- © 2008 by American Society of Clinical Oncology
Sustained Remissions of High-Risk Acute Myeloid Leukemia and Myelodysplastic Syndrome After Reduced-Intensity Conditioning Allogeneic Hematopoietic Transplantation: Chronic Graft-Versus-Host Disease Is the Strongest Factor Improving Survival
- David Valcárcel,
- Rodrigo Martino,
- Dolores Caballero,
- Jesus Martin,
- Christelle Ferra,
- Jose B. Nieto,
- Antonia Sampol,
- M. Teresa Bernal,
- Jose L. Piñana,
- Lourdes Vazquez,
- Jose M. Ribera,
- Joan Besalduch,
- Jose M. Moraleda,
- Dolores Carrera,
- M. Salut Brunet,
- Jose A. Perez-Simón and
- Jorge Sierra
- From the Hospital de la Santa Creu i Sant Pau (Universitat Autonoma Barcelona), Barcelona; Hospital Clínico Universitario Salamanca, Salamanca; Hospital Germans Trias i Pujol, Badalona; Hospital Morales Meseguer, Murcia; Hospital Son Dureta, Palma de Mallorca, Mallorca; Hospital Central de Asturias, Ovieda, Spain
- Corresponding author: David Valcárcel, MD, H. Santa Creu i Sant Pau, Sant Antoni Maia Claret 167, Barcelona, Spain 08025; e-mail: dvalcarcel{at}santpau.es
Abstract
Purpose Reduced-intensity conditioning (RIC) for allogeneic stem-cell transplantation (allo-SCT) reduces nonrelapse mortality (NRM). This reduction makes it possible for patients who are ineligible for high-dose myeloablative conditioning allo-SCT to benefit from graft-versus-leukemia reaction. In this multicenter, prospective study of patients with acute myeloid leukemia (AML) and high-risk myelodysplastic syndrome (MDS), we investigated the efficacy of RIC allo-SCT from a human leukocyte antigen–identical sibling by using a regimen that uses fludarabine and busulfan.
Patients and Methods Ninety-three patients with AML (n = 59) and MDS (n = 34) were included, and the median age was of 53 years. Follow-up for survivors was 43 months (range, 3 to 89 months). The conditioning regimen consisted of fludarabine (150 mg/m2) and oral busulfan (8 to 10 mg/kg). All except one patient received mobilized peripheral blood stem cells. Graft-versus-host disease (GVHD) prophylaxis consisted of cyslosporine and methotrexate or mycophenolate mofetil.
Results The 100-day, 1-year, and 4-year incidences of NRM were 8, 16%, and 21%, respectively. The 1- and 4-year relapse cumulative incidences were 23% and 37%, respectively, and leukemia recurrence was the main cause of death. The 4-year disease-free survival (DFS) and overall survival (OS) rates were 43% and 45%, respectively. The 4-year cumulative incidence of chronic GVHD was 53% (45% extensive), and its development was the major factor associated with lower relapse incidence and improved DFS and OS.
Conclusion Our results confirm the capacity of this RIC regimen to obtain long-term remissions in patients ineligible for a conventional allo-SCT. The results suggest an important role of the development of chronic GVHD in reducing relapse and improving DFS and OS.
INTRODUCTION
Initial uncertainties about reduced-intensity conditioning (RIC) regimens for allogeneic transplantation (allo-SCT) have been convincingly answered, by the demonstration of their efficacy in allowing engraftment and in decreasing nonrelapse mortality (NRM) in patients ineligible for conventional (high-dose conditioning) allo-SCT.1-3 Currently, a key aspect to be investigated is whether RIC regimens have a graft-versus-leukemia (GvL) effect that is strong enough to cure high-risk patients, even those who have relapsed after autologous SCT.
