Biol Blood Marrow Transplant 25 (2019) 183190 Biology of Blood and Marrow Transplantation journal homepage: www.bbmt.org Allocation to Matched Related or Unrelated Donor Results in Similar Clinical Outcomes without Increased Risk of Failure to Proceed to Transplant among Patients with Acute Myeloid Leukemia: A Retrospective Analysis from the Time of Transplant Approval Eduardo Rodríguez-Arbolí1,y, Francisco Jose Marquez-Malaver1,y, Nancy Rodríguez-Torres1, Teresa Caballero-Velazquez1, Virginia Escamilla-Gomez1, Cristina Caldero n-Cabrera1,  Francisco Falantes-Gonza Jose 1 -Rodríguez , Patricia García-Ramírez1, lez , María Sole 2 María Moya-Arnao , Enric Carreras , Ildefonso Espigado-Tocino1, Jose 1 3  Antonio Pe rez-Simo n1,* 1 Department of Hematology, Hospital Universitario Virgen del Rocío, Instituto de Biomedicina de Sevilla, Universidad de Sevilla, Seville, Spain 2 Department of Hematology, Hospital Juan Ramon Jimenez, Huelva, Spain 3 Spanish Bone Marrow Donor Registry, Foundation and Research Institute Josep Carreras Against Leukemia, Barcelona, Spain Article history: A B S T R A C T Received 31 May 2018 Clinical outcomes after allogeneic hematopoietic stem cell transplantation (allo-SCT) from unrelated donors Accepted 19 August 2018 (URDs) approach those of matched related donor (MRD) transplants in patients with acute myeloid leukemia (AML). Yet, available data fail to account for differences in pretransplantation outcomes between these donor Keywords: selection strategies. In this regard, URD allo-HSCT is associated with longer waiting times to transplantation, Acute myeloid leukemia potentially resulting in higher probabilities of failure to reach transplant. We retrospectively analyzed 108 AML Matched related donor patients accepted for first allo-HSCT from the time of approval to proceed to transplant. Fifty-eight (54%) patients Unrelated donor were initially allocated to MRD, while URD search was initiated in 50 (46%) patients. Time to transplant was longer in patients allocated to a URD when compared with patients assigned to an MRD (median 142 days versus 100 days; p < .001). Forty-three of 58 (74%) patients in the MRD group and 35 of 50 (70%) patients in the URD group underwent transplantation (odds ratio [OR], 1.22; p = .63). Advanced disease status at the time of allo-HSCT approval was the only predictor of failure to reach transplantation in the multivariate analysis (OR, 4.78; p = .001). Disease progression was the most common cause of failure to reach allo-HSCT (66.7%) in both the MRD and URD groups. With a median follow-up from transplantation of 14.5 (interquartile range, 5 to 29) months, the 2-year estimate of overall survival (OS) from allo-HSCT was 46% in the MRD group and 57% in the URD group (p = .54). There were no differences in OS according to donor type allocation in the multivariate analysis (hazard ratio, 1.01; p = .83). When including patients from the time of transplant approval, 2-year OS was 39% in the MRD group ver- sus 42% in the URD group. Our study suggests that allocation of AML patients to URDs may result in comparable clinical outcomes to MRD assignment without a significant increase in the risk of failure to reach transplant. © 2018 American Society for Blood and Marrow Transplantation. INTRODUCTION arrival of emerging targeted therapies to the field, allo-HSCT Acute myeloid leukemia (AML) represents the most com- remains the first-choice consolidation strategy for fit patients mon indication for allogeneic hematopoietic stem cell trans- in first remission who are at a high risk of relapse [2,3,4]. More- plantation (allo-HSCT) worldwide [1]. Despite major over, allo-HSCT is the only therapeutic option resulting in sig- progresses in our understanding of AML biology and the recent nificant long-term survival rates in patients with refractory or relapsed disease [5]. Matched related donors (MRDs) offer the best clinical outcomes and are thus the preferred graft source Financial disclosure: See Acknowledgments on page 189. * Correspondence and reprint requests: Jose  Antonio Pe rez-Simo n, MD, PhD, in this setting, but about two thirds of the patients who require Department of Hematology, Hospital Universitario Virgen del Rocío, Instituto de a transplant will not have such a donor. Together with advan- Biomedicina de Sevilla, Universidad de Sevilla (IBIS/CSIC/CIBERONC), Seville, Spain, ces in conditioning regimens, graft-versus-host disease Av. Manuel Siurot, s/n, 41013 Seville, Spain. (GVHD) prophylaxis, and supportive care, the growing use of E-mail address: josea.perez.simon.sspa@juntadeandalucia.es rez-Simo (J.A. Pe  n). alternative donors has extended allo-HSCT eligibility in this y These authors contributed equally to this work. population of AML patients without an available MRD. While https://doi.org/10.1016/j.bbmt.2018.08.019 1083-8791/© 2018 American Society for Blood and Marrow Transplantation. 