The optimal intensity of reduced-intensity conditioning (RIC) before allogeneic hematopoietic stem cell transplantation (allo-HSCT) remains uncertain.
The optimal intensity of reduced-intensity conditioning (RIC) before allogeneic hematopoietic stem cell transplantation (allo-HSCT) remains uncertain.
In this centrally randomized phase 2 study, the authors compared 2 different strategies of RIC. In total, 139 patients (median age, 54 years; range, 21-65 years) with hematologic malignancies underwent allo-HSCT from a human leukocyte antigen-identical sibling after conditioning combining fludarabine with either busulfan and rabbit antithymocyte-globulin (BU-rATG) (n = 69) or total body irradiation (TBI) (n = 70). Postgraft immunosuppression consisted of cyclosporin A in all patients with the addition of mycophenolate-mophetil after TBI.
The median follow-up was 54 months (range, 26-88 months). One-year overall survival rate was identical in both groups. Four patients experienced graft-failure after TBI. The incidence of grade 2 through 4 acute graft-versus-host-disease was greater after BU-rATG than after TBI (47% vs 27%; P = .01), whereas no difference was observed with chronic graft-versus-host-disease. The BU-rATG group had a higher objective response rate (65% vs 46%; P = .05) and a lower relapse rate (27% vs 54%; P < .01). However, the nonrelapse mortality rate was higher after BU-rATG than after TBI (38% vs 22%; P = .027). At 5 years, the overall and progression-free survival rates were 41% and 29%, respectively, and did not differ statistically between groups. A detrimental effect on some parameters of quality of life was more pronounced after BU-rATG, but recovery was identical in both groups. The mean total cost per patient, including the cost to treat disease progression post-transplantation, did not differ statistically between groups.
Five years after transplantation, the BU-rATG regimen was associated with greater disease control. However, because of the higher nonrelapse mortality rate, this did not translate into better overall or progression-free survival. Cancer 2013. © 2012 American Cancer Society.
Allogeneic hematopoietic stem cell transplantation (allo-HSCT) is an efficient form of adoptive immunotherapy for the treatment of patients with hematologic malignancies.1 Initially, high-dose cytotoxic conditioning was delivered for effective disease control. Because of the perceived need for this cytotoxicity, which eventually was accompanied by impairment of patient survival, initially, only a limited subset of patients underwent allo-HSCT. Adverse effect was even more apparent in fragile patients. Subsequent studies demonstrated that the procedure relied on graft-derived immunocompetent donor cells, which can target both tumor and nontumor tissues.2 Subsequently, Holler and colleagues observed that the cytotoxicity of the conditioning generated proinflammatory cytokines, which represent a potential starting point for the negative effects of donor-derived cell cytotoxicity. This led to a hypothesis that reducing cytokine generation may result in decreased toxicity.3 This hypothesis was verified when low cytotoxic immunosuppressive agents were combined as the preparative regimen before allo-HSCT to test whether a graft-versus-leukemia response would be sufficient to achieve a curative antitumor effect.4-6
These findings have been confirmed by several studies, and so-called reduced intensity conditionings (RIC) are now widely used and associated with a dramatic reduction in nonrelapse mortality (NRM). However, the optimal combination of antitumor cytotoxicity and immunosuppression, including postgraft immunosuppression, remains uncertain. Three major questions remain unresolved regarding 1) the respective importance of chemotherapy dose intensity and allogeneic effect for disease control, 2) the adaptation of partial in vivo T-cell depletion to reduced toxicity approaches, and 3) the impact of postgraft immunosuppression on the entire strategy. To address some of these issues, in 2002, we conducted a prospective randomized study to assess survival after 2 popular but markedly different (and never prospectively compared) RIC strategies that initially were developed by S. Slavin at the Hadassah Hospital (Jerusalem, Israel) and R. Storb at the Fred Hutchinson Cancer Research Center (Seattle, Wash).
A single study protocol was reviewed and approved by each institutional review board at the 5 participating institutions, the Marseille II Ethical Committee, and the Cellular Therapy Committee of the French Agency for the Safety of Health Products. Written informed consent was obtained from eligible patients and their donors before randomization.
