AUTHORS
Andrew J. Menssen, Ph.D., HCLD(ABB), Amenech B. Ashango2, Todd A. Alonzo, PhD3, Robert B. Gerbing, Ph.D4, Maria Pardo5, Fan-Chi Hsu, PhD6, Loren L. Lott7, Fangyan Dai8, Denise A. Wells, MD9, Todd Cooper I, DO10, Jessica A. Pollard I, MD11, Katherine Tarlock I, MD12, Soheil Meshinchi, MD, PhD13, Richard Aplenc I, MD, PhD14, Lisa Eidenschink Brodersen, PhD15, Michael Loken, PhD16, Chad A. Hudson, MD, Ph.D17
AFFILIATIONS
1 Hematologics. Inc., Seattle, Washington, United States
2 Hematologics. Inc., Seattle, Washington, United States
3 Children’s Oncology Group, Arcadia, California, United States, University of Southern California, Los Angeles, California, United States
4 Children’s Oncology Group, Arcadia, California, United States
5 Fred Hutchinson Cancer Center, Seattle, Washington, United States, Hematologics. Inc., Seattle, Washington, United States
6 Hematologics. Inc., Seattle, Washington, United States
7 Hematologics. Inc., Seattle, Washington, United States
8 Hematologics. Inc., Seattle, Washington, United States
9 Hematologics. Inc., Seattle, Washington, United States
10 Seattle Children's Hospital, Seattle, Washington, United States
11 Dana-Farber Cancer Institute, Boston, Massachusetts, United States
12 Seattle Children's Hospital, Seattle, Washington, United States
13 Fred Hutchinson Cancer Center, Seattle, Washington, United States
14 Children's Hospital of Philadelphia, Philadelphia, Pennsylvania, United States
15 Children's Hospital of Philadelphia, Philadelphia, Pennsylvania, United States, University of Pennsylvania, Philadelphia, Pennsylvania, United States
16 Hematologics. Inc., Seattle, Washington, United States
17 Hematologics. Inc., Seattle, Washington, United States
ABSTRACT
Introduction: Pediatric acute myeloid leukemia (AML) is a heterogeneous disease characterized by diverse genotypic and phenotypic characteristics with variable clinical outcomes. To better understand the impact of phenotype in AML we analyzed diagnostic samples from the Children’s Oncology Group AAML1031 trial and identified immunophenotypic clusters based on antigen expression. We hypothesized that comparing clusters with genetic drivers, clinical features, and outcomes would reveal additional, and enhance understanding of existing, subtypes of pediatric AML.
Methods: Children with de novo AML enrolled in AAML1031 who submitted diagnostic bone marrow specimens were included (n=1034). AML antigen expression was assessed by multidimensional flow cytometry and quantified as previously published. 14 antigens plus side scatter were used for unsupervised hierarchical clustering and clusters were identified by blinded manual review.
Results: Ten distinct AML subgroups (A-J) of varying size (range: 11-321 patients) were identified. Clusters A-F were broadly CD34 positive while G-J were largely CD34 negative. Additional defining features included general immaturity (cluster A), elevated CD123/CD64 without CD117 (B), increased CD13/CD56 (C), monocytic antigens (D and J), decreased HLA-DR/CD38 (E), RAM phenotype (F), decreased HLA-DR with dim CD117 (G), and monocytic with (I) or without (H) CD56 and CD117.
Comparison to clinical characteristics revealed distinct features. Clusters A, C, and E demonstrated increased age (median: 12.1, 12.5, and 12.3 years, respectively) while B, D, and F were younger (2.0, 1.7, and 1.4). WBC counts were elevated in C, E, and J (30.6, 35.5, and 36.8 ×10³/µL). MRD varied, with D, H, I, and J demonstrating lower rates post induction 1 (11%, 6%, 20% and 15%) while E and F had higher rates (45% and 86%).
Comparison of genetic drivers to clusters revealed numerous distinct associations. Among these were NUP98 fusions, which were enriched in clusters A (44% total NUP98 patients) and G (40%) and showed fusion-specific segregation. 30/43 NUP98::NSD1 (NSD1) cases were in A while all 13 NUP98::KDM5A cases were in G along with the remaining NSD1 cases, though in entirely separate subclusters. Review of NSD1 cases showed that CD34 expression drove separation (A vs. G). Cluster A also had a predominance of CEBPA mutations (77% CEBPA mutated patients) and FLT3-ITD (Allelic Ratio >0.1) (39% FLT3-ITD) which like NUP98 was also enriched in G (45%). FLT3-ITD rarely co-occurred with CEBPA (9/52 CEBPA mutated) but frequently with NSD1 (44% of FLT3-ITD, 52% of NSD1 in cluster A; 16%, 67% respectively in G) and NPM1 (60% of FLT3-ITD in G).
Given the co-occurrence of mutations and similar cluster localization we compared outcomes of patients with FLT3-ITD with or without NPM1 or NSD1, all of which were high risk by trial protocol. Patients in cluster A with FLT3-ITD and NSD1 had a trend towards decreased EFS (19%) compared to cluster A patients with FLT3-ITD alone (37%, p=0.1) but similar OS (56% vs. 58%). In cluster G, cases with FLT3-ITD and NPM1 mutations (EFS and OS: 79% and 85%) had a trend towards improved outcomes compared to cluster G FLT3-ITD alone (EFS and OS: 54% and 63%, p=0.09 and 0.16) as well as improved outcomes compared to all FLT3-ITD patients in cluster A, regardless of NSD1 fusion status. The improved outcomes of patients with FLT3-ITD alone in cluster G vs. cluster A raises the question about a possible role of CD34 expression in outcome for these patients.
Conclusions: Unsupervised hierarchical clustering of AAML1031 revealed subgroups with distinct immunophenotypic and clinical features. Analysis of prognostic genetic drivers demonstrated strong relationships with individual clusters. As discussed here, FLT3-ITD and NUP98 rearrangements were enriched in clusters A and G, and NPM1 mutations in cluster G. Furthermore, patients with similar co-occurring mutations clustered together, both between and within clusters. A trend of improving outcomes in FLT3-ITD patients was observed, with FLT3-ITD/NUP98::NSD1 patients doing the worst, followed by cluster A FLT3-ITD, then cluster G FLT3-ITD, and finally FLT3-ITD/NPM1. This finding suggests that phenotype may impact outcome in patients with some genetic lesions, as well as a possible reevaluation of the prognostic impact of co-occurring mutations in the setting of FLT3-ITD, specifically patients with co-occurring NPM1.
DISCLOSURES
Andrew J. Menssen, Ph.D., HCLD(ABB): Hematologics Inc - Current Employment