184 E. Rodríguez-Arbolí et al. / Biol Blood Marrow Transplant 25 (2019) 183190 this expansion of the donor pool has allowed that nearly all “transplantation approval at diagnosis”). Patients with advanced disease patients will currently find a potential donor, optimal alterna- were evaluated at the discretion of the referring physician, irrespective of dis- ease status and planned treatment before allo-HSCT at the time of referral. tive donor selection is still controversial. Those patients who did not fulfill standard criteria to undergo allo-HSCT A number of mainly retrospective studies have reported based on risk stratification according to international [2] and PETHEMA outcomes after unrelated donor (URD) [6,7,8,9,10,11], haploi- Group recommendations, as well as those considered unfit for allo-HSCT after dentical [12,13], or umbilical cord blood (UCB) [14,15] trans- evaluation by the transplantation board, were excluded from the analysis. There were no patients denied to proceed to allo-HSCT approval due to medi- plantation approaching those of MRD in adult patients with cal insurance restrictions. Patients with acute promyelocytic leukemia were AML, although with conflicting results with regard to the rela- excluded from the study. All patients consented to the use of their clinical tive virtues and disadvantages of each alternative donor data and the study was approved by the Hospital Universitario Virgen del source. Recent data from the European Society for Blood and Rocío Institutional Review Board. Data were retrospectively collected from Marrow Transplantation showed a survival benefit for MRD, electronic records. fully matched URD or haploidentical allo-HSCT over UCB or Donor Selection Algorithm partially mismatched URD allo-HSCT [16]. Prior large observa- HLA typing of first-degree relatives was generally performed early after tional studies have, on the other hand, reported comparable the diagnosis of AML. Patients were assigned to receive an allo-HSCT from an overall survival (OS) after MRD, 8/8 HLA-matched URD MRD if available. Nonsibling matched family donors were considered on an individual basis. Search for URD was initiated in the remaining cases. When (MURD), and 7/8 mismatched URD (MMURD) allo-HSCT, albeit not done before, high-resolution HLA-A, HLA-B, HLA-C, and DRB1 typing was at the expense of higher GVHD incidences in the latter 2 performed immediately after the confirmation of MRD unavailability. The pri- groups [9,11]. Single-center and registry data also suggest that mary donor choice was 8/8 allele matched URD for patients without an MRD, haploidentical transplantation with post-transplant cyclophos- with 7/8 matched URD being selected as a secondary choice. URD search was not started until transplant approval, once confirmed that an MRD was not phamide prophylaxis may achieve similar OS and lower GVHD available. Patients were assigned to search for a haploidentical donor if a suit- incidence as compared with MURD transplants [12,13]. Like- able URD was not identified after a maximum of 4 months from the date of wise, UCB transplantation appears to provide similar outcomes search initiation. as those reported after MURD transplants in some studies [14,15], as well as possibly lower relapse rates among patients Study Endpoints and Definitions OS was measured separately from the date of approval to undergo allo- with positive pretransplant minimal residual disease [15]. HSCT and from the date of allo-HSCT. Relapse was defined as morphological Overall, significant between-study heterogeneity and the evidence of leukemia recurrence (bone marrow blasts 5%, reappearance of inherent difficulties in isolating donor effects from the impact blasts in the blood, or development of extramedullary disease). Relapse inci- of associated transplantation schemes have so far precluded dence (RI) and leukemia-free survival (LFS) were calculated from the date of transplant approval and from the date of allo-HSCT for patients in complete the definition of a definite hierarchy of alternative donor sour- remission (CR), with death without evidence of relapse or treatment failure ces in patients with AML. (relapse or death from any cause) as competing events, respectively. Patients Notably, available data are restricted to those subsets of with active disease were included in the RI and LFS analyses from the date of patients who have successfully undergone transplantation and achievement of CR. Nonrelapse mortality (NRM) was calculated considering relapse as a competing event. Acute GVHD (aGVHD) was graded using the therefore fail to account for differences in pretransplantation Glucksberg criteria [20]. Chronic GVHD (cGVHD) was evaluated according to outcomes between donor selection strategies. In this regard, the National Institutes of Health consensus criteria [21]. Patients alive at day URD allo-HSCT is associated