Patients between ages 18 and 65 years were eligible for the study if they had a hematologic malignancy for which an allo-HSCT from a matched-related donor (MRD) was indicated. Enrollment criteria for patients and donors included the usual parameters, as previously reported.7 Patients were assigned randomly to 1 of 2 treatment groups after stratification according to treatment center, disease stage, and whether or not patients were in complete remission (CR).
The primary objective of this randomized phase 2 study was to select a RIC for further development. The primary endpoint was 1-year overall survival (OS). By using a phase 2 selection design,8,9 the target sample size was 82 patients per group to reach a 90% probability of correctly selecting the more successful treatment if the true hazard ratio (HR) was 1.3, assuming a 50% 1-year OS rate. The first patient was registered in January 2003. Exclusion criteria included excessive engraftment failure and NRM. Data were reviewed each year by an independent data-safety monitoring board. In 2008, the data-safety monitoring board recommended stopping the study, because additional patients were unlikely to change the primary endpoint. The last patient was included in May 2008. Analysis was performed after 2010 to allow for sufficient follow-up.
Patients in group A received fludarabine with busulfan and rabbit antithymocyte-globulin (FLU-BU-rATG), which included: 1) fludarabine (Fludara; Schering AG, Lys-Les-Lannoy, France) 30 mg/m2 daily on days −5 to −1, 2) oral busulfan (Myleran; Glaxo-Smith-Kline, Marly-le-Roi, France) 1 mg/kg 4 times daily on days −4 and −3, and 3) rabbit-antithymocyte-globulin (rATG) (Thymoglobulin; Genzyme, St.-Germain-en-Laye, France) 2.5 mg/kg on day −3. Patients in group B received fludarabine and total body irradiation (FLU-TBI), which included: 1) fludarabine 30 mg/m2 daily on days −5 to −2, and 2) 1 session of TBI (2 grays) on day 0. Peripheral blood stem cells were harvested from MRD after the administration of granulocyte-colony–stimulating factor, according to standard procedure.7 All patients received cyclosporin as postgraft immunosuppression, as previously reported.7 In addition, in group B (FLU-TBI), patients received oral mycophenolate-mophetil at a fixed dose of 2 g daily on days 0 through 28 that was decreased through day 60. Supportive care, including anti-infectious drugs and blood product transfusions, was administered according to usual policies to prevent or treat anemia; bleeding; and bacterial, viral, and fungal infections and was similar for all study patients in each center.
The neutrophil and platelet engraftments were defined as previously reported.7 T-cell chimerism was assessed by polymerase chain reaction analysis of polymorphic microsatellite regions as previously described10 on days 30, 60, 90, 120, and 180 after transplantation. Full donor chimerism was defined as the presence of >95% cells of donor origin. Acute graft-versus-host-disease (aGVHD) and chronic graft-versus-host-disease (cGVHD) were defined according to usual criteria.11 Patients who relapsed, those who had evidence of disease progression or persistent disease without any sign of GVHD after immunosuppressive therapy withdrawal, or those who had mixed hematopoietic chimerism beyond 3 months were candidates for donor lymphocyte infusion (1 × 105 and 1 × 107T-cell coreceptor [CD3]-positive cells/kg of the patient's body weight; the higher dose was received by patients who had more active disease).
Health-related quality of life (HRQL) was measured prospectively using the European Organization for Research and Treatment of Cancer Core Quality-of-Life Questionnaire.12 Patients received questionnaires 7 days before transplantation and on days 30, 80, 180, and 360 after transplantation. In patients who relapsed, the patient HRQL study was discontinued.
A cost analysis was performed from the hospital's point of view. Direct medical transplantation costs were estimated by microcosting based on patient files from both groups from day 0 to 18 months after transplantation. The actual daily cost was calculated based on the Paoli-Calmettes Institute's analytic records for all relevant hospitalization units. Costs of treatment for disease progression were estimated for the 22 patients who received treatment at the Paoli-Calmettes Institute and whose disease progressed within 5 years after transplantation. The estimated cost of treating disease relapse or progression included hospitalization costs and high-cost therapies, such as lenalidomide, bortezomib, and donor lymphocyte infusion. This estimated cost ultimately was applied to the other patients in the trial who presented with progressive disease, according to their diagnosis.
