Survival In Acute Myeloid Leukemia Research Articles

Integration of measurable residual disease by WT1 gene expression and flow cytometry identifies pediatric patients with high risk of relapse in acute myeloid leukemia.

Molecular testing plays a pivotal role in monitoring measurable residual disease (MRD) in acute myeloid leukemia (AML), aiding in the refinement of risk stratification and treatment guidance. Wilms tumor gene 1 (WT1) is frequently upregulated in pediatric AML and serves as a potential molecular marker for MRD. This study aimed to evaluate WT1 predictive value as an MRD marker and its impact on disease prognosis. Quantification of WT1 expression levels was analyzed using the standardized European Leukemia Network real-time quantitative polymerase chain reaction assay (qRT-PCR) among a cohort of 146 pediatric AML patients. Post-induction I and intensification I, MRD response by WT1 was assessed. Patients achieving a ≥2 log reduction in WT1MRD were categorized as good responders, while those failing to reach this threshold were classified as poor responders. At diagnosis, WT1 overexpression was observed in 112 out of 146 (76.7%) patients. Significantly high levels were found in patients with M4- FAB subtype (p=0.018) and core binding fusion transcript (CBF) (RUNX1::RUNX1T1, p=0.018, CBFB::MYH11, p=0.016). Following induction treatment, good responders exhibited a reduced risk of relapse (2-year cumulative incidence of relapse [CIR] 7.9% vs 33.2%, p=0.008). Conversely, poor responders' post-intensification I showed significantly lower overall survival (OS) (51% vs 93.2%, p<0.001), event-free survival (EFS) (33.3% vs 82.6%, p<0.001), and higher CIR (66.6% vs 10.6%, p<0.001) at 24 months compared to good responders. Even after adjusting for potential confounders, it remained an independent adverse prognostic factor for OS (p=0.04) and EFS (p=0.008). High concordance rates between WT1-based MRD response and molecular MRD were observed in CBF patients. Furthermore, failure to achieve either a 3-log reduction by RT-PCR or a 2-log reduction by WT1 indicated a high risk of relapse. Combining MFC-based and WT1-based MRD results among the intermediate-risk group identified patients with unfavorable prognosis (positive predictive value [PPV] 100%, negative predictive value [NPV] 85%, and accuracy 87.5%). WT1MRD response post-intensification I serves as an independent prognostic factor for survival in pediatric AML. Integration of WT1 and MFC-based MRD results enhances the reliability of MRD-based prognostic stratification, particularly in patients lacking specific leukemic markers, thereby influencing treatment strategies.

T cell-mediated tumor killing sensitivity gene signature-based prognostic score for acute myeloid leukemia.

Acute myeloid leukemia (AML) is an aggressive, heterogenous hematopoetic malignancies with poor long-term prognosis. T-cell mediated tumor killing plays a key role in tumor immunity. Here, we explored the prognostic performance and functional significance of a T-cell mediated tumor killing sensitivity gene (GSTTK)-based prognostic score (TTKPI). Publicly available transcriptomic data for AML were obtained from TCGA and NCBI-GEO. GSTTK were identified from the TISIDB database. Signature GSTTK for AML were identified by differential expression analysis, COX proportional hazards and LASSO regression analysis and a comprehensive TTKPI score was constructed. Prognostic performance of the TTKPI was examined using Kaplan-Meier survival analysis, Receiver operating curves, and nomogram analysis. Association of TTKPI with clinical phenotypes, tumor immune cell infiltration patterns, checkpoint expression patterns were analysed. Drug docking was used to identify important candidate drugs based on the TTKPI-component genes. From 401 differentially expressed GSTTK in AML, 24 genes were identified as signature genes and used to construct the TTKPI score. High-TTKPI risk score predicted worse survival and good prognostic accuracy with AUC values ranging from 75 to 96%. Higher TTKPI scores were associated with older age and cancer stage, which showed improved prognostic performance when combined with TTKPI. High TTKPI was associated with lower naïve CD4 T cell and follicular helper T cell infiltrates and higher M2 macrophages/monocyte infiltration. Distinct patterns of immune checkpoint expression corresponded with TTKPI score groups. Three agents; DB11791 (Capmatinib), DB12886 (GSK-1521498) and DB14773 (Lifirafenib) were identified as candidates for AML. A T-cell mediated killing sensitivity gene-based prognostic score TTKPI showed good accuracy in predicting survival in AML. TTKPI corresponded to functional and immunological features of the tumor microenvironment including checkpoint expression patterns and should be investigated for precision medicine approaches.

Abstract 6471: Trends in chemotherapy utilization in acute myeloid leukemia (AML) in last 15 years

Abstract Introduction: Chemotherapy (Chemo) can relieve symptoms and improve quality of life and survival in AML, including in many older and frail adults. However, in a prior study from the US, almost a quarter of patients (pts) with new diagnoses of AML between years (yrs) 2003 and 2011 did not receive chemo (Bhatt et al., Blood Advances, 2018). Our study examined trends in chemo utilization in the last 15 yrs to test the hypothesis that chemo use has increased in more recent yrs. We also examined any disparities in chemo use. Methods: We utilized the National Cancer Database to identify 78,864 pts diagnosed with AML between 2004 and 2019 and divided them into 2 cohorts based on date of diagnosis, 2004-2010 and 2011-2019. We used logistic regression to estimate effect of demographic and disease characteristics on odds of receiving chemo and evaluate differences in these effects between the constructed patient cohorts. Results: Seventy-seven percent of total pts received chemo. Compared to 73% in 2004, 81% of pts received chemo in 2019. The receipt of chemo declined with increasing age with higher use of chemo in all groups in 2011-2019 compared to 2004-2010: &amp;gt;80 (yrs)= 40 vs 33%, 71-80 yrs= 72 vs 63%, 60-70 yrs= 87 vs 84%, 41-59 yrs= 93 vs 92%, and 18-40 yrs = 95 vs 94%.The receipt of chemo was lowest in white pts overall; in black pts, receipt of chemo increased in 2011-2019 compared to 2004-2010: black= 83 vs 76%, white= 78 vs 75%, and others 83 vs 83%. The receipt of chemo was lowest in therapy-related (t)-AML in 2011-2019 and 2004-2010: t-AML 64 vs 60%, acute promyelocytic leukemia= 88 vs 87%, core binding factor AML 93 vs 88% and other 77 vs 74%. In multivariable analysis, the odds of receiving chemo increased in 2011-2019 compared to 2004-2010 in all pts, particularly in older adults. Compared to white, black pts had similar odds of receiving chemo in 2004-2010, but odds were higher in 2011-2019 (Odds ratio, 1.2, 95% confidence interval, 1.1-1.3. Pts with t-AML were less likely to receive chemo compared to all other subtypes in 2004-2010 and 2011-2019. Female sex, lower income, public insurance, higher comorbidity score, and shorter distance to treatment facility were associated with a lower likelihood of receiving chemo; however, chemo utilization did not differ over the two time periods based on these variables. Conclusion: To our knowledge, this is the largest scale analysis of chemo utilization in AML in real-world practice. Chemo receipt increased significantly over the yrs, particularly among older adults, black pts, core binding factor, and t-AML. These recent trends are encouraging and represent novel findings. Despite these results, healthcare disparity based on sex, socioeconomic, and healthcare system factors continues to persist. In addition to developing safer and more effective chemo regimens, other innovative strategies to tackle these disparities are crucial to improve outcomes. Citation Format: Prajwal Dhakal, Bradley Loeffler, Aditya Ravindra, Avantika Pyakuryal, Vijaya Raj Bhatt. Trends in chemotherapy utilization in acute myeloid leukemia (AML) in last 15 years [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2024; Part 1 (Regular Abstracts); 2024 Apr 5-10; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2024;84(6_Suppl):Abstract nr 6471.