with limited donor availability for 100 were evaluable for cGVHD/late-onset aGVHD. aGVHD and cGVHD/late- patients who belong to ethnic minority groups and longer onset aGVHD incidences were estimated with death without aGVHD within 100 days or death without cGVHD/late-onset aGVHD as competing risks, waiting times to transplantation, potentially resulting in respectively. Genetic risk was assigned based on cytogenetic data and accord- higher probabilities of failure to reach transplantation and ing to the 2017 European LeukemiaNet risk stratification, with molecular data poorer transplant outcomes [17,18,19]. Conversely, the near (FLT3-ITD allelic ratio and/or NPM1 status) considered when available. universal and expedited access to haploidentical donors or UCB units may favor these graft sources when short-lived Statistical Analysis The primary purpose of our analysis was the comparison of 2 donor allo- remission duration is expected or patient ethnicity predicts a cation strategies from the time of transplant approval. All comparisons low a priori likelihood of finding a suitable URD. However, the between patient groups were thus made based on the initial assignment to precise role of alternative donor selection on transplant rates donor type (MRD versus URD) at the time of inclusion into the allo-HSCT pro- and pretransplant clinical outcomes remains speculative. In gram, irrespective of the final donor source in the patients that reached allo- HSCT. Those patients that underwent haploidentical transplantation after this context, current practice is hence largely based on institu- URD search failure were therefore included in the URD group. Standardized tional experience rather than on evidence-based approaches mean differences (SMDs) were used as an effect size measure of differences to alternative donor selection. in baseline characteristics. Nonparametric cumulative incidences of allo- In the present study we have evaluated the impact of donor HSCT from time of transplant approval were estimated for all patients accepted into the program. A multivariate logistic regression model was fit- selection on both pretransplant and post-transplant clinical ted to assess the impact of donor type assignment on the probabilities of outcomes based on a donor selection strategy consisting in the transplantation and to identify independent predictors of failure to reach preferential allocation of AML patients without MRD to MURD allo-HSCT after transplant approval. All variables with a p value <.2 on uni- or MMURD. variate logistic regression analysis were evaluated in the multivariate analy- sis. An interaction term was added to assess the potential effect modification METHODS of donor assignment on the association between disease status and failure to Data Source and Patient Population reach allo-HSCT. Probabilities of OS and LFS were estimated using the The study population included patients with AML who received approval Kaplan-Meier method. Nonparametric cumulative incidences were estimated for first-time allo-HSCT at our institution between January 2011 and February to assess RI, NRM, aGVHD, and cGVHD. Kaplan-Meier and cumulative inci- 2017. Transplantation board meetings were held on a weekly basis. Trans- dence curves were compared using the log-rank and the Gray tests, respec- plant approval refers to the specific date when the board issued a positive tively. Cox proportional hazards or Fine-Gray multivariate regression models decision regarding the indication of transplantation for any patient and, were constructed to estimate the effect of donor type assignment on clinical accordingly, the acceptance of such a patient to proceed to allo-HSCT in our outcomes. Reduced submodels were compared with reference models includ- HSCT unit. Patients with a frontline allo-HSCT indication were referred to the ing all the theoretically relevant covariates with a p value <.2 on univariate transplantation board for allo-HSCT approval concomitantly with the admin- analysis. The final models were selected based on the parsimonious principle istration of frontline chemotherapy. All these patients were administered at (ie, those with the minimum number of covariates and the lowest estimate least 1 cycle of consolidation chemotherapy before proceeding to transplant, standard error, allowing for a 10% estimate change from the reference and a proportion of them received allo-HSCT approval before the assessment model). Data management and statistical analyses were conducted in Stata of postinduction response (subsequently referred in the text as 13 (StataCorp, College Station, TX) and R version 3.3.1 (www.r-project.org). E. Rodríguez-Arbolí et al. / Biol Blood Marrow Transplant 25 (2019) 183190 185 RESULTS or further line treatment or CR2) failed to reach transplanta- Patient Baseline Characteristics tion, as compared with only 10 of 65 (15.4%) among newly