Up-to-date, conventional allo-SCT is considered the best approach for patients with acute myeloid leukemia (AML) and high-risk myelodysplastic syndrome (MDS), because it combines high-dose chemotherapy with an immune-mediated GvL reaction.4-7 However, this approach is not recommended in many patients because of an unacceptable high risk of NRM (eg, from second transplants, poor performance status, other comorbidities, or advanced age). An option for this group is autologous SCT, but the advantage compared with chemotherapy in high-risk AML is unclear.8,9 In contrast, it would be of interest to know if these patients could benefit from a GvL effect, especially when the tumor burden is low.10
To estimate the potential of this approach, Lazarus et al11 have reviewed the experiences from several groups that used RIC regimens in AML.12-18 These studies present several problems that hamper firm conclusions: most studies are retrospective; include diverse diagnoses and usually only a low number of AML/MDS; use different conditioning regimens and are often heterogeneous; pool sibling and unrelated transplants in the same study; and include bone marrow transplantation (BMT) and peripheral blood stem-cell transplantation (PBSCT) as sources (with or without in vivo or ex vivo T-cell depletion) in the same study. These studies report an NRM that ranges from 5% to 51%, a relapse incidence between 11% and 45%, an overall survival (OS) between 17% and 68%, and a disease-free survival (DFS) that ranges from 31% to 80%.
The present study reports the outcome of patients with AML/MDS who underwent RIC allo-SCT with busulfan (8 to 10 mg/m2) and fludarabine (150 mg/m2) in six Spanish centers. The prospective design of the study and the long follow-up (median, 43 months) allow the analysis of patients who experienced late treatment failures with this RIC allo-SCT. All the patients were candidates for allo-SCT, but—for different reasons, mainly advanced age—the anticipated early high NRM precluded the use of conventional conditioning. Our results of 45% OS despite the presence of several high-risk factors in most patients (with 20% of them undergoing SCT with refractory disease) are encouraging. Of note, our study confirms that chronic GVHD is the most important factor for reducing relapses and for improving DFS and OS after RIC allo-SCT for myeloid malignancies.
PATIENTS AND METHODS
Patients
Ninety-three patients with AML/MDS received fludarabine and busulfan for RIC allo-SCT from human leukocyte antigen (HLA)–identical siblings on a phase II prospective trial conducted in six Spanish centers between 1998 and 2005. The inclusion criteria were an AML considered for allo-SCT but at high risk for NRM with conventional conditioning; and provision of written informed consent. Exclusion criteria were creatinine more than than twice the upper limit of normal (ULN); AST and/or ALT more than three times the ULN; bilirubin more than twice the ULN; other nonhematologic malignancies; or HIV positive status. Patient characteristics are listed in Table 1.19 The median age was 53 years (range, 21 to 70 years), and 38 patients (41%) were aged ≥ 60 years. The diagnoses were AML in 59 patients (63%) and MDS in 34 patients (37%). Patients who had MDS with increased blasts underwent the transplantation as first-line therapy if they had less than 10% bone marrow blasts. If they had more than 10% blasts, they received AML-type chemotherapy first and then proceeded to allo-SCT.
Patient Characteristics
Patients were enrolled after protocol approval by the Spanish Drug Regulator Agency and by local research ethical committees, and written informed consent was obtained from patients and donors before conditioning.
Conditioning Regimen and Graft-Versus-Host- Disease Prophylaxis
The conditioning regimen consisted of intravenous (IV) fludarabine 150 mg/m2 (30 mg/m2 on days −9 to −5) and oral busulfan 8 to 10 mg/kg (1 mg/kg every 6 hours on days −6 to −4); stem cells were infused on day 0. All patients received mobilized peripheral blood stem cells (except for one patient, who received bone marrow) from an HLA-identical sibling.
Graft-versus-host disease (GVHD) prophylaxis consisted of cyslosporine (CsA) in all patients and a short course of methotrexate (MTX) (on days 1, 3, and 6) followed by folinic acid rescue for 82 patients. Mycophenolate mofetil (MMF) 1 g every 8 hours was used instead of MTX in 11 patients. The CsA dose was maintained until 3 months after transplantation and then was tapered if no acute GVHD appeared. In 2004, there was a modification in the conditioning regimen and the GVHD prophylaxis to further reduce drug-related toxicity. Patients previously reported as having the highest risk of 1-year NRM received busulfan 8 mg/kg instead of 10 mg/kg and received MMF instead of MTX.20 Because all outcomes were identical in these 11 patients, they are described together with the other 82 patient cases.