CLEC11A methylation is correlated to AML subtypes and cytogenetic risk factors but not patient demographics.

Acute myeloid leukemia (AML) is an aggressive and lethal cancer of the blood, which leads to the death of over 11,000 patients in the United States each year. Research on identifying, characterizing, and treating AML is crucial in the fight against this deadly disease. Recent studies have examined the role of CLEC11A in cancer, including AML. However, there have been conflicting reports related to tumor progression and survival. Because survival is based on a variety of factors, including classification of the tumor, genetic risk factors, and demographics, it is imperative that we determine what role CLEC11A may have in cancer survival. Therefore, utilizing data from the Genomic Data Commons, we analyzed CLEC11A methylation in 108 AML patients compared to FAB classification, cytogenetic risk factors, age, race, and gender. Our results show statistically significant correlations between methylation of CLEC11A and FAB classification as well as poor genetic risk factors. However, no difference was observed in CLEC11A methylation when compared to demographic data. Our results, matched with a known biological function of CLEC11A in early hematopoiesis, indicate that CLEC11A may be an important marker for AML diagnosis and prognosis and provide relevant data in the ongoing search for novel therapeutics to improve AML survival.

Prognostic impact of cooccurring mutations in FLT3-ITD pediatric acute myeloid leukemia

AbstractWe sought to define the cooccurring mutational profile of FLT3-ITD–positive (ITDpos) acute myeloid leukemia (AML) in pediatric and young adult patients and to define the prognostic impact of cooperating mutations. We identified 464 patients with FLT3-ITD mutations treated on Children’s Oncology Group trials with available sequencing and outcome data. Overall survival, event-free survival (EFS), and relapse risk were determined according to the presence of cooccurring risk stratifying mutations. Among the cohort, 79% of patients had cooccurring alterations across 239 different genes that were altered through mutations or fusions. Evaluation of the prognostic impact of the cooccurring mutations demonstrated that patients with ITDpos AML experienced significantly different outcomes according to the cooccurring mutational profile. Patients with ITDpos AML harboring a cooccurring favorable-risk mutation of NPM1, CEBPA, t(8;21), or inv(16) experienced a 5-year EFS of 64%, which was significantly superior to of 22.2% for patients with ITDpos AML and poor-risk mutations of WT1, UBTF, or NUP98::NSD1 as well to 40.9% for those who lacked either favorable-risk or poor-risk mutation (ITDpos intermediate; P < .001 for both). Multivariable analysis demonstrated that cooccurring mutations had significant prognostic impact, whereas allelic ratio had no impact. Therapy intensification, specifically consolidation transplant in remission, resulted in significant improvements in survival for ITDpos AML. However, patients with ITDpos/NUP98::NSD1 continued to have poor outcomes with intensified therapy, including sorafenib. Cooccurring mutational profile in ITDpos AML has significant prognostic impacts and is critical to determining risk stratification and therapeutic allocation. These clinical trials were registered at www.clinicaltrials.gov as NCT00002798, NCT00070174, NCT00372593, and NCT01371981.

RNA-binding protein RBM5 plays an essential role in acute myeloid leukemia by activating the oncogenic protein HOXA9

BackgroundThe oncogenic protein HOXA9 plays a critical role in leukemia transformation and maintenance, and its aberrant expression is a hallmark of most aggressive acute leukemia. Although inhibiting the upstream regulators of HOXA9 has been proven as a significant therapeutic intervention, the comprehensive regulation network controlling HOXA9 expression in leukemia has not been systematically investigated.ResultsHere, we perform genome-wide CRISPR/Cas9 screening in the HOXA9-driven reporter acute leukemia cells. We identify a poorly characterized RNA-binding protein, RBM5, as the top candidate gene required to maintain leukemia cell fitness. RBM5 is highly overexpressed in acute myeloid leukemia (AML) patients compared to healthy individuals. RBM5 loss triggered by CRISPR knockout and shRNA knockdown significantly impairs leukemia maintenance in vitro and in vivo. Through domain CRISPR screening, we reveal that RBM5 functions through a noncanonical transcriptional regulation circuitry rather than RNA splicing, such an effect depending on DNA-binding domains. By integrative analysis and functional assays, we identify HOXA9 as the downstream target of RBM5. Ectopic expression of HOXA9 rescues impaired leukemia cell proliferation upon RBM5 loss. Importantly, acute protein degradation of RBM5 through auxin-inducible degron system immediately reduces HOXA9 transcription.ConclusionsWe identify RBM5 as a new upstream regulator of HOXA9 and reveal its essential role in controlling the survival of AML. These functional and molecular mechanisms further support RBM5 as a promising therapeutic target for myeloid leukemia treatment.

A Comprehensive Metabolism-Related Gene Signature Predicts the Survival of Patients with Acute Myeloid Leukemia.

(1) Background: Acute myeloid leukemia (AML) is a clonal malignancy with heterogeneity in genomics and clinical outcome. Metabolism reprogramming has been increasingly recognized to play an important role in the leukemogenesis and prognosis in AML. A comprehensive prognostic model based on metabolism signatures has not yet been developed. (2) Methods: We applied Cox regression analysis and the least absolute shrinkage and selection operator (LASSO) normalization to establish a metabolism-related prognostic gene signature based on glycolysis, fatty acid metabolism, and the tricarboxylic acid cycle gene signatures. The Cancer Genome Atlas-Acute Myeloid Leukemia-like (TCGA-LAML) cohort was set as the training dataset for model construction. Three independent AML cohorts (GSE37642, GSE10358, and GSE12417) combined from Gene Expression Omnibus (GEO) datasets and the Beat-AML dataset were retrieved as two validation sets to test the robustness of the model. The transcriptome data and clinic information of the cohorts were enrolled for the analysis. (3) Results: Divided by the median value of the metabolism risk score, the five-year overall survival (OS) of the high-risk and low-risk groups in the training set were 8.2% and 41.3% (p < 0.001), respectively. The five-year OS of the high-risk and low-risk groups in the combined GEO cohort were 25.5% and 37.3% (p = 0.002), respectively. In the Beat-AML cohort, the three-year OS of the high-risk and low-risk groups were 16.2% and 40.2% (p = 0.0035), respectively. The metabolism risk score showed a significantly negative association with the long-term survival of AML. Furthermore, this metabolism risk score was an independent unfavorable factor for OS by univariate analysis and multivariate analysis. (4) Conclusions: Our study constructed a comprehensive metabolism-related signature with twelve metabolism-related genes for the risk stratification and outcome prediction of AML. This novel signature might contribute to a better use of metabolism reprogramming factors as prognostic markers and provide novel insights into potential metabolism targets for AML treatment.

Establishment and validation of a prognostic immune-related lncRNA risk model for acute myeloid leukemia.

Acute myeloid leukemia (AML) is a cancer arising in the bone marrow and is the most common type of adult leukemia. AML has a poor prognosis, and currently, its prognosis evaluation does not include immune status assessment. This study established an immune-related long non-coding RNA (lncRNA) prognostic risk model for AML based on immune lncRNAs screening. To construct training and validation cohorts, The Cancer Genome Atlas (TCGA) and Gene Expression Omnibus (GEO) public databases were accessed to obtain gene expression profiles and clinical data. The correlation between lncRNAs and immunity genes was analyzed using the "limma" package, and the immune-related lncRNAs were obtained. Through least absolute shrinkage and selection operator regression, a prognostic model was established with immune-related lncRNAs. Using the median risk score, patients were divided into high- and low-risk groups. The Kaplan-Meier method was used for survival analysis, whereas the accuracy of the risk model was evaluated using time-dependent receiver operating characteristic curves, risk score distribution, survival status, and risk heat maps. We utilized univariate and multivariate Cox regression to examine the association between risk score and clinical variables and AML survival and prognosis. In the immune-related lncRNA prognostic risk model, the prognosis was better for low-risk than for high-risk patients, indicating risk score of this model as an independent indicator of prognosis. The area under the curve value for 1-, 3-, and 5-year survival of TCGA patients was 0.817, 0.859, and 0.909, respectively, whereas that of GEO patients (of dataset GPL96-GSE37642) was 0.603, 0.652, and 0.624, respectively. Gene set enrichment analysis revealed the enrichment of multiple pathways, such as antigen processing, B-cell receptor signaling pathway, natural killer cell-mediated cytotoxicity, and chemokines, in high-risk patients. In this study, immune-related lncRNA prognostic risk models effectively predicted AML survival and provided potential treatment targets.