Baseline characteristics of the study population are summa- diagnosed patients or those in CR1 after frontline chemother- rized in Table 1. Between January 2011 and February 2017, 108 apy at the time of acceptance into the program. Age, sex, diag- eligible AML patients were accepted for first allo-HSCT at our nosis (secondary versus de novo AML), and genetic risk center. Fifty-eight (53.7%) patients were initially allocated to category were associated with the probability of transplanta- MRDs, while URD search was initiated in 50 (46.3%) patients. tion. In multivariate analysis, advanced disease status at the Median age at inclusion into the program was 47 (range, 17 to time of transplant approval remained as the only predictor of 68) years. Patients assigned to URDs were more likely to har- failure to reach transplantation (OR, 4.52; 95% CI, 1.79 to bor adverse risk genetics (30.0% versus 19.0%; SMD = .31) and 11.42; p = .001) (Table 2). to have advanced disease status at the time of transplant No interaction was detected between donor selection and approval (48.0% versus 32.6%; SMD = .52). Conversely, therapy- disease status in the final model (p = .89). related AML was more frequent among patients allocated to Time to transplant was significantly longer in patients allo- MRD (13.8% versus 6.0%; SMD = .28). cated to URDs when compared with patients assigned to MRD (median 142 [IQR, 115 to 164] days versus 100 [IQR, 73 to 127] days; p < .001). Reflecting the additional time required to com- Predictors of Failure to Reach Transplantation and plete frontline chemotherapy, median time to transplant from Pretransplantation Clinical Outcomes the date of transplant approval was 149 (IQR, 100 to 189) days Probabilities to reach transplantation were similar irrespec- for patients included at diagnosis, as compared with 115 (IQR, tive of initial donor assignment status (Figure 1A). Forty-three 83 to 141) days for the rest of the patients (p = .03). The proba- of 58 (74.1%) patients in the MRD group and 35 of 50 (70.0%) bility to reach allo-HSCT among subjects included at diagnosis patients in the URD group underwent transplantation (odds (81.8%) was similar to that of patients with early disease who ratio [OR] 1.22; 95% confidence interval [CI], .53 to 2.86; were included after achievement of CR1 (86.0%) (OR, .73; 95% p = .63). Patients with advanced disease at the time of allo- CI, .18 to 2.91; p = .66). Median times to exclusion from trans- HSCT approval were more likely to fail to reach transplantation plant were 63 (IQR, 23 to 124) days and 113 (IQR, 63 to 163) than those with early disease status (OR, 4.78; 95% CI, 1.94 to days for patients failing to reach allo-HSCT in the MRD and 11.78; p = .001) (Figure 1B), regardless of donor assignment URD cohorts, respectively (p = .47) (Figure 2). (for those with MRD: OR, 4.95 [95% CI, 1.42 to 17.31]; for those Disease progression was the most common cause of failure allocated to URD search: OR, 4.65 [95% CI, 1.23 to 17.67]; p for to reach allo-HSCT: 10 of 15 (66.7%) in both the MRD and URD the interaction = .95). Twenty of 43 (46.5%) patients with groups, while intercurrent complications from transplant advanced disease (12 patients in partial remission [PR] or with approval accounted for the remaining cases. By the time of fail- refractory/relapsed disease and 8 patients in CR1 after second ure to reach allo-HSCT, a suitable URD was available for 9 patients and donor search was still ongoing in 4 cases assigned Table 1 to URDs, whereas URD search had been unsuccessful for 2 Baseline Characteristics of Patients Accepted into the HSCT Program patients (both of whom had available haploidentical donors). Additionally, 4 AML relapses that did not ultimately preclude Donor Type transplantation were reported among newly diagnosed Variable MRD (n = 58) URD (n = 50) SMD p patients or those in CR at the time of allo-HSCT approval (2 of Age category .08 .91 43 [4.7%] in the MRD group and 2 of 35 [5.7%] in the URD 17-39 yr 18 (31.0) 17 (34.0) group; p = .54). 40-59 yr 33 (56.9) 28 (56.0) 60 yr 7 (12.1) 5 (10.0) Transplantation Characteristics Sex .20 .30 Male 29 (50.0) 20 (40.0) Seventy-eight eligible patients underwent a first allo- Female 29 (50.0) 30 (60.0) HSCT during the study period. All 43 patients in the MRD Diagnosis .28 .56 group received an HLA-identical sibling transplant. In the De novo 35 (60.3) 33 (66.0) URD group, 19 (54.3%) patients received an 8/8 matched Therapy-related 8 (13.8) 3 (6.0) Pre-existing MDS/MPN or AML 13 (22.4) 13 (26.0) transplant, 14 (40.0%) were assigned to a 7/8 URD, and 2 with MRC (5.7%) patients underwent a haploidentical transplant after Not reported 2 (3.4) 1 (2.0) URD search failure. The graft source was peripheral blood Genetic risk category .31 .46 stem cells in all MRD recipients, while 7 (20%) patients in the Favorable 6 (10.3) 7 (14.0) URD group received a bone