Myelofibrosis (MF) is a rare Philadelphia chromosome– negative myeloproliferative neoplasm with an annual incidence of ~ 0.5 per 100,000 and a median age of approximately 65 years at diagnosis [1, 2]. It is characterized by progressive bone marrow fibrosis, extramedullary hematopoiesis with splenomegaly, constitutional symptoms, and a risk of leukemic transformation [1, 3]. The prognosis is poor in comparison with other myeloproliferative neoplasms, with median survival ranging from > 10 years in low-risk patients to 2–3 years in highrisk patients [2, 4]. The pathogenesis of MF involves the clonal proliferation of hematopoietic stem cells, frequently with Janus kinase 2 (JAK2), calreticulin (CALR), or myeloproliferative leukemia protein (MPL) mutations, and an altered bone marrow microenvironment that produces pro-fibrotic cytokines such as transforming growth factor beta 1 (TGF-β1) and bone morphogenetic protein 2 (BMP-2) [5].

Risk stratification of patients with MF is most commonly performed using the Dynamic International Prognostic Scoring System (DIPSS), which incorporates age, hemoglobin level, leukocyte count, peripheral blasts, and constitutional symptoms, with an anemia-weighted double score [4]. High-molecular-risk (HMR) mutations (e.g., ASXL transcriptional regulator 1 [ASXL1], serineand arginine-rich splicing factor 2 [SRSF2], enhancer of zeste 2 polycomb repressive complex 2 subunit [EZH2], and isocitrate dehydrogenase 1 and 2 [IDH1/2]) further worsen the prognosis [6]. Treatment is risk-adapted. Low-risk or asymptomatic patients may be observed, while intermediate-2, high-risk, or symptomatic patients typically receive Janus kinase (JAK) inhibitors such as ruxolitinib [7], fedratinib [8], pacritinib [9], or momelotinib [10]. These agents improve symptoms and splenomegaly but rarely achieve molecular remission or significantly extend survival. Allogeneic hematopoietic stem cell transplantation (allo-HSCT) remains the only potential curative therapy [11].

Recent EBMT/ELN consensus recommendations emphasize allo-HSCT as the only potentially curative option and support its consideration in medically fit patients with Intermediate-2 or High-risk MF (and selected patients showing Intermediate-1 disease with additional adverse features) while integrating comorbidity and molecular risk assessment into decision-making [12]. Transplant-specific prognostic tools such as the Myelofibrosis Transplant Scoring System (MTSS) have been proposed to refine post-transplant risk estimation [13]. Notably, the current consensus suggests that chronological age alone should not constitute an absolute contraindication and that reduced-intensity conditioning (RIC) may expand the feasibility of HSCT in selected older patients [14, 15]. In this study, we retrospectively described the outcomes of 42 patients with myelofibrosis who underwent allo-HSCT at a single center between 2014 and 2023. We focused on the pretransplant DIPSS risk category as the primary stratification variable and evaluated age and other transplant- and disease-related factors in secondary exploratory analyses.

In this single-center retrospective cohort study, DIPSS risk at transplantation was the only clinical variable significantly associated with post-transplant OS. All 14 patients with DIPSS High-risk disease died during follow-up after transplantation, with a median OS of only 6.4 months, while the corresponding value was 15.2 months in the Intermediate-risk group. In the exploratory multivariable models, DIPSS High-risk disease was independently associated with inferior OS (adjusted HR, 2.91; P = 0.011), whereas no other variable, including age, HCT-CI, and donor type, reached statistical significance. Notably, all patients with DIPSS High-risk disease were aged ≥ 60 years, precluding the reliable separation of age and disease-risk effects. These findings support the concept of a therapeutic window for allo-HSCT in MF, and suggest that transplantation should ideally be pursued before progression to DIPSS High-risk disease [16, 17].