Neighborhood Disadvantage Is Associated with Inferior Overall Survival for Acute Myeloid Leukemia Patients Treated with Standard Intensive Induction

Background: Identifying biologic and socioeconomic contributors to disparities in outcomes is critical for the management of acute myeloid leukemia (AML). Prior studies in AML revealed that Black patients (pts) have inferior survival compared to White pts, with more striking disparity for Black pts aged &amp;lt;60. We previously explored our large and diverse AML cohort from urban and large rural areas and did not identify inferior survival for pts with self-reported Black race or Hispanic ethnicity. Although no surrogates for social determinants of health (SDOH) or insurance status were recognized as contributing to survival differences, oncology nurse navigation (ONN) was associated with a survival benefit, with no significant difference in access to ONN among racial or ethnic groups. To further evaluate socioeconomic drivers of disparities in this population supported by ONN, we aim to use the Area Deprivation Index (ADI) to determine the impact of neighborhood disadvantage on survival in AML. Methods: AMLpts diagnosed at Levine Cancer Institute between 12/2015 and 12/2020 were included. Pts with acute promyelocytic leukemia (APL) were excluded from this analysis. Details regarding insurance coverage, ONN services, induction therapy, stem cell transplant (SCT) status, and survival outcomes were collected. Race and ethnicity were self-reported. European LeukemiaNet (ELN) 2017 criteria were used to assign AML risk. Pt and disease characteristics were summarized, described, and compared with Fisher exact tests and Wilcoxon rank-sum tests for categorical and continuous data, respectively. Pts were grouped into 4 quartiles based on ADI national percentile. Subsequent models were fit with ADI dichotomized at the median after adjacent ADI groups were removed with backward elimination. Associations between overall survival (OS) and pt and disease characteristics, insurance status, race, ethnicity, ONN services, and ADI were investigated. OS was defined as the time from diagnosis (dx) to death from any cause, with surviving subjects censored at the time of last follow-up. Kaplan-Meier methods were used to summarize OS and calculate landmark survival rates. Survival was compared with unadjusted log-rank tests. Cox proportional hazards models were used to estimate hazard ratios (HR) and 95% confidence intervals (CIs) associated with OS. Results: Three hundred eighty pts with available baseline characteristics were included in our analysis ( Table 1) . For pts &amp;lt;60 who received intensive induction and had available ADI (N=86), there was a significant difference in OS amongst the quartiles of ADI (P=0.007), with differences even more pronounced with dichotomized ADI (P&amp;lt;0.001) ( Figure 1). With a median follow-up of 38.1 months for the intensively treated pts &amp;lt;60, the estimated two-year OS was 96.6% for more advantaged/low ADI pts and 55.4% for less advantaged/high ADI pts (HR 5.21; 95% CI 1.82 to 14.95). Within the cohort of pts &amp;lt;60 who received intensive induction, sixty-eight percent were Non-Hispanic White and 28% were Non-Hispanic Black. Many (66%) were privately insured at dx, with 24% uninsured or with Medicaid. ONN services were accessed by most (72%) pts with known ADI. Care barriers were expressed by pts of all ADI quartiles, with transportation being the only barrier most reported by pts of increased disadvantage (P&amp;lt;0.01). Forty-two percent of pts received SCT (N=38). Nearly 60% of socially advantaged pts (ADI≤51) received SCT versus 33% of socially disadvantaged pts (P=0.03). In univariate models, receipt of SCT (P=0.016) was associated with improved OS while receiving navigation services (P=0.32) was not. Conclusions: Higher ADI was associated with inferior OS in AML pts &amp;lt;60 who received intensive induction. Receipt of SCT was observed more often in pts with more neighborhood advantage, which is of great importance since SCT improved OS in this cohort. Although insurance status is often linked to SCT access, insurance status did not significantly impact OS. Of the ONN reported barriers to care, transportation issues had a greater impact on pts with more neighborhood disadvantage. Expanding access to SCT and addressing barriers to care, such as transportation, is necessary to overcome the detrimental impact of neighborhood disadvantage on survival in AML.

Selenium Deficiency and Chronic Anemia in an Outpatient Hematology Clinic

Background Selenium (Se) is an essential micronutrient with roles in reproduction, metabolism, immunity, and the cellular cycle, including DNA synthesis and prevention of cell death. Its role in malignancy has also been studied, specifically for esophageal, gastric, lung, prostate cancer. Studies have shown benefit with Se supplementation for those malignancies. There is evidence that low Se levels are independently associated with worse outcomes regarding treatment response and long-term survival for Non-Hodgkin's Lymphoma (NHL) and cutaneous T-cell lymphoma, and additional studies on its association in acute myeloid leukemia (AML), follicular lymphoma (FL), and Hodgkin's lymphoma (HL) have revealed worsening outcomes as well. However, Se was not predicative of treatment response or survival for AML, FL and HL, therefore it's role in the development of hematological malignancies is not clear. Previously, the role of Se in myelodysplastic syndrome (MDS) was investigated, indicating the importance of Se for apoptosis in MDS cell lines. There is evidence that a causal relationship with decreased Se levels and chronic anemia might exist; however, there is a paucity of data regarding its role in the development of chronic anemia and hematological disorders. This descriptive single center study describes Se deficiency in patients with chronic anemia. Objectives Primary endpoint was to determine if there was evidence of Se deficiency in patients with chronic anemia in an academic hematology outpatient clinic. Methods We herein present a descriptive study on 18 patients with hematological disorders and selenium deficiency. Upon initial detection of Se deficiency, levels of hemoglobin (Hb), mean corpuscular volume (MCV), Copper (Cu), Zinc (Zn) and iron (Fe) closest to time of detection were noted if tested. Average values with standard deviation were calculated and association with malignancies were noted. Results Of the 18 patients, 13 had pre-malignant or malignant conditions, including 7/13 who had bone marrow failure/MDS, 5/13 patients with lymphoid disorders, and 1/13 with plasma cell disorders. The 5 remaining patients had chronic gastrointestinal disorders. Averages on detection of Se deficiency were noted to be: Hb (9.9 ± 2.0), MCV (96.0 ± 11.2), Se (100.5 ± 11.6), Cu (95.8 ± 28.8), Zn (70.1 ± 21.5) and Fe (81.6 ± 56.7). No association was noted between Se and other trace elements. 72% (13/18) of patients with Se deficiency also had malignant conditions. Of those with malignancy, 54% were noted to have bone marrow failure. Discussion Currently there are no clinical guidelines that recommend following levels of Se routinely in patients with chronic anemia. Our data might indicate a possible association between chronic anemia and Se deficiency. Given the high number of MDS patients in our study, there might be a link between Se deficiency and MDS. Further investigation is warranted into the role of Se in the development of chronic anemia disorders, as supplementation with Se and routine follow-up may be warranted and can improve clinical outcomes.

