Biol Blood Marrow Transplant 26 (2020) 22372244 Biology of Blood and Marrow Transplantation journal homepage: www.bbmt.org Allogeneic  Adult Predicting Survival after Allogeneic Hematopoietic Cell Transplantation in Myelofibrosis: Performance of the Myelofibrosis Transplant Scoring System (MTSS) and Development of a New Prognostic Model Juan-Carlos Herna ndez-Boluda1,*, Arturo Pereira2, Alberto Alvarez-Larran3, Ana-Africa Martín4, Ana Benzaquen , Lourdes Aguirre5, Elvira Mora6, Pedro Gonza 1 lez7, Jorge Mora8, Nieves Dorado9, Antonia Sampol10, Valentín García-Gutie rrez11, Oriana Lo pez-Godino12, María-Laura Fox13, 14 Juan Luis Reguera , Manuel Pe rez-Encinas , María-Jesu 15  s Pascual16, Blanca Xicoy17, Rocío Parody18, Leslie Gonzalez-Pinedo , Ignacio Espan 19 ~ ol , Alejandro Avendan 20 ~ o4, Juan-Gonzalo Correa3, Carlos Vallejo , Manuel Jurado , Irene García-Cadenas , Santiago Osorio9, María-Antonia Dura 5 7 8 n10, Fermín Sanchez-Guijo4, Francisco Cervantes3, Jose -Luis Pin~ ana6 1 Hematology Department, Hospital Clínico Universitario, INCLIVA, University of Valencia, Valencia, Spain 2 Hemotherapy and Hemostasis Department, Hospital Clinic, Barcelona, Spain 3 Hematopoietic Stem Cell Transplantation Unit, Hospital Clinic, IDIBAPS, University of Barcelona, Barcelona Institut Josep Carreras, Barcelona, Spain 4 Hematology Department, Hospital Universitario de Salamanca, IBSAL, Salamanca, Spain 5 Hematology Department, Hospital Universitario de Donostia, San Sebastia n, Spain 6 Hematology Department, Hospital La Fe, IIS La Fe, Valencia, Spain 7 Hematology Department, Hospital Universitario Virgen de las Nieves, Granada, Spain 8 Hematology Department, Hospital de la Santa Creu i Sant Pau, Biomedical Research Institute (IIB Sant-Pau) and Jose Carreras Leukemia Research Institute, Autonomous University of Barcelona (UAB), Barcelona, Spain 9 Hematology Department, Hospital Gregorio Maran ~on, Gregorio Maran~o n Health Research Institute (IiSGM), Madrid, Spain 10 Hematology Department, Hospital Son Espases, Mallorca, Spain 11 Hematology Department, Hospital Ramo n y Cajal, Madrid, Spain 12 Hematology and Medical Oncology Department, Hospital Universitario Morales Meseguer, Centro Regional de Hemodonacio n, Instituto Murciano de Investigacio n Biomedica-Arrixaca, Murcia, Spain 13 Hematology Department, Hospital Universitari Vall d’Hebron, Vall d’Hebron Institute of Oncology (VHIO), Department of Medicine, Universitat Auto noma de Barcelona, Barcelona, Spain 14 Hematology Department, Hospital Virgen del Rocío, Sevilla, Spain 15 Hematology Department, Hospital Clínico Universitario, Santiago de Compostela, Spain 16 Hematology Department, Hospital Regional, Ma laga, Spain 17 Hematology Department, Institut Catala d'Oncologia-Hospital Germans Trias i Pujol, Josep Carreras Leukemia Research Institute, Autonomous University of Barcelona, Badalona, Spain 18 Hematology Department, Institut Catala d’Oncologia, Hospital Duran i Reynals, Hospitalet de Llobregat, Barcelona, Spain 19 Hematology Department, Hospital Dr Negrín, Las Palmas de Gran Canaria, Spain 20 Hematology Department, Hospital Virgen de la Arrixaca, Murcia, Spain Article history: A B S T R A C T Received 9 June 2020 Accurate prognostic tools are crucial to assess the risk/benefit ratio of allogeneic hematopoietic cell transplanta- Accepted 19 July 2020 tion (allo-HCT) in patients with myelofibrosis (MF). We aimed to evaluate the performance of the Myelofibrosis Transplant Scoring System (MTSS) and identify risk factors for survival in a multicenter series of 197 patients with Key Words: MF undergoing allo-HCT. After a median follow-up of 3.1 years, 47% of patients had died, and the estimated 5- Myelofibrosis year survival rate was 51%. Projected 5-year risk of nonrelapse mortality and relapse incidence was 30% and 20%, Risk factors respectively. Factors independently associated with increased mortality were a hematopoietic cell transplanta- Prognostication tion-specific comorbidity index (HCT-CI) 3 and receiving a graft from an HLA-mismatched unrelated donor or Survival cord blood, whereas post-transplant cyclophosphamide (PT-Cy) was associated with improved survival. Donor Transplantation type was the only parameter included in the MTSS model with independent prognostic value for survival. Accord- ing to the MTSS, 3-year survival was 62%, 66%, 37%, and 17% for low-, intermediate-, high-, and very high-risk Financial disclosure: See Acknowledgments on page 2244. *Correspondence and reprint requests: Juan Carlos Herna ndez-Boluda, Hematology Department, Hospital Clínico Universitario, INCLIVA Research Institute, Depart- ment of Medicine, University of Valencia, Avd. Blasco Iban ~ ez 17, 46010, Valencia, Spain. E-mail address: hernandez_jca@gva.es (J.-C. Herna ndez-Boluda). https://doi.org/10.1016/j.bbmt.2020.07.022 1083-8791/© 2020 American Society for Transplantation and Cellular Therapy. Published by Elsevier Inc. 2238 ndez-Boluda et al. / Biol Blood Marrow Transplant 26 (2020) 22372244 J.