Pre-transplant molecular international prognostic scoring system score reduction is not associated with improved outcomes in myelodysplastic syndromes Authors: Masataka Sakashita, Junya Kanda, Keiichiro Hattori, Nobuhiro Hiramoto, Nobuhiro Kanemura, Senji Kasahara, Mitsumasa Watanabe, Satoshi Yoshihara, Michiko Ichii, Nobuo Sezaki, Masaaki Tsuji, Takeshi Maeda, Ko Sasaki, Toshiyuki Kitano, Yasuko Miyahara, Tadakazu Kondo, Mamiko Sakata-Yanagimoto, Akifumi Takaori-Kondo, Seishi Ogawa, Yasuhito Nannya Published: 2026 Dear Editor, Myelodysplastic syndromes (MDS) are hematopoietic disorders characterized by clonal hematopoiesis, recurrent genetic abnormalities, myelodysplasia, and a risk of progression to acute myeloid leukemia (AML) [1–3]. Despite the availability of various therapeutic interventions such as chemotherapy and hypomethylating agents (HMAs) to delay disease progression, allogeneic hematopoietic stem cell transplantation (HSCT) remains the only curative therapy [1]. For over a decade, the International Prognostic Scoring System (IPSS) and its revised version (IPSS-R) have served as the standard for risk stratification and clinical decision-making, relying on clinical parameters and cytogenetic abnormalities [45]. The Molecular IPSS (IPSS-M), which incorporates somatic mutations across 31 MDS-related genes, provides superior prognostic discrimination and has been shown to re-stratify nearly 46% of patients compared to the IPSS-R [6]. Whether high-risk MDS patients require cytoreductive therapy before proceeding to HSCT has long been debated [7]. Recent studies suggest that apparent pre-transplant disease modification may not necessarily overcome the adverse impact of baseline disease biology [8–10]. However, the impact of IPSS-M score reduction through bridging therapy (BT) remains largely unexplored. We analyzed 3413 MDS cases from 55 collaborating institutes in Japan, using targeted-capture sequencing in our laboratory between January 2001 and December 2024. Among these cases, patients who underwent HSCT primarily for MDS and had available information on IPSS-M-related variables and treatment status were included. Both pre- and post-BT IPSS-M scores were calculated using bone marrow samples in all patients. The same NGS panel was used for both diagnostic and follow-up sequencing. Among patients with IPSS-M data before HSCT, only those with measurements obtained within 200 days prior to transplantation were included. MDS was diagnosed according to the revised 4th edition of the World Health Organization classification of myeloid neoplasms [11]. Clinical data, next-generation sequencing (NGS) measurement status, and pre-transplant treatment were reviewed. Cases that progressed to AML during bridging therapy were included in the analysis, and IPSS-M scores were calculated using an R-based IPSS-M calculator (see Supplementary Information). Patients were classified into four groups according to the timing of IPSS-M assessment: (a) those who received BT and had IPSS-M scores available both before and after BT (Pre-post); (b) those who received BT and had IPSS-M available only before BT (Pre-only); (c) those who received BT and had IPSS-M available only after BT (Post-only); and (d) those who did not receive BT and had IPSS-M available before HSCT (Direct-transplant). Among patients who received BT, IPSS-M pre-BT was defined as the first available IPSS-M score obtained prior to BT, reflecting the IPSS-M score at diagnosis. IPSS-M post-BT was defined as the last available IPSS-M score obtained after BT and prior to HSCT, representing the IPSS-M score before transplantation. Patient subsets differed according to the timing of IPSS-M assessment: Pre–post, Pre-only, and Direct-transplant groups were included in analyses at diagnosis; Pre–post, Post-only, and Direct-transplant groups in analyses before HSCT; and only the Pre–post group in analyses of IPSS-M score