Defining a Targetable Leukemia Intrinsic Dependency on Noncanonical PI3Kgamma Signaling

There are overlapping clinical and pathologic characteristics between poor prognosis subsets of acute myeloid leukemia (AML) and blastic plasmacytoid dendritic cell neoplasm (BPDCN), but common dependencies are ill-defined. We hypothesized that identifying novel shared targets and biomarkers could elucidate biology and nominate new therapies for both malignancies. We performed genome-wide CRISPR interference in BPDCN cells and identified 327 dependency genes, which were then analyzed informatically and functionally in AML, ALL, and BPDCN cells and patient-derived xenografts (PDXs). We defined a leukemia subset characterized by PIK3R5 activation and sensitivity to phosphoinositide 3-kinase gamma (PI3Kγ) inhibition via a noncanonical downstream pathway. The PI3Kγ complex includes an enzymatic subunit p110γ encoded by PIK3CG and a regulatory subunit p101 encoded by PIK3R5. Intriguingly, despite elevation of both p110γ and p101 proteins in this leukemia subset, only the PIK3R5 mRNA is overexpressed. To investigate this disconnection, we measured protein half-life and found that p110γ stability is positively regulated by PIK3R5/p101, associated with reduced p110γ ubiquitination. This suggested that increased expression of PIK3R5/p101 leads to posttranslational p110γ elevation by protecting it from ubiquitin proteasome-mediated degradation. To identify the upstream activator of PIK3R5, we analyzed RNA-seq data in &amp;gt;200 leukemia PDXs and found that AMLs with activated PIK3R5 are enriched for an innate immune response signature (IIRS). Supporting this model and indicating potential clinical relevance, analysis of TCGA AML patients revealed that the IIRS, which includes PIK3R5, was associated with M4/M5 monocytic AML and poor prognosis, independent of age, cytogenetics, or somatic mutations. Furthermore, in cells with low baseline expression, PIK3R5 was induced by the Toll-like receptor agonist resiquimod, showing that PIK3R5 is activated by inflammatory signaling and consistent with a monocytic/dendritic lineage association. By integrated analysis of ATAC-seq and ChIP-seq in PIK3R5 high vs low leukemias, as well as gene dependency scores, we identified the transcription factor PU.1 as responsible for PIK3R5 activation. Treatment with resiquimod enhanced the interaction between PU.1 and the PIK3R5 promoter, while depletion of PU.1 nullified the ability of resiquimod to activate PIK3R5. To elucidate molecular events downstream of p101/PI3Kγ, we studied consequences of pathway inactivation using PIK3R5 interference or the PI3Kγ-selective inhibitor eganelisib. Surprisingly, there was no effect of PI3Kγ inactivation on AKT phosphorylation or canonical PI3K-AKT-mTOR mRNA targets. Therefore, we performed phosphoproteomics and found that PAK1 S144, a phosphorylation site not previously associated with PI3K signaling, was suppressed by PIK3R5/PI3Kγ inhibition ( panel a). Selective inhibitors of PAK1, but not AKT, replicated the effect of eganelisib. Furthermore, introduction of constitutively active PAK1, but not inactive PAK1, eliminated sensitivity of leukemias to eganelisib, supporting the functional relationship between PAK1 phosphorylation and PI3Kγ dependency. PI3Kγ inhibitors are being tested in solid tumor trials to repolarize macrophages and potentiate immune checkpoint blockade, but their cancer cell-intrinsic role has not been explored. To assess PI3Kγ inhibition in vivo, we first performed intradermal xenografting assays using BPDCN derived from a patient's skin tumor. Eganelisib achieved comparable efficacy to cytarabine chemotherapy and the combination of eganelisib and cytarabine exerted significant synergy (combination index=0.63, p&amp;lt;0.0001). Next, we measured the effect of PI3Kγ inhibition on survival in disseminated PDX models. As predicted, single agent eganelisib significantly prolonged survival in AML, ALL, and BPDCN with high PIK3R5, while having no benefit in cases with low PIK3R5. To our surprise, the combination of eganelisib/cytarabine synergistically increased survival in all cases, regardless of baseline PIK3R5 ( panel b). We conclude that inflammatory signaling-activated PIK3R5 is a biomarker for sensitivity to PI3Kγ inhibition via an unappreciated PI3Kγ-PAK1 axis. Synergy with cytarabine unmasked eganelisib sensitivity in additional leukemias, expanding the therapeutic potential of targeting PI3Kγ.

Harnessing Artificial Intelligence for Risk Stratification in Acute Myeloid Leukemia (AML): Evaluating the Utility of Longitudinal Electronic Health Record (EHR) Data Via Graph Neural Networks

Background: AML is a life-threatening disease, and to determine which patients need allogeneic stem cell transplantation, hematologists risk-stratify each case. However, standard risk stratification using the European LeukemiaNet (ELN) criteria is focused on baseline mutations and chromosomal aberrations, and the risk estimate is not updated during a patient's course. In other blood cancers, recalculating the risk with treatment response data can help guide the need for more intensive therapy (Kurtz, et al, Cell, 2019). Furthermore, deep learning graph neural networks (GNN) applied to EHR data have strong predictive power in a hematology context (Fouladvand, et al, J Biomed Inform, 2023). Thus, we evaluated the power of a GNN to predict survival in AML using longitudinal EHR data, specifically with labs and histological features that are not included in the ELN but may capture the treatment response. Methods: Patients who were seen at the Stanford Cancer Institute, had EHR data available within six months of diagnosis, and were diagnosed with AML between June 1998 and January 2021 were included in this retrospective analysis. The GNN was trained to predict survival at two years from diagnosis using the first six months of clinical data. Patients were excluded if they were lost to follow-up before two years or died before six months. Data were collected from structured databases associated with Stanford's EHR, except that diagnosis dates were from Stanford's Cancer Registry, and survival data was supplemented with other databases including the Social Security Death Index. Dysplasia, bone marrow cellularity, and bone marrow blast percentages from pathology reports (“pathology report data”) were extracted using text processing algorithms and weakly supervised machine learning (Ratner, et al, ArXiv, 2017). To represent time series information, we framed each patient's timeline as a network (or “graph”) of events. The primary GNN model was a heterogenous graph transformer classifier with two node types: complete blood count (CBC) data and pathology report data (Hu, et al, ArXiv, 2020). Data from the same week were assumed to be from the same timeframe and connected with bidirectional edges. Data separated by longer time periods were connected with unidirectional edges of a separate edge type. The independent test dataset consisted of patients whose ELN 2022 classification was available, and to train the model, the remaining data were divided into train/validation splits of 0.9/0.1. Results: Of the 2,535 patients with survival data, 1,029 met inclusion criteria. Table 1 summarizes the data available in the EHR for each variable, and nearly all patients had CBC and pathology report data. The area under the receiver operating characteristic (AUROC) using the ELN 2022 criteria for predicting survival in the test dataset was 0.79. The AUROC curve for the GNN model was comparable at 0.76, despite not using any variables from the ELN criteria, and the model effectively stratified patients' disease into high- and low-risk in the independent test dataset (hazard ratio [HR] 3.0, log-rank p = 0.0009). Interestingly, despite not having access to mutation or cytogenetic data, the high-risk cases were enriched in known high-risk mutations, like TP53 and RUNX1, and in high-risk chromosomal aberrations, like 5q deletion (Table 1). Although the model predictions correlated with the ELN criteria in some ways, they also stratified the ELN intermediate-risk AML cases into high and low risk (HR 6.1 for model-predicted high risk among ELN intermediate cases, p = 0.07). Conclusions: Risk stratification using artificial intelligence and longitudinal data from the EHR performed comparably to the ELN 2022 criteria and has the potential to further stratify the ELN categories. The model performed well despite only using histological features and lab values, which are more readily available and more frequently updated than next-generation sequencing results. In the future, this approach may further improve with a larger sample size and additional variables, such as measurable residual disease and treatment information. Given the heterogeneity and increasing complexity of AML classification, leveraging artificial intelligence to assist with classification will be crucial, and these results are a step towards a future where data are automatically extracted from the EHR and used for continuously updated risk stratification.

