Dear Editor,

In primary myelofibrosis (PMF), there is a stepwise evolution from an initial prefibrotic (Pre)/early stage, characterized by hypercellular bone marrow with absent or minimal reticulin fibrosis, to an overt fibrotic (Overt) stage with marked reticulin or collagen fibrosis in the bone marrow [1] and 5–30% of patients develop to blast phase [2,3,4,5,6]. Although gene mutations were confirmed as the important prognostic factor in PMF [7,8,9,10,11,12], whether or not these gene mutations can predict fibrosis progression and leukemic transformation is still unclear. The aim of this study was to explore if some non-driver mutations play a key role in disease progression based on the mutational landscape in the progression of PMF. Through analyzing the non-driver mutations in different stages of PMF patients, we found that ASXL1 mutations were significantly associated with both the exacerbation of fibrosis and the leukemic transformation.

According to the World Health Organization (WHO) 2016 classification [13], 258 consecutive PMF patients from July 7, 2015 to December 23, 2021 were enrolled. All cases were reviewed by two experienced pathologists (PZ, QS) to confirm the diagnosis of PMF, evaluate disease progression and exclude post-PV/ET MF according to the WHO 2016 classification of MPN [13]. A total of 275 samples from different stages of disease were included in the analysis: 69 Pre-PMF, 161 Overt-PMF and 45 PMF-AP/BP (details in Supplementary Fig. 1). At least two-time points serial samples from 17 patients during disease progression (4 patients progressed from Pre-PMF to Overt-PMF and 13 patients progressed from Overt-PMF to PMF-AP/BP) were sequenced. The median interval from PMF diagnosis to AP/BP transformation was 28.5 (1–144) months. Patient samples were obtained with written informed consent by the Declaration of Helsinki, and the study was approved by the human research ethics committee at the Institute of Hematology and Blood Disease Hospital, Chinese Academy of Medical Sciences (CAMS) and Peking Union Medical Collage (PUMC) according to the guidelines of the Declaration of Helsinki.

Twenty-seven genes (Supplementary Table 1) were employed to investigate the frequency and enrichments of non-driver mutations in different stages of PMF. High molecular risk (HMR) mutations were considered as mutations in any one of the six genes: ASXL1, EZH2, IDH1, IDH2, SRSF2 and U2AF1Q157 [11]. More detailed information is described in the Supplementary Methods.

Comparison of the baseline clinical and laboratory characteristics of patients classified in different stages are summarized in Supplementary Table 4. Driver mutations were distributed as follows: JAK2V617F (55.8%, n = 144), CALR Exon9 (24%, n = 62), MPLW515 (3.5%, n = 9), double mutations (CALR Exon9 and MPLW515) in 1 patient (0.4%). Forty two (16.3%) patients did not have any driver mutations (triple-negative). Only the variant allele frequency(VAF) of JAK2V617F had a significant increment between Pre-PMF and PMF-AP/BP patients (median of VAF: 38.3% vs. 50.9%, P = 0.031), but no difference between Overt-PMF and PMF-AP/BP (median of VAF: 45.4% vs.50.9%, P = 0.333) (Supplementary Fig. 2A–C). The distribution and the number of non-driver mutations in different stages of PMF were shown in Supplementary Fig. 2D–G.

Mutations in the ASXL1 (31.1% vs. 10.1%, P = 0.001) and U2AF1 (13.7% vs. 1.4%, P = 0.003) were more frequent in Overt-PMF compared with Pre-PMF (Fig. 1A). Moreover, mutations in ASXL1 (51.1% vs. 31.1%, P = 0.013), SRSF2 (26.7% vs. 11.2%, P = 0.009), RUNX1 (22.2% vs. 5.6%, P = 0.001), SETBP1 (17.8% vs. 7.5%, P = 0.039), NRAS (15.6% vs. 3.1%, P = 0.002) and EZH2 (13.3% vs. 5%, P = 0.049) were significantly more frequent in PMF-AP/BP compared with Overt-PMF (Fig. 1B). Consistent with prior studies [14, 15], ASXL1 mutations were most common in PMF and frameshift and nonsense mutations are the major ASXL1 mutation types (Supplementary Fig. 3). From the whole course of the disease, the frequency of ASXL1 mutations significantly increased during the progression (10.1% vs. 31.1% vs. 51.1%) (Fig. 1C). However, there was no significant change in the VAF of ASXL1 mutations in different stages (Median of VAF: 42.1% vs. 30.6% vs. 37.8%) (Supplementary Fig. 2H). In addition, the frequency of HMR mutations also significantly increased during the disease progression (14.5% vs. 45.3% vs.73.3%) (Supplementary Fig. 2I) as expected.

Fig. 1: The association between non-driver mutation and different stages of PMF.
figure 1

Comparison of mutation frequency in Pre-PMF vs. Overt-PMF (A) and Overt-PMF vs. PMF-AP/BP patients (B). C The frequency of ASXL1 mutations in different stages of PMF. D Enrichment of non-driver mutations in PMF-AP/BP and Overt-PMF relative to Overt-PMF and Pre-PMF, respectively. Enrichment is expressed as an odds ratio (OR) of mutation rates in PMF-AP/BP vs. Overt-PMF on the x-axis and Overt-PMF vs. Pre-PMF on the y-axis. Mutations showing significant enrichment in either comparison are indicated by colors according to OR 95% CI limits being above (if OR > 1) or below (if OR < 1). *P < 0 .05; **P < 0.01; ***P < 0.001.

Full size image

To clarify the relationship between non-driver mutations and disease progression, we looked at the enrichments of major non-driver mutations in different stages of PMF.