The comparability of baseline characteristics between the 2 randomized groups was assessed using chi-square and Student t test statistics. OS was calculated from the date of transplantation to the date of death from any cause. Progression-free survival (PFS) was calculated from the date of transplantation to the date of either disease progression or death. Patients who survived without disease progression and patients who remained alive at the end of follow-up were censored at the date of last contact. Survival estimates were calculated using the Kaplan-Meier method13 with 95% confidence intervals (CIs), and differences between treatment groups were assessed using the log-rank test and a Cox proportional hazards model. Cumulative incidences of NRM, progression-related mortality, relapse or progression, aGVHD, and cGVHD were analyzed using Prentice estimation, the Gray test, and a Fine and Gray regression model. To take into account the heterogeneity of the population characteristics, further studies were performed. NRM and OS probabilities were adjusted according to the European Group for Blood and Marrow Transplantation outcome risk score.14 Relapse and PFS probabilities were adjusted according to the risk of relapse.15 Like deaths after disease progression were considered in the incidence of NRM, deaths without evidence of progression were treated as competing events in analyses of the incidence of relapse. Similarly, graft failure, relapse or progression, and deaths were treated as competing risks when analyzing the incidence of GVHD. A multistate modeling approach was used to estimate the prevalence of cGVHD in both groups.16 Linear mixed-model analysis was performed to test differences in HRQL outcomes within and between the 2 groups over time. A group-by-time interaction term was tested to explore whether differences in HRQL scores were a function of group, conditional of time. If this term was not significant, then the main effects (treatment and time) were evaluated for significant differences. Differences in costs between treatment groups were assessed using Student t test statistics from the costs logarithm. All statistical analyses were carried out using SAS version 9.1 (SAS Institute, Inc., Cary, NC) or R version 2.10.1 (R Project for Statistical Computing, Vienna, Austria; available at: http://www.r-project.org [accessed March 27, 2012]).
A schematic of the trial is provided in Figure 1. The study included 139 patients with a median follow-up of 54 months (range, 26–88 months). The primary evaluation endpoint, 1-year OS, was identical between the 2 groups.
There were no significant differences between the 2 groups of patients with respect to baseline characteristics (Table 1). All but 4 patients achieved sustained engraftment, with granulocyte and platelet recovery confirmed at a median of 12 days (range, 0–56 days) and 3 days (range, 0–121 days), respectively (Table 2). Full donor chimerism was documented in 98 of 124 evaluable patients (79%) at a median of 72 days (range, 27–207 days) after transplantation without any differences observed between the 2 groups (Table 2). Four patients presented after TBI with graft failure at a median of 64 days (range, 28–190 days) after transplantation.
|No. of Patients (%)|
|Characteristic||FLU-BU-rATG, N = 69||FLU-TBI, N = 70||P|
|Age : Median [range], y||54 [21-65]||53 [34-65]||.81|
|Men with male donor||16/67 (24)||22/70 (31)|
|Men with female donor||26/67 (39)||24/70 (34)|
|Women, with female donor||15/67 (23)||10/70 (14)|