Correlation between Number of Molecular Mutations and Outcomes in De Novo Acute Myeloid Leukemia

Introduction Genomic sequencing of cancer has led to a deeper genetic understanding of the pathogenesis of acute myeloid leukemia (AML). Studies in myelodysplastic syndrome have shown that as the number of oncogenic mutations increases, patient outcomes progressively worsen. We hypothesized that patients with AML may also have a worse prognosis if they have a higher number of somatic mutations. To test this hypothesis, we examined the somatic mutational profile of 225 patients with previously untreated AML. Methods A single-center retrospective analysis was done utilizing the Kentucky Cancer Registry (KCR) and evaluated patients with de novo AML diagnosed between January 2015 to January 2021. Patient characteristics, cytogenetics, and molecular data were collected by chart review. Patients with available cytogenetic and somatic mutations profile from next-generation sequencing (NGS) were included. Clinically significant (CS) mutations for AML were detected by NGS performed on the diagnostic bone marrow specimen. The list of clinically significant mutations was obtained from My Cancer Genome and was compared to the mutations present in our patients. Statistical analysis was done using the Mann-Whitney U test for continuous variables and proportions were compared using Chi-square test. Overall survival (OS) was calculated by the Kaplan-Meier method and compared using a log-rank test. P value was considered significant if &amp;lt; 0.05. Results A total of 225 De Novo AML patients met inclusion criteria. The median age at diagnosis was 61 (range, 21-86). The population was fairly distributed among males (52%) and females (48%). Approximately 51% (n=115) of patients had 0-2 CS mutations, 43% (n=97) had 3-5 mutations, and 6 % (n=13) had &amp;gt; 5 CS mutations. The median number of CS mutations for the entire cohort was 2 (range: 0-9). Survival was worse in the group with &amp;gt;5 mutations when compared to those with 5 or less mutations (23 vs 7 months, p= 0.012, Figure 1). When analyzing the number of CS mutations per cytogenetic group, those with intermediate risk cytogenetics had a higher median number of CS mutations compared to the favorable and adverse risk group combined (3 versus 1, p &amp;lt; 0.001). In patients with &amp;gt; 5 CS mutations, 11/13 had a mutation associated with an adverse prognosis. To determine whether overall survival was impacted by the presence of an adverse mutation, a subgroup analysis was done using 43 patients with a normal karyotype and at least 1 adverse mutation. The cohort with &amp;gt; 5 CS mutations had a significantly shorter OS compared to patients with ≤ 5 CS mutations (43 vs 9 months, p= 0.001, Figure 2). Conclusion In summary, we found that patients with intermediate-risk cytogenetics have a high number of CS mutations. When patients were stratified by number of mutations, we found that a higher number of mutations (&amp;gt;5) led to statistically significant difference in survival compared to those with ≤5 mutations. To determine whether this difference in survival was a result of the presence of adverse risk mutations vs. number of mutations alone, we analyzed our patients who had normal karyotype with at least one adverse genetic mutation and stratified them based on number of mutations. The presence of more than 5 CS mutations was associated with a worse OS compared to ≤ 5 mutations. It appears that the number of clinically significant mutations may be an independent predictor of overall survival in AML. This can serve as another tool in stratifying AML risk with normal karyotype. This finding will need to be validated in a large prospective clinical trial.

A High Percentage of DNAM-1+ and Low Percentage of Tactile+ NK Cells at Diagnosis Correlate with Complete Response and Survival in AML: A Sub-Analysis of the Ven-a-Qui Pethema Trial

Background: Natural killer (NK) immune status in Acute Myeloid Leukemia (AML) could influence therapeutic results and natural history of the disease. Objectives: In the phase I/II VEN-A-QUI (Clinical trials NCT04687761) clinical trial for unfit newly diagnosed AML patients (venetoclax+Quizartinib+azacytidine or low-dose cytarabine), it was planned a prospective study of NK populations and their influence in the main clinical outcomes. NK cells were analyzed at the diagnosis in a central laboratory. Material and methods: We analyzed NK cell subpopulations based in the expression of DNAM-1, TIGIT, TACTILE, CD8, PD-1, TIM-3, LAG-3, NKp44, NKp46, NKp30, CD16, CD85j,KIR2D, NKG2C, NKG2A, CD57, CD6, CD244, NKp80, Perforin, Granulolysin and Granzyme B. We compare composite Complete Remission (CRc) results of the Venetoclax/Quizatinib with Aza or Cytarabine vs. not achieving CRc using t Student. MaxStat statistics was used to test the significant limit of those populations with a significant difference (R-statistics version 4.2.2). Survival analysis comparing significant differences between patients in CRc was performed using LogRank test and Kaplan Maier. Cox survival analysis was performed to evaluate if the presence of this populations could be an independent factor against the European Leukemia Net classification of 2017 and p53. Data of molecular characteristics were obtained with an NGS panel of 40 genes. RESULTS: The trial included 76 ineligible patients for intensive chemotherapy (37 and 39 were in Aza and LDAC arms respectively). cCR was obtained in 39 patients (51.3%). ELN2017 risk was favorable in 8, intermediate in 15, and Adverse in 53. In 70 patients NK cell populations were analyzed (4 patients with very low percentage of NK cells, 2 with no sample at diagnosis were excluded from the study). There were significant differences between patients who achieve CRc and those who did not, according to DNAM-1 (74.4% vs 61.9, p=0.0018) and TACTILE (62% vs. 72%, p=0.02) expression on NK cells. The rest of antigens analyzed on NK cells were not significant. We found a cutoff of 62% of DNAM-1+ and less than 81% of TACTILE to differentiate populations in terms of survival using long-rank test. The number of DNAM-1 positive was 47 (60%) and TACTILE- was 55 (78%). The median OS of DNAM-1+ was better than DNAM-1− (18.4 vs 4.7 months, p=0.0001), and between TACTILE− vs TACTILE+ (17,36 vs. 4.6 months, p=0.005). Using an Cox model comparing karyotype, p53 and DNAM-1+ and TACTILE-, only DNAM-1+ was associated with better outcome (HR:0.33 (0.17-0.66) and TACTILE+ with worse (HR: 2.44 (1.20-4.99) in multivariate analysis. Conclusions: DNAM-1(CD226) positive and TACTILE (CD96) negative NK cell populations are associated to CR and better OS. These findings need to be confirmed in larger studies. Interestingly, DNAM-1 is an activating receptor and TACTILE is an inhibitory receptor for NK cells that share the same CD155 ligand but display opposite function.

