RIC versus MAC UCBT in adults with AML: A report from Eurocord, the ALWP and the CTIWP of the EBMT

Nonrelapse mortality (NRM) is the first cause of treatment failure after unrelated cord blood transplantation (UCBT) following myeloablative conditioning (MAC). In the last decade, reduced-intensity conditioning (RIC) regimens have been developed with the aim of reducing NRM and allowing older patients and those with medical comorbidities to benefit from UCBT. The aim of the current retrospective study was to compare transplantation outcomes of acute myeloid leukemia (AML) patients given UCBT after either RIC or MAC. Data from 894 adults with AML receiving a single or double UCBT as first allograft from 2004 to 2013 at EBMT centers were included in this study. 415 patients were given UCBT after RIC while 479 patients following a MAC. In comparison to MAC recipients, RIC recipients had a similar incidence of neutrophil engraftment and of acute and chronic graft-versus-host disease (GVHD). However, RIC recipients had a higher incidence of disease relapse and a lower NRM, translating to comparable leukemia-free (LFS), GVHD-free, relapse-free survival (GRFS) and overall survival (OS). These observations remained qualitatively similar after adjusting for differences between groups in multivariate analyses. In conclusion, these data suggest that LFS and OS are similar with RIC or with MAC in adults AML patients transplanted with UCBT. These observations could serve as basis for a future prospective randomized study.


INTRODUCTION
Allogeneic hematopoietic stem cell transplantation (allo-HCT) from HLA-identical sibling is the treatment of choice for selected patients with acute myeloid leukemia (AML) [1][2][3]. For AML patients who lack a suitable human leukocyte antigen (HLA)-identical sibling, unrelated cord blood transplantation (UCBT) is an adequate alternative to HLA-matched unrelated bone marrow/peripheral blood stem cell (PBSC) transplantation, particularly for patients at high risk of rapid disease relapse who urgently need a transplantation [4,5].
Despite major improvements in the field, nonrelapse mortality (NRM) has remained the main cause of failure of UCBT for AML [5]. In the last decade, reduced-intensity conditionings (RIC) for UCBT have been developed with the aim of reducing NRM and allowing older/unfit patients to benefit from UCBT [6][7][8][9]. Although a recent study demonstrated low NRM after RIC UCBT for AML [10], a high incidence of disease relapse has also been observed with this approach [10][11][12]. This prompted us to perform the current retrospective registry study aimed at assessing the impact of the conditioning intensity on transplantation outcomes in patients receiving UCBT as treatment for AML.

Patient, disease and transplant characteristics
Patients and disease characteristics are described in Table 1. Briefly, 415 patients were given UCB after this study. 415 patients were given UCBT after RIC while 479 patients following a MAC. In comparison to MAC recipients, RIC recipients had a similar incidence of neutrophil engraftment and of acute and chronic graft-versus-host disease (GVHD). However, RIC recipients had a higher incidence of disease relapse and a lower NRM, translating to comparable leukemia-free (LFS), GVHD-free, relapse-free survival (GRFS) and overall survival (OS). These observations remained qualitatively similar after adjusting for differences between groups in multivariate analyses. In conclusion, these data suggest that LFS and OS are similar with RIC or with MAC in adults AML patients transplanted with UCBT. These observations could serve as basis for a future prospective randomized study. show the unadjusted curves for MAC patients and the adjusted curves for RIC recipients. Curves were adjusted for age at transplantation, recipient gender, year of transplantation, disease status, TBF conditioning or not, TCF conditioning, or not, and the use of ATG. LFS, leukemia-free survival; OS, overall survival; RI, relapse incidence and NRM, nonrelapse mortality.

Engraftment and GVHD
Overall, CI of neutrophil engraftment at day 100 was not different in RIC (89%) and MAC (88%) recipients (P = 0.8) with the limitation that we did not systematically collect chimerism data and that autologous reconstitution is possible in RIC recipients. Median times for reaching 0.5 x 10 9 /L neutrophils were 21 (range, 3-66) days in RIC patients versus 23 (range, 1-106) days in MAC patients, respectively.
In univariate analysis, there was a higher incidence of grade II-IV acute GVHD in RIC recipients (35% versus 26%, P = 0.009) while the incidence of grade III-IV acute GVHD was similar in both groups of patients (11% and 11%, P = 0.9). However, in multivariate analyses adjusting for single versus double UCBT, gender combination, use of ATG and HLA-compatibility the incidence of grade II-IV acute GVHD was comparable in RIC and MAC recipients (HR = 1.09, P = 0.65).