In univariate comparison, ASXL1 (OR = 3.99, 95% CI 1.71–9.33; P = 0.001) and U2AF1 (OR = 10.76, 95% CI 1.42–81.53; P = 0.005) mutations were significantly enriched in Overt-PMF compared to Pre-PMF. Mutations in five genes, including ASXL1(OR = 2.32, 95% CI 1.18–4.55; P = 0.013), SETBP1 (OR = 2.69, 95% CI 1.02–7.64; P = 0.039), RUNX1 (OR = 4.83, 95% CI 1.82–12.76; P = 0.001), NRAS (OR = 5.75, 95% CI 1.73–19.10; P = 0.002) and SRSF2 (OR = 2.89, 95% CI 1.27–6.58; P = 0.009) were significantly enriched in PMF-AP/BP compared to Overt-PMF (Supplementary Fig. 2J and Supplementary Table 5).

To exclude the effects of the co-occurrence of non-driver mutations that might confound the result, we performed a multivariate analysis. In the comparison between Overt-PMF and Pre-PMF, ASXL1 (OR = 3.65, 95% CI 1.54–8.62; P = 0.003) and U2AF1 (OR = 9.21, 95% CI 1.20–70.69; P = 0.033) mutations were still significantly enriched in Overt-PMF. Meanwhile, in the comparison between PMF-AP/BP and Overt-PMF, only RUNX1 (OR = 5.11, 95% CI 1.66–15.74; P = 0.004), NRAS(OR = 4.06, 95% CI 1.05–15.60; P = 0.042) and ASXL1 (OR = 2.88, 95% CI 1.36–6.10; P = 0.006) mutations were strongly enriched in PMF-AP/BP (Fig. 1D and Supplementary Table 5). These results indicate that ASXL1 mutations might play a critical role in myelofibrosis progression and blast phase evolution during the whole course of PMF progression.

As mentioned above, ASXL1 mutations play a critical role both in the progression from Pre-PMF to Overt-PMF and through Overt-PMF to PMF-AP/BP. To evaluate the status of ASXL1 mutations in different stages, we analyzed 17 patients who had at least two-time points serial samples during disease progression (Supplementary Table 6). Firstly, to explore how ASXL1 mutations promote the progression from the chronic stage to the AP/BP stage, we analyzed clonal evolution in serial samples collected from Overt-PMF to PMF-AP/BP. Excepting newly acquired ASXL1 mutations at the AP/BP stage, all ASXL1 mutations that occurred in the Overt-PMF phase kept relatively stable with less than 10% of fluctuation (Fig. 2A). Contrary to ASXL1, RUNX1 mutations (4 in 5, 80%), RAS pathway mutations (7 in 8, 87.5%) and TP53 mutations (1 in 1,100%) were more likely to be freshly acquired during the transformation from Overt-PMF to PMF-AP/BP (Fig. 2B and Supplementary Fig. 4A and B). Interestingly, SRSF2 mutations tended to decrease clone size during disease progression (Supplementary Fig. 4C). Then, to answer the question that why AP/BP transformation did not accompany with obvious ASXL1 mutations expansion, we analyzed the co-mutations in PMF-AP/BP patients who had ASXL1 mutations at the chronic stage. We found that ASXL1 mutations were more likely to co-occur with RAS pathway mutations (NRAS 20% and NF1 50%) and ETV6 (50%) (Fig. 2C). Data of serial samples from chronic and AP/BP stages shed the light on that different non-driver mutations should play different roles in the AP/BP transformation: 1. ASXL1 mutations are not a direct event for leukemogenesis but accelerate the possibility of accumulation of direct leukemogenic factors, such as RAS pathway mutations. 2. RUNX1 mutations are direct and independent factors in AP/BP transformation. Moreover, ASXL1 mutations even cannot be significantly removed by hypomethylating agent (HMA) therapy in 3 PMF-AP/BP patients (Fig. 2D), but co-occurred RUNX1 and SETBP1 mutations burden were obviously decreased (Supplementary Fig. 4D).

Fig. 2: Dynamics of mutations in the AP/BP transformation in PMF.
figure 2

A VAF of mutations in 7 paired serial samples from PMF patients with ASXL1 mutation. B VAF of mutations in 5 paired serial samples from PMF patients with RUNX1 mutation. In (A) and (B), the left panel showed the VAF of ASXL1 or RUNX1 mutation and the right panel showed the VAF of other mutations in each patient. C The number and proportion of patients with newly acquired and VAF-increasing mutations during the AP/BP transformation were indicated in the hot plot. Newly acquired mutations were depicted in red and VAF-increasing mutations (>10%) were depicted in blue, and untested genes were depicted in green. Patients with and without ASXL1 mutation in Overt-PMF were labeled in brown and blackish green, respectively. D VAF of ASXL1 mutation, percentage of peripheral blood/bone marrow blasts as well as white blood cell counts before and after HMA therapy in PMF-AP/BP.

Full size image

In summary, our study provides a clue that ASXL1 mutations may play a critical role in the whole course of PMF, aggravating myelofibrosis and conferring the tractive force for leukemia transformation. Due to the high frequency and critical role of ASXL1 mutations in PMF, ASXL1 could potentially act as a therapeutic intervention point. Treatments targeting the functional effect of mutant ASXL1 in PMF, such as BET (Bromodomain and Extraterminal domain) inhibitors, BAP1(BRCA1-Associated Protein 1) inhibitors, TNFR (Tumor Necrosis Factor Receptor) inhibitors, et al., could be used in the future. Because of a small cohort with limited paired serial samples from a single center, our findings also need to be validated in a large cohort with more paired serial samples.