Genetic Evolution from Chronic Myeloproliferative Neoplasms to Acute Myeloid Leukemia: An Analysis of Forty-Six Paired Samples

Background: Secondary acute myeloid leukemia (sAML) may arise from chronic myeloproliferative neoplasm (MPN). The genetic evolution pattern may be understood by paired sample analysis. Knowledge of genetic change in MPN/sAML is still limited and more extensive studies should be done.

Patients and Methods: Forty-six paired samples (27 males) at diagnosis of MPN (12 polycythemia vera, 6 essential thrombocythemia, 25 primary myelofibrosis and 3 MPN-unclassified) and sAML transformation were available. Mutational analyses of 27 genes ( JAK2V617F, CALR, MPL, ASXL1, IDH1, IDH2, TET2, DNMT3A, EZH2, SRSF2, U2AF1, SF3B1, ZRSR2, SMC1, STAG2, SMC3, RUNX1, SETBP1, IKZF1, WT1, K-RAS, N-RAS, C-CBL, TP53, RAD21, BCOR and BCORL1) were performed. Mutations and their variant allele frequencies (VAF) were measured by next generation sequencing.

Results: At baseline (MPN), the mutation number was 2 (maximal 5). Forty-five samples harbored driver gene mutations (31 JAK2V617F, 13 CALR and 1 MPL). Twenty-eight samples harbored at least one epigenetic regulator mutations (11 TET2, 8 DNMT3A,7 ASXL1, 5 EZH2, 2 IDH2, 1 IDH1). Twelve samples harbored spliceosome mutations which were mutually exclusive (6 SRSF2, 3 SF3B1, 2 U2AF1 1 ZRSR2). Other mutations were RUNX1, BCORL1 (n=2 each), CBL and BCOR (n=1 each). In sAML, the median mutation number was 3 (maximal 7). Loss of mutation was found for JAK2V617F (9/31), TET2 (1/11) and SRSF2 (1/6). Acquisition was most common for RUNX1 (n=9), followed by TP53, ASXL1 (n=5 each), K-RAS (n=4) , ZRSR2, STAG2 (n=3 each). The median number of acquired mutation was 1 (maximal 5). The frequencies of gene mutations are summarized in Figure 1A.

The baseline VAF of JAK2V617F was high (>35%) in 80.6% (25/31). In progression to sAML, VAF was constant in 29.0% (9/31, VAF changes <10%), increased in 20.0% (9/31, increased VAF> 10%), decreased in 41.9% (13/31, decreased VAF>10%) (Figure 1B). The baseline VAF in CALR mutation was high (>35%) in 91.3% (12/13). VAF remained stable in 76.9% (10/13), and loss of CALR mutation was not observed (Figure 1B). TET2 mutations tended to be stable (7/16), but clonal expansion (5/16 of all mutation events at MPN), or acquisition (n=2) could be found during transformation to sAML. DNMT3A mutations remained stable (5/8), exhibited clonal expansion (2/8), or acquisition (n=1) in sAML progression. ASXL1 mutations remained stable (6/7), with clonal expansion (1/7), or acquisition (n=5) during MPN/sAML progression. All EZH2 mutants had clonal expansion (5/5) and one acquired in sAML (Figure 1B).

We found that CALR mutationwas often the founding clone with high VAF at MPN phase, stable clone during sAML, and might acquire mutations in signaling pathways (1 CBL, 1 KRAS), epigenetic regulators (2 ASXL1, 1 IDH2), transcription factors (2 RUNX1, 1 BCOR), spliceosome (2 ZRSR2, 1 SF3B1), and cohesin complex (1 STAG2). JAK2 mutationwas often preceded by other ancestral mutations such as TET2, ASXL1, DNMT3A, EZH2, IDH2, SRSF2 or ZRSR2. During evolution, leukemia arose from non- JAK2 mutated clones with mutations in EZH2, DNMT3A, RUNX1, EZH2, SETBP1, or acquisition of TP53, SF3B1, BCOR and STAG2 mutations) in sAML.

In outcome, the median time from MPN to sAML was 82 (range 3.7 to 338.5) months. Female gender, loss of JAK2V617F, presence of U2AF1 and absence of IDH1 were associated with shorter time to sAML transformation. No other genetic mutation in our gene panel or the number of mutations harbored by patients in our cohort had significant impact on time to sAML.

Marked increase of VAF was observed for JAK2 (7/22) , DNMT3A (1/7) , IDH1 (2/3), IDH2 (1/3), TET2 (3/10) , EZH2 (4/5), C-CBL (1/1) and SRSF2 (1/4). Marked decrease of VAF was observed for JAK2 (4/22), MPL (1/1), DNMT3A (1/7) , ASXL1 (1/8), SF3B1 (2/3), SRSF2 (1/4) and U2AF1 (1/2).

Multiple mutational clones were detected for TET2 (two clones in 4 samples and 3 clones in 1). Clonal expansion and reduction were observed in two pairs of samples upon sAML transformation.

Conclusions: At the MPN stage, most patients had driver mutations with frequent co-mutations of epigenetic modifier or spliceosome genes. CALR mutant clones were stable while for JAK2, sAML often arose from clones without JAK2V617F which acquired RUNX1, TP53, ASXL1 and K-RAS.

Research funding:Ministryof Science and Technology Council,Taiwan (NSTC 112-2314-B-182-055) and Ministry of Health and Welfare, Taiwan (112-TDU-B-222-124001)