Consistent with the updated EBMT/ELN recommendations, risk stratification is central to transplant decision-making in MF [12]. In our cohort, DIPSS High-risk disease was associated with inferior OS in both unadjusted and exploratory adjusted analyses. Transplantspecific scoring systems such as the MTSS may further refine post-transplant risk estimation [13]. However, the MTSS requires platelet count and performance status data that were not systematically captured in our retrospective dataset, precluding its calculation in the present study. Furthermore, comprehensive molecular annotation was limited because NGS data were available for only 13 of the 42 patients (31.0%), which precluded the evaluation of HMR features at the cohort level. Integrated clinical-molecular scoring systems such as MIPSS70 [18] incorporate HMR mutations that were not systematically captured in our cohort. In an exploratory analysis restricted to the NGS-tested subset, the presence of ≥ 2 HMR mutations appeared to be associated with poor outcomes, although the small sample size precluded formal statistical testing. Future studies with comprehensive molecular annotations are warranted to determine whether molecular risk features refine the prognostic impact of the DIPSS category in a transplant setting [19].

Conditioning intensity differed between the two groups: all patients with DIPSS High-risk disease received RIC (BuFlu2), whereas 50% of those with DIPSS Intermediate-risk disease received MAC (BuFlu4). This allocation reflects risk-adapted clinical practice, wherein patients with a greater disease burden are preferentially offered RIC to reduce early transplant-related toxicity. Although conditioning intensity can influence posttransplant outcomes in MF [20, 21], a sensitivity analysis restricted to the RIC-only subset (n = 28) preserved the prognostic separation between the DIPSS risk groups (3-year OS: 50.0% vs. 7.1%; log-rank P = 0.018; Cox HR: 2.88, 95% CI 1.15–7.22), indicating that the observed survival difference was not primarily driven by differential conditioning intensity. Nevertheless, the RIC-only subset remained small and non-randomized, and residual confounding from unmeasured factors cannot be fully excluded.

Driver mutation status and spleen size were not significantly associated with OS in our cohort, likely reflecting their limited statistical power [22–25]. Rather than performing splenectomy or splenic irradiation solely to reduce spleen size [26, 27], JAK inhibitors can be reasonably used primarily as a pretransplant bridging therapy, with splenic reduction as a secondary benefit [28]. Alternative approaches should be considered for patients who do not respond to JAK inhibitors [29, 30].

Infection was the leading cause of death in both DIPSS risk groups, with nine of 13 early deaths (within 6 months) being infection-related (69.2%). All patients underwent rabbit anti-human thymocyte globulin (rATG)-based conditioning, which is known to delay immune reconstitution and increase susceptibility to opportunistic infections. The higher proportion of haploidentical donors in the High-risk group (57.1% vs. 28.6%) may have further contributed to the risk of infection, given the additional immunosuppression required in an HLA-mismatched setting. These observations underscore the need for intensified infection surveillance and antimicrobial prophylaxis strategies, particularly in the first six months after HSCT. However, our dataset did not capture specific pathogens or infection sites, and the absence of competing-risk analyses limited further interpretation.

Our study had several limitations. The small sample size and retrospective single-center design limited the statistical power and generalizability of the findings. The age-stratified comparisons should be interpreted with caution because the conditioning regimen, JAK inhibitor use, and DIPSS risk were not uniformly distributed among the age groups. The only statistically significant association observed was the relationship between OS and DIPSS risk category; all other comparisons should be regarded as hypothesis-generating. When transplantation is considered in patients with DIPSS High-risk disease, pretransplant risk optimization, including JAK inhibitor bridging therapy, may be reasonable. If meaningful disease-risk reduction cannot be achieved, nontransplant approaches and supportive care should be discussed regardless of the patient’s age.

In this single-center retrospective analysis, DIPSS Highrisk disease at transplantation was associated with uniformly poor post-transplant survival, with 100% mortality even in patients who underwent allo-HSCT. These findings highlight the importance of transplant timing and suggest that allo-HSCT should be considered before progression to DIPSS High-risk disease. Given the limitations of this small retrospective cohort, these hypothesis-generating observations warrant validation in multicenter studies that incorporate integrated clinicalmolecular transplant risk models.