What question did this study set out to answer?

To investigate the relationship between clonal architecture and treatment outcomes in myelofibrosis patients receiving JAK inhibitors.

December 8, 2025Open Access

Clonal architecture inferred from routine NGS data predicts outcome and response to JAK-inhibitors in myelofibrosis

Key Points

To investigate the relationship between clonal architecture and treatment outcomes in myelofibrosis patients receiving JAK inhibitors.
Analyzed JAKi-treated myelofibrosis patients from a clinical database.
Conducted mutation analysis of 40 myeloid neoplasm-associated genes via next-generation sequencing.
Defined spleen response and loss of response per established criteria.
Categorized patients based on the order of mutation acquisition using variant allele frequencies.
Out of 250 patients, 47.8% achieved spleen response.
Age, spleen size, hemoglobin levels, and specific mutations were associated with spleen response.
MFC-first patients had significantly poorer outcomes compared to DM-first patients.

Abstract

Abstract Background: Intra-tumor heterogeneity (ITH) and mutated variant allele frequency (VAF) were associated with poor outcomes in cancers, including myeloid neoplasms. However, information from integrated analysis of clonal structure with response to treatment and outcome are scanty. Single-cell or colony assays offer accurate assessment of clonal architecture, but are not amenable to clinical practice, prompting development of surrogate methods for assessing ITH from bulk sequencing. In myelofibrosis (MF), JAK inhibitors (JAKi) improve splenomegaly, but many patients (pts) experience loss of response (LoR) that has been associated with clonal evolution and shortened overall survival (OS). Aim To investigate relationships between clonal architecture and outcomes in JAKi-treated MF pts. Methods All JAKi-treated MF pts available in our database were analyzed. Spleen response (SR) and LoR were defined per IWG-MRT criteria (Tefferi A, Blood 2013). Mutation analysis of 40 myeloid neoplasm-associated genes was performed by NGS before JAKi start. To infer the order of mutation acquisition, genes were classified into mutational functional categories (MFC), including epigenetic modifiers (TET2, EZH2, ASXL1, DNMT3A, IDH2), RAS pathway genes (NRAS, KRAS, CBL), and spliceasome (SF3B1, U2AF1, ZRSR2, SRSF2). For each MFC, pts who concurrently harbored a driver mutation (DM) were analyzed. Based on the VAF of MFC and DM, pts were categorized as either “DM-first” (DM VAF MFC VAF) or “MFC-first” (MFC VAF DM VAF) (Benard BA, Nat Commun 2021). Subsequently, for each pt, VAFs were used as input for PhyClone (Hurtado E, Bioinfo 2025) to infer the number of distinct clones and their relative abundance based on shared VAF patterns. The resulting information was used to compute the Shannon Diversity Index (SDI), which provides an estimate of clonal diversity (Cerrano M, Leukemia 2021). Median SDI value was calculated by normalizing the SDI on mutations' number for each pt. We then binned pts as those with SDI above (“clonal equilibrium”) or below (“clonal dominance”) the median. Results MFC were represented as follows: epigenetic modifier (50%), RAS pathway(14.8%), spliceasome(20%). 250 JAKi-treated pts were included; 140 primary-MF (56%), 110 secondary-MF (44%). 216 pts received ruxolitinib (86.4%), 15 fedratinib (6%), 19 momelotinib (7.6%). 101 pts died (40.4%); median OS from treatment start was 2.9y (range 0.7-13.7). SR was achieved in 119 pts (47.8%). Baseline variables associated with SR were age 65y (p=0.02), spleen10 cm from LCM (p=0.012), Hb10 g/dL (p=0.009), absence of ASXL1mut(p0.001), EZH2mut(p=0.04), NRASmut(p=0.04) or KRAS mut(p=0.012). Multivariable analysis confirmed the predictive value of age (HR 1.9, p=0.034), spleen (HR 2.3, p=0.012), absence of ASXL1mut(HR 1.8, p=0.034) and KRASmut(HR 9.5, p=0.049). Notably, ASXL1mutmutation was a strong predictor of NR (22.5% vs 77.5%, p0.001). 34 pts (28.6%) experienced LoR after a median of 21.3mo (0.9–134.8). Presence of spliceosome mutations was associated with shorter SR duration (1.3 vs 4.9y; HR 3.4; p=0.001); particularly, SF3B1mutconferred increased LoR risk (HR 6.9; p=0.002). Regarding order of mutation acquisition, among the 105 pts in the epigenetic modifiers MFC, 73 (70%) were classified as DM-first and 32 (30%) as MFC-first. Notably, the MFC-first pts had significantly poorer outcomes, with shorter OS (5.5 vs 11.3y, p0.001) and markedly lower SR (19.4% vs 80.6%, HR 0.3, p=0.007), underlining the importance of mutation acquisition sequence, not just their presence, as prognostic factor. Conversely, order of acquisition of spliceasome MFC did not influence SR (p=0.5) nor OS (p=0.8). RAS pathway alterations were all found to arise subsequently to DM, as reported (Coltro G, Blood Adv. 2020), therefore they were not informative. Finally, we estimated clonal diversity using the SDI. In Cox regression analysis, equilibrium status was significantly associated with worse OS (8.9 vs 12.7 yrs, HR 1.8 p=0.010), whereas no impact on SR was noted.Conclusion: This study highlights the impact of clonal architecture profile on response to JAKi in MF pts, particularly the role of mutations in epigenetic regulators when acquired as first-hit event. These findings were generated using an inferential method based on VAF calculation from routine bulk sequencing, that can be easily implemented in practice and used to guide therapeutic strategies.

Read Full Paperexternally

AIに質問

Bookmark

View Full Paper