Leukemia https://doi.org/10.1038/s41375-020-0917-7 ARTICLE Myelodysplastic syndrome Therapy-related myelodysplastic syndromes deserve specific diagnostic sub-classification and risk-stratification—an approach to classification of patients with t-MDS A. Kuendgen 1 M. Nomdedeu2 H. Tuechler 3 G. Garcia-Manero4 R. S. Komrokji5 M. A. Sekeres6 ● ● ● ● ● ● M. G. Della Porta7 M. Cazzola 8 A. E. DeZern9 G. J. Roboz10 D. P. Steensma 11 A. A. Van de Loosdrecht12 ● ● ● ● ● ● R. F. Schlenk 13,14,15 J. Grau2 X. Calvo 16 S. Blum 17 A. Pereira18 P. Valent19 D. Costa20 A. Giagounidis21 ● ● ● ● ● ● ● ● B. Xicoy22 H. Döhner13 U. Platzbecker23 C. Pedro24 M. Lübbert25 I. Oiartzabal26 M. Díez-Campelo27 ● ● ● ● ● ● ● M. T. Cedena28 S. Machherndl-Spandl29 M. López-Pavía30 C. D. Baldus31 M. Martinez-de-Sola32 R. Stauder33 ● ● ● ● ● ● B. Merchan34 A. List5 C. Ganster 35 T. Schroeder1 M. T. Voso 36 M. Pfeilstöcker37 H. Sill38 B. Hildebrandt39 ● ● ● ● ● ● ● ● J. Esteve40 B. Nomdedeu40 F. Cobo41 R. Haas1 F. Sole42 U. Germing1 P. L. Greenberg43 D. Haase35 ● ● ● ● ● ● ● ● G. Sanz 44 Received: 15 May 2020 / Revised: 23 May 2020 / Accepted: 5 June 2020 1234567890();,: © The Author(s) 2020. This article is published with open access 1234567890();,: Abstract In the current World Health Organization (WHO)-classification, therapy-related myelodysplastic syndromes (t-MDS) are categorized together with therapy-related acute myeloid leukemia (AML) and t-myelodysplastic/myeloproliferative neoplasms into one subgroup independent of morphologic or prognostic features. Analyzing data of 2087 t-MDS patients from different international MDS groups to evaluate classification and prognostication tools we found that applying the WHO classification for p-MDS successfully predicts time to transformation and survival (both p < 0.001). The results regarding carefully reviewed cytogenetic data, classifications, and prognostic scores confirmed that t-MDS are similarly heterogeneous as p-MDS and therefore deserve the same careful differentiation regarding risk. As reference, these results were compared with 4593 primary MDS (p-MDS) patients represented in the International Working Group for Prognosis in MDS database (IWG-PM). Although a less favorable clinical outcome occurred in each t-MDS subset compared with p- MDS subgroups, FAB and WHO-classification, IPSS-R, and WPSS-R separated t-MDS patients into differing risk groups effectively, indicating that all established risk factors for p-MDS maintained relevance in t-MDS, with cytogenetic features having enhanced predictive power. These data strongly argue to classify t-MDS as a separate entity distinct from other WHO-classified t-myeloid neoplasms, which would enhance treatment decisions and facilitate the inclusion of t-MDS patients into clinical studies. Introduction chemotherapy and/or radiation administered for an ante- cedent neoplastic or non-neoplastic disorder. According to Therapy-related myelodysplastic syndromes (t-MDS) are the last World Health Organization (WHO)-classification, defined as MDS occurring as a complication of cytotoxic they belong to the group of therapy-related myeloid neo- plasms (t-MNs) [1, 2]. From the first WHO-classification of myeloid disorders in 2001 [3], the WHO 2008 [4], to the current classification from 2016 [1, 2] patients with t-MDS These authors contributed equally: A. Kuendgen, M. Nomdedeu have been considered as having a generally poor prognosis. Supplementary information The online version of this article (https:// From the beginning, t-MDS patients were placed toge- doi.org/10.1038/s41375-020-0917-7) contains supplementary ther into the large group of therapy-related myeloid neo- material, which is available to authorized users. plasms (t-MNs), independent of blast count and * A. Kuendgen morphologic features such as cellularity or dysplasia. In the kuendgen@med.uni-duesseldorf.de first WHO-classification, t-MNs were sub-classified based Extended author information available on the last page of the article on causative agents: alkylating agent/ radiation-related A. Kuendgen et al. versus topoisomerase-II inhibitor-related [3]. This subdivi- in total. By the contributing centers all patients with a sion was removed from the next WHO-classification [4], diagnosis of MDS according to WHO and/or FAB were because in clinical practice it was difficult to apply, as many included if they had a history of an antecedent disease patients received combined chemotherapeutic regimens leading to chemo- (including alkylating agents, (alkylators, topoisomerase-II-inhibitors, antimetabolites, topoisomerase-II inhibitors, antimetabolites, and antitubulin antitubulin agents) and/or radiation therapy. agents) and/or radiotherapy [1–4, 25, 26]. To test the The 2016 WHO-classification recognizes the fact that applicability and performance of different scoring systems t-MNs can be sub-classified morphologically into t-MDS, as well as classifications strict inclusion/exclusion criteria t-MDS/MPN, and t-AML, but considers it best to distin- were applied. Selection criteria included information about guish them collectively from p-MNs as “a unique clinical primary disease, pretreatment (at least chemotherapy or syndrome” [1, 2], insinuating again that all t-MDS have a radiation), valid data to calculate the IPSS-R, survival data, uniformly poor prognosis. If this were considered true, all AML transformation, valid stratification variables age, sex, treatable patients would need to receive a recommendation and year of diagnosis. Patients were excluded if age <16 for an intensive/disease-modifying treatment approach, years, AML-defining cytogenetic abnormalities [inv(16), including allogeneic transplantation, chemotherapy, or t(15;17), t(8;21)], peripheral blasts >19%, normal kar- hypomethylating agents. yotypes based on <10 metaphases analyzed, proliferative Most prognostic tools for MDS have been developed CMML, AML as a primary diagnosis, survival or time to excluding patients with t-MDS [5–7]. An exception is the AML < 2 months occurred, and if the primary disease was MD Anderson Prognostic Scoring System (MDAPSS) [8]. in progression, to focus on the prognostic impact of the Following the categorization of t-MDS as a non-separate MDS itself. Karyotype was documented within the ISCN subgroup of t-MN, most publications have analyzed a formula [27] after cytogenetic review (performed by DH, conglomerate of t-MDS and t-AML, usually including both FS, JG, and BH). Treatment in MDS-phase, including treated and untreated patients [9–17]. Other publications intensive AML-type chemotherapy and allogeneic stem cell [18–20] have focused on t-AML. In contrast, publications transplantation, was not an exclusion criterion, but the on t-MDS only are rare [21–23]. Not categorizing t-MDS as analyses were repeated with untreated patients