|Women with male donor||10/67 (15)||14/70 (20)|
|Karnofsky performance status: Median [range]||90 [60-100]||90 [70-100]||.28|
|CMV seronegative/seronegative||4 (6)||4 (6)||.98|
|Acute myeloblastic leukemia||9 (13)||12 (17)|
|Acute lymphoblastic leukemia||3 (4)||0 (0)|
|Chronic myeloid leukemia||1 (1)||1 (1)|
|Myelodysplastic syndrome||9 (13)||5 (7)|
|Primary myelofibrosis||3 (4)||1 (4)|
|Non-Hodgkin lymphoma||17 (25)||15 (21)|
|Hodgkin lymphoma||3 (4)||3 (4)|
|Multiple myeloma||22 (32)||32 (46)|
|Chronic lymphocytic leukemia||1 (1)||1 (1)|
|Waldenstrom disease||1 (1)||0 (0)|
|Complete remission/chronic phase||26 (38)||25 (36)|
|Partial remission/stable disease||35 (50)||40 (57)|
|Refractory disease||8 (12)||5 (7)|
|Risk of relapse according to disease characteristicsa||.124|
|Low risk||30 (44)||41 (59)|
|Standard risk||18 (26)||17 (24)|
|High risk||21 (30)||12 (17)|
|Outcome according to EBMT risk scoreb||.55|
|1||2 (3)||1 (1)|
|2||22 (32)||29 (41)|
|3||33 (8)||32 (46)|
|4||12 (17)||8 (12)|
|Characteristic||FLU-BU-rATG, N = 69||FLU-TBI, N = 70||P|
|Days to reach|
|ANC 5 × 109/L||14 [2-28]||10 [0-56]||< .001|
|Platelets 20 × 109/L||3 [0-121]||0 [0-37]||.048|
|No. of erythrocyte transfusions||4 [0-84]||2 [0-25]||.002|
|No. of platelet transfusions||1 [0-112]||0 [0-43]||< .001|
|Achievement of lymphoid donor chimerism|
|No. of evaluable patients||60||64|
|No. who achieved full chimerism (%)||49 (82)||49 (77)||.49|
The cumulative incidence of grade 2 through 4 aGVHD was greater after busulfan (47%; 95% CI, 35%–59%) than after TBI (27%; 95% CI, 18%–36%; HR, 0.49; 95% CI, 0.28–0.85; P = .01) (Fig. 2A). In contrast, the cumulative incidence of extensive cGVHD at 1 year did not differ between groups (BU: 47% [95% CI, 34%–60%]; TBI: 38% [95% CI, 25%–51%]; HR, 0.73; 95% CI, 0.45–1.17; P = .155) (Fig. 2B). The prevalence of cGVHD in each treatment group over time is plotted in Figure 2C.
Of the initial 139 patients, 81 (58%) died during follow-up, mostly (49%) from recurrent disease (Table 3). There were more deaths after disease progression in the TBI group (HR, 2.01; 95% CI, 0.95–2.83; P = .052). The cumulative incidence of NRM was statistically different between the 2 groups (HR, 0.48; 95% CI, 0.25–0.92; P = .027) (Fig. 3A). At 1 year, the disease-related mortality and NRM rates were 14% (95% CI, 6%–22%) and 17% (95% CI, 9%–26%) after BU-FLU-rATG, respectively, compared with 7% (95% CI, 1%–13%) and 11% (95% CI, 4%–18%) after FLU-TBI, respectively.
|No. of Patients (%)|
|Characteristic||FLU-BU-rATG, N = 69||FLU-TBI, N = 70|
|Grade 2||16 (23)||8 (11)|
|Grade 3-4||17 (24)||12 (17)|
|Time to grade 3-4 aGVHD: Median/range, d||42/14-84||53/21-84|
|Limited grade||10 (18)||15 (25)|
|Extensive grade||34 (61)||27 (46)|
|Time to cGVHD: Median/range, d||112/91-420||130/77-486|
|Cause of death|
|Five-y probability [95% CI], %|
|Relapse/progression||27 [17-37]||54 [44-64]|
|Progression-related mortality||21 [11-31]||37 [36-48]|
|Transplantation-related mortality||38 [27-48]||22 [12-32]|
|Survival||41 [29-53]||41 [28-53]|
|Progression-free survival||35 [23-47]||23 [13-35]|
Of 89 patients with measurable disease, 54 had an objective response at a median of 122 days (range, 28–604 days). The 1-year cumulative incidence of objective response was 65% (95% CI, 51%–79%) after FLU-BU-rATG compared with 46% (95% CI, 31%–60%) after FLU-TBI (HR, 0.58; 95% CI, 0.34–1.00; P = .05). The cumulative incidence of progression or relapse at 1 year was 14% after FLU-BU-rATG compared with 37% after FLU-TBI (HR, 2.49; 95% CI, 1.45–4.29; P < .01) (Fig. 3B).