Additional Cox analyses for OS and LFS
To further dissect the impact of conditioning intensity on UCBT, we performed additional Cox analyses comparing UCBT outcomes among patients conditioned with RIC or MAC regimen separately for pre-transplant variables. The results of these analyses are presented graphically using Forest plots in Figures 2-3.
RIC regimens were associated with lower NRM in each subgroup categories with median HR ranging from 0.3 to 0.9 (Figure 2A). This was particularly the case in the subgroup of younger patients (HR = 0.3, 95% CI: 0.2-0.6) and in those not given ATG (HR = 0.6, 95% CI: 0.4-0.9). However, RIC regimen was also associated with a MAC, myeloablative conditioning; RIC, reduced-intensity conditioning; Y, year; M, male; UCBT, umbilical cord blood transplantation; CR, complete remission; #, number of patients; ATG, anti-thymocyte globulins; TNC, total nucleated cells; tacro, tacrolimus; CSP, cyclosporine A; MMF, mycophenolate mofetil. TCF, total body irradiation (TBI), cyclophosphamide and fludarabine; TBF, Thiothepa, busulfan and fludarabine. 1 , calculated with c 2 statistics for categorical variables and Mann-Whitney test for continuous variables; 2 , defined as t(8;21), t(15;17), inv or del (16), or acute promyelocyticleukemia, these abnormalities only or combined with others; 3 , defined as all cytogenetics not belonging to the good or high risk (including trisomias); 4 , defined as 11q23 abnormalities, complex karyotype, abnormalities of chromosomes 5 and 7; 5 , classified as RIC if the dose of TBI was < 6 Gy; 6 classified as RIC when the busulfan total dose was ≤ 8 mg/kg). www.impactjournals.com/oncotarget  higher risk of relapse in each subgroup categories with HR ranging from 1.2 to 2.5 ( Figure 2B). Interestingly, the association between RIC regimen and higher risk of relapse was not less pronounced in patients in CR1 at transplantation (HR = 2.5, 95% CI: 1.7-3.7) than in those in CR3+ or advanced disease (HR = 1.4, 95% CI HR 0.9-2.1), and was also observed in the subgroup of patients transplanted following TCF regimen (HR = 2.0, 95% CI 1.2-3.2). As shown in Figure 3A, RIC and MAC patients had comparable LFS in most transplantation variable subgroup. Interestingly, the assessment of heterogeneity according to age group evidenced a I 2 = 4.3%, demonstrating a very low impact of age on the association between conditioning intensity and LFS. However Finally, RIC and MAC patients had comparable OS in each transplantation variable subgroup ( Figure 3B).

DISCUSSION
The impact of dose intensity on outcomes in AML patients has been the focus of many studies since the development of non-myeloablative/RIC regimens in the last 2 decades [13][14][15][16][17][18]. These studies have focussed mainly on patients given PBSC as stem cell source. Large registry studies observed that the use of RIC regimen was associated with a higher risk of relapse, but also a lower incidence of NRM translating to similar OS and LFS [13][14][15][16]. More recently, the BMT-CTN performed a randomized study comparing RIC versus MAC in patients with MDS (N = 54) or AML (N = 218) (18-65 years, HCT-specific comorbidity index score [19] ≤ 4) who had < 5% marrow myeloblasts at allo-HCT [20]. As observed in registry studies, the use of RIC regimen was associated with higher risk of relapse (48 versus 14% at 18 months, P < 0.01) but also lower NRM (4% versus 16% at 18-months, P = 0.02). However, because the incidence of NRM was already low in MAC recipients, the amplitude of the reduction of NRM with RIC was insufficient to offset its negative impact on relapse, translating to significantly worse LFS in RIC recipients (47% vs 68% at 18 months, P < 0.01).
In contrast to allo-HCT using other stem cell source where relapse is usually the first cause of transplant failure in AML patients, NRM has remained the leading cause of death in UCB recipients given myeloablative conditioning [21]. Further, recent findings in humanized mice suggest that graft-versus-tumor effects might be higher with UCB than with PBSC [22], confirming prior clinical observations [23]. This could suggest that dose intensity might be less important following UCBT than following PBSCT in regards to risk of relapse. Based on these findings, we hypothesized that the use of RIC versus MAC might be beneficial in the UCBT setting.
This prompted us to perform a retrospective study in the EBMT/Eurocord registries comparing UCBT outcomes in patients administered a RIC or a MAC regimen. Several observations were made.
First, despite prior evidence of strong graft-versustumor effects after UCBT [23], RIC UCBT recipients had a significantly higher risk of relapse than MAC UCBT recipients (HR = 1.6, 95% CI: 1.2-2.2). This confirms the results of a prior single center study [24] and demonstrates the importance of dose intensity for preventing AML relapse even in the setting of UCBT. As expected, NRM was lower after RIC than after MAC UCBT. As a net result, LFS, GRFS and OS were comparable in RIC and MAC recipients, rejecting our hypothesis that RIC regimens would provide better outcomes than MAC in AML patients undergoing UCBT.
The current study also confirmed a detrimental impact of ATG on NRM as recently reported in a study including data from patents given UCB after MAC conditioning [25] or RIC in the double UCBT setting [26]. Further, despite ATG not only induces in vivo T-cell depletion of the graft but also promotes the generation of regulatory T cells [27,28], ATG failed to prevent GVHD in the current study. However ATG had no impact either on relapse incidence, in agreement with recent observations in the PBSC transplantation setting [29,30]. Further, as previously observed in the UCB setting [5,11], older age was associated with worse LFS and OS.
There are some limitation in our study including its design (retrospective registry survey), and the imbalance of the two groups for risk factors known to be associated with outcome: RIC patients were almost 2 decades older and were more often given UCB with ≥2 HLA-mismatches, but they received more often double UCBT and consequently received more cells while they were given less frequently ATG. These differences were carefully adjusted for in multivariate analysis while where forest plots demonstrated comparable OS with RIC and with MAC in each pre-transplant subgroups. Interestingly, a trend for better LFS was observed in MAC recipients in the subgroup of patients transplanted in first CR, in those who received ATG, and in patients infused with > 3.2 x10 7 TNC/Kg.
In summary, we observed that LFS and OS were similar with RIC and with MAC in adult AML patients offered UCBT. These observations could serve as basis for a future prospective randomized study. In the meantime, Oncotarget 43034 www.impactjournals.com/oncotarget Oncotarget 43035 www.impactjournals.com/oncotarget recent advances in the field of UCBT such as optimization of myeloablative regimen for UCBT, expansion of UCB hematopoietic stem/progenitor cells, and post-transplant administration of chimeric antigen receptor T cells are likely to improve outcomes of UCBT both in the RIC and in the MAC setting [31][32][33].