Acute Myeloid Leukemia Differentiation State and Genotype Influence Anti-Apoptotic Protein Expression, Venetoclax Sensitivity, and Survival in AML

Background: Somatic mutations (mut) and acute myeloid leukemia (AML) differentiation state (phenotype) influence sensitivity and response to treatment. However the interaction between genotype and phenotype with respect to sensitivity or resistance to venetoclax (VEN) based therapy has yet to be fully evaluated. Methods: Samples from patients (pts.) with newly diagnosed (ND) AML enrolled in Beat AML with available bulk RNA sequencing (RNAseq), whole exome sequencing, and ex vivo drug sensitivity testing (N=255) were analyzed for correlation between AML phenotype (defined using RNAseq [van Galen et. al. Cell 2019] and surface immunophenotype via multiparameter flow cytometry [MFC]), genotype, and VEN sensitivity. A monocytic score corresponding to monocytic gene expression and myeloid differentiation for each sample was included as previously defined (Bottomly et. al. Cancer Cell 2022). Clinical outcomes were assessed in an independent retrospective patient cohort (N=97) treated with frontline hypomethylating agents (HMA)+VEN at Oregon Health &amp; Science University. Categorical and continuous variables were analyzed using Fisher's or Wilcoxon rank sum testing. Time to event outcomes utilized the log-rank method with cox multivariable regression modeling. Results: Median patient (pt) age at diagnosis in the surveyed cohort was 61 years (range 20-88). AML phenotype determined using proportional deconvolution of RNAseq data was HSC/progenitor (primitive) in 31% (N=78), mixed/intermediate in 31% (N=79), cDC/monocytic (mature) in 29% (N=73), and unknown in 10% (N=25) of pts. Muts frequent in primitive vs. mature AML samples included transcription factor ( RUNX1, CEBPA, GATA2, GATA1, ETV6, IKZF1; 30% vs. 15%, p=0.050), tumor suppressor ( TP53, PHF6, WT1; 21% vs. 7%, p=0.019), and IDH1/2 (27% vs. 12%, p= 0.027). Active signaling mut ( FLT3-ITD/TKD, K/NRAS, PTPN11, BRAF) were more frequent in mature vs. primitive samples (78% vs. 44%, p &amp;lt; 0.0001), underscoring the correlation between genotype and myeloid differentiation (Fig.A). Expression of anti-apoptotic protein and myeloid differentiation markers measured using RNAseq varied by AML genotype and phenotype and corresponded to VEN resistance. The ratios of BCL2A1 (pearson R=0.62, p &amp;lt; 0.0001), CD11b (R=0.65, p &amp;lt; 0.0001), CD14 (R=0.73, p &amp;lt; 0.0001), and CLEC7A (R=0.72, p &amp;lt; 0.0001) to BCL2 positively correlated with VEN resistance (i.e., increasing area under the curve [AUC]). While IDH1/2 and NPM1 mutated samples demonstrated the highest BCL2 expression and lowest BCL2A1: BCL2 ratio and remained largely sensitive to VEN, myeloid differentiation (in particular monocytic differentiation) increased VEN resistance irrespective of genotype. For instance, classification of a primitive vs. mature phenotype further stratified ex vivo VEN sensitivity in pts with IDH1 (median AUC: 61 vs. 195, p: 0.004), IDH2 (median AUC: 61 vs. 185, p: 0.002), or NPM1 (median AUC: 60 vs. 220, p &amp;lt; 0.0001) mut (Fig.B). Overall survival (OS) following HMA+VEN was evaluated based on the presence or absence of surface immunophenotypic markers measured via MFC that correlated with myeloid differentiation. CD117+ AML (HR: 0.30 95% CI: 0.14 - 0.65, p=0.002) associated with improved OS, while inferior OS was observed with CD11b+ (HR: 2.03 95% CI: 1.07 - 3.86, p=0.031) and CD64+ AML (HR:1.86 95% CI: 0.90 - 3.85, p=0.095). When assessing the combined influence of genotype and immunophenotype on OS in CD11b+ or CD64+ AML, IDH1/2 mut (N=8) numerically improved OS compared to IDH1/2 wild-type cases (N=17) (median 12.3 vs. 5.8 months, p: 0.17). In multivariate analysis adjusted for CD11b or CD64 positivity, age, IDH1/2 mut, and secondary or therapy-related AML, CD11b or CD64 positivity (HR: 1.98 95% CI: 1.08-3.66, p=0.028) retained the strongest impact on OS. Conclusions: Genotype and phenotype both modulate VEN sensitivity in AML. Myeloid differentiation is associated with increased alternative anti-apoptotic protein expression and VEN resistance irrespective of genotype; however certain mutations (i.e., IDH1/2) may retain prognostic significance independent of myeloid differentiation. Given both factors influence VEN sensitivity, prognostic models incorporating AML phenotype and genotype may further improve risk stratification and predict response to VEN based therapy in AML.

Comprehensive Molecular Stratification of Patients with AML Treated with CPX-351

Background: Liposomal cytarabine-daunorubicin (CPX-351) is approved for treatment of acute myeloid leukemia (AML) with myelodysplasia (MDS)-related changes and therapy-related AML. CPX-351 induction has been observed to result in longer periods of bone marrow suppression than with 7+3 therapy. Thus, careful consideration is warranted when selecting patients for whom CPX-351 would provide an optimal risk-to-benefit ratio. It is unknown whether any specific genetic mutations impact outcomes from CPX-351 therapy. Methods: Forty-one subsequent patients with AML at VCU Massey Comprehensive Cancer Center from 2017 to 2022 were treated with CPX-351 (liposomal cytarabine-daunorubicin). Genetic mutation profiles were obtained from our institutional multi-gene panel next generation sequencing assay. Multivariate analysis was performed using Cox proportional hazards regression. Overall survival (OS) was analyzed via the Kaplan-Meier method. Composite complete response (CCR) was defined as CR + CR with incomplete hematologic recovery + CR with partial hematologic recovery. Results: The median age of patients treated with CPX-351 was 65.2 years (range, 20.1-78.5). Median ECOG performance status was 1 (range, 0-2) and Charlson Comorbidity Index was 4 (range,2-14). The majority (65.9%) of patients had adverse risk per ELN 2022 criteria, mostly with MDS-related and complex cytogenetics (17.1% favorable risk; 14.6% intermediate risk). Multivariate analysis identified mutated PTPN11 ( PTPN11mut)and IDH2 ( IDH2mut) as independent risk factors for decreased OS in patients treated with CPX-351. The hazard ratio (HR) for death of PTPN11 mut was 2438 (p = 0.017). The HR for IDH2mut was 124.2 (p = 0.022). Mutated TP53 was also a risk factor for reduced OS, with HR 3.21, but this was not statistically significant. Mutated SRSF2 ( SRSF2mut) was an independent prognostic factor for improved OS with CPX-351, with a HR of 0.0019 (p = 0.017). ASXL1, TET2, IDH1 and FLT3-ITD mutations were associated with improved OS, but these results did not reach statistical significance. For patients with SRSF2mut treated with CPX-351, median OS was 20.5 months, compared to 8.9 months for those without SRSF2mut (HR 0.49 [95% CI 0.28-0.88], p=0.037). The CCR rate was not significantly different between patients with and without SRSF2mut (p=0.66), and measurable residual disease (MRD) status could not be assessed due to unavailable historical data. Patients with PTPN11mut had a median OS of 11.2 months, whereas those without PTPN11mut had a median OS of 21.6 months, although this comparison was not statistically significant (HR 1.51 [95% CI 0.70-3.26], p=0.207). CCR rates were not significantly different between patients with and without PTPN11 mut. The median OS for patients with IDH2mut was 16.4 months, compared to 21.6 months for patients without IDH2mut but this difference was not statistically significant. Conclusions: Among patients with AML treated with CPX-351, PTPN11 and IDH2 mutations were independently associated with inferior overall survival compared to those without these mutations. The PTPN11 gene is involved in RAS pathway regulation and has also been associated with shorter survival in AML in other studies. Here, mutated SRSF2 was associated with improved survival after CPX-351 induction. SRSF2 is an RNA splicing factor gene that is frequently found in AML with antecedent MDS but has previously been associated with adverse outcomes after conventional intensive induction chemotherapy. Interestingly, CCR and MRD negativity rates did not differ for patients with these mutations compared to those without at our institution despite the noted changes in survival. These findings remain speculative and warrant further investigation with a greater number of patients in the era of routine MRD assessment. Moreover, controlling for confounders (such as death due to toxicity rather than disease or relapse), the impact of HCT, and frailty were not accounted for in this retrospective analysis. Future work includes identifying the potential impact of cooperating mutations.