-C. Herna groups, respectively. By pooling together the low- and intermediate-risk groups, as well as the high- and very high-risk groups, we pinpointed 2 categories: standard risk and high risk (25% of the series). Three-year survival was 62% in standard-risk and 25% in high-risk categories (P < .001). We derived a risk score based on the 3 independent risk factors for survival in our series (donor type, HCT-CI, and PT-Cy). The corresponding 5-year survival for the low-, intermediate-, and high-risk categories was 79%, 55%, and 32%, respectively (P < .001). In conclusion, the MTSS model failed to clearly delineate 4 prognostic groups in our series but may still be useful to identify a subset of patients with poor outcome. We provide a simple prognostic scoring system for risk/benefit considerations before transplantation in patients with MF. © 2020 American Society for Transplantation and Cellular Therapy. Published by Elsevier Inc. INTRODUCTION METHODS Study Population Myelofibrosis (MF) is a chronic myeloproliferative neo- This retrospective study included all adult patients who underwent first plasm with a highly heterogeneous but potentially aggressive allo-HCT for MF between April 2003 and December 2019 in 19 Spanish hospi- clinical course. Median survival is around 6 years [1]. Alloge- tals. Patients who underwent transplantation after leukemic transformation neic hematopoietic cell transplantation (allo-HCT) remains the were excluded. The study was approved by the Ethics Committee of Hospital Clínico Universitario of Valencia and registered by the AEMPS (Agencia Espa- only curative treatment, yet the advanced age of patients with ~ ola de Medicamentos y Productos Sanitarios), with the reference number n MF and significant transplant-related mortality limit the appli- INC-TPH-2019-01. Informed consent for inclusion in the study was obtained cability of this procedure [2]. in accordance with local ethics committee requirements. The indications and optimal timing of allo-HCT in patients with MF are particularly challenging in daily clinical practice. Cur- Study Definitions and Variables rent recommendations indicate that patients with MF with Assessment of hematologic recovery was performed only in patients who survived more than 28 days after graft infusion. Primary graft failure was expected survival of less than 5 years should be considered candi- defined as failing to reach neutrophil count >0.5 £ 109/L within the first dates for transplantation [3,4]. In order to identify such patients, a 28 days after stem cell infusion or documentation of autologous reconstitution number of prognostic models [5-12] are available to accurately by chimerism analysis in the absence of relapse [16]. Cases with relapse within predict disease risk. However, only a minority of high-risk 1 month of graft failure were classified as no graft failure [17]. Acute graft-ver- sus-host disease (aGVHD) was graded according to Glucksberg et al. [18] and patients with MF actually undergo allo-HCT. In a Spanish nation- chronic graft-versus-host disease (cGVHD) according to Shulman et al. [19] cri- wide series of 544 patients with MF aged 70 years followed up teria. Disease progression/relapse was defined as disease recurrence or persis- for a median of 3 years, only 17% of those in intermediate 2 or tence in patients who survived more than 28 days after transplantation [20]. high-risk groups by the International Prognostic Scoring System In patients who died after disease relapse, this was considered the primary cause of failure, regardless of the immediate cause of death [21]. [5] at diagnosis were subsequently allografted [13]. Factors such Variables investigated for prognostic significance were selected on the as patient comorbidities or poor performance status, lack of a suit- basis of their inclusion in the MTSS model and their clinical meaningfulness. able donor, or, particularly, patient or physician perceptions [14] They included year of transplantation (2009 versus later), patient sex, age regarding potential risks of transplantation may have precluded 57 years, KPS <90%, hematopoietic cell transplantation-specific comorbidity recommending curative treatment in most patients. It seems index (HCT-CI) [22], MF subtype (primary versus secondary), prior splenec- tomy, spleen size, platelet count <150 £ 109/L, leukocyte count >25 £ 109/L, therefore essential to develop reliable prognostic tools that help hemoglobin <10 g/dL, circulating blasts 5%, ferritin levels >1000 ng/mL, appraise the risk/benefit ratio of HCT in patients with MF [3]. bone marrow osteosclerosis, Dynamic International Prognostic Scoring System To this end, a clinical-molecular Myelofibrosis Transplant risk category [6], transfusion dependence, driver mutations (CALR/MPL Scoring System (MTSS) integrating patient-, disease-, and mutated versus others), additional somatic mutations (in ASXL1, SRSF2, IDH1, IDH2, or U2AF1 genes), prior ruxolitinib treatment, intensity of conditioning transplant-specific factors has recently been proposed to esti- regimen (myeloablative