only. These a subgroup of MDS limits proper clinical decision-making, results are given in the supplement. interferes with epidemiological/ biological research, and FAB, WHO-2016, IPSS-R, and WPSS-R classifications supports the established practice of excluding t-MDS from were calculated. For the WHO-classification RCUD, clinical studies [24], thereby potentially preventing ther- RARS, and MDS del(5q) were grouped together as in the apeutic improvements. WPSS. To test and quantify differentiating abilities of these In cooperation with centers from the International tools the stratified log-rank test and the stratified Dxy- Working Group for Prognosis in MDS (IWG-PM) as well coefficient [28] were applied. Dxy is a concordance coef- as the U.S. MDS Clinical Research Consortium we have ficient varying between −1 and 1, with 0 representing no compiled a database comprising 2087 patients. After a monotone discriminative ability and 1 perfect monotone detailed review of all ISCN formulas, strict criteria were discrimination of a tool with respect to the time of interest applied regarding the cytogenetic data included in this (transformation free survival, overall survival, time to analysis. Information existed on 1245 patients for overall AML). As the main risk criterion, we used transformation- survival (OS) and AML progression with complete free survival since as a combined endpoint it described the reviewed data to apply IPSS-R and WHO-classification. For clinically relevant disease-related risk more appropriately a comparison to primary MDS (p-MDS), we used data from than overall survival. For completeness, Dxy values for 4593 patients from the IPSS-R database that was limited to overall survival and time to AML transformation were also the institutions contributing data to both projects. included. In detail, time variables were defined as follows: These two very large databases on both t- and p-MDS all start with diagnosis of MDS and are censored, if no enabled us to gather comparative prognostic data in t-MDS event occurred until the end of follow up. Transformation related to p-MDS, test the performance of currently existing free survival ends when transformation or death without tools for classification and prognostication, and finally transformation occur. Time to AML ends with transforma- improve the current stratification systems for use in t-MDS. tion but is censored in case of death without transformation. The results are based on stratified analyses compensating for possible confounding influences of sex, age, center, and Material and methods year of diagnosis. Except for the influence of the primary diagnosis on the outcomes analyzed, we also stratified for Eight different study groups in the US, Germany, Spain, primary diagnosis. Event time was calculated from time of Italy, Austria, and the Netherlands contributed 2087 patients MDS-diagnosis. Time to AML transformation was analyzed Therapy-related myelodysplastic syndromes deserve specific diagnostic sub-classification and. . . by censoring at time of death and in addition by treating Table 1 Patient characteristics of therapy-related (t-MDS) and primary (p-MDS) myelodysplastic syndromes (additional information on death as competing event. Cumulative incidence curves for patient characteristics regarding participating centers and year of death with and without transformation are shown in the diagnosis and a comparison of untreated patients only is shown in supplement. Supplementary Tables 1 and 2). All analyses were conducted with the statistics software Characteristics t-MDS p-MDS p R 3.4.3, including the package “survival” [29, 30]. Two- patients patients sided P values less than 0.05 were considered significant. In (n = 1245) (n = 4593) line with the essentially exploratory nature of the study, no n % n % adjustment for multiple testing was applied. Data from t-MDS patients were compared with a cohort MDS treatment of 4593 untreated p-MDS patients from six different study Treateda 715 63% 0 0% groups within the IPSS-R database, including only patients Untreated 422 37% 4593 100% from centers that also contributed cases to the t-MDS Total with information 1137 91% 4593 100% project. Stem cell transplantation Yes 210 19% 0 0% No 906 81% 4593 100% Results Total 1116 90% 4593 Age (years) <0.001 Patient characteristics ≤60 342 28% 1053 23% >60 to ≤70 395 32% 1267 27% Of the total number of 2087 t-MDS patients, 1245 fulfilled >70 to ≤80 404 32% 1601 35% all relevant selection criteria. Patients with high-risk (22%) >80 104 8% 672 15% and very high-risk (31%) IPSS-R score were more frequent Median 68 70 among t-MDS than in the p-MDS group. Although less Total 1245 100% 4593 100% frequent, a considerable number (8% and 21% patients) had Gender <0.001 an IPSS-R of very low and low-risk, respectively (p < Male 680 55% 2854 62% 0.001). Concordantly, as expected from previous publica- Female 565 45% 1739 38% tions [8], 30% t-MDS patients had a very poor and 15% a Total 1245 100% 4593 100% poor-risk cytogenetic score, and 27% had ≥5 abnormalities. FAB <0.001 Conversely, 2% and 37% patients were diagnosed with a RA 490 40% 1707 37% very good or good-risk karyotype, respectively: 30% pre- RARS 110 9% 839 18% sented with a normal karyotype and 21% with only a single RAEB 455 38% 1217 27% aberration. Patient characteristics as well as a comparison to p-MDS can be found in Table 1 and Supplementary Table 1 RAEB-T 81 7% 328 7% (the same analysis, limited to untreated patients is shown in CMML 44 4% 435 9% Supplementary Table 2). Regarding WHO, the size of the Unclassified 29 2% 67 2% four different risk groups was almost equal: RCUD, RARS, Total 1209 97% 4593 100% and MDS del(5q) 25%, RCMD 32%, RAEB-1 23%, and WHO <0.001 RAEB-2 21% and the distribution of the sub-groups were RCUD 183 17% 639 17% relatively similar to p-MDS (34%, 29%, 17%, and 20%). RARS 66 6% 507 13% However, patients with RAEB-I were more frequent and RCMD 335 31% 1097 29% RARS (6 vs. 13%), MDS del(5q) (1 vs. 4%), as well as RAEB-1 246 22% 627 16% MDS/MPN (CMML, excluding proliferative CMML) (4 vs. RAEB-2 219 20% 748 19% 10%) were less frequent (p < 0.001). The median overall MDS (del5q) 13 1% 143 4% survival was 18 months with a median follow up of MDS-U 29 3% 90 2% 60 months for the t-MDS patients. The group of untreated Total 1091 88% 3851 84% p-MDS patients from the IPSS-R database had a median IPSS-R <0.001 survival of 41 months. Median follow up was 49 months for Very low 105 8% 893 19% this cohort, respectively. Low 260 21% 1644 36% The t-MDS patients’ primary diagnoses were a solid Intermediate 225 18% 882 19% tumor in 54% and a hematological disease in 43%. The High 275 22% 628 14% remaining 13% of patients had received treatment for a benign immunological disease. Treatment included A. Kuendgen et al. Table 1 (continued) Table 