changes. See Supplementary Information for additional methodological details. A schematic overview of the study design is shown in Fig. 1A. A total of 213 patients were included in the analysis (Table 1). Sixty (28%) patients were classified as Pre-post, 71 (34%) as Pre-only, 33 (15%) as Post-only, and 49 (23%) as Direct-transplant. BT consisted of azacitidine (AZA) in 140 patients (66%) and intensive chemotherapy (IC) in 13 patients (6%). A complex karyotype was observed in 68 patients (38%), and TP53 biallelic alterations were identified in 43 patients (24%). Other genetic abnormalities observed in the cohort are shown in Supplementary Fig. S1. The median interval from the last IPSS-M assessment after BT to HSCT was 18 days in the Pre-post group (Table 1). Other patient characteristics are summarized in Table 1.Fig. 1Association between IPSS-M score reduction and post-transplant outcomes.A Schematic overview of the study design. B Post-transplant overall survival (OS) stratified by IPSS-M risk at diagnosis. C Distribution of changes in IPSS-M risk category following BT. D OS after transplantation according to changes in IPSS-M risk category following BT. E Number of patients whose IPSS-M score improved or worsened according to individual genetic and cytogenetic abnormalities. F Distribution of changes in IPSS-M score comparing patients with TP53 biallelic alterations and those without. G Overall survival after transplantation according to changes in IPSS-M risk category following BT, with patients harboring biallelic TP53 alterations treated as a distinct subgroup.Table 1Patient characteristics stratified by IPSS-M measurement pattern.Measurement patternOverallDirect-transplantPost-onlyPre-onlyPre-postValue%Value%Value%Value%Value%p valueNumber of patients (n, %)2131004923331571336028Male sex (n, %)1517137762267496943720.818Age at HSCT (median [IQR])58 [50–64]56 [48–61]57 [47–62]59 [51–64]60 [51 – 66]0.082Diagnosis (n, %)0.002 MDS with dysplasia552623472617241322 MDS with excess blasts133621837278247664168 MDS with ring sideroblasts136510134635 MDS, unclassifiable7300263423 MDS/MPN5236130012Laboratory data at diagnosis** Bone marrow blasts (median, % [IQR])4.6 [1.2 – 8.65]3.4 [0.8–5.4]–5.4 [0.8–9.6]5.8 [2.15–9.05]0.205 Hemoglobin (median, g/dL [IQR])8.2 [6.8 – 9.93]7.8 [6.5–9.4]–8.2 [6.95–10.5]8.2 [7.3 – 9.83]0.156 Platelet count (median, ×10^9/L [IQR])9.05 [4.18 – 15.62]7.0 [3–15.6]–9.6 [4.5 – 16.4]9.8 [5 – 14.28]0.829 White blood cell count (median, /μL [IQR])2525 [1750 – 3803]2580 [1760–4300]–2330 [1600–3535]2765 [1935–4125]0.258 Neutrophil count (median, /μL [IQR])1013 [492 – 1854]1012 [290–2091]–924 [532–1756]1101 [496–1965]0.793Interval from first IPSS-M assessment to HSCT (median, days [IQR])124 [68–179]84 [30–136]62 [22– 115]139 [102–197.5]164 [122–228.75]<0.001Interval from last IPSS-M assessment to HSCT (median, days [IQR])57 [17–132]37 [17–112]26 [13–63]139 [102–197.5]18 [11–41.25]<0.001Genetic features at diagnosis (n, %)** Complex karyotype68381327––263729480.063 TP53 biallelic mutation4324714––182518300.149Bridging therapy (n, %)<0.001 No bridging therapy492349100000000 Azacitidine-based1406600288557805592 Chemotherapy13600266858 Others0000000000 Unknown115003981100Donor source (n, %)0.477 Bone marrow74352245113322311932 Peripheral blood stem cells7133173582423322338 Cord blood6531918133925351830 Unknown3112131100Disease status at HSCT (n, %)0.131 CR78371122144223323050 Non-CR126593673185544622847 Unknown9424134623Best response to bridging therapy (n, %)***- CR4024––92715211627 PR127––2634712 SD6037––92725352643 PD2716––927913915 Unknown2515––412192723Conditioning (n, %)0.544 MAC89422245154524342847 RIC117552653164844623152 Unknown7312263412IPSS-R risk at diagnosis (n, %)**- Very Low0000––0000 Low1810918––5747 Intermediate2715714––152158 High57321429––23322033 Very High78431939––28393152IPSS-M category at diagnosis (n, %)**- Very Low0000––0000 Low2212918––710610 Moderate Low181048––111535 Moderate High2011816––71058 High45251224––17241627 Very High75421633––29413050* Percentages were calculated using the number of patients in each measurement-pattern group as the denominator, unless otherwise indicated.