Blast Phase of Myeloproliferative Neoplasm Resembles Acute Myeloid Leukemia, Myelodysplasia-Related, in Clinical Presentation, Cytogenetic Pattern, and Genomic Profile, and Often Undergoes Reversion to Second Chronic Phase Status After Induction Chemotherapy

Following approval of the study by the Institutional Review Board of Duke University Medical Center (Durham, North Carolina), we retrieved 21 cases of MPN that subsequently developed blast phase. All cases were retrospectively reviewed by 2 of the authors (Y.Z. and E.W.), and the diagnoses of MPN and subsequent blast phase were confirmed according to the 5th edition of World Health Organization Classification of Haematolymphoid Tumours: Myeloid and Histiocytic/Dendritic Neoplasms,⁹  supplemented by 2022 International Consensus Classification of Myeloid Neoplasm and Acute Leukemia.²  In addition, 3 cases of de novo acute myeloid leukemia (AML) with JAK2^V617F mutation were collected, and the diagnosis was confirmed in each case according to the World Health Organization (WHO) and International Consensus Classification classifications. The clinical history and laboratory data, including complete blood cell counts, serum lactate dehydrogenase, flow cytometric analyses, cytogenetic results, and NGS mutational profiles, were collected from corresponding clinical notes and test results in our medical informatics system, and retrospectively reviewed and analyzed.

Peripheral blood smears were stained with Wright stain, bone marrow aspirate smears and biopsy touch imprints were stained with Wright-Giemsa stain, and bone marrow core biopsies and clot sections were stained with hematoxylin-eosin stain. Bone marrow cellularity was evaluated on core biopsy, or on clot section in cases where the biopsy quality was suboptimal.

Genomic DNA was extracted from peripheral blood or bone marrow aspirate sample to perform NGS. Specific regions of 75 target genes, including 1500 hotspot mutation loci, were sequenced using the Illumina NextSeq 500. The resulting DNA sequence was assembled and aligned to the reference human genome version GRCh37/hg19. Sequence analysis was performed using the Archer Analysis bioinformatics pipeline. The lower limit of detection for this assay was 5%, meaning that a variant allele can be detected in a 95% wild-type allele background (or 10% neoplastic cell content). Sensitivity was impacted by the depth of sequence coverage, variant type, and allelic frequency of the variant. A depth of coverage of less than 100× was considered inadequate.

The statistical analyses were performed with SAS, version 9 (SAS Institute, Cary, North Carolina). The Student t test, Wilcoxon-Mann-Whitney test, χ² test, and Kaplan-Meier survival analysis were used to test the statistical significance of differences between the groups. P value of <.05 was considered to show statistical significance of differences between the compared groups.

The clinical features in 24 cases of MPN blast phase and de novo AML with JAK2^V617F are summarized in Supplemental Table 1 (see supplemental digital content containing 2 tables and 2 figures, at https://meridian.allenpress.com/aplm in the December 2024 table of contents), the pathologic characteristics are summarized in Supplemental Table 2, and the clinicopathologic features and statistical analyses stratified by types of initial myeloid neoplasm are summarized in the Table.