1 (continued) Characteristics t-MDS p-MDS p Characteristics t-MDS p-MDS p patients patients patients patients (n = 1245) (n = 4593) (n = 1245) (n = 4593) n % n % n % n % Very high 380 31% 546 12% Radiation alone 243 19% Total 1245 100% 4593 100% Radioactive iodine 11 1% WPSS-R <0.001 Chemotherapy alone 568 45% Very low 83 8% 822 22% Radiation and chemotherapy 431 35% Low 164 15% 1036 28% Total 1243 99,8% Intermediate 228 21% 654 17% Alkylating agents High 399 38% 916 24% Yes 536 65% Very high 188 18% 325 9% No 292 35% Total 1062 85% 3753 82% Total 828 83% Cytogenetic risk categories <0.001 Topoisomerase II inhibitors (IPSS-R)-cipssr Yes 356 43% Very good 24 2% 150 3% No 472 57% Good 460 37% 3261 71% Total 828 83% Intermediate 198 16% 622 14% Antitubulin agents Poor 184 15% 197 4% Yes 340 41% Very poor 379 30% 363 8% No 488 59% Total 1245 100% 4593 100% Total 828 83% Number of cytogenetic <0.001 Antimetabolites aberrations Yes 313 43% 0 377 30% 2753 73% No 515 57% 1 266 21% 706 19% Total 828 83% 2 144 12% 136 4% a 3 77 6% 55 2% HMAs, chemotherapy, and/or allogeneic HSC transplant. 4 50 4% 28 1% ≥5 331 27% 76 2% chemotherapy only in 46%, radiation only in 19%, and both Total 1245 100% 3754 82% in 35% of the patients. Primary diagnosis Hematological 529 43% Application of classification and prognostic scoring Breast 203 16% systems Prostate 125 10% All classification and prognostic scoring systems applied Other solid tumor 342 28% were able to discriminate different risk groups within our Non-malignant disease 43 3% t-MDS cohort, although the performance of the scoring Total 1242 99,8% systems was inferior when compared with p-MDS. The Years from primary diagnosis to MDS prognostic power measured by Dxy for the different clas- Median (years) 6.9 sification and scoring systems in t- and p-MDS is given in ≤2 130 11% Table 2 (the same analysis, limited to untreated patients >2 to ≤4 216 18% only, is shown in Supplementary Table 3). The FAB- >4 to ≤8 347 29% classification (Supplementary Fig. 1a) could only dis- >8 to ≤16 356 29% criminate two different adjacent risk groups (RA vs. RAEB >16 166 14% p < 0.001). There was no significant prognostic difference Total 1215 98% between RA and RARS (p = 0.8) as well as RAEB and Therapy for primary disease RAEB-T (p = 0.291). However, these results are in line All chemotherapy 1000 80% with results obtained in primary MDS and are thus not All radiation including 676 54% t-MDS specific. To examine the WHO-classification, we radioiodine decided to use the categorization used within the WPSS, as there was no expectation regarding risk differences between Therapy-related myelodysplastic syndromes deserve specific diagnostic sub-classification and. . . Table 2 Dxys for the different scoring systems and outcomes patients with other, non-malignant disease appeared to have presented for t- and p-MDS (therapy-related and primary myelodysplastic syndromes): FAB (French-American-British a better outcome (Supplementary Fig. 1d, p = 0.051). classification), WHO (World Health Organization classification), Cumulative incidence of death with and without transfor- IPSS-R (International Prognostic Scoring System-revised), WPSS-R mation as an addition to Fig. 1a–c are shown in Supple- (WHO-based Prognostic Scoring System-revised, cipssr (cytogenetic mentary Fig. 2a–c. component of the IPSS-R), number of aberrations, and primary Since the prognostic values of all classification and diagnosis. scoring systems as well as single variables, except the Score Transformation Overall survival Time to AML cipssr, were inferior compared with p-MDS (Dxy (IPSS-R) free survival in t-MDS 0.38 for OS and 0.35 for time to AML), we t-MDS p-MDS t-MDS p-MDS t-MDS p-MDS analyzed if this was influenced by the fact that our cohort FAB 0.19 0.30 0.17 0.28 0.24 0.42 was a mixture of treated and untreated patients. This hypothesis could be verified since all scoring systems per- WHO 0.24 0.29 0.19 0.26 0.41 0.44 formed much better in the subsample of untreated t-MDS IPSS-R 0.37 0.41 0.38 0.40 0.36 0.53 patients (see Fig. 2a–c and Supplementary Fig. 3a–c). In WPSS-R 0.35 0.38 0.33 0.36 0.40 0.51 this subgroup the Dxy for the IPSS-R was not inferior (0.45 cipssr 0.30 0.23 0.32 0.23 0.23 0.28 for OS and 0.48 for time to AML) when compared with the Number of 0.29 0.13 0.32 0.13 0.22 0.14 aberrations p-MDS cohort (0.40 for OS and 0.53 for time to AML) and for the cipssr 0.30 and 0.54 in t-MDS versus 0.23 and 0.28 Primary 0.05 / 0.05 / 0.03 / diagnosis in p-MDS (Supplementary Table 3). Although patients with t-MDS in general could be clas- sified by the WHO-classification system developed for RCUD, RARS, and MDS (del5q). With this combined low- p-MDS, its performance differed within specific t-MDS risk group a better separation versus the remaining cate- subgroups. The prognostic power was almost comparable to gories could be achieved, although there was no statistical p-MDS in patients with a solid tumor as primary disease as significant difference between MDS with unilineage dys- well as in patients after radiotherapy only. In patients with a plasia (RCUD plus RARS plus MDS(del5q)) and MDS history of a hematologic disease or after chemotherapy, the with multilineage dysplasia (p = 0.389), while RCMD vs. prognostic power was lower. However, even in these sub- RAEB I (p < 0.001) and RAEB I vs RAEB II (p < 0.001) groups, patients with different outcomes could be separated differed significantly (Fig. 1a). The same observation could (see Fig. 3a–c and Supplementary Fig. 4a–c). be made in p-MDS (RCUD, RARS + MDS(del5q) vs. RCMD p = 0.403; RCMD vs. RAEB I p < 0.001; RAEB I vs RAEB II p < 0.001). Discussion The two prognostic scoring systems performed both very well. The IPSS-R could separate five different risk groups In this collaborative IWG-PM project, we were able to for all outcomes tested (Fig. 1b), while the WPSS separated assemble the largest database on t-MDS to date, offering the five risk groups regarding OS, but only four regarding PFS analysis of well-characterized and clinically annotated data and AML transformation (see Supplementary Fig. 1b). The with a long follow up. Our analyses of the data presented influence of IPSSR(A) is shown in Supplementary Fig. 1c. here showed that patients with t-MDS benefited in a major Regarding these outcomes, the difference did not reach way from receiving differentiated classification and prog- statistical significance between low and very low-risk, nostic evaluation, distinct from t-MN. Our scrupulous likely because the very low-risk group was relatively small evaluation of the patients’ earlier treatment for malignant or (p = 0.146). non-malignant disease contributed to our case-finding. When we analyzed the performance of the cytogenetic T-MDS has been relatively neglected regarding classifi- component of the