Molecular Profiling in Extramedullary Acute Myeloid Leukemia Reveals Distinct Subgroups Defined By Mutations in KIT and RUNX1

Background: Between 2 to 23% of patients with acute myeloid leukemia manifest extramedullary involvement including myeloid sarcoma. While the genomic landscape of AML and the respective clinical and outcome correlations are well characterized, our understanding of the molecular underpinnings of extramedullary AML remains limited. Various cytogenetic abnormalities have been described in association with extramedullary AML, including core binding factor (CBF) translocations t(8;21) and inv(16). Recent studies have also applied targeted gene sequencing approaches (21 to 39 genes) in small patient cohorts (6-18) to generate limited mutational profiles of extramedullary AML [Li et al. Leukemia 2015; Pastoret et al. Leukemia & Lymphoma 2016; Kashofer et al. Leukemia & Lymphoma]. Here we utilize a comprehensive next generation sequencing (NGS) panel interrogating 585 genes implicated in hematological malignancies in 17 patients and provide the most comprehensive mutational profile of extramedullary AML to date.

Methods: 17 pathologically confirmed cases of extramedullary AML were identified. Genomic DNA extraction was performed from formalin-fixed, paraffin-embedded (FFPE) tumor samples. Isolated DNA was sent for sequencing using the HemePACT panel, a custom hybridization capture-based NGS assay encompassing 585 genes related to hematologic malignancies. All HemePACT data were compared to an unmatched normal sample. Data were passed to standard pipelines used to detect candidate substitutions and indels (Caveman and pindel respectively). Candidate variants were manually reviewed to eliminate single nucleotide polymorphisms and artifacts and to find a subset of high-confidence putative oncogenic variants. Cytogenetic data was obtained from standard clinical karyotyping and fluorescence in-situ hybridization assays from bone marrow aspirates from patients with bone marrow AML involvement.

Results: Demographic data for sequenced extramedullary AML patients is provided in Table 1. We identified a total of 80 mutations. The median number of mutations /sample is 4 (range:1-11). The most frequently mutated genes are KIT found in 7/17 patients (41%) and RUNX1 found in 6/17 (35%) patients. Intriguingly, KIT and RUNX1 mutations are mutually exclusive(p<0.05), and 13/17 (76.5%) sequenced extramedullary AML samples have mutations in either KIT or RUNX1 . (Figure 1) DNMT3A mutations are found in 3/17 (17%) patients. FLT3-tyrosine kinase domain (TKD) mutations are found in 4/17 (23%) of patients. DNMT3A and FLT3-TKD mutations co-segregate with RUNX1 (2/3 and 2/4 respectively) but not with KIT mutations. As only 2/17 patients underwent cytogenetic evaluation from extramedullary AML, mutational data from extramedullary AML was largely integrated with cytogenetic data from bone marrow aspirates. 7/17 (41.1%) of sequenced patients had CBF translocations (6/7 with inv(16)/t(16;16), and 1/7 with t(8;21) present in bone marrow cytogenetic evaluation by karyotype or FISH. Out of 7 patients with CBF translocations, 6/7 (85.7%) have KIT mutations present in extramedullary AML. 1 patient with inv(16) found by karyotyping from myeloid sarcoma did not have a KIT mutation in extramedullary AML. No patients with RUNX1 mutations in extramedullary AML (0/6) have CBF translocations. Kaplan-Meier analysis from this small cohort did not show a statistically significant difference in overall survival between patients with KIT and RUNX1 mutations (p=0.57).Overall, our data suggest the potential for distinct and specific mutational profiles associated with AML extramedullary involvement.

Conclusions: Our study provides the most comprehensive mutational profiling of extramedullary AML to date. Combined molecular and cytogenetic profiling demonstrate that extramedullary AML is characterized by a distinct repertoire of acquired mutations enriched for mutations in KIT and RUNX1. Importantly our data identify two distinct and non-overlapping molecular subgroups. One subgroup is defined by CBF alterations and KIT mutations, while the second most prominent subtype is defined by mutations in RUNX1. Recognition of these subtypes and their co-mutations may form the basis for genotype-clinical correlations, as well as experimental models to explore the biology of extramedullary AML. Additional extramedullary AML patient samples are being sequenced and will be presented.

View largeDownload slide

View largeDownload slide

Close modal