conditioning versus reduced-intensity conditioning), mate survival of patients with MF undergoing allo-HCT [15]. graft-versus-host disease (GVHD) prophylaxis, donor/patient cytomegalovirus This prognostic scoring system was derived from a series of serostatus, donor/patient sex (female donor to male recipient versus any 205 patients with either primary MF or post-essential throm- other), donor type, donor/recipient HLA match, source of progenitor cells, and CD34+ cell dose. The prognostic score according to the MTSS model was cen- bocythemia or polycythemia vera MF from Hamburg and Paris. trally calculated for each patient as previously described [15]. The MTSS model defines 4 risk categories (low, intermediate, Outcome variables (graft failure, GVHD, disease progression/relapse) were high, and very high) based on 7 adverse prognostic factors considered only after the first transplant. For survival analysis, patients who evaluated at the time of transplantation (patient age received a second allo-HCT were not censored at the time of this transplant. 57 years, Karnofsky performance status [KPS] <90%, leuko- cytes >25 £ 109/L, platelet counts <150 £ 109/L, non-CALR/ Statistical Methods The primary study endpoint was survival. Secondary endpoints included MPLdriver mutation genotype status, ASXL1 mutation, and nonrelapse mortality (NRM), graft failure, disease progression/relapse, aGVHD, HLA-mismatched unrelated donor). In an external validation and cGVHD. NRM was defined as death without evidence of relapse or disease cohort of 156 patients with MF from 2 German centers, 5-year progression. Survival was estimated by the Kaplan-Meier method and com- survival was 83%, 64%, 37%, and 22% for the low-, intermedi- pared by log-rank test. Factors associated with overall survival at a significance level of P  0.1 in univariable analysis were entered in a Cox proportional haz- ate-, high-, and very high-risk categories, respectively. ards model, using stepwise selection procedures to assess the independent Although the MTSS may eventually prove useful in trans- effect of each covariate when controlled for the others. A weighted score was plant decision making, any new prognostic model should be assigned to each significant variable for survival on the basis of the hazard thoroughly tested for external validity and discrimination ratios (HRs) obtained from multivariable analysis. A new prognostic scoring power before widespread use in daily clinical practice. More- system was subsequently developed based on the sum of risk points to dis- criminate patient risk groups with significant differences in survival. over, external testing can help fine-tune the model for Median follow-up was determined using reverse Kaplan-Meier method. enhanced clinical usefulness. Therefore, our primary goal was Risk of NRM, graft failure, disease progression/relapse, aGVHD, and cGVHD to evaluate the performance of the MTSS in a multicenter was evaluated in the context of competing risk and statistically compared by series of 197 transplanted patients with MF. In addition, we estimating the sub-hazard ratio (SHR) [23]. Death and relapse were taken as competing risks for acute and chronic GVHD. Continuous and count data aimed to identify risk factors for survival that could improve were summarized as median and interquartile range. Categorical variables prognostic assessment in this clinical setting. were represented as frequencies and percentages. The 95% confidence ndez-Boluda et al. / Biol Blood Marrow Transplant 26 (2020) 22372244 J.-C. Herna 2239 Table 1 Table 1 (Continued) Baseline Characteristics in 197 Patients with Myelofibrosis Undergoing Alloge- neic Hematopoietic Cell Transplantation Characteristic No. of Value Evaluable Characteristic No. of Value Cases Evaluable PT-Cy 197 55 (28) Cases ATGy 197 46 (23.5) Age, yr* 197 58 (52-62) Splenectomy before transplant 197 13 (7) Male sex 197 116 (59) JAK inhibitors before transplant 197 111 (56) Primary myelofibrosis 197 110 (56) Data are given as frequency (%) unless otherwise specified. Percentages are Palpable spleen 10 cm 153 42 (27.5) calculated over the number of patients who had the data. Spleen size 22 cm by imaging 128 34 (27) DIPSS indicates Dynamic International Prognostic Scoring System; CMV, cyto- Blood levels* megalovirus; PB, peripheral blood; ATG, antihuman T-lymphocyte immunoglobulin. Hemoglobin, g/dL 193 8.9 (7.7-10.5) * Median (interquartile range). Leukocytes, £ 109/L 194 5.9 (3.2-11.2) y One patient received alemtuzumab instead of ATG. Platelets, £ 109/L 194 113 (54-225) Peripheral blasts 171 1 (0-3) Ferritin levels, ng/mL* 164 612 (167-1359) Bone marrow osteosclerosis 139 41 (29.5) interval (CI) was used to inform about the precision of estimates. P values of less than or equal to .05 were considered statistically significant in all tests. RBC transfusion dependence 195 111 (57) All statistical analyses were performed with IBM SPSS 22.0 (SPSS, Inc., Chi- KPS <90% 190 61 (32) cago, IL) and Stata 11 (StataCorp, College Station, TX). HCT-CI 3 