IPSS-R (cipssr), the prognostic power cation and differentiated prognostication. All versions of the was already very high on its own. Only the difference WHO-classification, including the latest from 2016, do not between the very low and low-risk group did not reach classify t-MDS within the group of MDS [1, 2]. Instead, statistical significance (p = 0.210), but this might very t-MDS patients are currently still placed together with t- likely be a matter of the size of the very low-risk group (n AML and t-MPD in a combined category of therapy-related = 24). The performance of the cipssr was at least equal in t- myeloid neoplasms [1, 2]. Moreover, most established MDS compared with p-MDS (Fig. 1c). scoring systems were developed excluding t-MDS patients. Other possible, t-MDS specific influences like the pri- The only t-MDS specific score was published by Quintas- mary diagnosis (Dxy 0.05) or type of prior treatment did Cardama et al. including variables such as performance not influence the different outcomes significantly. Only status and age, which determine patient-related but not A. Kuendgen et al. Fig. 1 Outcome of patients with t-MDS according to different tools for classification and prognosis. a outcome according to WHO 2016, b outcome according to IPSS-R, c Outcome according to cytogenetic IPSS-R risk categories. disease-specific risk features, and performance status might results might be explained partially by the difference in size not be readily available in all databases [23]. compared with our patient cohort. Furthermore, the paper Until now, only one paper analyzed the impact of the includes patients not in remission from their primary dis- WHO-classification for p-MDS in t-MDS [21]. This pub- ease. We excluded such patients to obtain a cleaner esti- lication by Singh et al. [21] included 155 patients with mation of the specific MDS-related risk. t-MDS or t-AML of whom only 81 patients were t-MDS. Earlier publications on t-MDS demonstrated mostly No differences were found in median survival times among high-risk karyotypes in these patients, mainly including patients classified into the different WHO subgroups and chromosome 5 and 7 abnormalities as well as complex Singh and coworkers described a uniformly poor outcome karyotypes, in more than 90% of patients [9, 31, 32]. in t-MDS regardless of morphologic classification. These Although these features are still a hallmark of t-MDS in Therapy-related myelodysplastic syndromes deserve specific diagnostic sub-classification and. . . Fig. 2 Outcome of patients with t-MDS according WHO-classification depending on treatment in MDS phase and comparison to p-MDS. a Outcome according to WHO-classification for treated patients. b Outcome according to WHO-classification for untreated patients. c Outcome according to WHO: p-MDS. general, they do not represent all patients with a history of Even in the most recent WHO-classification [1, 2] the chemo- and/or radiotherapy. Our data demonstrated an prognosis of t-MN is described as being generally poor and unexpectedly high percentage of good-risk and normal that prognosis of these patients is influenced strongly by cytogenetics, which are concordant with other karyotype as well as by the primary disease. Although this more recently published data [13, 19, 23]. Regarding these important information is given in the Revised Fourth Edi- early t-MDS publications, a reporting bias may have con- tion of the WHO series on histological and genetic typing of tributed since in some cases conspicuous cytogenetics human tumors it remains purely descriptive, as it provides might have been required for the question about previous no consequences for the resulting t-MN classification. treatments. Chromosome 5 and/or 7 abnormalities, TP53 mutations, A. Kuendgen et al. Fig. 3 Outcome of patients with t-MDS according WHO-classification depending on the primary disease and comparison to p-MDS. a Outcome according to WHO: primary disease hematologic. b Outcome according to WHO: primary disease solid tumor c Outcome according to WHO: p-MDS. and complex karyotype are indicated by the WHO as hav- involving chromosome band 11q23, the localization of the ing a particularly poor outcome, while patients with MLL/KMT2A-gene [9, 33–37]. These translocations can, in balanced translocations were stated to have a good prog- contrast to the typical good risk “AML defining transloca- nosis, albeit not as good as in p-MN. However, exceptions tions” t(8;21), t(15;17), and inv(16), occur as t-MDS and are to these generalizations are notable. In our database we associated with an extremely poor prognosis as previously observed some patients with isolated del(5q) and a history described [9, 33–37]. These categorization problems relate of exposure to mutagenic agents that have a good prognosis, in part from combining t-MDS and t-AML leading to as is the case in p-MDS [33]. Conversely, considered the use of AML cytogenetic classification systems for the typical for t-MDS is the group with balanced translocations entire group of t-MN. In our study we have demonstrated Therapy-related myelodysplastic syndromes deserve specific diagnostic sub-classification and. . . that the influence of cytogenetics is very high in t-MDS and effects. We find a relevant shift towards better risk the prognostic power of the cipssr is in t-MDS at least as patients in our t-MDS cohort if we look at untreated good as in p-MDS, due to a high proportion of abnormal patients only. It is likely, that there will be a similar shift karyotypes. Without a proper morphological classification in the p-MDS cohort as well, since treated p-MDS patients and separation from t-AML, t-MDS patients are also with- were not included in our reference database (compare held the most powerful tool for prognostic evaluation. Figs. 4 and 5). Results in t- and p-MDS would possibly be In addition, prognostic factors for t-MDS, other than better comparable if the analysis were restricted to cytogenetics, are not recognized by the WHO-classification. untreated patients only. However, regarding the main In practice, this means that a patient with a blast count of issue of differentiating and sub-classifying tMDS exclu- 0% (t-MDS) is in the same risk category as a patient with a sion of treated cases would bias those conclusions, since blast count of 99% (t-AML). Also, with the current WHO- we observed a relevant shift towards older age and lower- classification irrelevant for therapeutic decision-making is risk disease among untreated patients (Fig. 5, Supple- whether a patient has dysplastic (t-MDS) or proliferative mentary Table 2). disease (t-MDS/MPD). These issues have major implica- Further, in addition to patient and method-related factors, tions for patient therapy. Thus, proper morphologic and biological differences between the