Anti-infective measures were determined by local protocols and included (in all instances) the following: trimethoprim/sulfamethoxazole or nebulized pentamidine as prophylaxis for Pneumocystis jirovecii infections; and twice-weekly monitoring for cytomegalovirus (CMV) reactivation, with administration of pre-emptive ganciclovir or foscarnet therapy when monitoring resulted in positive results.
Donor lymphocyte infusions (DLIs) were given because of relapse or because of persistent mixed chimerism after CsA tapering.
Statistical Considerations and Definitions
The main end points of the study were OS, DFS, NRM, and relapse. The probabilities of OS and DFS were estimated from the time of transplantation by using Kaplan-Meier curves,21 whereas NRM and relapse were calculated by using cumulative incidence estimates and by taking into account the competing risk structure.22,23 Other post-transplantation outcomes that were calculated by using cumulative incidence estimates were hematopoietic recovery and acute and chronic GVHD. Univariate analyses of the association of various clinical risk factors with these post-transplantation outcomes were calculated using univariate Cox regression models, whereas the log-rank test was used for OS and DFS.22,23 Multivariate analyses were performed by Cox proportional hazards regression, and variables with a P value < .1 in the prior univariate testing were included. The assumption of proportional hazards over time was tested for all explanatory covariates by using a time-dependent covariate. Quantitative variables that were found to have an impact on any outcome were reanalyzed as categoric variables. Tests of significance were two-sided and had a significance level of .05 or less. All statistical analyses were performed with SPSS version 13.0 (SPSS, Chicago, IL) with the exception of the cumulative incidence analyses, which were carried out with NCSS 2004 (Number Cruncher Statistical System; NCSS, Kaysville, UT). The variables studied in the univariate analysis are listed in Table 2. If acute or chronic GVHD (included individually as time-dependent covariates) were found to have an impact on post-transplantation outcomes in the final multivariate Cox analysis, plots were constructed to illustrate visually the effects of GVHD on these outcomes, as described elsewhere.24
Univariate and Multivariate Analyses of Outcomes
Chemotherapy sensitivity was defined as the achievement of complete remission (CR) after the last course of chemotherapy (CT), whereas chemotherapy refractoriness included responses different than CR. Those patients who did not receive CT before allo-SCT were described as untreated.
An early disease status was defined according to accepted criteria.25,26 Specifically, this category included patients with poor-prognosis AML or refractory anemia with excess blasts–type 2 in first CR after induction chemotherapy and with untreated refractory anemia with excess blasts–type 1 (bone marrow blasts < 10%).
Neutrophil and platelet engraftment were defined as the first of three consecutive days with an absolute neutrophil count greater than 0.5 × 109/L and an untransfused platelet count greater than 20 × 109/L. Acute and chronic GVHD were graded according to standard criteria27; the diagnoses of acute and chronic GVHD were not made on the basis of the classical 100-day cutoff only, but diagnoses were made by the clinical presentation and the histopathologic findings of biopsies, as recommended recently by an NIH working group.28 Response and relapse were defined according to standard hematologic criteria. All patients who survived at least 15 days were evaluated for hematologic recovery.
Chimerism analyses was performed by using a polymerase chain reaction (PCR) of informative minisatellite loci, as previously specified.29 Complete donor chimerism (CDC) was defined as the presence of 100% donor DNA in the sample analyzed. CDC was defined on the basis of T-cell chimerism, if available (n = 46), or on unfractionated nucleated cells (n = 86).
RESULTS
Engraftment and Chimerism
All but two patients, who died early (day 2 and 6), were assessable for hematologic reconstitution. Neutrophil recovery occurred in all patients at a median of 17 days (range, 7 to 27 days), and platelet recovery occurred at a median of 11 days (range, 0 to 59 days). There were neither primary nor secondary graft failures.
Eighty-six patients (92%) were assessable for chimerism studies. The median time to reach complete donor chimerism (CDC) was shorter in total nucleated cells than in T lymphocytes (median, 30 days [range, 17 to 240 days] v median, 100 days [range, 17 to 731 days], respectively; P < .01).
