-Author name in bold denotes the presenting author
-Asterisk * with author name denotes a Non-ASH member
Clinically Relevant Abstract denotes an abstract that is clinically relevant.

PhD Trainee denotes that this is a recommended PHD Trainee Session.

Ticketed Session denotes that this is a ticketed session.

1534 Identification of Prognostic Signatures and Potential New Therapies in NPM1 Mutant Adult Acute Myeloid Leukemia Patients Using Proteomics

Program: Oral and Poster Abstracts
Session: 618. Acute Myeloid Leukemias: Biomarkers and Molecular Markers in Diagnosis and Prognosis: Poster I
Hematology Disease Topics & Pathways:
AML, Adult, Acute Myeloid Malignancies, Research, Translational Research, Chemotherapy, Diseases, Non-Biological therapies, Myeloid Malignancies, Technology and Procedures, Study Population, Human, Omics technologies
Saturday, December 7, 2024, 5:30 PM-7:30 PM

Eduardo Sabino De Camargo Magalhaes, MD, PhD1*, Brandon D. Brown, MD2*, Yihua Qiu, MD3* and Steven M. Kornblau, MD3

1Department of Ageing Biology/ERIBA, University Medical Center Groningen, Groningen, Netherlands
2Department of Pediatrics, MD Anderson Cancer Center, Houston, TX
3Department of Leukemia, The University of Texas MD Anderson Cancer Center, Houston, TX

INTRODUCTION: NPM1 mutations are common in adult Acute Myeloid Leukemia (AML), but currently there is no targeted-based therapy for it, and many patients relapse. We analyzed the proteome of NPM1 mutant AML patients to identify prognostic signatures and new therapeutic opportunities.

METHODS: Reverse Phase Protein Arrays were used to measure the expression of 429 proteins (339 total, 90 post-translational modified) in fresh samples of 805 newly diagnosed AML patients. Protein expression was normalized to normal CD34+ cells. We selected 112 patients with NPM1 mutations, which were treated with either AraC-based (N=78) or Hypomethylating agent-based (HMA, N=34) therapies. Individually prognostic proteins were identified with Uni-(UV) and Multi-variate (MV) Cox proportional hazards models (CoxPH) for Overall Survival (OS) and Remission Duration (RD). Proteins that showed best cluster separation with unbiased hierarchical clustering were selected for further analyses. Variables were compared with Kruskal-Wallis (KS) tests or Fisher’s Exact tests with simulated p-values (10000 replicates) whenever applied. CoxPH models were generated for UV and MV analysis of cluster prognosis. Differentially expression analysis was performed with Wilcoxon tests with p-values adjusted with FDR (p<0.01) and a Log2-fold-change cutoff of 1.

RESULTS: 34 proteins were prognostic for OS, and formed 4 patient clusters (C1-C4). The biological processes related to those proteins were: epigenetics, DNA damage response and cell cycle, heatshock proteins, apoptosis and autophagy, growth and metabolic regulation, ribosomal and transcriptional activity, adhesion and cytoskeleton, and other signaling transduction pathways, respectively with 6, 6, 5, 4, 3, 3, 2 and 5 proteins. Clinical and molecular features were not biased towards any cluster, except for: age, with younger patients in C1 (median=55yo) and older in C2 (median=65yo); white blood cell count and percentage of blasts, both higher in C3; trisomy 8, higher in C4; M1 cases less common and M5 ones more frequent in C1; and more Venetoclax therapy in C2. Regarding NPM1 co-segregating mutations, ASLX1 were more common in C4, and IDH1 less frequent in C1.

The OS and RD of clusters were biased towards therapy, with C1, C2 and C3 clusters doing better with AraC-based therapy (median OS for C1, C2, C3 +AraC: >120mo, >108mo, =39.6mo, respectively vs +HMA: 8.5mo, 11.5mo, 3.6mo; median RD for C1, C2, C3 +AraC: >120mo, >120mo, >108mo, respectively vs +HMA: 6.8mo, 24mo, 3mo), and C4 patients having a superior outcome with HMA-based therapy (OS/RD C4: +HMA 84mo/48mo vs. +AraC 16mo/8mo. OS p<0.0001, RD p<0.0001). The same OS pattern was observed when the 4 clusters were filtered by NPM1+FLT3-ITD mutation or in 41-55yo patients. Regarding RD, filtering on caucasians, primary AML, intermediary cytogenetic risk and diploid karyotype showed a similar pattern.

In CoxPH models, clusters with similar prognosis were grouped, generating 3 prognostic groups (gp1-3) instead of 8. Considering gp1 as reference group of clusters (ref), all groups were predictive for OS in the UV analysis (HR=1, 5.2, 16.5; p=ref, 0.025, <0.001), along with age, secondary AML, and ASLX1, CEBPA, IDH1 and KIT mutations. In the MV analysis, all groups (HR=1, 4.3, 9.8, 4.6; p=ref, 0.049, 0.003) and secondary AML predicted OS. Regarding RD, only gp1 and gp3 were predictive in the UV model (HR=1, 4.3, 23; p=ref, 0.17, 0.002), as well as, hemoglobin levels, and ASLX1 and DNMT3a mutations. Similarly, gp1 and gp3 were predictive in the MV model (HR=1, 3.3, 18.8; p=ref, 0.28, 0.009), along with DNMT3a mutation.

Differentially expression analysis comparing clusters identified 24 proteins, among 429, which may be potential therapeutic targets: AKT1_2_3.pT308, ITGAL and SYK for C1; NOTCH1.cle and TGM2 for C2; CASP3 and FN1 for C3; and EZH2, BIRC5 and CHEK1 for C4. Of note, SYK inhibitors, such as entospletinib, and EZH2 inhibitors, such as romidepsin, are being evaluated for myeloid malignancies in clinical trials.

CONCLUSION: Proteomics in NPM1 mutant AML identified expression profiles that benefit from AraC-based treatment, while others do better with HMA-based therapies. Protein signatures are independently prognostic in UV/MV analysis, and could triage patients according to treatment. Moreover, new therapeutic targets that could guide additional therapy to improve outcomes were unraveled.

Disclosures: No relevant conflicts of interest to declare.

*signifies non-member of ASH