leukemic stem cells cytogenetic classification as well as separation from t-AML, might contribute as well, since inferior outcome of patients provide t-MDS patients a most powerful method for treat- with t-MDS seems not only limited to survival. Especially ment considerations. in the good and intermediate-risk group, we observe this Our data demonstrated that t-MDS patients can be sub- phenomenon with regard to AML transformation (see divided by diagnostic procedures into groups with clearly Fig. 6a–c and Supplementary Fig 5a-c). It is possible that, varying prognoses. These findings are underlined by results even within each subgroup there is a shift to higher risk from smaller group analyses focusing on the value of cytogenetic or molecular abnormalities. This will be an prognostic scoring systems [22, 23, 38]. It should be con- important comparison for future analyzes. The impact of sidered in particular that sporadic MDS cases might be cytogenetics appears to be even greater in t-MDS compared assigned to the group of t-MN based on their therapeutic with p-MDS. The major reason might be that the proportion history although this could be only coincidental, and they of patients with aberrant karyotype was higher in t-MDS. In actually belong to p-MDS. Observational data on MDS p-MDS, about half of the patients presented with a normal patients cannot prove a causal link between therapy and a karyotype and 71% belonged to the large cipssr good risk later developing MDS. This fact is already implied by the group, whereas in t-MDS it was only 30 and 37%, term “therapy-related”. As we cannot differentiate respectively (see Table 1). between the two situations this should not be taken as an Regarding molecular differences, Singhal et al. found argument for a less scrupulous diagnostic classification of that t-MDS patients with ≥15% ringed sideroblasts had a the concerned patients. Quite the contrary, such low frequency of SF3B1 mutations, but a much higher spontaneously developed MDS within the t-MDS group frequency of TP53 mutations compared with patients with have the greatest disadvantage from not being classified as p-MDS [41]. In line with this observation the same analysis p-MDS. showed a generally much higher frequency of TP53 muta- In addition, we note that the survival of t-MDS patients tions in patients with t-MDS and about half of the patients in different WHO groups was inferior to p-MDS, similar had <5% marrow blasts at the time of t-MDS diagnosis to findings in t- vs p-AML [18, 19]. This might have [41]. Clonal hematopoiesis or germline predisposition can several reasons, including patient-related factors asso- be found in hematopoietic cells of patients who develop ciated with the primary disease (relapse/progression of the t-MN, even before treatment of the prior disease, repre- primary disease, cumulative toxicity of primary and sec- senting a sign of increased chromosomal instability and ondary therapy on bone marrow reserve as well as other high-risk disease [16, 41–44]. However, preceding clonal organ function), as previously discussed hematopoiesis or genetic predisposition can also be found [10, 19, 22, 39, 40]. A second point for discussion might before the development of p-MDS [45–49]. be the retrospective nature of our, as well as previous Pedersen-Bjergaard et al. [50] suggested 25 years ago studies. And third, in the present database, we can only different genetic pathways for t-MDS and t-AML due to compare data from a t-MDS group that is heterogeneous their differing distribution of genetic abnormalities. While regarding treatment to an untreated group of p-MDS this is the case for some t-MNs, in other cases, as has been patients. As we have demonstrated, this influences the evident with more recently developed mutational data, a power of prognostic models and likewise will influence biological continuum exists from clonal hematopoiesis to patient outcomes in general as well, although further t-MDS and “secondary” t-AML with or without a MDS pre- analyses are needed to understand the exact nature of such phase and MDS-related features [16, 20, 41, 42]. These A. Kuendgen et al. Fig. 4 Distribution of risk groups according to different classification and prognostic tools for p- versus t-MDS. a Distribution of IPSS-R subgroups in p- and t-MDS. b Distribution of WHO- subtypes (according to WPSS + RAEB-T) in p- and t-MDS. c Distribution of cytogenetic IPSS-R subgroups in p- and t-MDS. d Distribution of number of aberrations in p- and t-MDS. Fig. 5 Distribution of risk groups according to different classifi- patients. b Comparison of IPSS-R cytogenetic risk-groups between cation and prognostic tools for treated versus untreated patients. treated and untreated patients. c Comparison of WHO subgroups a Comparison of IPSS-R risk-groups between treated and untreated between treated and untreated patients. different mechanisms of leukemogenesis occur in therapy- Knowledge of the diagnostic and prognostic evaluation related as in primary MNs. of t-MN patients according to pathogenesis and disease Our data showed that, transformation-free survival (TFS) characteristics is required for selecting patients who can was poorer in t-MDS subgroups vs those in p-MDS: range be cured or will at least benefit from active treatment 8–22 months, compared with 13–63 months, respectively. The [10, 24] and would facilitate clinical as well as epide- impact of blast count and the performance of morphological miological research. An updated classification should classifications in t-MDS was somewhat less than in p-MDS, separate t-MDS from t-AML and t-MDS/MPN as but remained substantial, in contrast to earlier, smaller pub- although these diseases might share some overlapping lications [21]. Thus, all patients can and should be subdivided features, they exhibit differences in clinical presentation into different diagnostic subcategories and risk groups. and molecular and cytogenetic characteristics [51, 52]. As Therapy-related myelodysplastic syndromes deserve specific diagnostic sub-classification and. . . Fig. 6 Comparison of outcome according to different tools for classification or prognostic evaluation t- versus p-MDS. a Comparison p- and t-MDS according to WHO-classification. b Comparison t- and p-MDS according to IPSS-R risk-categories. c Comparison t- and p-MDS according to cytogenetic IPSS-R categories. for t-MDS, the feasibility of further sub-classification has Based on our findings, we believe this newly established been demonstrated by publications on t-AML as well disease category t-MDS deserves further p-MDS-like sub- [9, 18, 19, 37, 53–56]. Regarding t-MDS/MPN, only 4% categorization. We propose an approach using the WHO of the patients in our database presented with t-CMML classification for p-MDS, but preceding each subgroup with dysplastic, as