models if not statistically significant. Differences in the distributions of the risk factors in the two groups of "planned" and "unplanned" were evaluated by a Survival in the planned and unplanned groups, was first estimated using the Kaplan-Meier method. The analysis for the comparison between the two groups, adjusted for the main risk factors, is based on a standard (multivariate) Cox model, and the results are reported as adjusted HRs with 95% CIs; the change in the effect in the long-term (so allowing the HR to vary over time) is illustrated introducing a time-dependent covariate. Since data used for this study are incomplete for many prognostic factors (Table 1), and removal of all cases with missing values would markedly decrease the number of available patients, we chose instead to add the category "missing" as a further level of the variables (where necessary), keeping therefore all cases in the study. Since for each variable the missing category was equivalent with respect to the major outcome measures to the "known" category in the multivariate analysis, this procedure allowed unbiased estimates of the HR of interest This simple analysis, based on the intention-to-treat principle, mimics the statistical analysis carried out in a prospective randomized clinical trial; however, as data are collected retrospectively, care is needed when interpreting any relationships. The analysis of relapse and TRM is made estimating two separate (multivariate) Cox models for each outcome while censoring for the other. The occurrence of second transplantation is included as a (categorial) time-dependent covariate, which switches from its initial value of 0 to the appropriate value depending on the time to transplantation (within 6 months, 6 to 12 months, after 12 months), at the time of transplantation. Obviously, due to the end point being considered, this variable is triggered only by the elective second transplantation (ie, the one performed before relapse). To complete the description of the course of the disease before relapse or death, a model for the time to elective transplantation has also been estimated.10 All prognostic factors (except TBI, which is part of the selected therapeutical strategy) have been included, together with a center effect (the latter as a random effect). The analysis of survival after relapse is carried out only on relapsed patients using a Cox model with left truncation at relapse. Occurrence of second transplantation before and after relapse is included as fixed and time-dependent covariates, respectively. Also in this case, a model for time to second transplantation (performed after relapse) has been estimated for patients not having an elective tandem transplantation. The descriptive analyses and the survival curve plots were created using SPSS version 10.0, while the Cox models were estimated in S-Plus version 6.0; the latter includes the routines by T. Therneau to estimate Cox models with random effects (frailty models).11
Patient Characteristics The Med B data for the 7,452 patients on whom the statistical analysis was carried out is presented in Table 1 for the following parameters: age, stage at diagnosis, type of myeloma, immunoglobulin type, light chain type, B2m at diagnosis, TBI (given/not given), and response after first-line therapy. The number of patients in whom data are missing is also shown; Table 1 also compares planned and unplanned groups, reporting the  2 test data for the equality of the distributions.
Survival and Hazard Ratios for Planned and Unplanned Figure 1 shows the (unadjusted) Kaplan-Meier survival functions for planned double transplantations and unplanned patients. The median survival from first transplantation of the planned group is 60 months (95% CI, 55 to 64), and for the unplanned it is 51 months (95% CI, 47 to 54). The (adjusted) HR is 0.90 (95% CI, 0.80 to 1.01; Table 2, Model 1). It can be seen that the planned and unplanned curves cross at time equal to 70 months. Whereas the HR of the planned group is 0.89 (95% CI, 0.79 to 1.00; P = .05) up to 70 months, this effect seems to be reversed in the longer term, with the HR of the planned group after 70 months being 3.01 (95% CI, 1.07 to 8.46; P = .04; Table 2, Model 2). Inspection of the multivariate Cox model shows that older age, stage III disease, unresponsive disease, immunoglobulin A paraprotein, lambda light chain, and raised B2m were statistically significant adverse factors for survival. TBI conditioning showed a detrimental effect, only marginally failing to reach the critical boundary of 0.05. Bence Jones myeloma, nonsecretory myeloma, and year of transplantation had no significant effect on the HR (Table 3).