marrow transplant. Tacrolimus/ Intermediate 33 (56.9) 23 (46.0) Adverse 11 (19.0) 15 (30.0) methotrexate was the first-choice immunoprophylactic regi- Not reported 8 (13.8) 5 (10.0) men among those patients receiving myeloablative condi- Disease status at transplantation .52 .33 tioning, with antithymocyte globulin if receiving 7/8 approval matched URD. Tacrolimus/sirolimus was the GVHD prophy- Transplantation approval at 11 (19.0) 11 (22.0) diagnosis laxis most frequently used for those patients receiving CR1 (first-line treatment) 28 (48.3) 15 (30.0) reduced-intensity conditioning. Other patient characteris- CR1 (second or later-line 5 (8.6) 7 (14.0) tics are shown in Table 3. treatment) CR2 1 (1.7) 3 (6.0) Transplantation Outcomes Partial response 2 (3.4) 2 (4.0) Refractory disease 6 (10.3) 3 (6.0) OS and LFS Relapsed disease 5 (8.6) 9 (18.0) With a median follow-up from transplantation of 14.5 Values are n (%). (IQR, 5.4 to 28.5) months, the 2-year unadjusted estimate MDS indicates myelodysplastic syndrome; MPN, myeloproliferative neoplasm; of OS was 45.9% (95% CI, 29.8% to 60.6%) in the MRD group MRC, myelodysplasia-related changes. and 57.1% (95% CI, 38.3% to 72.1%) in the URD group 186 E. Rodríguez-Arbolí et al. / Biol Blood Marrow Transplant 25 (2019) 183190 Figure 1. Cumulative incidences of transplantation from the time of approval to proceed to transplant by (A) donor group allocation and (B) disease status at the time of transplantation approval. Early disease refers to CR1 after frontline chemotherapy or transplantation approval at diagnosis, while advanced disease includes CR1 after second or further line treatment, >CR1, PR, or refractory/relapsed (R/R) disease. Table 2 Predictors of Failure to Reach Transplantation Univariate Multivariate Variable OR p 95% CI OR p 95% CI, Age Per 10-yr increase 1.27 .18 .90-1.80 1.01 .46 .98-1.05 Sex Male versus female .61 .26 .26-1.45 - - - Diagnosis Secondary* versus de novo 1.11 .82 .45-2.76 - - - Genetic risk category Adverse versus favorable/intermediate .80 .54 .39-1.63 - - - Disease status at HSCT approval Advanced versus early diseasey 4.78 .001 1.94-11.78 4.52 .001 1.79-11.42 Donor type assignment URD versus MRD 1.22 .63 .53-2.86 .97 .95 .39-2.41 R/R indicates refractory/relapsed. * Therapy-related AML, AML with MRC, or MPN-AML. y Early disease: CR1 after frontline chemotherapy or HSCT approval at diagnosis; advanced disease: CR1 after second or further line treatment, >CR1, PR, or R/R disease. (Figure 3A). Similarly, the 2-year unadjusted estimate of LFS Graft-versus-Host Disease was 46.8% (95% CI, 30.9% to 61.3%) in the MRD group and The unadjusted cumulative incidence of grade II to IV 55.0% (95% CI, 35.8% to 70.7%) in the URD group aGVHD was 35.7% (95% CI, 20.9% to 50.7%) in the MRD group (Figure 3B). There were no differences in OS (hazard ratio and 50.1% (95% CI, 32.1% to 65.8%) in the URD group [HR], 1.01; 95% CI, .54 to 2.16; p = .83) or LFS (HR, .92; 95% (Figure 3E). The risk of grade III or IV aGVHD was similar in CI, .46 to 1.86; p = .83) in the multivariate analysis (Table 4). both groups (18.6% [95% CI, 8.6% to 31.6%] in the MRD group When including patients from the time of transplant and 20.0% [95% CI, 8.7% to 34.7%] in the URD group). Regarding approval, 2-year OS was 38.9% (95% CI, 26.3% to 51.3%) in cGVHD, the 2-year unadjusted cumulative incidence (including the MRD group versus 42.3% (95% CI, 28.2% to 55.7%) in the 3 late-onset aGVHD events) did not differ significantly URD group, while 2-year LFS was 46.1% (95% CI, 30.5% to between the MRD group (47.0% [95% CI, 28.9% to 63.2%] and 60.4%) in the MRD group and 49.0% (95% CI, 32.5% to 63.6%) the URD group (39.7% [95% CI, 21.9% to 56.9%]) (Figure 3F). The in the URD group) (Figures 4A and 4B). risks of grade II to IV aGVHD (sHR, 1.56; 95% CI, .78 to 3.13; p = .21) and cGVHD (sHR, .84; 95% CI, .37 to 1.87; p = .66) were RI and NRM similar in both groups in the multivariate analysis. The 2-year unadjusted estimate of RI was 34.9% (95% CI, Outcomes in the subgroup analysis of recipients of 8/8 or 7/ 20.2% to 50.0%) in the MRD group and 16.3% (95% CI, 5.7% 8 matched URD transplantation are shown in the supplemen- to 31.8%) in the URD group (Figure 3C), while the 2-year tary material (Supplementary Table S1 and Supplementary unadjusted estimate of NRM was 21.5% (95% CI, 10.5% to Figure S1). 