Prognostic impact of ASXL1 mutations on primary and secondary myelofibrosis: A FIM study

Myelofibrosis (MF) is a complex hematologic disorder characterized by dysregulation of the bone marrow stroma developing a reticulin fibrosis. High-throughput sequencing has demonstrated the various functions of somatic nondriver mutations, or additional mutations, such as splicing, signalization, and transcription factors, and therefore ‘high-risk’ genes are now included in the new scoring systems in primary myelofibrosis (PMF).

Although the prognostic function of additional mutations has been confirmed in PMF, there is limited data on secondary myelofibrosis (SMF). A genomic analysis study by Paz et al. investigating the prognostic impact of the mutational landscape in PMF and SMF, was published in Blood Advances and co-authored by two MPN Hub Steering Committee members, Stéphane Giraudier and Jean-Jacques Kiladjian.¹ The key findings are summarized below.

The MPN Hub has previously reported  on the gene expression profile in MF subtypes.

Methods

This was a genomic analysis study analyzing DNAs from unselected patients (N = 479) with PMF or SMF, registered in the French Intergroup of Myeloproliferative Neoplasms (FIM) observatory of MF. Samples of whole blood, purified blood granulocytes, or whole bone marrow were collected. Clinical and biological data were extracted at diagnosis and follow-up.

• Validation cohort was derived from data published by Grinfeld et al.,² where the mutational landscape of 309 patients with PMF or SMF was investigated. Also, the multistage model for prognosis was performed on 276 patients with MF. Events in the validation cohort included leukemic transformation and deaths.
• Next-generation sequencing (NGS) with a custom RNA-baits panel was used to cover all exons of 77 genes of interest. A bioinformatic pipeline for demultiplexing and aligning, and variant calling, was developed to analyze sequencing data.
• A Bayesian network analysis combined with hierarchical clustering analysis was performed to characterize homogeneous clusters of genes. A multistate model was used to decrypt the natural history of MF.

Baseline characteristics

Median time between diagnosis and sampling was 17 days (range, 0–88 months). Patient characteristics are summarized in Table 1. Of all the patients with SMF, 70 had post-polycythemia vera (post-PV) MF, and 104 had post-essential thrombocythemia (post-ET) MF.

Table 1. Baseline characteristics*

M, male; MF, myelofibrosis; F, female; PMF, primary myelofibrosis; SMF, secondary myelofibrosis. *Adapted from Paz et al.1
Characteristic	MF (N = 479)	PMF (n = 305)	SMF (n = 174)
Sex (M/F), %	64/36	71/29	52/48
Age at diagnosis, mean, years (range)	66 (18−89)	68 (18−88)	67 (31−89)
Hemoglobin (g/dL), mean (range)	11.1 (6.2−18)	11.1 (6.3−17.2)	11.1 (6.2−18)
Platelets at diagnosis (109/L), mean (range)	272 (16−999)	250 (16−999)	323 (19−980)
Leukocytes at diagnosis (109/L), mean (range)	9.7 (1.3−76)	9.5 (1.7−76)	9.8 (1.3−68)
Driver mutations, %
JAK2 V617F	65	61	70
CALR	23	24	21
MPL	5	6	3
Double mutation	2	1	2
Triple negative	6	7	3

Results

Mutational landscape

• A total of 1,385 additional variants were detected including 861 pathogenic variants, 179 likely pathogenic variants, and 345 variants of unknown significance. Median of 3 (range, 0−11) additional variants, and 2 (range, 0−10) additional mutations per patient were identified.
• The most common mutations were ASXL1, TET2, SRSF2, U2AF1, and EZH2.
• SRSF2 and U2AF1 mutations were more frequently encountered in patients with PMF (vs SMF) and those with the JAK2 mutation (vs CALR mutation). Patients with SMF showed higher frequency of NFE2 mutations.
• ATM, KMT2C, KMT2D, and NOTCH1 mutations were variants of unknown significance and considered rare polymorphisms.
• Additional mutations displayed different profiles depending on the genes and allele burden distribution. An allele burden between 40% to 50% harbored SRSF2 and U2AF1, 0% to 100% harbored EZH2 and TP53, and an allele burden with a bimodal distribution harbored ASXL1 and TET2.
• For pathogenic and likely pathogenic mutations, additional mutations showed no difference based on sex (p = 0.08). However, the ≥1 additional mutation were more frequent in older patients (p = 0.0009).

Prognostic impact

Four prognostic groups based on their impact on overall survival (OS) were developed (Table 2).