its frequency is lower in t- compared with a t-; for example, t-MDS-SLD, t-MDS-MLD, etc. We p-MDS [57]. Patients with t-CMML proliferative were suggest restricting the use of prognostic systems to t-MDS excluded from this analysis. Our data indicate for all patients in remission from their primary disease so as not to t-MDS an increased influence of karyotype on prognosis confound the results of risk factors for the MDS. For those compared with other prognostic variables. This finding is not in remission it would be important to incorporate the similar to that demonstrated for p-CMML and t-CMML confounding risk of the primary disease into clinical wherein the frequency of cytogenetic abnormalities is decision-making. much higher in t-CMML, while the frequency of most In summary, our data demonstrated that classification molecular abnormalities seems to be comparable [57]. tools established in p-MDS were effective for stratifying Based on this important publication, Patnaik et al. suggest subgroups in t-MDS and indicated the high prognostic that, due to the unique biological pattern and dismal relevance of cytogenetics in t-MDS. These findings from prognostic impact, t-CMML should be considered as the largest t-MDS database to date should initiate a dis- separate subtype in the classification scheme for both cussion of a potential revision of the WHO-classification CMML and t-MN [57]. and encourage clinicians to use the existing tools for risk A. Kuendgen et al. assessment and treatment decisions for patients with that accounts for events not considered in the original Interna- therapy-related disease. tional Prognostic Scoring System. Cancer. 2008;113:1351–61. 9. Smith SM, Le Beau MM, Huo D, Karrison T, Sobecks RM, Anastasi J, et al. Clinical-cytogenetic associations in 306 patients Acknowledgements We thank the International Working Group for with therapy-related myelodysplasia and myeloid leukemia: the Prognosis in MDS and the U.S. MDS Clinical Research Consortium as University of Chicago series. Blood. 2003;102:43–52. well as all participating centers for the contribution of patient data to 10. Kim SA, Hong J, Park WC, Shin DY, Koh Y, Kim I, et al. Better compile the databases analyzed in this paper. MTV was supported by treatment outcomes in patients with actively treated therapy- AIRC 5 × 1000 call “Metastatic disease: the key unmet need in related myeloid neoplasms harboring a normal karyotype. PLoS oncology” to MYNERVA project (project #21267 http://www. ONE. 2018;13:e0209800. progettoagimm.it). PV was supported by the Austrian Science Fund, 11. Ok CY, Patel KP, Garcia-Manero G, Routbort MJ, Fu B, Tang G, grant F4704-B20. Open access funding provided by Projekt DEAL. et al. Mutational profiling of therapy-related myelodysplastic syndromes and acute myeloid leukemia by next generation Compliance with ethical standards sequencing, a comparison with de novo diseases. Leuk Res. 2015;39:348–54. Conflict of interest The authors declare that they have no conflict of 12. Hartmann L, Nadarajah N, Meggendorfer M, Höllein A, Vetro C, interest. Kern W, et al. Molecular characterization of a second myeloid neoplasm developing after treatment for acute myeloid leukemia. Leukemia. 2019. https://doi.org/10.1038/s41375-019-0633-3. Publisher’s note Springer Nature remains neutral with regard to 13. Kuzmanovic T, Patel BJ, Sanikommu SR, Nagata Y, Awada H, Kerr jurisdictional claims in published maps and institutional affiliations. CM, et al. Genomics of therapy-related myeloid neoplasms. Hae- matologica. 2019. https://doi.org/10.3324/haematol.2019.219352. Open Access This article is licensed under a Creative Commons 14. Guru Murthy GS, Hamadani M, Dhakal B, Hari P, Atallah E. Attribution 4.0 International License, which permits use, sharing, Incidence and survival of therapy related myeloid neoplasm in adaptation, distribution and reproduction in any medium or format, as United States. Leuk Res. 2018;71:95–9. long as you give appropriate credit to the original author(s) and the 15. Shih AH, Chung SS, Dolezal EK, Zhang SJ, Abdel-Wahab OI, source, provide a link to the Creative Commons license, and indicate if Park CY, et al. Mutational analysis of therapy-related myelodys- changes were made. The images or other third party material in this plastic syndromes and acute myelogenous leukemia. Haematolo- article are included in the article’s Creative Commons license, unless gica. 2013;98:908–12. indicated otherwise in a credit line to the material. If material is not 16. Gillis NK, Ball M, Zhang Q, Ma Z, Zhao Y, Yoder SJ, et al. included in the article’s Creative Commons license and your intended Clonal haemopoiesis and therapy-related myeloid malignancies in use is not permitted by statutory regulation or exceeds the permitted elderly patients: a proof-of-concept, case-control study. Lancet use, you will need to obtain permission directly from the copyright Oncol. 2017;18:112–21. holder. To view a copy of this license, visit http://creativecommons. 17. Schroeder T, Kuendgen A, Kayser S, Kröger N, Braulke F, org/licenses/by/4.0/. Platzbecker U, et al. Therapy-related myeloid neoplasms follow- ing treatment with radioiodine. Haematologica. 2012;97:206–12. References 18. Kern W, Haferlach T, Schnittger S, Hiddemann W, Schoch C. Prognosis in therapy-related acute myeloid leukemia and impact of karyotype. J Clin Oncol. 2004;22:2510–1. 1. Arber DA, Orazi A, Hasserjian R, Thiele J, Borowitz MJ, Le Beau 19. Kayser S, Döhner K, Krauter J, Köhne CH, Horst HA, Held G, MM, et al. The 2016 revision to the World Health Organization et al. The impact of therapy-related acute myeloid leukemia classification of myeloid neoplasms and acute leukemia. Blood. (AML) on outcome in 2853 adult patients with newly diagnosed 2016;127:2391–405. AML. Blood. 2011;117:2137–45. 2. Swerdlow SH, Campo E, Harris NL, Jaffe ES, Pileri SA, Stein H, 20. Lindsley RC, Mar BG, Mazzola E, Grauman PV, Shareef S, Allen et al. (eds). WHO classification of tumours of haematopoietic and SL, et al. Acute myeloid leukemia ontogeny is defined by distinct lymphoid tissues. 4th ed. Lyon, France: International Agency for somatic mutations. Blood. 2015;125:1367–76. Research on Cancer; 2017. 21. Singh ZN, Huo D, Anastasi J, Smith SM, Karrison T, Le Beau 3. Vardiman JW, Harris NL, Brunning RD. The World Health MM, et al. Therapy-related myelodysplastic syndrome: morpho- Organization (WHO) classification of the myeloid neoplasms. logic subclassification may not be clinically relevant. Am J Clin Blood. 2002;100:2292–302. Pathol. 2007;127:197–205. 4. Vardiman JW, Thiele J, Arber DA, Brunning RD, Borowitz MJ, 22. Zeidan AM, Al Ali N, Barnard J, Padron E, Lancet JE, Sekeres Porwit A, et al. The 2008 revision of the World Health Organization MA, et al. Comparison of clinical outcomes and prognostic utility (WHO) classification of myeloid neoplasms and acute leukemia: of risk stratification tools in patients with therapy-related vs de rationale and important changes. Blood. 