35.2%) in the MRD group and 30.8% (95% CI, 15.7% to 47.3%) in the URD group (Figure 3D). In the multivariate analysis DISCUSSION there was a lower risk of relapse in the URD group (subdis- A number of studies have analyzed the advantages and tribution HR [sHR], .26; 95% CI, .07 to .91; p = .04), accom- disadvantages of the different alternative donor sources for panied by a significant increase in the risk of NRM (sHR, adult AML patients without an MRD. While URD, haploi- 2.85; 95% CI, 1.04 to 7.80; p = .04) (Table 4). Differences in dentical and UCB transplantation may lead to similar clini- 2-year RI were abrogated when the analysis was performed cal outcomes, waiting times to transplantation vary from the time of transplant approval (36.0% [95% CI, 21.4% significantly between graft sources. In fact, delays associ- to 50.8%] in the MRD group and 28.0 [95% CI, 14.8% to ated to URD search have been cited among the reasons that 42.8%] in the URD group) (Figure 4C). may justify the preferential use of other alternative donors E. Rodríguez-Arbolí et al. / Biol Blood Marrow Transplant 25 (2019) 183190 187 Table 3 Characteristics of Patients who Underwent Transplantation Donor Type SMD p Variable MRD (n = 43) URD (n = 35) Age category .22 .74 17-39 yr 14 (32.6) 13 (37.1) 40-59 yr 24 (55.8) 20 (57.1) 60 yr 5 (11.6) 2 (5.7) Sex .11 .63 Male 22 (51.2) 16 (45.7) Female 21 (48.8) 19 (54.3) Diagnosis .31 .42 De novo 28 (65.1) 23 (65.7) Therapy-related 6 (14.0) 2 (5.7) Pre-existing MDS/MPN or 9 (20.9) 10 (28.6) AML with MRC Genetic risk category .50 .20 Favorable 4 (9.3) 4 (11.4) Figure 2. Cumulative incidences of failure to reach transplantation from the Intermediate 27 (62.8) 14 (40.0) time of approval to proceed to transplant by donor group allocation. Adverse 7 (16.3) 12 (34.3) Not reported 5 (11.6) 5 (14.3) Disease status at transplantation .64 .06 approval CR1 with negative MiRD 28 (65.1) 20 (57.1) [22]. Yet, potential differences in failure rates to reach CR1 with positive MiRD 7 (16.3) 3 (8.6) transplantation by donor type assignment have not been CR2 2 (4.7) 9 (25.7) taken into consideration in previous analyses. To our best Refractory or relapsed disease 6 (14.0) 3 (8.6) knowledge, this study represents the first attempt to com- HCT comorbidity index .12 .97 Low 9 (20.9) 6 (17.1) pare outcomes from the time of allo-HSCT approval Intermediate 14 (32.6) 12 (34.3) between 2 different donor allocation strategies, MRD and High 17 (39.5) 15 (42.9) URD, in patients with AML. Not reported 3 (7.0) 2 (5.7) In our series, 27% of patients receiving approval to Conditioning regimen .14 .55 undergo allo-HSCT could not reach transplant due to either Myeloablative 25 (58.1) 18 (51.4) Reduced intensity 18 (41.9) 17 (48.6) disease progression or onset of other disease-related com- Donor type - - plications. This finding is in line with the pretransplantation HLA-matched sibling 43 (100) 0 attrition rate reported among high-risk patients selected for Unrelated donor (8/8) 0 19 (54.3) allo-HSCT in the German-Austrian AMLHD98A trial [10]. As Unrelated donor (7/8) 0 14 (40.0) Haploidentical 0 2 (5.7) expected, URD allocation was associated with a 42-day Stem cell source .71 .003 increase in median time to transplant with respect to MRD Peripheral blood 43 (100) 28 (80.0) assignment. Despite this substantial delay, a similar propor- Bone marrow 0 7 (20.0) tion of patients were able to proceed to allo-HSCT after GVHD prophylaxis* - - assignment to either group in our cohort. Advanced disease CNI/methotrexate 32 (74.4) 8 (22.9) CNI/methotrexate/antithymocyte 0 8 (22.9) at the time of the approval of transplant indication emerged globulin as a major predictor of failure to reach transplant in our Tacrolimus/sirolimus 11 (25.6) 16 (45.7) analysis, while no association was found for other baseline Other 0 3 (8.6) characteristics, including type of donor. Recipient/donor CMV match .93 .002 Pos/pos 33 (76.2) 14 (43.6) Disease relapse or refractoriness to treatment was the Pos/neg 6 (14.3) 14 (38.5) primary driver of failure to reach allo-HSCT among the Neg/pos 0 2 (5.1) patients included in the study, accounting for 70% of cases. Neg/neg 0 2 (5.1) URD search failure was a rare event on the other hand, Not reported 4 (9.5) 3 (7.7) being reported in just 2 patients. Disease progression and Recipient/donor sex match .76 .02 Male/male 11 (25.6) 14 (40.0) other complications impeding allo-HSCT often occurred Male/female 11 (25.6) 2 (5.7) early after the approval of the transplant indication. Female/male 12 (27.9) 16 (45.7) Median time to proceed to allo-HSCT was 120 days, while Female/female 9 (20.9) 3 (8.6) median time to exit from the transplantation program was Values are n (%). just 89 days. This pattern of early failures may partly MiRD indicates minimal residual disease; HCT, hematopoietic cell transplanta- explain the lack of impact of donor type allocation on trans- tion; CNI, calcineurin inhibitor; CMV, cytomegalovirus; pos, positive; neg, negative. plantation rates despite the significant delays associated *Group comparison is not shown as GVHD prophylaxis schemes differ by with URD search. protocol according to donor source. There were no detectable differences with regards to transplantation outcomes in 2-year OS or LFS estimates between donor type groups. In contrast, a significantly lower 2-year RI was counterbalanced by an increased 2-year NRM no significant differences in terms of