Planned and Unplanned Second Transplantation: Further Analyses Table 4 presents the number and percentage of patients in the planned and unplanned groups who received a second transplantation, while Table 5 details the relationship between first and second transplantation, relapse, and death. Whereas only 69 patients in the planned group had their second transplantation after relapse, 256 patients in the unplanned group had their second transplantation before their first relapse, with only 199 having the second transplantation postrelapse, indicating that the planned and unplanned group allocation as inferred from the Med B forms is not fully consistent with the observed relationship between first relapse and second transplantation. An analysis of time to transplantation confirms this: the data on the planned transplantations show many patients with a very long time from first to second transplantation (38 patients having a time greater than 12 months and 15 even greater than 24 months), casting doubt on the assumption that a tandem transplantation was really planned at the time of the first transplantation. In contrast, of the 256 unplanned who had their second transplantation actually before relapse, 143 had a time to transplantation of less than 6 months, and 58, less than 3 months.
The time to second transplantation in patients with an elective second transplantation was also analyzed by calendar year, and the results are shown in Table 6. A marked change in practice can be seen. While there is little variation in the planned group between 1993 and 2001, in the unplanned group, there is a marked decrease in the time to the second transplantation, suggesting that in the last 4 years, most of these unplanned transplantations are in fact planned.
Impact of Elective Second Transplantation on TRM and Relapse As presented in Table 7, patients undergoing transplantation a second time had a significantly lower HR for relapse than those who did not undergo transplantation for a second time, while the protective effect is decreasing somewhat with increasing time between first and second transplantation (HR varying from 0.43 to 0.64). As might be expected, every second transplantation patient had a higher risk of TRM, in particular, those transplanted more than 1 year after their first transplantation.
Impact of Second Transplantation on Survival After First Relapse As presented in Table 8, compared with relapsed patients who did not have a second transplantation, those who did before the occurrence of the relapse had a higher hazard of dying (HR, 1.70; 95% CI, 1.40 to 2.05; P < .0001; notice that having an elective second transplantation was associated with a lower risk of relapsing; Table 7), while it seems that a transplantation performed after relapse produces no significant effect on survival (HR, 1.06; 95% CI, 0.87 to 1.29; P = .6; Table 8).
Factors Influencing Survival, Relapse, TRM, and the Decision to Undertake Second Transplantation The effects of the prognostic factors included in the multivariate analyses detailed above are summarized in Table 3 as HR estimates, with significant effects indicated by an asterisk. It can be seen that only age retains a significance in all situations, though with a small HR per year. For relapsed patients, a late relapse (> 24 months) is associated with a better survival (HR, 0.71; 95% CI, 0.57 to 0.88; P = .002). Factors associated with early occurrence of elective second transplantation were younger age, advanced stage at diagnosis, immunoglobulin A, recent calendar year, and center effect. The time between relapse and a subsequent transplantation was influenced by the same factors (except immunoglobulin type), plus long time to relapse.
In the first analysis in this study, analysis of groups with a stated intention to treat has identified a problem with crossover as seen in "gold-standard" randomized trials 1 with less than 60% of patients receiving the planned treatment while over 9% of the control group received a second transplantation—about half of these being salvage transplantations postrelapse, and the remainder being protocol violations. Nevertheless, despite this apparently adverse starting point, Kaplan-Meier estimation shows the median survival of the single (control) and planned double transplantation groups (51 months and 60 months, respectively) is similar to the results of the study carried out by the Intergroupe Francais du Myelome (IFM; IFM '94),4 which was the only randomized trial of double versus single transplantation to have reached maturity. Although the Bologna '96 Trial12 also shows a superior event free survival, this trial has not matured sufficiently to show an advantage in OS if it follows the pattern of IFM'94. However, as the Kaplan-Meier analysis indicates crossing of the survival curves, the data were analyzed using the time-dependent model which suggests that the benefit for double transplantation is lost in the long-term.. While similar outcomes with a significant proportion of long-term