183 47 (26) Driver mutation RESULTS JAK2 188 117 (62) Patient Characteristics at Time of Transplant CALR 174 33 (19) Table 1 summarizes the main clinical and transplant- MPL 170 8 (5) related characteristics of the series. Median age was 58 years, Triple negative 170 15 (9) and 59% of patients were male. The majority of patients (56%) Additional somatic mutations had been diagnosed with primary MF, and median time from ASXL1 71 26 (37) MF diagnosis to transplant was 19 months. Most transplants SRSF2 60 6 (10) (87%) were performed from 2010 to 2019. Compared with the IDH1 50 2 (4) original cohort of the MTSS, the present series included a IDH2 49 3 (6) higher proportion of transplants from HLA-matched related U2AF1 49 4 (8) donors (43% versus 26%), haploidentical donors (9% versus 1%), and cord blood (2.5% versus 0%). In contrast, fewer patients DIPSS 179 received a graft from an HLA-mismatched unrelated donor in Low 3 (2) our series (13% versus 26%). A total of 55 patients (28%) Intermediate 1 43 (24) received post-transplant cyclophosphamide (PT-Cy) as GVHD Intermediate 2 95 (53) prophylaxis. Finally, 111 patients (56%) had received JAK High 38 (21) inhibitors (mainly ruxolitinib) before allo-HCT. Time to transplant, mo* 196 19 (9-43.5) Year of transplant 197 Survival, Clinical Evolution, and Causes of Death 2000-2009 25 (13) After a median follow-up of 3.1 years (95% CI, 2.5 to 3.5), 92 2010-2019 172 (87) (47%) patients had died (8 within 28 days after graft infusion). Female donor to male recipient 195 44 (23) Estimated survival rates at 1, 3, and 5 years were 60% (95% CI, CMV serostatus patient/donor 187 53% to 67%), 52% (95% CI, 45% to 59%), and 51% (95% CI, 43% to +/+ 101 (54) +/ 52 (28) /+ 12 (6) / 22 (12) Myeloablative conditioning 197 64 (33) PB as graft source 196 183 (93) CD34+ cell dose, £ 106/kg* 193 5.4 (4.3-6.8) Donor type 197 HLA-matched related 85 (43) HLA-matched unrelated 62 (31.5) Single-allele mismatched related 2 (1) Haploidentical related 18 (9) HLA-mismatched unrelated 25 (13) Cord blood 5 (2.5) GVHD prophylaxis 191 Cyclosporin based 103 (54) Tacrolimus based 76 (40) Rapamycin based 12 (6) Figure 1. Projected survival of 197 patients with myelofibrosis undergoing (continued) allogeneic hematopoietic cell transplantation. 2240 ndez-Boluda et al. / Biol Blood Marrow Transplant 26 (2020) 22372244 J.-C. Herna Table 2 Univariable Analysis of Factors Predicting Mortality in 197 Patients Undergo- ing Allo-HCT. Factor Univariable Analysis HR (95% CI) P Value Age 57 yr 1.27 (0.84-1.94) .26 Male sex 1.13 (0.74-1.72 .57 Myelofibrosis subtype 0.89 (0.59-1.34) .56 Palpable spleen 10 cm 1.55 (0.94-2.54) .083 Spleen size 22 cm by imaging 1.11 (0.61-2.0) .74 Prior splenectomy 0.57 (0.23-1.42) .23 KPS <90% 1.59 (1.03-2.45) .037 HCT-CI 3 1.63 (1.02-2.61) .041 Hemoglobin <10 g/dL 1.42 (0.88-2.27) .15 Leukocyte counts >25 £ 109/L 1.27 (0.62-2.64) .51 Platelet counts <150 £ 109/L 1.18 (0.78-1.80) .44 Peripheral blasts 5% 1.45 (0.86-2.45) .17 Ferritin level >1000 ng/mL 1.48 (0.92-2.37) .10 Bone marrow osteosclerosis 1.0 (0.58-1.73) .99 Transfusion dependency 1.38 (0.90-2.11) .14 CALR or MPL unmutated 1.13 (0.62-2.07) .69 ASXL1 mutated 1.08 (0.51-2.27) .84 SRSF2 mutated 2.40 (0.81-7.12) .12 IDH1 mutated 1.37 (0.18-10.43) .76 IDH2 mutated 1.76 (0.40-7.71) .45 U2AF1 mutated 1.51 (0.35-6.50) .58 Intermediate 2/high-risk DIPSS 1.25 (0.73-2.14) .41 Type of donor HLA-matched related Reference HLA-matched unrelated 1.18 (0.71-1.94) .52 Haploidentical related 0.92 (0.60-1.42) .71 HLA-mismatched unrelated 1.21 (0.99-1.48) .067 Cord blood 1.59 (1.25-2.03) <.001 HLA-mismatched unrelated/cord blood 1.41 (1.10-1.80) .007 Female donor/male recipient 1.25 (0.79-2.0) .34 CMV serostatus patient/donor (+/) 1.16 (0.73-1.86) .52 Myeloablative conditioning 0.85 (0.54-1.33) .47 CD34+ cell dose 5 £ 106/kg 0.74 (0.49-1.12) .15 PT-Cy 0.49 (0.27-0.86) .013 ATG use 1.74 (1.12-2.72) .015 JAK inhibitors before transplant 0.76 (0.50-1.15) .19 Transplant before year 2010 1.70 (1.01-2.87) .046 59%), respectively (Figure 1). The projected risk of NRM at 1, 3, and 5 years was 27%, 29%, and 30%, respectively. During follow-up, 108 (55%) patients were diagnosed with Figure 2. Unadjusted survival curves by risk factor in 197 patients with mye- aGVHD (grade III to IV in 45, 23%) and 78 (40%) with cGVHD lofibrosis who underwent transplantation: HCT-CI 3 (A), HLA-mismatched (extensive in 40, 20%). Cumulative incidence of grade II to IV unrelated donor/cord blood (B), and use of post-transplant cyclophosphamide aGVHD at 180 days was 40%. Cumulative incidence of any (C). grade cGVHD and extensive cGVHD at 2 years was 33% and 24%, respectively. Disease progression or relapse occurred in 35 (18%) Prognostic Factors for Survival patients and graft failure in 10 (5%). Cumulative incidence of Factors associated with survival are summarized in Table 2. relapse/progression at 1, 3, and 5 years was 16%, 19%, and 20%, In univariable analysis, the only 2 parameters included in the respectively. A second transplant was performed