survival or relapse rates in the URD group. These results were upheld in subsequent were observed upon analyzing outcomes from the time of multivariate analyses. Taken together, these data are largely transplant approval. consistent with previous reports showing comparable out- Our analysis has several limitations. First, the sample size comes after MRD or URD transplantation [611]. Moreover, of the study compromised our ability to perform in-depth 188 E. Rodríguez-Arbolí et al. / Biol Blood Marrow Transplant 25 (2019) 183190 Figure 3. Clinical outcomes from the time of transplant by donor type assignment: (A) OS, (B) LFS, (C) RI, (D) NRM, (E) grade II to IV aGVHD, and (F) cGVHD. Log-rank or Gray's p values are shown. subgroup analyses. We were therefore unable to thoroughly result in inferior clinical outcomes in our cohort. Second, the explore whether allocation to URDs may lead to lower trans- homogenous ethnic composition of our patient population plantation rates in specific patient subpopulations. Although of primarily Caucasian descent limits the generalizability of we did not find an interaction between disease status at the our conclusions. Analyses based on the U.S. National Marrow time of allo-HSCT approval and donor type assignment, Donor Program data estimate that the likelihoods of finding these results require replication in larger studies. For similar an 8/8 or a 7/8 URD may be as low as 18% and 71% among reasons, we could not formally compare outcomes between sub-Saharan Africans, as compared with probabilities of 75% MURD and MMURD recipients. A rough comparison between and 97%, respectively, for Caucasians [17]. The potential groups indicates, however, that MMURD allo-HSCT did not impact of ethnicity on transplantation rates and whether E. Rodríguez-Arbolí et al. / Biol Blood Marrow Transplant 25 (2019) 183190 189 Table 4 Multivariate Regression Analysis of Transplantation Clinical Outcomes by Donor Type Outcome (s)HR (95% CI) p OS .83 MRD 1 URD 1.01 (.54-2.16) LFS .83 MRD 1 URD .92 (.46-1.86) RI .04 MRD 1 URD .26 (.07-.91) NRM .04 MRD 1 URD 2.85 (1.04-7.80) Grade II-IV aGVHD .21 MRD 1 URD 1.56 (.78-3.13) cGVHD .66 MRD 1 URD .84 (.37-1.87) OS and LFS were adjusted for disease status at transplantation; RI was adjusted for diagnosis (de novo versus secondary AML) and genetic risk category; NRM was adjusted for disease status at transplantation and CMV mismatch; cGVHD was adjusted for the HCT comorbidity index; no additional covariates were selected to be included in the final model for grade II to IV aGVHD. patients belonging to certain ethnic groups may benefit from early allocation to haploidentical or UCB transplants rather than to URDs search remain key questions that could not be addressed in the present study. Third, waiting times from the time of approval to proceed to transplant are dependent on the efficiency of transplantation programs and differ across health systems and institutions. Thus, the degree to which our data can be extrapolated to other settings needs to be further examined. In this regard, the median time to identify an available URD in the experience of the REDMO (Registro Espan ~ ol de Donantes de Me  dula Osea) was 37 (range, 32 to 51) days during the study period, with a reported 80% probability of finding a suitable donor in the first 3 months since the start of the search. Finally, these results may not apply to diagnoses other than AML. In this respect, considering that most patients with AML who are candidates to receive an allo-HSCT can be identified as early as at the time of diagnosis or at the time of disease response evaluation after first induction therapy, the 42-day median delay to proceed to transplant between patients assigned to URDs as compared with MRD allocation had no impact on Figure 4. Clinical outcomes from the time of transplant approval by donor the probability to reach allo-HSCT. This might not be the type assignment. (A) OS, (B) LFS, and (C) RI. Log-rank or Gray's p values are case in other clinical scenarios where the transplant indica- shown. tion cannot be anticipated. For instance, even within AML, patients may not be considered candidates to proceed to transplantation a priori but later in the disease course due to poor response to treatment. Notwithstanding these limitations, our study indicates patients undergoing allo-HSCT. Further validation of these that allocation of AML patients to URDs may result in com- results and comparison with outcomes obtained with other parable clinical outcomes to MRD assignment without an alternative graft sources are warranted. associated increased risk of failure to proceed to transplant. Overall, these data highlight the need for integration of ACKNOWLEDGMENTS both pretransplantation and post-transplantation data in Financial Disclosure: This research was partially supported future studies evaluating donor selection strategies in by a CIBERONC (CB16/12/00480) grant from the Instituto de 190 E. Rodríguez-Arbolí et al. / Biol Blood Marrow Transplant 25 (2019) 183190 Salud Carlos III (Spanish Ministry of Economy, Industry and 10. Schlenk RF, Do € hner K, Mack S, et al. Prospective evaluation of allogeneic Competitiveness). hematopoietic stem-cell transplantation from matched related and matched unrelated donors in younger adults with high-risk acute myeloid Conflict of Interest Statement: There are no conflicts of inter- leukemia: German-Austrian trial AMLHD98A. J Clin Oncol. 