Clinicopathologic Correlates and Natural History of Atypical Chronic Myeloid Leukemia

INTRODUCTION: Atypical chronic myeloid leukemia (aCML) is a rare subtype of myelodysplastic/myeloproliferative neoplasms (MDS/MPN) associated with shorter survival and higher risk of transformation to acute myeloid leukemia (AML) than other MDS/MPNs. However, the clonal mechanisms underlying transformation to leukemia remain unclear. There is a need to develop predictive models and identify the optimal therapeutic management of these pts.

METHODS: We evaluated all consecutive pts with aCML treated at the University of Texas MD Anderson Cancer Center from 2005 to 2020. Whole bone marrow (BM) DNA was subject to 28 or 81 gene targeted next-generation sequencing (NGS) analysis in a subset of pts. Variant allele frequency (VAF) estimates were used to evaluate clonal relationships within each sample using Pearson goodness-of-fit tests and VAF differences. Response to therapy was assessed following MDS/MPN IWG response criteria. Cox proportional hazards regression was used to study association of variables with survival.

RESULTS: A total of 65 pts were identified. Median age was 67 years (range 46-89). Median WBC, Hgb and platelets were 44.5x10⁹/L (5.9-474.9x10⁹/L) 10.0g/dL (5.7-14.7g/dL) 93x10⁹/L (12-560x10⁹/L), respectively. Median neutrophil, promyelocyte, myelocyte and metamyelocyte percentages were 64%, 0%, 0% and 16%. Forty-one (63%) pts had normal karyotype, 5 (8%) trisomy 8, 2 (3%) i(17q) and 2 (3%) del(20q). Splenomegaly was observed in 26 (40%) pts and 7 (11%) had extramedullary disease. NGS data was available in 35 (54%) pts. The most frequently mutated genes included ASXL1 in 83%, SRSF2 in 68% and SETBP1 in 58%. Frequency and VAF of identified mutations is shown in Figure 1A. Mutations in SETBP1, SRSF2, TET2 and GATA2 tended to appear within dominant clones while other RAS pathway mutations were more likely to appear as minor clones. SRSF2 and SETBP1 tended to be co-dominant while ASXL1 appeared within minor clones in up to 50% of pts (Figure 1B). Therapy consisted of single agent hypomethylating agent (HMA) in 19 (29%), hydroxyurea in 8 (12%), HMA in combination with ruxolitinib in 7 (11%), other HMA combinations in 5 (8%), ruxolitinib single agent in 5 (8), induction chemotherapy in 3 (5%) and other investigational agents in 1 (2%) pts. Response outcomes by therapy are detailed in Figure 1C. With a median follow up of 35.6 months (95% CI 28.2-43.1) 18 (28%) of pts experienced transformation to AML within a median of 18 months (1-123 months). Median survival after transformation of 8.3 months (95% CI 5.5-11.0 months). NGS at the time of transformation was available in 12 (67%) pts with matched NGS at diagnosis of aCML and AML in 8 (44%) pts. Acquisition of new previously undetectable mutations was observed in 5 pts the most common involving signaling pathway mutations (Figure 1D). Acquisition of new cytogenetic abnormalities was observed in 9/14 pts (Figure 1D) the most frequent involving i(17q). The median OS was 25 months (95% CI 20.0-30.0) with pts who received intensive chemotherapy having significantly worse OS than those receiving HMA-based therapy or other agents such as ruxolitinib or hydroxyurea (p=0.012, Figure 1E). By multivariate analysis for survival, age, platelet count, BM blast percentage and serum LDH levels influenced prognosis. Based on these factors we developed a multivariable Cox model to generate a nomogram which assigned a score to each of the prognostic variables and allowed to predict 1-year and 3-year OS based on the total score among all prognostic variables (Figure 1F).

CONCLUSIONS: aCML is characterized by high frequency of co-dominant SRSF2 and SETBP1 mutations. HMA therapy is associated with the best response outcomes. Clinicopathological features can help predict outcomes of these pts.