The Atpase Domain of LONP1 Is Necessary for Mitochondrial Protein Solubility and the Viability of Acute Myeloid Leukemia (AML) Cells

Compared to normal hematopoietic cells, we and others have shown that AML cells have a heightened reliance on oxidative phosphorylation and mitochondrial metabolism for survival and proliferation. To support this unique metabolic phenotype, we previously demonstrated that AML cells have increased import of nuclear-encoded mitochondrial proteins. These newly imported proteins must be properly folded by mitochondrial chaperones, proteases, and heat shock proteins, and failure to properly process and fold these precursors leads to protein aggregation, mitochondrial dysfunction, and cell death. To evaluate the reliance of AML cells on this family of mitochondrial chaperones, proteases, and heat shock proteins, we assessed their dependencies using the gene dependency datasets (eg: depmap.org). From this analysis, we identified the mitochondrial protease, LONP1, as the top hit, and among the top 10% of all essential genes for AML. Localized to the mitochondrial matrix, LONP1 is a nuclear encoded AAA+ serine protease. Proteins are unfolded by its ATPase domain and degraded by its serine-catalyzed proteolytic domain. Compared to normal hematopoietic cells, LONP1 mRNA was overexpressed in AML across three publicly available datasets. Compared to normal hematopoietic cells (n=8) and CD34+ cells (n=3), LONP1 protein was increased in 16/30 primary AML samples by immunoblotting. Using shRNA and CRISPR, LONP1 knockdown and knockout reduced the growth and viability of OCI-AML2, OCI-M2, NB4, and TEX cells. Genetic knockdown or knockout of LONP1 increased levels of insoluble, aggregated mitochondrial proteins as measured by mass spectrometry, proteostat fluorescence and confocal microscopy, and immunoblotting of soluble and insoluble protein fractions. LONP1 knockdown/knockout also reduced mitochondrial respiration, depolarized the mitochondria, and induced AML cell death. We next determined the domain of LONP1 that was necessary for mitochondrial protein solubility, mitochondrial function, and AML survival. We over-expressed wild type, ATPase dead (E591A), or proteolytically dead (S855A) LONP1 cDNA in OCI-AML2 cells and knocked down endogenous LONP1 with shRNA targeting the 3'UTR of the endogenous gene. We then measured mitochondrial protein solubility/aggregation, mitochondrial respiration and cell viability. Wild type LONP1 and the proteolytically dead (S855A) mutant, but not the ATPase mutant (E591A), rescued mitochondrial protein solubility, mitochondrial respiration and cell viability, thus demonstrating that the ATPase domain of LONP1 is necessary for these functions. Bardoxolone methyl (CDDO-Me) is a synthetic triterpenoid that inhibits the ATPase activity of LONP1 by binding to an allosteric site near the ATPase domain of the enzyme. CDDO-Me killed OCI-AML2 and NB4 cells with IC 50 values of 178.5±29.7 and 156.5±39.7 nM, respectively. CDDO-Me (200nM) also killed &amp;gt;50% of cells in 3 out of 4 high-LONP1 expressing primary AML patient samples, but 10 of 10 tested primary AML samples with low levels of LONP1 were insensitive to the drug. Likewise, CDDO-Me induced mitochondrial protein aggregation in OCI-AML2 cells and in a primary AML patient sample with high LONP1 expression, while a primary AML patient sample with undetectable LONP1 expression showed no protein aggregation. In summary, the mitochondrial serine AAA+ protease LONP1 is over-expressed in a subset of AML cells and primary samples. The ATPase domain is necessary for LONP1's function in maintaining mitochondrial protein solubility. Selective inhibition of this domain leads to mitochondrial protein aggregation, impaired mitochondrial respiration and AML cell death. Thus, inhibiting the ATPase domain of LONP1 may be a novel therapeutic strategy for AML.

Evolving Racial/Ethnic Disparities in AML Survival in the Novel Therapy Era: Real-World Evidence of Improved Survival in Older Patients with the Largest Improvement in Non-Hispanic Black Patients

Introduction: The approval of venetoclax (VEN)-based frontline therapy for acute myeloid leukemia (AML) in November 2018 marks a new era of AML care. Racial/ethnic disparities have been increasingly recognized in AML with historically inferior outcomes reported in non-Hispanic Black (NHB) patients relative to non-Hispanic White (NHW) patients. The advances of novel therapeutics offer an unprecedented opportunity to narrow the care gaps between patient groups; but they also create new challenges related to testing, clinical decision-making, and care coordination. Currently, little is known about the impact of race/ethnicity on the recent advances in AML management. We performed a retrospective analysis leveraging real-world data to examine the magnitude and mechanisms of racial/ethnic disparities in modern AML care. Methods: Adult patients from Flatiron Health, a nationwide, electronic health record derived, de-identified database, with AML diagnosed between January 2014 to December 2018 (Pre era) and January 2019 to October 2022 (Post era) who identified as NHW, NHB or Hispanic and received at least one line of anti-leukemic therapy were included. Overall survival (OS) was defined from the time of therapy initiation to the time of death. Multivariable Cox regression models were used to compare the hazards of death for NHB and Hispanics relative to NHW. Predicted survival curves were then computed. Pre-planned sensitivity analyses including patients with 1) no ethnicity information, 2) no anti-leukemic therapy, and 3) care gap &amp;gt;100 days from diagnosis to induction were performed. Results: 2998 patients in Pre era (2566 NHW, 221 NHB, 211 Hispanics) and 2098 patients in Post era (1809 NHW, 162 NHB, 127 Hispanics) were included. Older patients with more comorbidities and worse disease biology (Table 1) were treated in Post era. After adjusting for race/ethnicity, age and Hematopoietic Cell Transplant Specific Comorbidity Index (HCT-CI), patients in Post era had a 10% decrease in the hazard of death compared to Pre era (HR 0.90, 95% CI 0.83-0.96). The difference was driven by OS benefits observed in patients aged &amp;gt;70 years (HR 0.82, 95% CI 0.75-0.91). NHB and Hispanics were younger at diagnosis than NHW in both eras. Lower socioeconomic status was observed in NHB and Hispanics. No difference in ELN2017 risk stratification was observed (Table 1). After adjusting for age and HCT-CI, NHB had a 22% increased hazard of death compared to NHW in Pre era (HR 1.22, 95% CI 1.04-1.43); whereas Hispanics had comparable outcomes (HR 1.01, 95% CI 0.84-1.21). Strikingly, the survival disparity between NHB and NHW in Pre era was not observed in Post era with predicted 2-yr OS rate being 45.3% and 39.9%, respectively (HR 0.86, 95% CI 0.69-1.08) (Figure 1). Predicted 2-yr OS rates improved for NHB patients in both ≤70 years (42.3% vs. 53.2%) and &amp;gt;70 years group (13.6% vs. 37.0%). No significant change in OS was observed in Hispanics between Pre and Post era (Figure 1). Results of all sensitivity analyses were consistent with primary analyses. Molecular testing at diagnosis significantly differed by eras across race/ethnicity groups (overall 66.4% vs. 82.9%, p&amp;lt;0.001). In contrast, molecular testing occurred more frequently in NHB at relapse in Post era (45.6% vs. 65.4%, p=0.020). Despite being older, more NHW received allogeneic transplant than NHB in Pre era (25.8% vs. 18.6%, p=0.046). In Post era, NHW were more likely to receive VEN-based induction therapy (46.9%, 37.0%, 35.4% in NHW, NHB and Hispanic, respectively, p= 0.003). Among patients who were tested and qualified, no difference was observed in the use of gemtuzumab ozogamicin, CPX-351, FLT3 inhibitors or IDH1/2 inhibitors across race/ethnicity. Conclusions: In this large, real-world dataset, we demonstrate OS improvements in the era of modern AML care, particularly among patients &amp;gt;70 years though their outcomes remain inferior to younger patients. Intriguingly, the largest OS improvement was observed in NHB, where the OS disparity between NHB and NHW in Pre era appears to have been mitigated. Hispanics have not experienced similar OS improvement in modern AML care. Ongoing analyses are focused on examining the mechanisms of survival trends observed with focus on assessing the contribution(s) of patient-, disease biology- and treatment-related factors in order to identify potential targets for intervention. • Dr. Kelly Getz and Dr. Catherine Lai contributed equally