2010;28 est to report. (30):4642–4648. 11. Saber W, Opie S, Rizzo JD, Zhang M, Horowitz MM, Schriber J. Outcomes after matched unrelated donor versus identical sibling hematopoietic cell SUPPLEMENTARY DATA transplantation in adults with acute myelogenous leukemia. Blood. 2012;119(17):3908–3916. Supplementary data associated with this article can be found 12. Di Stasi A, Milton DR, Poon LM, et al. Similar transplantation outcomes for in the online version at doi:10.1016/j.bbmt.2018.08.019. acute myeloid leukemia and myelodysplastic syndrome patients with haploidentical versus 10/10 human leukocyte antigen-matched unrelated and related donors. Biol Blood Marrow Transplant. 2014;20(12):1975– REFERENCES 1981. 1. Passweg JR, Baldomero H, Bader P, et al. Hematopoietic stem cell trans- 13. Ciurea SO, Zhang MJ, Bacigalupo AA, et al. Haploidentical transplant with plantation in Europe 2014: more than 40 000 transplants annually. Bone posttransplant cyclophosphamide vs matched unrelated donor transplant Marrow Transplant. 2016;51(6):786–792. for acute myeloid leukemia. Blood. 2015;126(8):1033–1040. 2. Do€ hner H, Estey E, Grimwade D, et al. Diagnosis and management of AML 14. Atsuta Y, Morishima Y, Suzuki R, et al. Comparison of unrelated cord blood in adults: 2017 ELN recommendations from an international expert panel. transplantation and HLA-mismatched unrelated bone marrow transplan- Blood. 2017;129(4):424–447. tation for adults with leukemia. Biol Blood Marrow Transplant. 2012;18 3. Cornelissen JJ, Blaise D. Hematopoietic stem cell transplantation for (5):780–787. patients with AML in first complete remission. Blood. 2016;127(1):62–70. 15. Milano F, Gooley T, Wood B, et al. Cord-blood transplantation in patients 4. Khwaja A, Bjorkholm M, Gale RE, et al. Acute myeloid leukaemia. Nat Rev with minimal residual disease. N Engl J Med. 2016;375(10):944–953. Dis Prim. 2016;2:16010. 16. Versluis J, Labopin M, Ruggeri A, et al. Alternative donors for allogeneic 5. Duval M, Klein JP, He W, et al. Hematopoietic stem-cell transplantation for hematopoietic stem cell transplantation in poor-risk AML in CR1. Blood acute leukemia in relapse or primary induction failure. J Clin Oncol. Adv. 2017;1(7):477–485. 2010;28(23):3730–3738. 17. Gragert L, Eapen M, Williams E, et al. HLA match likelihoods for hematopoi- 6. Moore J, Nivison-Smith I, Goh K, et al. Equivalent survival for sibling and etic stem-cell grafts in the U.S. registry. N Engl J Med. 2014;371(4):339–348. unrelated donor allogeneic stem cell transplantation for acute myeloge- 18. Frassoni F, Labopin M, Powles R, Mary J. Effect of centre on outcome of nous leukemia. Biol Blood Marrow Transplant. 2007;13(5):601–607. bone-marrow transplantation for acute myeloid leukaemia. Lancet. 7. Schetelig J, Bornha€user M, Schmid C, et al. Matched unrelated or matched 2000;355:1393–1398. sibling donors result in comparable survival after allogeneic stem-cell 19. Oudshoorn M, Cornelissen JJ, Fibbe WE, et al. Problems and possible solu- transplantation in elderly patients with acute myeloid leukemia: a report tions in finding an unrelated bone marrow donor. Results of consecutive from the cooperative German transplant study group. J Clin Oncol. searches for 240 Dutch patients. Bone Marrow Transplant. 1997;20 2008;26(32):5183–5191. (12):1011–1017. 8. Walter RB, Pagel JMGT. Comparison of matched unrelated and matched 20. Glucksberg H, Storb R, Fefer A, et al. Clinical manifestations of graft-ver- related donor myeloablative hematopoietic cell transplantation for adults sus-host disease in human recipients of marrow from HLA-matched sib- with acute myeloid leukemia in first remission. Leukemia. 2010;24 ling donors. Transplantation. 1974;18(4):295–304. (7):1276–1282. 21. Jagasia MH, Greinix HT, Arora M, et al. National Institutes of Health Con- 9. Gupta V, Tallman MS, He W, et al. Comparable survival after HLA-well- sensus Development Project on criteria for clinical trials in chronic graft- matched unrelated or matched sibling donor transplantation for acute versus-host disease. Biol Blood Marrow Transplant. 2015;21(3):389–401. myeloid leukemia in first remission with unfavorable cytogenetics at diag- 22. Shaw BE. Related haploidentical donors are a better choice than matched nosis. Blood. 2010;116(11):1839–1848. unrelated donors: counterpoint. Blood Adv. 2017;1(6):401–406.