Of the 24 patients, 17 were male and 7 were female. Ages ranged from 49 to 91 years, with a median of 70.5 years at the time when blast phase of MPN or de novo AML with JAK2^V617F was diagnosed. For the 21 patients with MPN blast phase, ages ranged from 33 to 89 years with a median age of 61 years at the time when the chronic phase of MPN was initially diagnosed. In 21 cases with MPN blast phase, subtypes of initial MPN included PV in 3 cases, ET in 9 cases, and PMF in 9 cases. MPN driver mutations were detected at chronic phase in all 24 cases, including JAK2^V617F in 19 (79.2%), CALR mutation in 4 (16.7%), and MPL mutation in 3 (12.5%) cases. Of these, 2 cases demonstrated concurrent mutations of CALR and MPL, an uncommon finding, because driver mutations are usually mutually exclusive. Median interval from diagnosis of chronic phase to blast phase was 48 months with a range of 7 to 384 months. Six of 8 PV and ET cases (75%) showed evidence of myelofibrosis before leukemic transformation, assessed by significant (MF-2 or more) reticulin fibrosis at pre-blast, blast, and/or second chronic phases. At the time of blast phase or de novo JAK2^V617F-positive AML, 9 of the 24 cases (37.5%) presented with leukocytosis with circulating blasts, 1 (4.2%) with thrombocytosis, 1 (4.2%) with leukocytosis and thrombocytosis, 8 (33.3%) with pancytopenia, and 5 cases (20.8%) with bicytopenia or unicytopenia. Of the 3 cases of de novo JAK2^V617F-positive AML, 2 (66.7%; cases 23 and 24) presented with leukocytosis with left-shifted neutrophilia and eosinophilia in addition to circulating blasts, and the remaining case (case 22) presented with pancytopenia. Of 23 cases, 22 (95.7%) showed elevated serum lactate dehydrogenase at the time of blast phase/AML, with a median of 648 U/L and range of 170 to 2925 U/L (normal reference range, 100–200 U/L). Of the 24 cases, 15 (62.5%) demonstrated morphologic evidence of a second chronic phase status after induction chemotherapy. Of these, 4 cases (26.7%) showed either leukocytosis and/or thrombocytosis, including thrombocytosis in 2 (cases 3 and 9), leukocytosis in 1 (case 17), and leukocytosis/thrombocytosis in 1 (case 1), and the rest (11 cases; 73.3%) presented with pancytopenia, bicytopenia, or unicytopenia at the time of second chronic phase (Supplemental Table 1 and Table).

A total of 22 cases had peripheral blood smears examined at the time of blast phase or de novo AML, and all demonstrated circulating blasts, with a median of 23% and range of 1% to 86%. Of these, 19 (86.4%) exhibited elevated circulating blasts of greater than 5%, including 3 cases with left-shifted neutrophilia, and the remaining 3 cases (13.6%) showed cytopenias with rare to less than 5% circulating blasts. Bone marrow aspirate smears demonstrated increased blasts in all 21 cases that had bone marrow examination performed, with a median marrow blast count of 40% (range, 12%–85%). In addition to a marked increase in blasts, bone marrow biopsy showed marked myelofibrosis (MF-3) in 9 (45%), moderate myelofibrosis (MF-2) in 2 (10%), mild myelofibrosis (MF-1) in 2 (10%), and no significant myelofibrosis (MF-0) in 7 (35%) of the 20 cases evaluated for the change. Of 8 cases of PV or ET that had reticulin stains performed, 5 biopsies (62.5%) exhibited moderate or severe myelofibrosis (MF-2 or MF-3) at blast phase. Of 3 cases of de novo JAK2^V617F-positive AML, 2 (cases 23 and 24) demonstrated left-shifted neutrophilia with eosinophilia in peripheral blood smears and showed severe myelofibrosis in bone marrow biopsies, in addition to a marked increase in blasts.

Strikingly, after induction chemotherapy, 15 (62.5% of 24; 75% if 4 patients receiving supportive care are excluded) cases underwent reversion to morphology suggestive of a second chronic phase (Figure 1 and Supplemental Figure 1, A through E), including 2 cases of de novo JAK2^V617F-positive AML (Figure 2). These changes in bone marrow biopsies included hypercellular bone marrow with myeloid hyperplasia and megakaryocytic hyperplasia in 8 of 15 cases (53.3%), megakaryocytic hyperplasia with clustering of atypical megakaryocytes in 4 cases (26.7%; Figure 1, D), and megakaryocytic hyperplasia with moderate to marked myelofibrosis (MF-2 to MF-3) in 8 cases (53.3%; Figure 2, E and F, and Supplemental Figure 1, D and E).