Metabolism-Related Features Identify the Combination Metformin Plus NAMPT Inhibitors As a Selective Treatment Strategy in Acute Myeloid Leukemia

Acute myeloid leukemia (AML) is a heterogeneous disease, which remains difficult to treat due to its heterogeneous nature. Recently, metabolic adaptation has been recognized as a trademark of leukemogenesis. Notably, altered function of mitochondria (mt) in leukemic stem cells (LSCs) has been linked to chemoresistance in some patients, causing relapse of disease. Thus, targeting the rewired energy metabolism of leukemic cells has gained a lot of attention in recent years. Here, using a large cohort of adult de novo AML samples (n=502 patients, all intensively treated with 3+7 scheme), we identified a subset of patients with increased mitochondrial DNA content (mtDNAcn, n=163) compared to age- and sex-matched healthy controls (n=308). mtDNAcn was significantly higher in patients harboring FLT3-ITD (31%) and IDH1 mutations (50%). Patients with high mtDNAcn were associated with worse clinical outcomes regardless of European Leukemia Net (ELN) risk stratification, sex, and age. In contrast, high mtDNAcn was not a predictive survival marker in a cohort of adult AML patients treated with venetoclax (VEN) + hypomethylating agents (HMA), indicating that these metabolic features are specifically linked to the resistance to intensive chemotherapy. Using label-free quantitative proteome analysis (11272 proteins) combined with metabolomic screening (173 metabolites) on sorted CD34 +/CD117 + AML blasts (n=26), we found that AMLs with high mtDNAcn were enriched for signatures like “GMP-like” and “mt complex I activity”, being metabolically enriched for long-chain fatty acids, suggesting that these cells rely more on oxidative phosphorylation (OXPHOS) metabolism. Functionally, AML blasts with high mtDNAcn displayed increased levels of BCL2, mt mass, and mt membrane potential, which was associated with increased oxygen consumption rate (OCR) measured by extracellular flux analysis. Considering the impact of mtDNAcn on AML survival in patients treated with intensive chemotherapy, which was not observed in the context of VEN+HMA, and considering the increased reliance of LSC on mtOXPHOS, we studied the impact of mtDNAcn on chemotherapy response. Ex vivo analysis of primary AML samples with high mtDNAcn indicated increased resistance to AraC-induced apoptosis, which was reverted upon knock-down of the DNA polymerase gamma gene (POLG) encoding the catalytic subunit of the mtDNA polymerase. Due to the link between mtDNAcn and mt complex I activity, we investigated if the treatment with metformin, known to inhibit complex I, would sensitize AML patients with high mtDNAcn to the standard of care AML drugs, such as Cytarabine (AraC), FLT3-inhibitors and VEN. Treatment with low dose of metformin (1 mM) effectively reduced the mtDNAcn in AML cells (n=40), resulting in decreased OCR, while glycolysis was increased, suggesting metabolic rewiring. Furthermore, the combination of metformin with AraC, VEN and FLT3-inhibitors synergistically increased drug-induced apoptosis in AML cells with high mtDNAcn and helped to overcome VEN resistance in AML cells. In vivo, combination of metformin with VEN prolonged the survival of mice transplanted with K562 cells (VEN resistant), resulting in decreased leukocyte counts and blood chimerism. Since treatment with metformin resulted in metabolic rewiring, with increased glycolysis and NAD + consumption, we wondered if the combination of metformin with glycolytic (2-DG/DAP) and NAMPT (KPT-9274) inhibitors would impede the metabolic rewiring, enhancing cytotoxicity. Among the different treatment combinations, we find that the combination of metformin and KPT-9274 gave the best therapeutic window by significantly increasing drug-induced apoptosis, with more pronounced effects in patients with high mtDNAcn, while sparing healthy CD34 + cells. Concomitantly, OXPHOS and glycolysis were diminished upon KPT-9274+metformin, suggesting that the combination was able to bypass the metabolic rewiring of the AML cells. In conclusion, we uncover a subgroup of patients with high mtDNAcn linked to increased mtOXPHOS that is resistant to chemotherapy-induced apoptosis and is characterized by poor clinical outcomes, which can be overcome by the inhibition of the mt complex I. Moreover, the quantification of mtDNAcn can be easily incorporated into clinical practice as a simple and cost-effective metabolic readout of AML patients to identify metabolic vulnerabilities.

Genome-Wide Methylome and Transcriptome Profiling of Acute Myeloid Leukemia Derived Bone Marrow Mesenchymal Cells Identify Age Group Specific Biological Pathway Dysregulation in the Bone Marrow Microenvironment of AML

Adult patients with acute myeloid leukemia (AML) have a proportionate increase in mortality from their disease and therapy as age increases. In great part, this mortality is due to disease relapse that occurs at a higher rate irrespective of genetic assigned risk. While intensification of therapy in patients with favorable genomics such as core-binding-factor AML (CBF-AML) has yielded a high proportion of long-term remission and potential cure, this benefit has been less frequently observed in patients with AML age 60 and older despite receipt of similar therapy. Moreover, even less benefit has been observed in patients with unfavorable genomics such when bearing a complex karyotype (karyotype with ≥ 3 chromosome abnormalities). Reasons for this unfavorable outcome among older patients is unknown. We hypothesized that tumor-induced aging in cells of bone marrow microenvironment, specifically, in bone marrow mesenchymal stromal cells (BMSCs) plays a significant role in AML survival and progression in older compared to younger patients and that the difference in prognosis may depend on specific aging induced changes in the BMSCs. We expanded BMSCs under physiologic hypoxic conditions (91% nitrogen, 4% oxygen and 5% CO 2) to mimic the bone marrow microenvironment, isolated DNA and RNA from early passaged cells and performed methylation analysis using 850k Illumina DNAmehtylation arrays and gene expression analysis using mRNAseq. We used the Horvath Skin and Blood clock to calculate epigenetic age (DNA methylation age). We determined differentially expressed genes with increasing epigenetic or chronological age in the different AML genetic risk groups (n=9 CBF-AML, n=11 CK-AML and n=7 Normal donor BMSC) as well as in older (&amp;gt;60) and younger (&amp;lt;60) AML-BMSCs, and performed Gene Set Enrichment Analysis (GSEA) for the Hallmarks as well as aging and senescence related genesets (n=33 genesets). We found that BMSCs derived from AML bone marrow cells are more age accelerated than normal donor BMSCs. Our GSEA analysis revealed enrichment for age-associated gene networks that reflected hallmarks of aging, including mitochondrial dysfunction, inflammation, and cellular senescence. Specifically, we observed a positive enrichment for the SenMayo gene signature ( Saul et al. 2022, Nature Communications) and a negative enrichment of proliferation and mitotic gene sets with increasing epigenetic and chronological age (padj&amp;lt;0.05). Gene sets that are related to inhibition of adipogenesis were positively enriched in the CBF-AML BMSCs compared to the CK-AML BMSCs. Our integrated analysis revealed novel observations regarding the cellular changes that occur in the AML stromal microenvironment with increasing age. Most interestingly, our findings suggest that AML derived BMSCs have increased epigenetic age compared to normal donor BMSCs. Additional experiments are ongoing to functionally validate the findings predicted by the enrichment analysis. Our study is the first to suggest an increased epigenetic age in the stromal bone marrow microenvironment of older AML patients.