2009;114:937–51. novo myelodysplastic syndromes: a report on behalf of the MDS 5. Greenberg P, Cox C, LeBeau MM, Fenaux P, Morel P, Sanz G, Clinical Research Consortium. Leukemia. 2017;31:1391–7. et al. International scoring system for evaluating prognosis in 23. Quintás-Cardama A, Daver N, Kim H, Dinardo C, Jabbour E, myelodysplastic syndromes. Blood. 1997;89:2079–88. Kadia T, et al. A prognostic model of therapy-related myelodys- 6. Greenberg PL, Tuechler H, Schanz J, Sanz G, Garcia-Manero G, plastic syndrome for predicting survival and transformation to Sole F, et al. Revised international prognostic scoring system for acute myeloid leukemia. Clin Lymphoma Myeloma Leuk. myelodysplastic syndromes. Blood. 2012;120:2454–65. 2014;14:401–10. 7. Malcovati L, Germing U, Kuendgen A, Della Porta MG, Pascutto 24. Borate U, Norris BA, Statler A, Fu R, Bucy T, Sekeres MA. C, Invernizzi R, et al. Time-dependent prognostic scoring system Representation of therapy-related myelodysplastic syndrome in for predicting survival and leukemic evolution in myelodysplastic clinical trials over the past 20 years. Blood Adv. 2019;3:2738–47. syndromes. J Clin Oncol. 2007;25:3503–10. 25. Bennett JM, Catovsky D, Daniel MT, Flandrin G, Galton DA, 8. Kantarjian H, O’Brien S, Ravandi F, Cortes J, Shan J, Bennett JM, Gralnick HR, et al. Proposals for the classification of the myelo- et al. Proposal for a new risk model in myelodysplastic syndrome dysplastic syndromes. Br J Haematol. 1982;51:189–99. Therapy-related myelodysplastic syndromes deserve specific diagnostic sub-classification and. . . 26. Valent P, Orazi A, Steensma DP, Ebert BL, Haase D, Malcovati syndrome but has a distinctive distribution. Leukemia. 2019;33: L, et al. Proposed minimal diagnostic criteria for myelodysplastic 2842–53. syndromes (MDS) and potential pre-MDS conditions. Oncotarget 42. Katagir S, Makishima H, Azuma K, Nannya Y, Saitoh Y, Yosh- 2017;43:73483–500. izawa S, et al. Predisposed genomic instability in pre-treatment 27. McGowan-Jordan J, Schmid M. ISCN 2016: An international bone marrow evolves to therapy-related myeloid neoplasms in system for human cytogenomic nomenclature. S. Karger, Basel, malignant lymphoma. Haematologica 2019. https://doi.org/10. 2016. 3324/haematol.2019.229856. 28. Harrell FE, Lee KL, Mark DB. Multivariable prognostic models: 43. Gibson CJ, Lindsley RC, Tchekmedyian V, Mar BG, Shi J, Issues in developing models, evaluating assumptions and ade- Jaiswal S, et al. Clonal hematopoiesis associated with adverse quacy, and measuring and reducing errors. Stat Med. 1996;15: outcomes after autologous stem-cell transplantation for lym- 361–87. phoma. J Clin Oncol. 2017;35:1598–605. 29. Core Team, R: a Language and Environment for Statistical 44. Schulz E, Valentin A, Ulz P, Beham-Schmid C, Lind K, Rupp V, Computing, R Foundation for Statistical Computing, Vienna, et al. Germline mutations in the DNA damage response genes Austria, 2018, https://www.R-project.org/. BRCA1, BRCA2, BARD1 and TP53 in patients with therapy 30. Therneau T, A Package for Survival Analysis in S. Version 2.38, related myeloid neoplasms. J Med Genet. 2012;49:422–8. 2015, http://CRAN.R-project.org/package=survival. 45. Furutani E, Shimamura A. Genetic predisposition to MDS: diag- 31. Le Beau MM, Albain KS, Larson RA, Vardiman JW, Davis EM, nosis and management. Hematol Am Soc Hematol Educ Program. Blough RR, et al. Clinical and cytogenetic correlations in 63 2019;2019:110–9. patients with therapy-related myelodysplastic syndromes and 46. Genovese G, Kähler AK, Handsaker RE, Lindberg J, Rose SA, acute nonlymphocytic leukemia: further evidence for character- Bakhoum SF, et al. Clonal hematopoiesis and blood-cancer risk istic abnormalities of chromosomes no. 5 and 7. J Clin Oncol. inferred from blood DNA sequence. N. Engl J Med. 2014;371: 1986;4:325–45. 2477–87. 32. Pedersen-Bjergaard J, Philip P, Larsen SO, Andersson M, Dau- 47. Genovese G, Jaiswal S, Ebert BL, McCarroll SA. Clonal hema- gaard G, Ersbøll J, et al. Therapy-related myelodysplasia and topoiesis and blood-cancer risk. N. Engl J Med. 2015;372:1071–2. acute myeloid leukemia. Cytogenetic characteristics of 115 con- 48. Sperling AS, Gibson CJ, Ebert BL. The genetics of myelodys- secutive cases and risk in seven cohorts of patients treated plastic syndrome: from clonal haematopoiesis to secondary leu- intensively for malignant diseases in the Copenhagen series. kaemia. Nat Rev Cancer. 2017;17:5–19. Leukemia. 1993;7:1975–86. 49. Jaiswal S, Fontanillas P, Flannick J, Manning A, Grauman PV, 33. Kuendgen A, Tuechler H, Nomdedeu M, Haase D, Garcia-Manero Mar BG, et al. Age-related clonal hematopoiesis associated with G, Komrokij RS, et al. Frequency and prognostic significance of adverse outcomes. N Engl J Med. 2014;371:2488–98. cytogenetic abnormalities in 1269 patients with therapy-related 50. Pedersen-Bjergaard J, Pedersen M, Roulston D, Philip P. Different myelodysplastic syndrome - a study of the international working genetic pathways in leukemogenesis for patients presenting with group (IWG-PM) for myelodysplastic syndromes (MDS). Blood. therapy-related myelodysplasia and therapy-related acute myeloid 2016;128:112. leukemia. Blood. 1995;86:3542–52. 34. Pui CH, Relling MV, Rivera GK, Hancock ML, Raimondi SC, 51. Bacher U, Haferlach C, Alpermann T, Schnittger S, Kern W, Heslop HE, et al. Epipodophyllotoxin-related acute myeloid leu- Haferlach T. Patients with therapy-related myelodysplastic syn- kemia: a study of 35 cases. Leukemia. 1995;9:1990–6. dromes and acute myeloid leukemia share genetic features but can 35. Andersen MK, Johansson B, Larsen SO, Pedersen-Bjergaard J. be separated by blast counts and cytogenetic risk profiles into Chromosomal abnormalities in secondary MDS and AML. Rela- prognostically relevant subgroups. Leuk Lymphoma. 2013;54: tionship to drugs and radiation with specific emphasis on the 639–42. balanced rearrangements. Haematologica. 1998;83:483–8. 52. Zhang L, Wang SA. A focused review of hematopoietic neo- 36. Zhang Y, Poetsch M, Weber-Matthiesen K, Rohde K, Winkemann plasms occurring in the therapy-related setting. Int J Clin Exp M, Haferlach T, Gassmann W, et al. Secondary acute leukaemias Pathol. 2014;7:3512–23. with 11q23 rearrangement: clinical, cytogenetic, FISH and FIC- 53. Kayser S, Krzykalla J, Elliott MA, Norsworthy K, Gonzales P, TION studies. Br J Haematol. 1996;92:673–80. Hills RK, et al. Characteristics and outcome of patients with 37. Bloomfield CD, Archer KJ, Mrózek K, et al. 11q23 balanced therapy-related acute promyelocytic leukemia front-line treated chromosome aberrations in treatment-related myelodysplastic with or without arsenic trioxide. Leukemia. 