GVHD
Thirty-three patients (35%) developed acute GVHD, and 14 of these (15%) were severe (grade 3 to 4). The cumulative incidences were 34% (95% CI, 26% to 46%) and 12% (95% CI, 7% to 21%), respectively.
Fifty-one patients developed chronic GVHD, which resulted in a 4-year cumulative incidence of 53% (95% CI, 44% to 65%). Thirty-nine patients (50%) developed extensive chronic GVHD, which resulted in a 4-year cumulative incidence of 45% (95% CI, 35% to 58%). The characteristics of chronic GVHD are listed in Table 3.
Chronic GVHD Characteristics
At the last follow-up, 29 patients (35%) had been completely withdrawn from immunosuppressive (IS) medication at a median of 6 months (range, 2 to 49 months) post-SCT. Nineteen (43%) of the 43 surviving patients were free of IS medication at the last follow-up, which was a median of 9 months (range, 3 to 41 months) post-transplantation.
Outcomes
Overall survival.
At the last follow-up, 44 patients (47%) were alive. Forty-nine patients died as consequence of the following: relapse or progression of the underlying disease (n = 27; 29%), GVHD (n = 12; 13%), infections (n = 5; 5%), or other toxicities (n = 5; 5%). The 4-year probability of OS was 45% (95% CI, 34% to 56%) for the entire group. OS for patients with AML and MDS were 42% (95% CI, 28% to 56%) and 49% (95% CI, 31% to 76%), respectively (P = .4 Fig 1A). Table 2 lists the variables associated with OS. In univariate analyses, chemotherapy-refractory disease, the presence of greater than 5% blasts in BM at SCT, high-risk cytogenetics, advanced-phase disease, and the absence of chronic GVHD were related with poorer OS. In multivariate analysis, only the presence of chronic GVHD improved OS (hazard ratio [HR] 6.1; 95% CI, 3.3 to 11.3; P < .001; Table 2; Fig 2).
(A and B) Overall and (C and D) disease-free survival. MDS, myelodysplastic syndrome; AML, acute myeloid leukemia.
Impact of chronic graft-versus-host disease (GVHD) on the outcome of the patients: (A) overall survival; (B) disease-free survival; and (C) relapse. (——) Patients who developed chronic GVHD. (···) Patients who did not develop chronic GVHD.
Relapse and DFS.
Thirty-three patients (35%) relapsed at a median of 4 months (range, 0.6 to 49 months) after SCT. The 3-month, 1-year, and 4-year cumulative incidences of relapse were 15% (95% CI, 9% to 25%), 29% (95% CI, 21% to 40%), and 37% (95% CI, 28% to 49%), respectively, for the entire group (Fig 3). The 4-year cumulative incidences for AML and MDS were 44% (95% CI, 32% to 59%) and 25% (95% CI, 14% to 46%), respectively.
Relapse.
In univariate analysis, variables associated with greater relapse risk were chemotherapy-refractory disease, high-risk cytogenetics, the presence of greater than than 5% blasts in BM at SCT, and the absence of chronic GVHD. In the multivariate analysis, risk factors were the absence of chronic GVHD (HR, 6.5; 95% CI, 3.1 to 13.8; P < .001) and an advanced disease status (HR, 2.2; 95% CI, 1.1 to 4.4; P = .03; Table 2; Fig 2).
The 4-year probability of DFS for the entire group was 43% (range, 32% to 54%). The DFS rates for AML and MDS were 39% (95% CI, 26% to 52%) and 49% (95% CI, 31% to 67%), respectively (P = .3; Fig 1B). In univariate analysis, the variables associated with poorer DFS were advanced-phase disease, chemotherapy-refractory disease, high-risk cytogenetics, the presence of greater than 5% of blast in BM at SCT, and the absence of chronic GVHD. In the multivariate analysis, the only variables that influenced DFS were the absence of chronic GVHD (HR, 5.3; 95% CI, 2.4 to 13.8; P < .001) and an advanced disease status (HR, 2.6; 95% CI, 1.2 to 5.4; P = .01; Table 2; Fig 2).