Of 22 cases with flow cytometry performed, 18 (81.8%) demonstrated myeloid differentiation in blasts, 3 (13.6%) showed monocytic or myelomonocytic differentiation, and 1 (4.6%) exhibited erythroid differentiation (acute erythroid leukemia by WHO 5^th edition⁹ ; pure erythroid leukemia by WHO 4^th revision¹⁰ ; case 2).

Chromosome analysis was performed in all 24 cases at the diagnosis of MPN blast phase, with 1 case having failed in harvesting of metaphase cells. Of 23 cases with successful karyotyping, 12 (52.2%; 50%, if the molecular cytogenetic result in the case with failed karyotyping is included in the denominator) cases demonstrated complex abnormalities with or without 7q-/5q-, 6 cases (26.1%) demonstrated either -7/7q- or -5/5q-, excluding those associated with complex karyotypes, 3 cases (13%) showed other abnormalities, and the remaining 3 cases (13%) exhibited normal karyotype. Interphase fluorescence in situ hybridization (FISH) was performed on aspirate samples in 22 cases, and the results were consistent with chromosome analysis for each of the individual cases. Of note, on the single case with failed chromosome analysis (case 23), FISH detected monosomy 7 in 89.5% of the interphase nuclei. Of 3 cases of de novo JAK2^V617F-positive AML, 1 had complex abnormalities with both 5q- and 7q-, 1 had monosomy 7 (by FISH), and the other had a normal karyotype.

Cytogenetic studies, either chromosome analysis, interphase FISH, and/or both tests, were successfully performed on 12 of 15 cases that underwent reversion to a second chronic phase and demonstrated cytogenetic abnormalities confirming persistence of the same clonal process in 8 cases (cases 3, 6, 8, 9, 14, 15, 17, and 20). Two cases (cases 1 and 10) demonstrated a normal karyotype in the second chronic phase, in contrast to the complex abnormalities seen at blast phase, suggesting loss of the blast phase–related clone, compatible with parallel evolution. One case (case 12) demonstrated a clonal karyotypic abnormality with interstitial deletion of 20q at blast phase, but showed isochromosome 17q (loss of 17p, TP53 locus) at the second chronic phase, a finding also consistent with parallel evolution determined by comparative analysis of genomic profiles. The remaining case (case 22) demonstrated normal karyotype at second chronic phase, which was also seen at the preceding blast phase (Supplemental Table 2).

Twenty-two cases had NGS analysis performed on blood or bone marrow samples at the time when MPN blast phase was diagnosed. In the remaining 2 cases without NGS analysis, allele-specific tests for JAK2^V617F and/or AML-related mutations were performed. Of these, 20 cases (87%) demonstrated retention of MPN driver mutations, JAK2^V617F, CALR, or MPL mutation, with variant allele frequency (VAF) ranging from less than 5% to 96% (<5% in 3 cases). Of these, 3 cases had driver mutations other than JAK2^V617F, including concurrent CALR/MPL mutations (VAFs, 47%/44.9% and 60.1%/19.2%, respectively) in 2 cases (cases 8 and 17) and isolated CALR mutation in 1 case (case 12). Of 23 cases with NGS/specific mutation analyses performed at blast phase, all (100%) had mutations other than MPN driver mutations detected, including 12 of 22 cases (54.5%) with 3 or more mutations other than the drivers detected. Among the subtypes of MPN or de novo JAK2^V617F-positive AML, those with a history of ET appeared to have the lowest frequency of multiple mutation hits (14.3% versus 66.7%–100%; Table), although the difference was not statistically significant, probably because of small sample size. Pathogenic mutations of TP53 were detected in 10 of 24 cases (41.7%; NGS in 22 and specific genomic analysis in 2 cases), with a median VAF of 79.2% (range, 61.4%–93.5%). Such high VAFs of TP53 mutations with all cases demonstrating mutant VAFs greater than 50% suggests the possibility of loss of heterozygosity. Although 9 of the TP53 mutated cases were associated with complex karyotypes, the remaining case exhibited a noncomplex change (del(20q)). Other somatic mutations involved TET2 in 10 (45.5%), SRSF2 in 5 (22.7%), ASXL1 in 4 (18.2%), and RUNX1 in 4 (18.2%) of 22 cases with NGS analysis, as well as sporadic occurrences of somatic mutations in other genes. The genes that were mutated included epigenetic regulators in 19 cases, kinase signaling factors in 8 cases, spliceosomes in 5 cases, transcription factors in 9 cases, and tumor suppressors in 12 cases (Figure 3).