2017;31:2347–54. syndromes and acute leukemia: report from an international 54. Pulsoni A, Pagano L, Lo Coco F, Avvisati G, Mele L, Di Bona E, workshop. Genes Chromosomes Cancer. 2002;33:362–78. et al. Clinicobiological features and outcome of acute promyelo- 38. Ok CY, Hasserjian RP, Fox PS, Stingo F, Zuo Z, Young KH, et al. cytic leukemia occurring as a second tumor: the GIMEMA Application of the international prognostic scoring system-revised experience. Blood. 2002;100:1972–6. in therapy-related myelodysplastic syndromes and oligoblastic 55. Ornstein MC, Mukherjee S, Mohan S, Elson P, Tiu RV, Saun- acute myeloid leukemia. Leukemia. 2014;28:185–9. thararajah Y, et al. Predictive factors for latency period and a 39. Nazha A, Seastone DP, Keng M, Hobson S, Kalaycio M, prognostic model for survival in patients with therapy-related Maciejewski JP, et al. The revised international prognostic scoring acute myeloid leukemia. Am J Hematol. 2014;89:168–73. system (IPSS-R) is not predictive of survival in patients with 56. Aldoss I, Pullarkat V. Therapy-related acute myeloid leukemia secondary myelodysplastic syndromes. Leuk Lymphoma. 2015; with favorable cytogenetics: still favorable? Leuk Res. 2012;36: 56:3437–9. 1547–51. 40. Ganser A, Heuser M. Therapy-related myeloid neoplasms. Curr 57. Patnaik MM, Vallapureddy R, Yalniz FF, Hanson CA, Ketterling Opin Hematol. 2017;24:152–8. RP, Lasho TL, et al. Therapy related-chronic myelomonocytic 41. Singhal D, Wee LYA, Kutyna MM, Chhetri R, Geoghegan J, leukemia (CMML): Molecular, cytogenetic, and clinical Schreiber AW, et al. The mutational burden of therapy-related distinctions from de novo CMML. Am J Hematol. 2018;93: myeloid neoplasms is similar to primary myelodysplastic 65–73. A. Kuendgen et al. Affiliations A. Kuendgen 1 M. Nomdedeu2 H. Tuechler 3 G. Garcia-Manero4 R. S. Komrokji5 M. A. Sekeres6 ● ● ● ● ● ● M. G. Della Porta7 M. Cazzola 8 A. E. DeZern9 G. J. Roboz10 D. P. Steensma 11 A. A. Van de Loosdrecht12 ● ● ● ● ● ● R. F. Schlenk 13,14,15 J. Grau2 X. Calvo 16 S. Blum 17 A. Pereira18 P. Valent19 D. Costa20 A. Giagounidis21 ● ● ● ● ● ● ● ● B. Xicoy22 H. Döhner13 U. Platzbecker23 C. Pedro24 M. Lübbert25 I. Oiartzabal26 M. Díez-Campelo27 ● ● ● ● ● ● ● M. T. Cedena28 S. Machherndl-Spandl29 M. López-Pavía30 C. D. Baldus31 M. Martinez-de-Sola32 R. Stauder33 ● ● ● ● ● ● B. Merchan34 A. List5 C. Ganster 35 T. Schroeder1 M. T. Voso 36 M. Pfeilstöcker37 H. Sill38 B. Hildebrandt39 ● ● ● ● ● ● ● ● J. Esteve40 B. Nomdedeu40 F. Cobo41 R. Haas1 F. Sole42 U. Germing1 P. L. Greenberg43 D. Haase35 ● ● ● ● ● ● ● ● G. Sanz 44 1 19 Department of Hematology, Oncology, and Clinical Immunology, Department of Internal Medicine I, Division of Hematology & University Hospital Duesseldorf, Duesseldorf, Germany Hemostaseology and Ludwig Boltzmann Institute for 2 Hematology and Oncology, Medical University of Vienna, Department of Laboratory Hematology, Institut Català Vienna, Austria d’Oncologia Hospital GermansTrias I Pujol, Badalona, Spain 20 3 Hematopathology Section, Hospital Clínic de Barcelona Boltzmann Institute for Leukemia Research, Hanusch Hospital, IDIBAPS, Barcelona, Spain Vienna, Austria 21 4 Department of Oncology, Hematology and Palliative Care, Department of Leukemia, MD Anderson Cancer Center, Marienhospital Duesseldorf, Duesseldorf, Germany Houston, TX, USA 22 5 Clinical Hematology Department, Institut Català d’Oncologia, Department of Malignant Hematology, H Lee Moffitt Cancer Hospital Germans Trias i Pujol, Badalona, Josep Carreras Center, Tampa, FL, USA Leukemia Research Institute, Universitat Autònoma de 6 Barcelona, Bellaterra, Spain Leukemia Program, Department of Hematology and Medical Oncology, Taussig Cancer Institute, Cleveland Clinic, 23 University Hospital Leipzig, Leipzig, Germany Cleveland, OH, USA 24 7 Clinical Hematology Department, Hospital del Mar, Cancer Center - IRCCS Humanitas Research Hospital & Barcelona, Spain Humanitas University, Rozzano - Milan, Italy 25 8 Department of Hematology, Oncology and Stem Cell Department of Hematology Oncology, IRCCS Policlinico San Transplantation, University Medical Center Freiburg, Faculty of Matteo Foundation, Pavia, Italy Medicine, Freiburg, Germany 9 Sidney Kimmel Comprehensive Cancer Center, Johns Hopkins 26 Clinical Hematology Department, Hospital Universitario Araba, University, Baltimore, MD, USA Vitoria-Gasteiz, Spain 10 Weill Cornell Medicine and The New York Presbyterian 27 Clinical Hematology Department, Hospital Universitario de Hospital, New York, NY, USA Salamanca (HUSA), Salamanca, Spain 11 Dana-Farber Cancer Institute, Boston, MA, USA 28 Clinical Hematology Department, Hospital Universitario 12 de 12 Octubre, Madrid, Spain Amsterdam UMC, Vrije Universiteit Amsterdam, Amsterdam, Netherlands 29 1st. Internal Department – Hematology with stem cell transplants, 13 Hemostaseology and Medical Oncology, Elisabethinen Hospital, Department of Internal Medicine III, University Hospital Ulm, Linz, Austria Ulm, Germany 30 14 Clinical Hematology Department, Hospital General Universitari National Center of Tumor Diseases-Trial Center, National Center de València, Valencia, Spain of Tumor Diseases, German Cancer Research Center, Heidelberg, Germany 31 Department of Hematology and Oncology, University Hospital 15 Schleswig-Holstein, Campus Kiel, Kiel, Germany Department of Internal Medicine V, Heidelberg University Hospital, Heidelberg, Germany 32 Clinical Hematology Department, Hospital Parc Taulí, 16 Sabadell, Spain Hematological Citology Laboratory, Pathology Department, Hospital del Mar, GRETNHE, IMIM Hospital del Mar Research 33 Department of Internal Medicine V (Hematology and Oncology), Institute, Barcelona, Spain Innsbruck Medical University, Innsbruck, Austria 17 Service of Hematology, University Hospital Lausanne, 34 Department of Hematology, University Hospital Vall d´Hebrón, Lausanne, Switzerland Barcelona, Spain 18 Hemotherapy and Hemostasis Department, Hospital Clínic de 35 Department of Hematology and Medical Oncology, University Barcelona IDIBAPS, Barcelona, Spain Medical Center Göttingen, Göttingen, Germany Therapy-related myelodysplastic syndromes deserve specific diagnostic sub-classification and. . . 36 41 Department of Biomedicine and Prevention, Tor Vergata Clinical Hematology Department, Hospital Quirón Teknon, University, Rome, Italy Barcelona, Spain 37 42 Hanusch Krankenhaus Wien, Vienna, Austria MDS Group, Josep Carreras Leukemia Research Institute, 38 Barcelona, Spain Medizinische Universität Graz, Graz, Austria 43 39 Stanford University Cancer Center, Stanford, CA, USA Institute of Human Genetics, University Duesseldorf, 44 Duesseldorf, Germany Clinical Hematology Department, Hospital Universitari I 40 Politècnic la Fe, Valencia, Spain Clinical Hematology Department, Hospital Clínic de Barcelona, IDIBAPS, Barcelona, Spain