in 11 (6%) MTSS model that were significantly associated with survival patients. Supplementary Figure S1 shows the cumulative inci- were KPS <90% and receiving a graft from an HLA-mismatched dence of NRM and relapse/disease progression in the context unrelated donor or cord blood. Age 57 years, platelets of competing risks. <150 £ 109/L, leukocytes >25 £ 109/L, ASXL1 mutations, and Causes of death were GVHD (n = 34, 37%), relapse/progres- non-CALR/MPL driver mutation were not significantly corre- sion (n = 26, 28%), infection (n = 15, 16%), organ failure/toxicity lated with survival. Six additional factors were included in (n = 8, 9%), graft failure (n = 6, 7%), and bleeding (n = 3, 3%). multivariable analysis based on their association with survival ndez-Boluda et al. / Biol Blood Marrow Transplant 26 (2020) 22372244 J.-C. Herna 2241 Figure 3. Cumulative incidence of grade II to IV aGVHD (A), extensive cGVHD (B), nonrelapse mortality (C), and disease relapse/progression with or without PT-Cy for GVHD prevention (D) in 197 patients with myelofibrosis who underwent transplantation. Death and relapse were competing events for aGVHD and cGVHD, whereas disease relapse/progression was competing with death without relapse. (P  .1): transplantation before the year 2010, HCT-CI 3, pal- pable spleen 10 cm below costal margin, ferritin >1000 ng/mL at transplant, use of antihuman T-lymphocyte immunoglobulin, and PT-Cy as GVHD prophylaxis. Factors that remained independently associated with increased mortality after stepwise selection were HCT-CI 3 at time of transplant (HR, 1.68; 95% CI, 1.05 to 2.69; P = .031) and receiving a graft from an HLA-mismatched unrelated donor/ cord blood (HR, 1.49; 95% CI, 1.13 to 1.97; P = .005), whereas PT-Cy for GVHD prevention was associated with improved sur- vival (HR, 0.51; 95% CI, 0.28 to 0.90; P = .022). Figure 2 shows the unadjusted survival curves for the 3 independent prognos- tic factors. Figure 3 depicts the cumulative incidence of grade II to IV aGVHD, extensive cGVHD, disease relapse/progression, and NRM depending on whether or not PT-Cy was used for GVHD prevention. In univariable testing, PT-Cy was signifi- cantly associated with reduced risk of disease relapse/progres- Figure 4. Survival curves of 55 patients with myelofibrosis undergoing allo- sion (SHR, 0.20; 95% CI, 0.06 to 0.64; P = .007), lower incidence HCT with post-transplant cyclophosphamide depending on the donor type: of grade II to IV aGVHD (SHR, 0.55; 95% CI, 0.33 to 0.92; HLA-identical sibling/unrelated donor (n = 31, group A) or haploidentical/HLA- mismatched unrelated donor (n = 24, group B). P = .02), and lower NRM (SHR, 0.46; 95% CI, 0.22 to 0.94; P = .03). With regard to extensive cGVHD, there was a nonsig- nificant trend for reduced incidence in patients who received seen, no significant difference on survival was observed PT-Cy (SHR, 0.46; 95% CI, 0.19 to 1.11; P = .08). Finally, Figure 4 (P = .78). shows the survival curves of the 55 patients undergoing allo- HCT with PT-Cy depending on the donor type: HLA-identical Performance of MTSS Model sibling/unrelated donor (n = 31, group A) or haploidentical/ Among the 110 patients for whom information was suffi- HLA-mismatched unrelated donor (n = 24, group B). As can be cient to calculate MTSS category at transplant, 31 (28%), 51 2242 ndez-Boluda et al. / Biol Blood Marrow Transplant 26 (2020) 22372244 J.-C. Herna Figure 5. Survival curves of the different risk groups by the MTSS model (A) and after pooling together patients assigned to the low- and intermediate-risk groups, as well as those within the high- and very high-risk groups (B). (46%), 13 (12%), and 15 (14%) cases were allocated to the low-, intermediate-, high-, and very high-risk groups, respectively. Overall, the proportion of patients assigned to each risk cate- gory was very similar to the original description of the MTSS. Estimated 3-year survival was 62% (95% CI, 43% to 81%), 66% (95% CI, 52% to 80%), 37% (95% CI, 6% to 68%), and 17% (95% CI, 0% to 37%) for low-, intermediate-, high-, and very high-risk groups, respectively (Figure 5A). As can be seen, the prognostic model was not able to discriminate 4 risk groups in our series. We then pooled together patients assigned to the low- and intermediate-risk groups, as well as those within the high- and very high-risk groups. On this basis, 2 categories could be identified: standard risk (n = 82) and high risk (n = 28, 25% of the series). Three-year overall survival was 62% (95% CI, 49% to 72%) in standard-risk and 25% (95% CI, 9% to 45%) in high-risk categories (P < .001) (Figure 5B). Figure 6. Survival curves of the different risk groups by the new prognostic Development of a New Prognostic Model model. Based on the HRs of prognostic factors, we derived a risk score by adding 1 point to each HCT-CI 3 and transplant from HLA-mismatched unrelated donor/cord blood and subtracting The main independent factors predicting lower survival were 1 point for use of PT-Cy. Risk scores ranged from 1 to 2 and an HCT-CI 3 