Eleven patients (12%) received at least one DLI, and all of them were for disease relapse. The median time from transplantation to DLI was 4.6 months (range, 3 to 43.5 months). Four (36%) of these patients received some type of chemotherapy before DLI. Complete response after DLI was obtained in five patients (42%), and four of these patients were alive and without AML/MDS at the last follow-up (1, 9, 10, and 35 months after DLI, respectively). Two patients received DLI in early relapse (ie, cytogenetic or molecular relapse without hematologic relapse), and both patients achieved CR and maintained CR at last follow-up. All patients who did not respond to DLI died as a result of AML/MDS.
Nonrelapse mortality.
Eighteen patients died as a result of nonrelapse causes at a median of 4.3 months (range, 0 to 21 months). The 3-month, 12-month, and 4-year cumulative incidences of acute NRM were 8% (95% CI, 4% to 15%), 16% (95% CI, 10% to 26%), and 20% (95% CI, 14% to 32%), respectively. The most frequent causes of NRM were GVHD (n = 13), infections without active GVHD (n = 5), and other toxicities (two episodes of CNS bleeding, one pancreatic adenocarcinoma, and one thromboembolism).
In univariate analyses, variables associated with higher NRM rates were acute GVHD, age older than 60 years, greater than 5% blasts in BM, and the absence of prior chemotherapy. In multivariate analyses, acute GVHD (grades 2 to 4) was the most important variable associated with NRM (HR, 3.7; 95% CI, 1.4 to 13; P = .005), whereas age older than 60 years showed a trend (HR, 2.6; 95% CI, 1 to 6.9; P = .059; Table 2).
Twenty seven (31%) of 86 assessable patients developed at least one CMV infection, and 7 patients (8%) developed CMV disease, which occurred in the setting of GVHD in all patients and was the cause of death in three patients (4%).
DISCUSSION
In the last 10 years, RIC allo-SCT has been increasingly used for providing the GvL effect to patients ineligible for conventional transplantation. Unfortunately, the GvL effect is closely linked to GVHD, and—so far—it is not possible to separate the two phenomena. The key finding of our study was that the development of chronic GVHD was the strongest variable that protected from relapse. Because chronic GVHD did not significantly increase NRM, this variable was also the strongest predictor for 4-year OS and DFS. Acute GVHD is also associated with the GvL effect, but in our study it led to an increased NRM and did not provide any benefit in OS or DFS.
In our study, as in most other series about RIC allo-SCT, relapse was the main cause of treatment failure.30,31 As in other studies that involved AML and high-risk MDS, most relapses occurred a few months post-transplantation. In our series, a median time to relapse of only 4 months was too short to allow a powerful and sustained GvL effect29,32 The platform for this effect is complete T-cell chimerism, which was observed after 100 days or more in most patients who underwent transplantations. Additionally, most patients who relapsed were still under the influence of IS drugs. Two alternatives may contribute to a decrease in early relapse after SCT. First, earlier tapering of IS medications to allow faster T-cell complete donor chimerism and an earlier GvL effect, despite severe acute GVHD, may increase NRM; this approach in the Seattle experience was unsuccessful.33 Second, the use of agents with antineoplastic activity to control leukemia progression until the GvL effect takes place may contribute.
RIC was designed for patients with a high-risk of NRM after conventional procedures. The observed NRM in this series was low (8% at 100 days and 16% at 1 year); thus we succeeded in achieving this initial goal. Because most occurrences of NRM resulted from acute GVHD, a simplistic approach could be to eliminate this complication with the in vivo use of alemtuzumab.34 However, this approach would also decrease occurrences of chronic GVHD, but GVHD was the strongest factor associated with improved outcome in our series.
The dose-intensity of chemotherapy used as a conditioning regimen also may have a relevant impact on relapse.30,31,35-37 In a group of 18 patients with AML who were in first CR and who received total-body irradiation 2Gy and fludarabine 90 mg/m2, Feinstein et al.33 recently reported seven of 10 deaths associated with relapse at a median of 198 days and a 1-year NRM of 16%. Conversely, the low 29% relapse rate observed in the current study could be explained partially by the more potent conditioning regimen used.