Of 15 cases that reverted to a second chronic phase, 11 had paired NGS performed at the diagnosis of second chronic phase, and all demonstrated MPN driver mutations, with a median VAF of 36.9% (range, 10%–56.1%). Of these, 8 cases (53.3%) demonstrated disappearance of mutations seen at blast phase, including mutations in NF1 (100%), FLT3 (100%), RUNX1 (66.7%), TP53 (50%), TET2 (16.7%), NPM1 (100%), IDH2 (100%), and KRAS (100%; Figure 3). All cases but one showed some somatic mutations, at least the driver mutations, identified in blast phase (Figure 5; Table). One case (case 10) demonstrated complete cross of mutation profiles between blast phase and second chronic phase, implying parallel clonal evolution (Figure 4, F). In 4 cases without paired NGS performed at second chronic phase, cytogenetic studies demonstrated some abnormalities detected either at initial chronic phase or blast phase in individual patients, confirming the presence of the same clonal processes.

Of 23 cases with NGS or driver mutation analysis performed, 15 (65.2%) demonstrated high VAFs of MPN driver mutations at blast phase, suggesting a linear evolution of the original MPN clone. Of these, 6 cases had paired NGS performed at second chronic phase and showed persistently high VAFs of driver mutation, confirming the stem-line nature of the driver mutation shared between chronic phase and blast phase (Figure 4, A through C). The remaining 8 cases (34.8%) showed either absence of detectable driver mutations (3 cases) or low level of the mutant allele frequencies (5 cases) at blast phase. Six cases of the latter group had paired NGS performed at second chronic phase, and all demonstrated high VAFs of driver mutations, including 3 (cases 1, 5, and 10) with undetectable drivers at blast phase and reappearance of the driver mutations at second chronic phase (Figure 4, D through F). One case (case 6) did not have NGS performed at blast phase or second chronic phase, but the associated cytogenetic analysis demonstrated the same clonal abnormalities across initial chronic phase, blast phase, and second chronic phase. In that case, it was unclear whether or not the CALR mutation identified at initial chronic phase was retained at the blast phase, although cytogenetic results suggested a clonal evolution (Table and Supplemental Table 2).

All 24 patients had treatment information available. Treatment included azacitadine (or decitabine)/venetoclax in 10 patients, combination chemotherapy in 4 patients (including 2 patients with de novo JAK2^V617F-positive AML), decitabine in 2 patients, hematopoietic stem cell transplantation (HSCT) in 3 patients, clinical trials in 2 patients (patient 12 failed intense chemotherapy before the trial), and palliative/supportive care in 4 patients. Median follow-up was 7.5 months (range, 0–57 months). Of 24 patients, 17 (70.8%) died of disease progression or treatment complications during the follow-up, and the remaining 7 patients (29.2%) were alive at last clinical visit. The median overall survival was 10 months, with a 95% CI of 4.6 to 15.4 months (Supplemental Figure 2, A). When stratified by types of cytogenetic abnormalities, those with complex abnormalities demonstrated significantly poorer clinical outcome than those without (estimated median survivals, 4 months [CI, 0–10 months] versus 35 months [CI, 12.8–57.2 months]; P = .004; Supplemental Figure 2, B). Although TP53 mutations correlated well with occurrence of complex cytogenetic abnormalities, and patients with TP53 mutations showed trends toward poorer clinical outcome than those without (estimated median survivals, 9 months [CI, 1–17 months] versus 10 months [CI, 0–33.5 months]; Supplemental Figure 2, C), the difference was not statistically significant (P = .14), likely because of the small sample size. Of 15 patients who underwent reversion to second chronic phase status after induction of chemotherapy, 9 (60%) died of subsequent blast phase relapse or disease progression. Although the patients who underwent reversion to second chronic phase demonstrated significantly prolonged overall survival (estimated median survival, 16 months, with CI of 6.6–25.4 months) compared with those without second chronic phase (estimated median survival, 1.5 months, with CI of 0.8–2.2 months; P < .001), the survival benefit was not durable for the former group (Supplemental Figure 2, D). For the latter group, most (8 of 9; 89%) of the patients died at the initial phase of chemotherapy. Of 3 patients receiving HSCT, 1 died at 11 months, 1 was censored at 25 months, and the remaining 1 was censored at 57 months after the diagnosis. Both surviving patients showed complete engraftment of donor cells, negative genomic markers detected at blast phase, and normalized complete blood cell counts. There was no statistically significant difference in survival when the analysis was stratified by type of initial MPN or pattern of clonal evolution.