Data collection
This survey is a retrospective study performed by the Acute Leukemia Working Party (ALWP) of the European group for Blood and Marrow Transplantation (EBMT) and by Eurocord. EBMT registry is a voluntary working group of more than 500 transplant centers, participants of which are required once a year to report all consecutive stem cell transplantations and followup. Eurocord collects data on UCBT performed in > 50 countries worldwide and > 500 transplant centers, mainly EBMT centers. Population selection criteria included adult recipients (defined as ≥ 18 years old at UCBT), primary AML, first allogeneic stem cell transplantation, single or double unit UCBT performed from 2004 to 2013. Grading of acute and chronic GVHD was performed using established criteria [34]. HLA-compatibility was based on antigenic level typing for HLA-A and -B, and allele-level typing for HLA-DRB1. For the purpose of this study, all necessary data were collected according to EBMT and Eurocord guidelines. RIC was defined as use of fludarabine (Flu) associated with < 6 Gy total-body irradiation (TBI), or busulfan ≤ 8 mg/kg, melphalan ≤ 140 mg/m 2 or other nonmyeloablative drugs, as previously reported [14,[35][36][37]. Specifically, the combination of total body irradiation, cyclophosphamide and fludarabine (TCF regimen) was classified as RIC when the TBI dose was < 6 Gy (most RIC patients were given 2 Gy TBI) and as MAC when the TBI dose was ≥ 6 Gy (most MAC patients were given > 10 Gy TBI). Similarly, the association of thiothepa, busulfan and fludarabine (TBF regimen) was classified as RIC or MAC based on the dose of busulfan received (≤ 8 mg/kg or > 8 mg/kg, respectively).

Ethics
The scientific boards of the ALWP of EBMT and of Eurocord approved this study.

Statistical analyses
Data from all patients meeting the inclusion/ exclusion criteria were included in the analyses. Start time was date of transplant for all endpoints. Neutrophil engraftment was defined as first of 3 consecutive days with a neutrophil count of at least 0.5 x 10 9 /L.
To evaluate the relapse incidence, patients dying either from direct toxicity of the procedure or from any other cause not related to leukemia were censored. NRM was defined as death without experiencing disease recurrence. Patients were censored at the time of relapse or of the last follow-up. Cumulative incidence functions (CIF) were used for relapse incidence and NRM in a competing risk setting, since death and relapse were competing together.
For estimating the cumulative incidence of chronic GVHD, death was considered as a competing event. OS and LFS were estimated using the Kaplan-Meier estimates. GVHD-free, relapse-free survival (GRFS) was defined as being alive with neither grade III-IV acute GVHD, severe chronic GVHD nor disease relapse [38]. Univariate analyses were done using Gray's test for CIF and log rank test for OS and LFS. Associations of patient and graft characteristics with grade II-IV acute GVHD were evaluated using multivariate logistic regression. Variables introduced in the multivariate logistic regression included conditioning intensity (RIC versus MAC), single or double UCBT, gender combination, use of ATG and HLA-compatibility. Associations of patient and graft characteristics with other outcomes (chronic GVHD, relapse, NRM, LFS and OS) were evaluated in multivariable analyses, using Cox proportional hazards. Variables introduced in the Cox models included conditioning intensity (RIC versus MAC), conditioning type (thiotepa, busulfan and fludarabine (TBF) versus other and TBI, Flu and cyclophosphamide (TCF) versus other), the use of ATG or not, recipient age, recipient gender, years of transplantation and disease status at transplantation. Exploratory analyses of the heterogeneity of RIC vs MAC among pre-transplant subgroups for OS and LFS were performed using Cox models. The results of these Cox models were presented graphically using forest plots [39]. Heterogeneity according to age group for LFS was assessed by calculating the I 2 = (Qstatistic-degre of freedom)/Qstatistic x 100.
All tests were two sided. The type I error rate was fixed at 0.05 for determination of factors associated with time to event outcomes. Statistical analyses were performed with SPSS 19 (SPSS Inc, Chicago, IL), and R 2.13.2 (R Development Core Team, Vienna, Austria) software packages.