** Patients in the post-only measurement-pattern group were excluded from analyses of variables assessed at diagnosis.*** Percentages for best response to bridging therapy were calculated using the number of patients who received bridging therapy as the denominator.BT bridging therapy, CR complete remission, HSCT hematopoietic stem cell transplantation, IPSS-R Revised International Prognostic Scoring System, MAC myeloablative conditioning, MDS/MPN myelodysplastic/myeloproliferative neoplasm, Non-CR not in complete remission, PD progressive disease, PR partial remission, RIC reduced-intensity conditioning, SD stable disease. Association between IPSS-M score reduction and post-transplant outcomes. A Schematic overview of the study design. B Post-transplant overall survival (OS) stratified by IPSS-M risk at diagnosis. C Distribution of changes in IPSS-M risk category following BT. D OS after transplantation according to changes in IPSS-M risk category following BT. E Number of patients whose IPSS-M score improved or worsened according to individual genetic and cytogenetic abnormalities. F Distribution of changes in IPSS-M score comparing patients with TP53 biallelic alterations and those without. G Overall survival after transplantation according to changes in IPSS-M risk category following BT, with patients harboring biallelic TP53 alterations treated as a distinct subgroup. Table 1Patient characteristics stratified by IPSS-M measurement pattern.Measurement patternOverallDirect-transplantPost-onlyPre-onlyPre-postValue%Value%Value%Value%Value%p valueNumber of patients (n, %)2131004923331571336028Male sex (n, %)1517137762267496943720.818Age at HSCT (median [IQR])58 [50–64]56 [48–61]57 [47–62]59 [51–64]60 [51 – 66]0.082Diagnosis (n, %)0.002 MDS with dysplasia552623472617241322 MDS with excess blasts133621837278247664168 MDS with ring sideroblasts136510134635 MDS, unclassifiable7300263423 MDS/MPN5236130012Laboratory data at diagnosis** Bone marrow blasts (median, % [IQR])4.6 [1.2 – 8.65]3.4 [0.8–5.4]–5.4 [0.8–9.6]5.8 [2.15–9.05]0.205 Hemoglobin (median, g/dL [IQR])8.2 [6.8 – 9.93]7.8 [6.5–9.4]–8.2 [6.95–10.5]8.2 [7.3 – 9.83]0.156 Platelet count (median, ×10^9/L [IQR])9.05 [4.18 – 15.62]7.0 [3–15.6]–9.6 [4.5 – 16.4]9.8 [5 – 14.28]0.829 White blood cell count (median, /μL [IQR])2525 [1750 – 3803]2580 [1760–4300]–2330 [1600–3535]2765 [1935–4125]0.258 Neutrophil count (median, /μL [IQR])1013 [492 – 1854]1012 [290–2091]–924 [532–1756]1101 [496–1965]0.793Interval from first IPSS-M assessment to HSCT (median, days [IQR])124 [68–179]84 [30–136]62 [22– 115]139 [102–197.5]164 [122–228.75]<0.001Interval from last IPSS-M assessment to HSCT (median, days [IQR])57 [17–132]37 [17–112]26 [13–63]139 [102–197.5]18 [11–41.25]<0.001Genetic features at diagnosis (n, %)** Complex karyotype68381327––263729480.063 TP53 biallelic mutation4324714––182518300.149Bridging therapy (n, %)<0.001 No bridging therapy492349100000000 Azacitidine-based1406600288557805592 Chemotherapy13600266858 Others0000000000 Unknown115003981100Donor source (n, %)0.477 Bone marrow74352245113322311932 Peripheral blood stem cells7133173582423322338 Cord blood6531918133925351830 Unknown3112131100Disease status at HSCT (n, %)0.131 CR78371122144223323050 Non-CR126593673185544622847 Unknown9424134623Best response to bridging therapy (n, %)***- CR4024––92715211627 PR127––2634712 SD6037––92725352643 PD2716––927913915 Unknown2515––412192723Conditioning (n, %)0.544 MAC89422245154524342847 RIC117552653164844623152 Unknown7312263412IPSS-R risk at diagnosis (n, %)**- Very Low0000––0000 Low1810918––5747 Intermediate2715714––152158 High57321429––23322033 Very High78431939––28393152IPSS-M category at diagnosis (n, %)**- Very Low0000––0000 Low2212918––710610 Moderate Low181048––111535 Moderate High2011816––71058 High45251224––17241627 Very High75421633––29413050* Percentages were calculated using the number of patients in each measurement-pattern group as the denominator, unless otherwise indicated.