The incidence of blast phase in BCR::ABL1-negative MPN is relatively low according to previous studies, with the estimated risk at 10% to 20% for PMF, 2% to 4% for PV, and ∼1% for ET, with 10-year follow-up.^5,11  Because the application of NGS became routine in the clinical workup of myeloid neoplasms, several genomic studies of MPNs and their blast phase have been published, providing a genomic landscape of clonal evolution in MPN.^6–8,12  Despite these sporadic reports, comprehensive pathologic studies of MPN blast phase, integrating pathologic features, genomic profile, and clinical findings, are still lacking. Here, we report a retrospective analysis of MPN blast phase, with a focus on dynamic changes of clinicopathologic features and genomic events during the disease course, particularly the changes mediated by induction chemotherapy, and clinical outcome in the era of current treatment modalities. In this series, we demonstrated a median interval of 48 months from MPN chronic phase to blast phase, and trends toward a longer latency for ET to develop blast phase compared with PMF, although this finding needs to be confirmed by larger cohorts.

Driver mutations in 3 genes have been identified for the 3 major BCR::ABL-negative MPNs. These include JAK2^V617F, JAK2 exon 12 mutation, 2 types of CALR deletional mutations, and MPL^W515L or MPL^W515K mutations.^3,4,13  Products of all 3 of these genes physiologically involve the JAK-STAT signal transduction pathway in regulating hematopoiesis, and the pathogenic mutations of these genes result in gain of function, leading to a disordered proliferation of hematopoietic elements.⁴  Because these driver mutations are highly associated with BCR::ABL1-negative MPN, detection of the mutations not only provides evidence of myeloid clonality in a patient with elevated blood components, but it also helps to determine the classification of the hematopoietic neoplasia in clinically and morphologically equivocal cases. Furthermore, these driver mutations can be used to monitor the neoplastic evolution during the disease course and following therapy, particularly as NGS becomes more routinely applied in the diagnosis of hematopoietic neoplasms. In this series, approximately two-thirds of the cases showed high allele frequencies of the driver mutations at blast phase, suggesting clonal evolution in a driver-mutated clone (Figure 5, route 2). On the other hand, the remaining one-third of the cases demonstrated absence of the driver mutations or low mutant allele burdens at blast phase, but they showed resurgence of the driver mutant alleles at a second chronic phase. In about half of the latter group, this was accompanied by a decline in blast phase–associated mutations, particularly TP53, NF1, FLT3 mutations, etc, at second chronic phase, giving a line-cross symbol between the driver mutations and blast-associated mutations (Figure 4, D through F). These findings support the notion that mutations governing the blast phase may occur alternatively on a driver wild-type clone that parallels the driver-mutated clone (Figure 5, route 3). Route 3 is in striking contrast to chronic myeloid leukemia (CML), in which the blast phase of the disease characteristically harbors BCR::ABL1, the hallmark genomic events in CML. This may be explained by certain early genomic events preceding MPN driver mutations in MPN,⁸  in contrast to an apparently pluripotent stem cell level of acquiring BCR::ABL1 in CML.^14,15  Lundberg et al,⁸  via genomic analysis of chronologically collected samples, demonstrated clonal architecture of somatic mutations in MPN, indicating a gain of mutations during leukemic transformation. Several somatic mutations in addition to the driver mutations are associated with blast phase, including deleterious changes in TP53, TET2, ASXL1, IDH1/2, etc, with TP53 mutations being frequently observed.