at the time of transplantation and receiving a were grouped into 3 risk categories based on separation graft from an HLA-mismatched unrelated donor or cord blood, between adjacent prognostic classes with respect to HRs and whereas PT-Cy for GVHD prevention was associated with overall survival. Low-risk category included score 1 (n = 35, improved survival. Of note, donor type was the only parameter 19%), intermediate-risk category included score 0 (n = 100, included in the MTSS model with independent prognostic 55%), and high-risk category included scores 1 and 2 (n = 48, value for survival. KPS <90% was statistically significant in uni- 26%). The corresponding 5-year overall survival for the low-, variable analysis, but its independent prognostic value was not intermediate-, and high-risk groups was 79% (95% CI, 63% to confirmed in the multivariable model. The original 4-group 95%), 55% (95% CI, 45% to 65%), and 32% (95% CI, 17% to 47%), MTSS model did not fit our patients’ survival, but its discrimi- respectively (P < .001) (Figure 6). Cumulative incidence of nation power increased substantially by collapsing the 4 NRM by risk group as defined by the MTSS and the Spanish groups into 2 risk categories. These findings indicate that the risk score is shown in Supplementary Figure S2. MTSS model has potential clinical utility but may require fur- Table 3 summarizes the projected survival of patients ther refinement. Finally, we developed an easily implemented belonging to each risk category in the MTSS and the new risk prognostic scoring system based on 3 independent risk factors. score, as well as the HRs when each category was compared The resulting model divided patients into 3 risk categories with the immediately adjacent, lower risk one. As can be seen, with clearly distinct survival after transplant. the MTSS model failed to show significant differences in sur- There are several potential explanations for the MTSS mod- vival risk categories, whereas the new prognostic model el’s inability to discriminate 4-level risk stratification in our defined 3 clearly separated risk groups. data set, the first being differences in MF populations or in transplant characteristics between both studies and possibly DISCUSSION the smaller size of our series. Hematologic parameters at time In the present study, we evaluated the MTSS scoring system of transplantation, particularly platelet counts at a threshold of and the risk factors associated with survival in a multicenter 150 £ 109/L, can vary depending on the treatment modality series of 197 patients with MF undergoing allo-HCT in Spain. given just before transplantation and may not necessarily ndez-Boluda et al. / Biol Blood Marrow Transplant 26 (2020) 22372244 J.-C. Herna 2243 Table 3 Projected Survival According to Risk Categories by Prognostic Scoring System in 197 Patients with Myelofibrosis Undergoing Transplantation Risk Category No. (%) of Patients Survival, Median Hazard Ratio* (95% CI) P Value (95% CI), yr MTSS Low 31 (28) NR Intermediate 51 (46) 10.25 0.97 (0.45-2.10) .94 High 13 (12) 1.7 1.81 (0.75-4.38) 0.19 Very high 15 (14) 0.5 2.36 (0.90-6.16) 0.079 New prognostic score Low 35 (19) NR Intermediate 100 (55) 10.25 2.57 (1.09-6.04) .031 High 48 (26) 0.8 1.80 (1.13-2.86) .013 Numbers in bold are those with a significant P value on the statistical analysis (P < .05). CI: confidence interval, NR: not reached. * The hazard ratio compares a risk category with the immediate adjacent, lower risk one. reflect disease biology. Moreover, healthy Mediterranean pop- prognostic model distinguished 3 risk groups with clearly dis- ulations already have lower platelet counts than people from tinct transplantation outcomes. Thus, the corresponding 5- northern or central Europe [24]. It is worth noting here that year overall survival for the low-, intermediate-, and high-risk previous studies have also observed the lack of prognostic sig- (26% of the series) groups was 79%, 55%, and 32%, respectively, nificance of leukocytes >25 £ 109/L [25,26], mutational status indicating that transplant results in patients assigned to the of myeloproliferative neoplasm driver genes [27,28], or ASXL1 high-risk category are suboptimal, and this treatment should mutations [27-29] in survival after allo-HCT. be weighted up in each particular case against the predicted In our view, although the MTSS model failed to discrimi- outcome without transplant. nate 4 risk groups in the present series, it could still have value In conclusion, the MTSS model failed to clearly delineate 4 in clinical practice. Indeed, 2 risk categories can be delineated prognostic groups in the present series. However, this prognostic by pooling together patients assigned to the low- and interme- model was still useful to identify a high-risk category of patients diate-risk groups, as well as those within the high- and very (about 25%) with poor outcomes in whom allo-HCT might not be high-risk groups: the “standard-risk” and the “high-risk” cate- the preferred treatment option. The favorable