The estimated probability of survival at 1 year was 75% (95% CI, 63%–84%) after FLU-BU-rATG and 74% (95% CI, 62%–83%) after FLU-TBI (HR, 0.91; 95% CI, 0.59–1.42; P = .69) (Fig. 4A). The estimated probability of PFS at 1 year was 68% (95% CI, 56%–78%) in the FLU-BU-rATG group compared with 51% (95% CI, 39%–62%) in the FLU-TBI group (HR for death, 1.38; 95% CI, 0.92–2.07; P = .11) (Fig. 4B). At 5 years, OS and PFS were not significantly different between the 2 groups (Table 3). Outcome comparison between the 2 groups did not differ after adjustment for population heterogeneity according to the risk of relapse and the outcome score (data not shown).
Of the 574 requested questionnaires, 439 (76%) were received. The HRQL data revealed that the highest symptom scores and the lowest functioning scores were experienced 1 month after stem cell transplantation (Fig. 5). Thereafter, the levels of symptoms decreased and the levels of functioning increased, tending to return to baseline. Hence, the P value for the time variable was < .05 for the global quality-of-life score for all functioning scales and for the symptoms pain, dyspnea, and loss of appetite. There was no significant group-by-time interaction except for the cognitive functioning scale, which revealed an enhanced negative impact of the FLU-BU-rATG conditioning regimen on cognitive function over time (P = .0225). We also observed a deleterious impact of the FLU-BU-rATG conditioning regimen on global quality of life, physical functioning, and role functioning independent of time (P = .004, P = .0075, and P = .0209, respectively). Fatigue, sleep disturbance, and loss of appetite were negatively impacted by FLU-BU-rATG independent of time (P = .0061, P = .0194, and P = .0145, respectively). The mean total cost per patient of allo-SCT was not significantly different between groups (Table 4).
|Cost Category||FLU-BU-rATG, N = 69||FLU-TBI, N = 70||Pb|
|Conditioning regimen||4114||1392||< .001|
|Hospitalization (including readmissions)||55,744||37,640||.017|
|Intensive care unit||2532||2930|
|Drugs (anti-infectious and anti-GVHD)||9904||10,295||NS|
|Treatment of progressionc||10,989||24,664||.001|
|Total cost at 5 y||84,289||75,150||NS|
To our knowledge, this study is the first prospective assessment of 2 different transplantation strategies with a long-term follow-up of 5 years. The treatment strategies differ not only in terms of myeloablation but also in terms of T-cell depletion of the graft and postgraft immunosuppression. Therefore, the results are to be interpreted according to the strategy rather than the conditioning alone.
Although OS and PFS were similar in the 2 groups, the post-transplantation course was markedly different, and each strategy exhibited benefits and issues for consideration. FLU-BU-rATG was associated with higher long-term NRM, whereas FLU-TBI was associated with graft failure, lower tumor response, and higher relapse rates. These results are in line with previous report suggesting that lower intensity conditioning resulted in lower NRM but a higher relapse rate.17 Altogether, because the outcomes from both conditioning regimens compared in this study were similar, with each group having specific strengths and weaknesses, a singular, definitive choice was difficult. However, because patients undergo allogeneic transplantation with the objective of treating malignancies, interpretations may differ slightly, notably from the perspective of further developments. Indeed, although a low toxicity profile is desirable, effective and successful control of disease also should be achieved and should be considered a priority in establishing credible therapy.
In this clinical trial, FLU-TBI conditioning was associated with lower NRM, as previously reported. However, the 54% 5-year disease progression probability is a major limitation. Indeed, although this high relapse incidence (Fig. 3B) may falsely minimize the long-term NRM results after FLU-TBI, this finding is in line with a recent report in patients with acute myeloid leukemia.18 The authors of that report pointed out that disease control is the Achilles heel of this regimen, especially in patients with advanced disease; and new or additional strategies, including treatments before and after allo-HSCT,19 are required to increase the graft-versus-leukemia effect without intensifying the conditioning. The impact of these new strategies remains to be defined and will require careful assessment of the NRM rate.
