Long-term outcome and risk factors for relapse after allogeneic hematopoietic stem cell transplantation in myelofibrosis: Results from a large multicenter cohort

Abstract

Abstract Background Allogeneic hematopoietic stem cell transplantation (alloHCT) is increasingly performed in Myelofibrosis (MF), where it remains the only potentially curative treatment. Disease relapse is a major contributor to post-transplant mortality and may occur even several years (yrs) after alloHCT in a relevant subset of patients (pts). Given the limited and heterogeneous data on long-term outcomes and relapse risk - particularly with respect to late relapse - we performed a landmark analysis in a large multicenter MF cohort to explore factors associated with late disease recurrence and post-transplant survival. Methods Clinical data were collected from primary MF (PMF) and secondary MF pts undergoing alloHCT between 2000-2023 at nine German centers. Pts surviving >100 days post-transplant were included for outcome analysis. Longitudinal chimerism data were available for all pts based on centralized short tandem repeat testing. Molecular profiling was performed on pre-transplant and relapse samples using a 60-gene myeloid next-generation sequencing (NGS) panel. Relapse was defined according to established criteria, including clinical, morphological or molecular recurrence prompting therapeutic intervention. Overall survival (OS) and relapse-free survival (RFS) were estimated using the Kaplan-Meier method. Multivariate analysis was performed using Cox regression with stepwise variable selection based on Akaike and Bayesian Information Criteria. Results Among the 234 included pts, median age at transplant was 59 yrs (range, 19-72); 66.7% were male, and 70.9% had PMF. Median follow-up was 74 months (IQR, 44–120). The cumulative incidence of relapse (CIR) was 15.4% at 1 year, 24.7% at 3 yrs, and 27.1% at 5 yrs post-transplant. Among pts alive and relapse-free at 5 yrs, the 5-year CIR from this landmark was 17.7%, indicating notable risk for late relapse. The 5-year OS from the time of relapse was 62% for early (<5 yrs) vs. 67% for late (≥5 yrs) relapse (p=.46). Among relapsed pts, salvage therapy was initiated in 87.5%. Of those, 77.8% achieved second remission, including 55.6% responding to donor lymphocyte infusion or immunosuppression withdrawal alone and 22.2% requiring a second alloHCT. Driver mutations included JAK2 (65%), CALR (21.7%), and MPL (11.5%). Additional mutations primarily affected epigenetic modifiers, with ASXL1 mutated in 52.1%, EZH2 in 13.4% and IDH1/2 in 6%. Spliceosome-related mutations (U2AF1, SRSF2, SF3B1, ZRSR2) were detected in 30.8%, RAS-pathway mutations (e.g., NRAS, KRAS, CBL, PTPN11) in 21.7%, cohesin complex mutations (STAG2, SMC1A, SMC3) in 6.5%, and TP53 mutations in 5.1%. Paired NGS (pre-transplant and at relapse; n=39) revealed stable mutational profiles - including both driver and co-occurring mutations - with no relevant evidence of clonal evolution, even in relapses beyond 10 yrs post-transplant. In univariate analysis, post-transplant survival showed no significant association with disease-related factors, including MF subtype (PMF vs. secondary MF) and pre-transplant risk scores (MIPSS70/MIPSS70+, MYSEC-PM), or transplant-related variables such as donor type, conditioning intensity, prior JAK inhibitor use, time from diagnosis to alloHCT, or transplant year (<2010 vs. ≥2010). Age ≥60 yrs was associated with inferior OS. Long-term mixed chimerism (<99% beyond day +200) predicted shorter RFS, but not OS. Limited chronic GvHD was associated with improved RFS and OS. Mutations in ASXL1 and EZH2 were linked to inferior RFS and OS; mutations in CBL and PTPN11 to shorter RFS. In multivariate analysis, only mutations in ASXL1 (HR 1.53, p=.045), CBL (HR 2.9, p=.003) and PTPN11 (HR 3.9, p<.001), as well as long-term mixed chimerism remained independently predictive for RFS, and only ASXL1 for OS (HR 2.2, p=.007). Conclusions In MF, relapse remains a long-term post-transplant risk and is predominantly observed without clonal evolution, indicating recurrence from residual clones rather than molecular progression. Mutations in ASXL1, CBL, and PTPN11 as well as long-term mixed chimerism were independently associated with inferior long-term outcomes. The favorable impact of limited chronic GvHD and responses to immunomodulatory salvage strategies underscore the potential of immune-mediated disease control in MF. Although limited by its retrospective, multicenter design, this study supports the relevance of molecular monitoring and immune-based interventions in long-term post-transplant care in MF.