^8,13,16  Our study demonstrated high frequencies of somatic mutations in TP53, TET2, ASXL1, and SRSF2, with high allele burdens of TP53 mutations suggestive of loss of heterozygosity, which is in agreement with what has been reported in the literature. In addition, via comparative genomic analysis, we were able to identify somatic mutations that were divided into either blast-related or blast-unrelated. The blast-related genomic events include TP53, NF1, TET2, RUNX1, NPM1, IDH2, KRAS, and FLT3 mutations per our analysis (Figure 3, and Figure 5, routes 2 and 3). Among these, somatic TP53 mutations are frequently associated with blast phase, and thus considered to be major genomic events driving the leukemic transformation. Interestingly, about half of our cases with TP53 mutation exhibited a high VAF at the blast phase and retained a high VAF of the mutation when the disease achieved second chronic phase (Figure 4, A, and Figure 5, dashed line). Although TP53 mutations in these cases appeared to be nonspecific for blasts, the chronologic acquisition of TP53 mutations before the blast phase is uncertain without genomic analysis of sequentially collected samples. Hypothetically, TP53 mutation might occur on a driver-mutated MPN clone without driving it into blast phase, and the mutated clone could enrich with time to replace the TP53 wild-type MPN clone before acquiring another “hit” on a cooperative gene, eventually driving the leukemic transformation. Therefore, TP53 mutation-mediated blast phase might follow different evolution pathways, resulting in 3 types of blast phase: blast-nonspecific/driver mutation–positive (route 1, which overlaps with the genomic profile of chronic phase), blast-specific/driver mutation–positive (route 2), and blast-specific/driver mutation–negative (route 3) blast phase (Figure 5). Of note, several somatic mutations, such as SRSF2, DNMT3A, ASXL1, TET2, etc, appeared to be shared between blast phase and second chronic phase, suggesting the possibility of stem-line genomic events before the driver mutations.^7,8  These early genomic events may create “preneoplastic” stem cells that expand into a repertoire before acquiring either the driver mutation to develop MPN or blast-related mutation to evolve to AML (blast phase, if driver mutation–positive MPN occurs earlier; Figure 5, route 1 and route 3, respectively). This hypothetical evolution model in MPN blast phase is similar to what was reported by Chen et al¹⁷  and Walter et al¹⁸  regarding MDS in transformation to AML using targeted deep sequencing combined with single-cell sequencing.^19,20  Of note, most TET2 mutations appeared to be early events, but in 1 of our cases (case 10), they were apparently blast related (Figure 4, B); therefore, the chronologic occurrence and type of mutations as well as coexisting cooperative mutations may be crucial for the biologic impact.

Previous studies have suggested that different MPN driver mutations are mutually exclusive.³  In contrast, our series demonstrated concurrence of CALR and MPL mutations (cases 8 and 17) in 2 cases and concurrence of JAK2^V617F and MPL in 1 case (case 1). In the former 2 cases, the 2 mutations were detected at blast phase, but NGS was not performed at second chronic phase, and thus the chronologic occurrence of each mutation is unclear. In the latter case, MPL mutation was detected at second chronic phase but was absent at blast phase. In 2 of these cases (cases 1 and 17), MPL mutation had much lower allele burden than the other driver mutation, apparently occurring as late subclone events, whereas the other case (case 8) showed about equal allele burdens of the 2 mutations, suggesting their occurrence on the same allele. The biologic significance of concurrent driver mutations remains to be studied.