The use of DLI in relapses of AML and MDS after conventional allo-SCT has limited efficacy.38,39 In our limited experience (11 patients), DLI appeared effective in a minority of patient cases. Of note, in the two patients who received DLI because of nonovert hematologic relapse (one molecular and one cytogenetic), a sustained CR was achieved. In contrast, only three of nine patients with overt leukemia achieved a CR. These data argue in favor of using DLI when a low tumor burden is detected (ie, by immunophenotyping, cytogenetics, or molecular studies). The feasibility of intervening in such early relapses is still uncertain. Marks et al40 reported their results with DLI after RIC allo-SCT, mostly with T-cell depletion (ie, alemtuzumab or CD 34+ selection) in various hematologic malignancies. They showed that the achievement of a response was closely associated with the development of acute and/or chronic GVHD. This finding is in agreement with our results, which showed that chronic GVHD was the most important factor associated with DFS and relapse.
The high relapse rate observed in patients who did not develop chronic GVHD may justify the use of DLI before relapse in these patients. However, because a significant proportion of patients developed severe GVHD after DLI, it is important to adequately select those patients who really need this intervention.
In conclusion, our study lends further support to the use of RIC allo-SCT in patients with AML/MDS who are not appropriate candidates for a standard conditioning regimen. The current study shows that the development of chronic GVHD not only reduces relapse but also improves 4-year survival, which—to the best of our knowledge—has not been reported to date in a large and homogeneous series of patients with AML/MDS who were treated with RIC allo-SCT. Nevertheless, patients with advanced AML and high-risk MDS require novel approaches to reduce early post-transplantation relapses.
AUTHORS’ DISCLOSURES OF POTENTIAL CONFLICTS OF INTEREST
The author(s) indicated no potential conflicts of interest.
AUTHOR CONTRIBUTIONS
Conception and design: David Valcárcel, Rodrigo Martino
Provision of study materials or patients: David Valcárcel, Rodrigo Martino, Dolores Caballero, Jesus Martin, Christelle Ferra, Jose B. Nieto, Antonia Sampol, M. Teresa Bernal, Jose L. Piñana, Lourdes Vazquez, Jose M. Ribera, Joan Besalduch, Jose M. Moraleda, Dolores Carrera, M. Salut Brunet, Jose A. Perez-Simón, Jorge Sierra
Collection and assembly of data: David Valcárcel, Rodrigo Martino, Dolores Caballero, Jesus Martin, Christelle Ferra, Jose B. Nieto, Antonia Sampol, M. Teresa Bernal, Jose L. Piñana, Jose M. Ribera, Joan Besalduch, Jose M. Moraleda, Dolores Carrera, Jose A. Perez-Simón, Jorge Sierra
Data analysis and interpretation: David Valcárcel, Rodrigo Martino
Manuscript writing: David Valcárcel, Rodrigo Martino
Final approval of manuscript: David Valcárcel, Rodrigo Martino, Dolores Caballero, Jesus Martin, Christelle Ferra, Jose B. Nieto, Antonia Sampol, M. Teresa Bernal, Jose L. Piñana, Lourdes Vazquez, Jose M. Ribera, Joan Besalduch, Jose M. Moraleda, Dolores Carrera, M. Salut Brunet, Jose A. Perez-Simón, Jorge Sierra
Footnotes
-
published online ahead of print at www.jco.org on December 17, 2007.
-
Supported in part by the Instituto de Salud Carlos III (expedient Grant No. CM06/00,139, Ministerio de Sanidad to J.L.P., and Grants No. PI052312 and RD06/0020/0707 to J.S.); the Instituto de Recerca Hospital de la Santa Creu I Sant Pau (J.L.P.); and DURSI (Grant No. 2005SGR01075 to J.S.), Catalonia, Spain.
-
Both D.V. and R.M. contributed equally to this work and should be considered co–first authors.
-
Authors’ disclosures of potential conflicts of interest and author contributions are found at the end of this article.
- Received March 9, 2007.
- Accepted October 24, 2007.