results achieved gories, with respective 3-year survival of 62% and 25%. Conse- with PT-Cy in different transplant modalities merit further investi- quently, transplantation outcome in patients assigned to the gation in prospective clinical trials. Finally, we herein provide a “high-risk” category (25% of the series) is predicted to be poor, simple prognostic scoring system that can help to assess the risk/ and this information will be useful for risk/benefit considera- benefit ratio of transplantation in potential candidates. Neverthe- tions before transplantation. less, additional work is still needed to develop better prognostica- Survival after allo-HCT in our patients could be predicted tion models for transplant decision making in MF. by a prognostic scoring system based on 3 pretransplant fac- tors: donor type, comorbidity index, and use of PT-Cy. It should ACKNOWLEDGMENTS be noted that the impact of patient comorbidities on survival Financial disclosure: The authors have nothing to disclose. was not evaluated in the original MTSS cohort [15] but has Conflict of interest statement: There are no conflicts of inter- proven an important risk factor for mortality and NRM in other est to report. studies including patients with MF [26,30,31]. Notably, patients with significant comorbidities did not have a worse SUPPLEMENTARY MATERIALS KPS at the time of transplantation (KPS <90% in 34% of patients Supplementary material associated with this article can be with HCT-CI 3 versus 32% in those with HCT-CI <3). The found in the online version at doi:10.1016/j.bbmt.2020.07.022. weak correlation between HCT-CI scores and the KPS index has already been described [32], suggesting that both scales REFERENCES should be used to evaluate patient risk before transplantation. 1. Cervantes F, Dupriez B, Passamonti F, et al. Improving survival trends in pri- On the other hand, the beneficial effect on survival of PT-Cy mary myelofibrosis: an international study. J Clin Oncol. 2012;30:2981–2987. 2. Devlin R, Gupta V. Myelofibrosis: to transplant or not to transplant? was remarkable. PT-Cy is currently being used with promising Hematology Am Soc Hematol Educ Program. 2016;2016:543–551. results for GVHD prevention outside the haploidentical donor 3. Kroger NM, Deeg JH, Olavarria E, et al. Indication and management of allo- transplantation setting [33-35]. In line with this trend, PT-Cy geneic stem cell transplantation in primary myelofibrosis: a consensus process by an EBMT/ELN international working group. Leukemia. was used after the year 2010 in all recipients of haploidentical 2015;29:2126–2133. donors (n = 18) but also in patients who underwent transplan- 4. Barbui T, Tefferi A, Vannucchi AM, et al. Philadelphia chromosome-nega- tation from HLA-identical siblings (n = 10) or unrelated donors tive classical myeloproliferative neoplasms: revised management recom- mendations from European LeukemiaNet. Leukemia. 2018;32:1057–1069. (n = 21), as well as from HLA-mismatched unrelated donors 5. Cervantes F, Dupriez B, Pereira A, et al. New prognostic scoring system for pri- (n = 6). Overall, patients receiving PT-Cy did not differ from mary myelofibrosis based on a study of the International Working Group for the remaining ones in their main clinical characteristics (age, Myelofibrosis Research and Treatment. Blood. 2009;113:2895–2901. 6. Passamonti F, Cervantes F, Vannucchi AM, et al. A dynamic prognostic spleen size, KPS, comorbidity index, proportion of high-risk model to predict survival in primary myelofibrosis: a study by the IWG- patients according to Dynamic International Prognostic Scor- MRT (International Working Group for Myeloproliferative Neoplasms ing System or MTSS) but had lower incidence of grade II to IV Research and Treatment). Blood. 2010;115:1703–1708. aGVHD, lower NRM, and less relapse/disease progression. 7. Gangat N, Caramazza D, Vaidya R, et al. DIPSS plus: a refined Dynamic International Prognostic Scoring System for primary myelofibrosis that Moreover, the use of PT-Cy was able to negate the potential incorporates prognostic information from karyotype, platelet count, and adverse effect of HLA mismatching in the present series. Our transfusion status. J Clin Oncol. 2011;29:392–397. 2244 ndez-Boluda et al. / Biol Blood Marrow Transplant 26 (2020) 22372244 J.-C. Herna 8. Guglielmelli P, Lasho TL, Rotunno G, et al. MIPSS70: mutation-enhanced 23. Fine JP, Gray RJ. A proportional hazards model for the subdistribution of a international prognostic score system for transplantation-age patients competing risk. J Am Stat Assoc. 1999:496–509. with primary myelofibrosis. J Clin Oncol. 2018;36:310–318. 24. Behrens WE. Mediterranean macrothrombocytopenia. Blood. 1975;46:199– 9. Tefferi A, Guglielmelli P, Lasho TL, et al. MIPSS70+ Version 2.0: mutation 208. and karyotype-enhanced international prognostic scoring system for pri- 25. Alchalby H, Yunus DR, Zabelina T, et al. Risk models predicting survival mary myelofibrosis. J Clin Oncol. 2018;36:1769–1770. after reduced-intensity transplantation for myelofibrosis. Br J Haematol. 