Table 2. Prognostic groups*

*Adapted from Paz et al.1
Additional mutations	Patients, n	Risk group
TP53-mutated	27	TP53
≥1 mutation in EZH2, CBL, U2AF1, SRSF2, IDH1, IDH2, NRAS, or KRAS	150	High-risk
ASXL1-only mutation (no associated mutation in TP53 or in high-risk genes)	74	ASXL1-only mutation
Mutations in NFE2, DNMT3A, TET2, and SF3B1	228	‘Other’ patients

• Median OS for TP53 and the high-risk group was 20 and 49 months, respectively, and median time to leukemic transformation was not reached and 110 months, respectively.
• In comparison, ASXL1-only and the ‘other patients’ group had a median OS of 89 and 116 months, respectively, and leukemic transformation was not reached (p < 0.0001).
• Leukemic transformation and risk of death without acute myeloid leukemia (AML) were significantly higher in the TP53 and high-risk mutations groups (see Table 3).
• Age at diagnosis, anemia, and leukocytosis were associated with a higher risk of death, whereas thrombocytopenia was associated with a higher risk of leukemic transformation.
• Validation cohort confirmed the findings and more specifically the prognostic impact of TP53 and high-risk mutations on leukemic transformation and death.
• It also confirmed the prognostic value of the ASXL1-only group. However, the protective effect of CALR mutations on leukemic transformation and reduced OS for TP53 mutations could not be reproduced in the validation cohort.

Table 3. Multistate model for AML and death*

AML, acute myeloid leukemia; HR, hazard ratio; MF, myelofibrosis; PMF, primary myelofibrosis. *Adapted from Paz et al.1 †Reference: JAK2. ‡Reference: ‘other patients’.
Characteristic	MF to AML		MF to death		AML to death
	HR (95% CI)	p value	HR (95% CI)	p value	HR (95% CI)	p value
Age, years	—­	—	1.05 (1.03–1.07)	<0.0001	—	—
Sex, male	—	—	1.41 (0.96–2.06)	0.077	—	—
Hemoglobin	—	—	0.81 (0.74–0.89)	<0.0001	—	—
Leukocytes	—	—	1.04 (1.02–1.05)	<0.0001	—	—
Platelets	0.99 (0.99–0.99)	0.012	—	—	—	—
Driver mutation†: CALR	0.21 (0.06-0.70)	0.011	—	—	—	—
Genomic‡
TP53	8.68 (3.32−22.73)	<0.0001	3.03 (1.66−5.56)	0.0003	1.21 (0.31−4.72)	0.784
High-risk	3.24 (1.58−6.64)	0.0013	1.77 (1.18−2.67)	0.006	0.88 (0.33−2.36)	0.801
ASXL1-only	2.45 (0.95−6.29)	0.063	1.17 (0.68−2.01)	0.579	1.22 (0.33−4.47)	0.767

ASXL1 mutations

• Sixty percent of cases with TP53 or high-risk gene mutations were associated with ASXL1 mutations and increased the risk of death (HR 1.5; 95% CI, 1.01−2.29; p = 0.044).
• High-risk mutations were associated with a higher allele burden of ASXL1 mutations. However, in the ASXL1-only group the allele burden of the ASXL1 mutation was not associated with acute leukemia (p= 0.293) or death (p = 0.763).
• ASXL1-only mutations had no prognostic value, which was supported by the validation cohort.

Comparison of prognostic performances

• The 4-tier genomic classification was superior to the International Prognostic Scoring System (IPSS) but inferior to Myelofibrosis Secondary to PV and ET-Prognostic Model (MYSEC-PM) for leukemic transformation.
• 4-tier genomic classification combined with the IPSS or MYSEC-PM remained superior to the Mutation-Enhanced International Prognostic Score System 70 (MIPSS70) and MIPSS70+v2 despite their good predictive values, in particular for death in PMF and SMF, and for leukemic transformation in PMF.
• The association of genomic groups with standard prognostic scoring systems was associated with improved prediction and accuracy of prognosis for both short- and long-term events.  

Conclusion

The study demonstrated that mutations of TP53 and high-risk genes (including EZH2, SRSF2, IDH1, IDH2, U2AF1, CBL, NRAS, and KRAS) were adverse prognostic factors in MF. When the prognostic impact of each gene in the high-risk group was investigated, it showed that all genes were associated and improved the prognostic evaluation. CALR mutations were found to be associated with a lower risk of leukemic transformation. The study also showed that isolated mutations of ASXL1 had no prognostic value consistent with the findings from the validation cohort, and the impact on leukemic transformation was limited. However, when ASXL1 mutations were associated with the TP53 mutation or high-risk genes, they conversed a worse prognosis.

These results suggest that the exclusion of ASXL1 mutations from the prognostic classification of somatic mutations in MF may be discussed where high-risk mutations such as TP53, U2AF1, CBL, NRAS and KRAS may be included. The findings from this study warrant for a new molecular classification that will need to be validated in large multicenter cohorts.