It has been demonstrated that the incidence of blast phase is much higher in patients with PMF than in those with PV or ET.¹³  In our series, we noted that approximately three-quarters of the cases with PV or ET developed blast phase intermediated by myelofibrosis, suggesting a critical step in leukemic transformation. Although no sequential analysis of somatic mutations was performed before blast phase in our patients, progressive accumulation of somatic mutations has been observed at the stage of myelofibrosis in PV and ET.⁴ 

Whether de novo JAK2^V617F-positive AML represents MPN blast phase with silent chronic phase has been an issue of debate.^6,21–23  Our series included 3 cases of JAK2^V617F-positive AML without explicit chronic phase. Of these, 1 had a complex karyotype, 1 had monosomy 7, and the other had a normal karyotype. All 3 cases contained multiple mutations in addition to JAK2^V617F, including 1 with TP53 mutation. Two of the cases demonstrated high allele burdens of JAK2^V617F mutation. The cytogenetic pattern and mutation profile appeared similar to those of MPN blast phase, and 2 patients died of the disease, correlating with their high-risk cytogenetic patterns. Interestingly, 2 cases demonstrated pathologic features suggestive of second chronic phase status after induction chemotherapy, including bone marrow morphology and genomic profile with persistent JAK2^V617F mutation. Although the issue may remain controversial, based on our findings, at least a proportion of de novo driver-mutated AML shows features suggestive of MPN blast phase with a clinically silent chronic phase.

A standard guideline for the treatment of MPN blast phase has not been established yet, and the clinical outcome remains generally poor.⁵  Treatment for our patients varied, with most being treated with relatively moderate chemotherapy or even palliative care because of older age and/or comorbidities. Nonetheless, a minority of the patients received intense chemotherapy and/or HSCT. Overall survival was dismal, with a median of 10 months, which is comparable to what has been reported in the literature.⁵  As seen in AML arising from myelodysplastic syndrome,^24,25  the survival of patients with MPN blast phase in our study correlated with the types of cytogenetic abnormalities, with complex changes showing significantly worse outcome. Although the cases with TP53 mutations correlated significantly with complex cytogenetic abnormalities and showed trends toward poorer outcome than the rest, the difference was not statistically significant, likely due to small sample size. Interestingly, about two-thirds of our cases underwent reversion to a second chronic phase status after induction chemotherapy. Although the patients with second chronic phase seemed to have a favorable survival compared with those without, the benefit was not durable, with approximately 63% of them dying from either another blast phase or disease progression. Of note, most of the cases exhibited persistence of extra mutations such as TP53 mutations seen in the blast phase when achieving second chronic phase, evidence of its biologic difference from the initial chronic phase. On the other hand, most of the patients without second chronic phase died at the initial phase of chemotherapy. Notably, of 3 patients receiving HSCT, 1 died at 11 months, 1 was censored at 25 months, and the remaining 1 was censored at 57 months. Therefore, HSCT may benefit this patient population as suggested by our findings and limited studies,⁵  although those receiving HSCT were apparently younger, with their ages averaging in the fifties. Most of our patients would not fulfill this age criteria. Future therapeutic studies may focus on how to maintain the second chronic phase or how to eradicate the driver-mutated stem-line clone at chronic phase by identifying more targetable pathogenic molecules, and thus effectively prevent the development of blast phase.

In conclusion, MPN blast phase resembles AML, myelodysplasia-related, in cytogenetic pattern, mutation profile, and clinical outcome. Somatic mutation in TP53 and subsequent loss of heterozygosity is a major “hit” causing genomic instability and driving leukemic transformation in MPN. Two patterns of clonal evolution are inferred by dynamic analysis of mutation profiles: linear and parallel clonal evolutions. Although overall survival was dismal, 62.5% of our cases underwent reversion to a second chronic phase status after induction chemotherapy, and showed better survival than those without second chronic phase. HSCT likely provides the best chance of cure or survival benefit, although the number of patients qualified for this procedure is limited by older age and comorbidities. Given the similar genomic profile and treatment response, MPN driver mutation–positive AML may represent MPN blast phase with a clinically silent chronic phase.