survivors have been reported in a number of single center studies,13,14 prospective, randomized multicenter studies by the Hovon15 and Myelome Autogreffe16 groups comparing intensified treatment with intensified followed by myeloablative therapy and single versus double transplantation respectively do not show any improvement in OS in the more intensive group. As this is a registry study, variables such as cytogenetics14,17 are not available in sufficient number to allow meaningful analysis. The marked center effect with respect to undertaking second transplantations suggested that there was no consensus on this topic during the study period. This is an important point as all our data are observational and do not take into account other factors including selection bias which may have caused the reporting physicians to choose a certain course of action and do not necessarily represent cause and effect. For patients given a second transplantation, it is possible that the second transplantation was only given to patients with good prognostic factors and thus the lower relapse rates would be due to selection and not the effect of the transplantation itself. Conversely, if the transplantation was given only to patients with poor prognostic factors the positive effect of a second transplantation will be underestimated. As this is a European study, reimbursement is unlikely to be an issue causing potential bias in contrast to centers in North America. While there have been no randomized trials to evaluate the benefit of second transplantation following relapse, it has been suggested that SCT (after first relapse) may achieve the same outcome as transplantation following initial therapy.18-20 Our data do not support an advantage to this strategy as the outcome of patients receiving second transplantation following first relapse did not differ significantly from those treated with non transplantation strategies, though the outcome is somewhat better for patients with a long time to first relapse. Does this mean that if the patient has relapsed there is no benefit of a second transplantation? This is certainly what has been observed (HR, 1.06; P = .6). However the interpretation can and should not be causal in this observational context: what kind of patients did and did not get another transplantation after relapse is not known as the reasons underlying the actual choice of the therapeutic strategy by the physician are unknown to us. If these reasons are actually related to the patient's prognosis and the latter is not entirely explained by the risk factors included in the analysis, it is not permissible to draw causal conclusions with respect to the different treatments since these conclusions could even be opposite to the truth. Moreover, this approach, though a weakness inherent in all registry studies, is sensitive to any errors or omissions in the information collected in the Med B forms about the therapeutic strategy. The time dependent multivariate Cox analysis for relapse and TRM with the other as additional censoring has yielded useful information. Transplantation will always be a risk factor for TRM (confirmed in Table 7). However, while the HR remains relatively constant between 0 and 6 months and 6 to 12 months, there is a marked increase in the HR over 12 months. In contrast, the HR for relapse is significantly reduced between 0 and 6 months and 6 to 12 months, and although reduced as well in the more than 12 months group, the HR is not reduced to the same degree and does not reach a P value of .05. We therefore tentatively suggest that second transplantation in patients who have not relapsed already is best performed within less than 6 to 12 months of the first transplantation. However, this suggestion should be taken cautiously since some of the results may be the results of a selection mechanism (eg, possibly only patients who were fit had an elective second transplantation early—these may be patients with good prognostic factors, so they would be expected to have a lower relapse rate). Until recently the options for relapsed posttransplantation patients were limited, particularly when the transplantation was not robust in the face of further chemotherapy. Newly introduced agents, in particular thalidomide, imids, and bortezomib22 alone or in combination with steroids or reduced doses of cytotoxics are likely to change the outlook for patients with myeloma, including postrelapse transplantation patients. This analysis may therefore be timely in setting a benchmark against which new strategies are developed; in particular posttransplantation consolidation or postrelapse reinduction therapy. Our data point to the need for more successful strategies for the elimination of myeloma cells, particularly when it is recognized that transplantation is a strategy available to less than 25% of all myeloma patients.22
The authors indicated no potential conflicts of interest.
Authors' disclosures of potential conflicts of interest are found at the end of this article.