** Patients in the post-only measurement-pattern group were excluded from analyses of variables assessed at diagnosis.*** Percentages for best response to bridging therapy were calculated using the number of patients who received bridging therapy as the denominator.BT bridging therapy, CR complete remission, HSCT hematopoietic stem cell transplantation, IPSS-R Revised International Prognostic Scoring System, MAC myeloablative conditioning, MDS/MPN myelodysplastic/myeloproliferative neoplasm, Non-CR not in complete remission, PD progressive disease, PR partial remission, RIC reduced-intensity conditioning, SD stable disease. Patient characteristics stratified by IPSS-M measurement pattern. * Percentages were calculated using the number of patients in each measurement-pattern group as the denominator, unless otherwise indicated. ** Patients in the post-only measurement-pattern group were excluded from analyses of variables assessed at diagnosis. *** Percentages for best response to bridging therapy were calculated using the number of patients who received bridging therapy as the denominator. BT bridging therapy, CR complete remission, HSCT hematopoietic stem cell transplantation, IPSS-R Revised International Prognostic Scoring System, MAC myeloablative conditioning, MDS/MPN myelodysplastic/myeloproliferative neoplasm, Non-CR not in complete remission, PD progressive disease, PR partial remission, RIC reduced-intensity conditioning, SD stable disease. At diagnosis, 42% of patients were classified as Very High risk (75/180), whereas this proportion was 30% before HSCT (43/142) (Supplementary Fig. S2A and B). The original six IPSS-M risk categories at diagnosis significantly stratified post-transplant overall survival (OS) in accordance with the ordered risk categories, whereas post-BT IPSS-M categories did not (log-rank test for trend, p = 0.0014 and 0.38, respectively) (Supplementary Fig. S2C and D). For simplicity, we reclassified patients into three groups: Very High, High/Moderate High, and all other categories as the Lower-risk group. The IPSS-M category at diagnosis significantly stratified post-transplant OS (log-rank test for trend, p = 0.001) (Fig. 1B and Supplementary Fig. S2E). For patients who underwent BT, the median number of treatment cycles was 3 cycles [IQR, 2–4] (AZA, median 3 cycles [IQR, 2–4]; intensive chemotherapy, median 1 cycle [IQR, 1–2.25]). When analyzed according to BT, post-transplant OS was comparable between patients treated with AZA and those who received IC (Supplementary Fig. S3A). Median IPSS-M score at diagnosis was 0.88 in the No-BT group, 1.34 in the AZA-based group, and 0.93 in the IC group. The IPSS-M score was significantly lower in patients who received no BT than in those who received AZA-based therapy (p = 0.04) (Supplementary Fig. S3B), reflecting risk-adapted treatment decisions in clinical practice. Of the 60 patients with Pre-post measurements, 37 patients (62%) showed improvement in their IPSS-M scores (Fig. 1C). Improvement in IPSS-M category was observed in 28 patients (47%), whereas no change and worsening were observed in 20 patients (33%) and 12 patients (20%), respectively. Mean IPSS-M score change was −0.813 [range, −6.15 to 2.00]. Component-level analysis revealed that changes in TP53 status contributed most substantially to improvement in the IPSS-M score, followed by changes in bone marrow blast percentage (Supplementary Fig. S4). No significant differences in OS were observed among the three groups (log-rank test, p = 0.33) (Fig. 1D). The number of steps of IPSS-M category improvement was also not