10. Tefferi A, Guglielmelli P, Nicolosi M, et al. GIPSS: genetically inspired prognostic 2012;157:75–85. scoring system for primary myelofibrosis. Leukemia. 2018;32:1631–1642. 26. Scott BL, Gooley TA, Sorror ML, et al. The Dynamic International Prognostic 11. Passamonti F, Giorgino T, Mora B, et al. A clinical-molecular prognostic model Scoring System for myelofibrosis predicts outcomes after hematopoietic to predict survival in patients with post polycythemia vera and post essential cell transplantation. Blood. 2012;119:2657–2664. thrombocythemia myelofibrosis. Leukemia. 2017;31:2726–2731. 27. Ali H, Aldoss I, Yang D, et al. MIPSS70+ v2.0 predicts long-term survival in 12. Grinfeld J, Nangalia J, Baxter EJ, et al. Classification and personalized prognosis myelofibrosis after allogeneic HCT with the Flu/Mel conditioning regimen. in myeloproliferative neoplasms. N Engl J Med. 2018;379:1416–1430. Blood Adv. 2019;3:83–95. 13. Hernandez-Boluda JC, Pereira A, Correa JG, et al. Prognostic risk models for 28. Tamari R, Rapaport F, Zhang N, et al. Impact of high-molecular-risk muta- transplant decision-making in myelofibrosis. Ann Hematol. 2018;97:813–820. tions on transplantation outcomes in patients with myelofibrosis. Biol 14. Palmer J, Scherber R, Girardo M, et al. Patient perspectives regarding allo- Blood Marrow Transplant. 2019;25:1142–1151. geneic bone marrow transplantation in myelofibrosis. Biol Blood Marrow 29. Stevens EA, Jenkins IC, Beppu LW, et al. Targeted sequencing improves DIPSS- Transplant. 2019;25:398–402. plus prognostic scoring in myelofibrosis patients undergoing allogeneic trans- 15. Gagelmann N, Ditschkowski M, Bogdanov R, et al. Comprehensive clinical- plantation. Biol Blood Marrow Transplant. 2020;26:1371–1374. molecular transplant scoring system for myelofibrosis undergoing stem 30. Hernandez-Boluda JC, Pereira A, Kroger N, et al. Determinants of survival cell transplantation. Blood. 2019;133:2233–2242. in myelofibrosis patients undergoing allogeneic hematopoietic cell trans- 16. Raj K, Eikema DJ, McLornan DP, et al. Family mismatched allogeneic stem plantation. Leukemia. 2020. https://doi.org/10.1038/s41375-020-0815-z. cell transplantation for myelofibrosis: report from the Chronic Malignan- Online ahead of print. cies Working Party of European Society for Blood and Marrow Transplan- 31. Samuelson Bannow BT, Salit RB, Storer BE, et al. Hematopoietic cell trans- tation. Biol Blood Marrow Transplant. 2019;25:522–528. plantation for myelofibrosis: the Dynamic International Prognostic Scor- 17. Olsson RF, Logan BR, Chaudhury S, et al. Primary graft failure after myeloa- ing System plus risk predicts post-transplant outcomes. Biol Blood Marrow blative allogeneic hematopoietic cell transplantation for hematologic Transplant. 2018;24:386–392. malignancies. Leukemia. 2015;29:1754–1762. 32. Sorror M, Storer B, Sandmaier BM, et al. Hematopoietic cell transplantation- 18. Glucksberg H, Storb R, Fefer A, et al. Clinical manifestations of graft-ver- comorbidity index and Karnofsky performance status are independent predic- sus-host disease in human recipients of marrow from HL-A-matched sib- tors of morbidity and mortality after allogeneic nonmyeloablative hematopoi- ling donors. Transplantation. 1974;18:295–304. etic cell transplantation. Cancer. 2008;112:1992–2001. 19. Shulman HM, Sullivan KM, Weiden PL, et al. Chronic graft-versus-host 33. Montoro J, Pinana JL, Hernandez-Boluda JC, et al. Uniform graft-versus- syndrome in man: a long-term clinicopathologic study of 20 Seattle host disease prophylaxis with posttransplant cyclophosphamide, siroli- patients. Am J Med. 1980;69:204–217. mus, and mycophenolate mofetil following hematopoietic stem cell trans- 20. Kroger N, Holler E, Kobbe G, et al. Allogeneic stem cell transplantation plantation from haploidentical, matched sibling and unrelated donors. after reduced-intensity conditioning in patients with myelofibrosis: a Bone Marrow Transplant. 2020;55:2147–2159. prospective, multicenter study of the Chronic Leukemia Working 34. Kwon M, Bailen R, Pascual-Cascon MJ, et al. Posttransplant cyclophospha- Party of the European Group for Blood and Marrow Transplantation. mide vs cyclosporin A and methotrexate as GVHD prophylaxis in matched Blood. 2009;114:5264–5270. sibling transplantation. Blood Adv. 2019;3:3351–3359. 21. Copelan E, Casper JT, Carter SL, et al. A scheme for defining cause of death 35. Solomon SR, Sanacore M, Zhang X, et al. Calcineurin inhibitor-free graft- and its application in the T cell depletion trial. Biol Blood Marrow Trans- versus-host disease prophylaxis with post-transplantation cyclophospha- plant. 2007;13:1469–1476. mide and brief-course sirolimus following reduced-intensity peripheral 22. Sorror ML, Maris MB, Storb R, et al. Hematopoietic cell transplantation blood stem cell transplantation. Biol Blood Marrow Transplant. (HCT)-specific comorbidity index: a new tool for risk assessment before 2014;20:1828–1834. allogeneic HCT. Blood. 2005;106:2912–2919.