1. Attal M, Harousseau JL, Stoppa AM, et al: A prospective, randomized trial of autologous bone marrow transplantation and chemotherapy in multiple myeloma. N Engl J Med 335:91-97, 1996
2. Barlogie B, Jagannath S, Vesole D, et al: Superiority of tandem autologous transplantation over standard therapy for previously untreated multiple myeloma. Blood 89:789-793, 1997
3. Lenhoff S, Hjorth M, Holmberg E, et al: Impact on survival of high dose therapy with autologous stem cell supporting patients younger than 60 years with newly diagnosed multiple myeloma: A population based study. Blood 95:7-11, 2000 4. Harousseau JL, Attal M: The role of stem cell transplantation in multiple myeloma. Blood Rev 16:245-253, 2002[CrossRef][Medline]
5. Child JA, Morgan GJ, Davies FE, et al: High-dose chemotherapy with hematopoietic stem-cell rescue for multiple myeloma. N Engl J Med 348:1875-1883, 2003 6. Tricot G, Jagannath S, Vesole DH, et al: Relapse of multiple myeloma after autologous transplantation: Survival after salvage therapy. Bone Marrow Transplantation 16:7-11, 1995[Medline] 7. Mehta J, Tricot G, Jagannath S, et al: Salvage autologous or allogenic transplantation for multiple myeloma refractory to or relapsing after a first-line autograft. Bone Marrow Transplantation 21:887-892, 1998[CrossRef][Medline] 8. Kulkarni S, Powles R, Singhal S, et al: Second autografts for relapsed multiple myeloma: Is tandem autotransplantation better? Clinical care: recurrence, secondary neoplasia and late complications after transplantation. Blood 92:344b, 1998 (suppl 1) 9. Bjorkstrand B: European Group for Blood and Marrow Transplantation Registry studies in multiple myeloma. Sem Haematol 38:219-225, 2001[CrossRef] 10. Keading JP, Klein JP, Horowitz HM: Multi state models and outcome prediction in bone marrow transplantation. Stat Med 20:1870-1885, 2000 11. Hougard P: Fraility models for survival data. Lifetime Data Anal 1:255-273, 1995[CrossRef][Medline] 12. Cavo M, Tosi P, Zamagni E, et al: The Bologna 96 clinical trial of single vs. double autotransplants for previously untreated multiple myeloma patients. Blood 100:179a, 2003 13. Moreau P, Misbahi R, Milpied N, et al: Long-term results (12 years) of high-dose therapy in 127 patients with de novo multiple myeloma. Leukemia 16:1838-1843, 2002[CrossRef][Medline]
14. Desikan R, Barlogie B, Sawyer J, et al: Results of high dose therapy for 1000 patients with multiple myeloma: Durable complete remission and superior survival in the absence of chromosome 13 abnormalities. Blood 95:4008-4010, 2000
15. Segren CM, Sonnereld P, van der Holt B, et al: Overall and event-free survived are not improved by the use of myeloablative therapy following intensified chemotherapy in previously untreated patients with multiple myeloma: A prospective randomised phase 3 study. Blood 101:2144-2151, 2003 16. Fermand J-P, Alberti C, Marolleau J-P: Single versus tandem high dose therapy (HDT) supported with autologous blood stem cell (ABSC) transplantation using unselected of CD34 enriched ABSC: Results of a two by two designed randomised trial in 230 young patients with multiple myeloma (MM). Haematol J 4:59, 2003 (supp 1) 17. Seong C, Delasalle K, Hayes K, et al: Prognostic value of cytogenetics in multiple myeloma. Br J Haematol 101:189-194, 1998[CrossRef][Medline] 18. Sirohi B, Powles R, Singhal S, et al: Second high-dose melphalan autograft for myeloma patients relapsing after one autograft: Results equivalent to tandem transplantation. Bone Marrow Transplantation 29:S12, 2002 (suppl 2)
19. Fermand JP, Ravaud P, Chevret S, et al: High-Dose Therapy and Autologous Peripheral Blood Stem Cell Transplantation in Multiple Myeloma: Up-front or Rescue Treatment? Results of a multicenter sequential randomized clinical trial. Blood 92:3131-3136, 1998 20. Gertz MA, Lacy MQ, Inwards DJ, et al: Delayed stem cell transplantation for the management of relapsed or refractory multiple myeloma. Bone Marrow Transplantation 26:45-50, 2000[CrossRef][Medline] 21. Hideshima T, Chauhan D, Podar K, et al: Novel therapies targeting the myeloma cell and its bone marrow microenvironment. Semin Oncol 28:607-612, 2001[CrossRef][Medline] 22. Morris TCM, Ranaghan E, Ogilvie C, et al: Population-based study of the 'young' myeloma population: Only a minority of potentially eligible patients with myeloma receive stem cell transplantation; a study of the contributing factors. Haematol J 4:203, 2003 (suppl 1)[CrossRef] Submitted July 2, 2003; accepted February 11, 2004. This article has been cited by other articles:
|
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
|
|||||||||||
|
|||||||||||
|
|
Copyright © 2004 by the American Society of Clinical Oncology, Online ISSN: 1527-7755. Print ISSN: 0732-183X
|
TOP
ABSTRACT
INTRODUCTION