associated with post-transplant outcomes (log-rank test, category improvement ≥1 versus others, p = 0.98; ≥2, p = 0.83; and ≥3, p = 0.84) (Supplementary Fig. S5). Similarly, changes in the IPSS-M score did not stratify OS after HSCT (log-rank test, score change < −1.0 versus others, p = 0.74) (Supplementary Fig. S6). Analysis using IPSS-R showed similar results (Supplementary Fig. S7). Among variables required for IPSS-M calculation, TP53-related alterations and complex karyotype were more frequent among patients who showed IPSS-M score reduction following BT (Fig. 1E). When patients with TP53 biallelic alterations were analyzed as a distinct subgroup, they exhibited significantly greater IPSS-M score reduction than patients without TP53 biallelic alterations (mean change, −1.94 versus −0.35, p = 0.014) (Fig. 1F). Patients with TP53 biallelic alterations showed markedly worse post-transplant outcomes compared with other risk categories (log-rank test, p < 0.0001) (Supplementary Fig. S8A). Among TP53 biallelic cases (n = 18), BT resulted in a substantial shift in IPSS-M categories, with Very High risk decreasing from 83% (15/18) to 39% (7/18). Notably, five patients were reclassified into Low or Very Low after BT (Supplementary Fig. S8B and Supplementary Table S2). Among patients without TP53 biallelic alterations, improvement in IPSS-M category was not associated with a survival benefit (log-rank test, p = 0.89) (Fig. 1G). Similarly, no survival advantage associated with IPSS-M category improvement was observed among patients with TP53 biallelic alterations (log-rank test, p = 0.29) (Fig. 1G). In a univariate analysis, TP53 biallelic status at diagnosis was associated with inferior OS (hazard ratio, 3.75; 95% confidence interval, 1.28–7.70; p < 0.001). In a multivariable Cox model after adjustment for selected covariates, the IPSS-M score change group was not significantly associated with OS (Supplementary Table S3). The distribution of best pre-transplant response differed according to IPSS-M score change, although the difference was not statistically significant (p = 0.153; Supplementary Table S4). Sensitivity analyses using cutoff windows of 200, 100, 60, and 30 days between the last IPSS-M assessment and HSCT yielded consistent results, supporting the robustness of the main findings (Supplementary Table S5 and Supplementary Fig. S9). Additional analyses of relapse incidence, non-relapse mortality, and relapse-free survival are provided in Supplementary Fig. S10. We acknowledge several limitations to this study. First, its retrospective nature introduces potential selection bias; for example, patients in the Direct-transplant group may have had more stable disease or more favorable clinical features. Second, because the study population was derived from multiple centers, transplant indications, bridging strategies, and conditioning regimens were heterogeneous, which may have influenced the survival analyses. Third, although information on the total number of treatment cycles administered before HSCT was available, the exact number of cycles administered before post-BT IPSS-M assessment was not consistently available. Fourth, the limited sample size precluded a detailed sub-analysis of the associations between specific BT strategies and outcomes. Future large-scale prospective analyses are warranted to clarify whether changes in IPSS-M before HSCT represent clinically meaningful modification of transplant risk or merely apparent score improvement that does not translate into better post-transplant outcomes. In conclusion, our findings demonstrate that improvement of IPSS-M following BT was not associated with improved post-transplant outcomes in patients with MDS. These results may suggest the importance of adequate risk stratification at diagnosis, rather than apparent treatment response. Supplementary information Supplementary Information Supplementary Information