Conversely, FLU-BU-rATG conditioning, although it was associated with higher NRM than FLU-TBI but with low NRM compared with standard myeloablative conditioning, delivered better disease control. The NRM may be the final consequence of a higher rate of severe aGVHD (Fig. 2A). The cytotoxicity induced by busulfan administration likely represents the precipitating event of this reaction.2 It is noteworthy that this protocol was initiated in 2002 at a time when the oral form of busulfan without dose adjustment was commonly used in France. A recent report revealed a promising association with use of the intravenous form of busulfan and a lower NRM, even in sicker or older populations, indicating that the intravenous form is likely more appropriate than oral busulfan.20
Additional immunosuppression is a critical aspect of these approaches. Mycophenolate-mophetil after RIC initially was introduced to sustain stable mixed chimeras and engraftment after FLU-TBI.21 With respect to lymphoid engraftment, full donor chimerism was documented equally in both groups (Table 2), although the only graft failures occurred after FLU-TBI. In addition, we confirmed that the reinforcement of postgraft immunosuppression allows for a low rate of severe aGVHD. However, it is also likely that the potency and length of this immunosuppression account for lower disease control. Conversely, it is surprising that the conditioning achieved with in vivo T-cell depletion was associated with more GVHD and fewer relapses. This may be related to the low thymoglobulin dose used in this trial; indeed, the optimal dose is controversial with regard to GVHD prevention and disease control.22,23 From that perspective, a recent analysis from the Center for International Blood and Marrow Transplant Research raised some concerns about the use of in vivo anti-T-cell antibodies as part of an RIC regimen.24 However, it should be noted that the median reported dose of r-ATG in that analysis was 7 mg/kg, which represents a rather high dose. Indeed, we previously demonstrated that doses between 7.5 mg/kg and 10 mg/kg in RIC were associated with profound immunosuppression that was responsible for sustained engraftment,25 a low rate of severe GVHD25,26 but a source of severe late infection,27 and an eventual reduction in disease control.26 Data published by the Calgary group support the finding that an intermediate dose (approximately 4.5 mg/kg) may be optimal both for preventing GVHD and for retaining an antitumor effect.28,2 In line with these data, we recently reported that, in patients with acute myeloid leukemia who underwent matched-related allo-HSCT after RIC, which included a dose of 2.5 mg/kg or 5 mg/kg,30 the higher dose produced dramatic reductions in aGVHD and cGVHD without a loss of disease control.
One of the unique features of our study was the inclusion of prospective socioeconomic data. Although survival was the absolute endpoint, the quality of life for surviving patients is a crucial factor, particularly when analyzing 2 groups that had similar survival outcomes. Our quality-of-life analysis revealed that FLU-BU-rATG had a stronger negative impact on patients' quality of life during treatment than FLU-TBI, which persisted until 1 year after allo-HSCT. This probably results from the persistence of extensive cGVHD in the FLU-BU-rATG group. However, because more patients relapsed in the FLU-TBI group, this resulted in an imbalance of patients who were evaluated for cGVHD, which may have affected the interpretation of these data. Finally, a cost analysis revealed no difference in the total cost of procedures, and the lower transplantation costs after FLU-TBI were offset by increased costs of treatment for disease progression after transplantation. Notably, cost estimates were lower than predicted costs if the study and treatments were performed in the United States. However, because the objective of the study to compare the 2 treatments, it is likely that the end result of the study would have been similar if it had been based in the United States. These results support current studies that address improvements in the treatment regimen while retaining a low NRM rate.
We conclude that the introduction of RIC regimens represents progress in stem cell therapy, and we have demonstrated that long-term survival is possible in a population of patients previously deemed unsuitable for allo-HSCT. The findings presented in this report provide important information for the design of future studies with the ultimate goal of defining an optimal regimen that can control malignant disease with a low toxicity profile. Presumably, choices of conditioning regimens also will depend on donor, disease type, and disease situation.
This work was supported by a grant from the Association pour la Recherche Contre le Cancer (Pole ARECA, 2001) and by a Programme Hospitalier de Recherche Clinique grant from the Ministry of Health (2001).
CONFLICT OF INTEREST DISCLOSURES
The authors made no disclosures.