NAMPT haploinsufficiency is a collateral lethal therapeutic vulnerability in high-risk myeloid malignancies with TP53 inactivation Open Access

Cytogenetic abnormalities occur in >50% of newly diagnosed acute myeloid leukemia (AML) cases and are a main predictor of patient outcomes.¹ Abnormalities involving chromosome 7, including deletions (del(7q)) and monosomy 7 (−7), fall under the high-risk category and are seen across various myeloid disorders.^2-4 Current treatments for high-risk patients with myelodysplastic syndrome, such as intensive chemotherapy with the demethylating azacitidine, are only effective in ∼39% of patients, with most experiencing relapse and modest survival improvements.⁵ For the 40% to 50% of patients who do not respond to azacitidine, the median overall survival is ∼6 months.⁶ Problematically, −7/del(7q) frequently coexists with additional unfavorable cytogenetic abnormalities, such as del(17p) (harboring TP53) and a complex karyotype (CK; with ≥3 independent cytogenetic abnormalities). Genetic lesions of TP53, which are independently associated with poor outcomes, occur in ∼43% to 56% of −7/del(7q) cases.^7,8 Patients with CK and −7 or del(7q) demonstrate significantly worse survival than those CK patients without −7/del7q.⁸ Therefore, these patients urgently require alternative, targeted, and effective treatments.

Loss of tumor suppressor genes on deleted chromosomal arms drives leukemogenesis, and efforts to restore their function have been largely unsuccessful.^9,10 An innovative approach to developing targeted therapy is to exploit the unique biological characteristics of −7/del(7q) cancerous cells that make them more vulnerable. Collateral lethality occurs when an essential passenger gene is codeleted with nearby tumor suppressors, causing haploinsufficiency and cancer-specific vulnerability.^11,12 These genes have been termed “CYCLOPS” (copy number alterations yielding cancer liabilities owing to partial loss) genes.¹² For example, CSNK1A1 on 5q32 is a haploinsufficient gene in del(5q) myelodysplastic syndrome. Its heterozygosity enables lenalidomide to selectively target malignant cells by degrading CK1α protein.¹³ 

In this study, we hypothesized that losing an essential chromosome 7 gene makes −7/del(7q) cells highly dependent on the remaining allele's protein product. We discovered nicotinamide phosphoribosyltransferase (NAMPT) as an AML essential gene located on chromosome band 7q22.3 using publicly available genome-wide CRISPR/Cas9 essentiality screens.^14-18,NAMPT encodes the rate-limiting enzyme in the NAD salvage pathway, converting nicotinamide to nicotinamide mononucleotide, a precursor to NAD. NAD is crucial for cellular energy, DNA repair, and genome integrity, supporting proteins such as PARP1 and SIRT1.¹⁹ CRISPR editing of a single allele of NAMPT in human AML cell lines led to lower NAMPT protein expression and increased dependence on residual NAMPT for AML survival, implicating NAMPT as a collateral lethal gene. Remarkably, we found that the TP53 mutational status did not attenuate the observed effects of NAMPT inhibition on cell death, indicating that NAMPT inhibitors are equally effective in −7 patients with or without TP53 lesions. Pharmacological inhibition of NAMPT was similarly efficacious in primary samples from patients with −7/del(7q) AML with and without TP53 defects. Finally, given the prior evidence for defects in DNA damage recognition and repair in −7/del(7q) malignancies,⁹ we evaluated the cooperative effects of NAMPT and PARP inhibitors and found synergy in the drug combination.

Haploinsufficient gene candidates were identified using publicly available CRISPR-Cas9 screen data from 23 AML cell lines (CRISPR_gene_dependency.csv; Dependency Map [DepMap]).¹⁷ DepMap classifies a cell line as dependent on a gene if the dependency probability exceeds 0.5. Gene knockout effects were quantified using Chronos scores, with values less than –0.5 indicating depletion and less than –1 indicating strong lethality. Additional CRISPR screen data from AML cell lines^15,16,18 were obtained from BioGRID ORCS (https://orcs.thebiogrid.org/). Pan-essential genes (CRISPRInferredEssentials.csv) from DepMap were used to distinguish essential from nonessential genes. Of 711 chromosome 7 genes evaluated, 552 were expressed in human hematopoietic stem cells.²⁰ Chronos scores were correlated with messenger RNA (mRNA) expression to identify CYCLOPS genes. Expression, relative, and absolute copy number data were downloaded from Cancer Cell Line Encyclopedia (CCLE) via DepMap²¹ (https://depmap.org/portal/download/custom/).

To unbiasedly identify hematopoietic essential genes on chromosome 7, we included only the genes that are expressed in human stem and progenitor cells (n = 552 genes; Figure 1A, step 1), the origin of myeloid neoplasms. Using the previously identified pan-essential gene list from cancer DepMap CRISPR screen data, we excluded nonessential genes on chromosome 7 (Figure 1A, step 2). Then, we analyzed genome-wide CRISPR-Cas9 screen data from 23 AML cell lines retrieved from the DepMap database^14,17 and identified 170 chromosome 7 essential genes (Figure 1A, step 3; supplemental Table 1). To verify the reproducibility of our DepMap data–mining results, we compared our findings with 22 independently published CRISPR screens involving 15 different AML cell lines^15,16,18 and identified 264 essential genes on chromosome 7 (Figure 1A, step 4; supplemental Table 1). By focusing on screens done in AML cell lines, we identified and differentiated genes on chromosome 7 that are pan-essential vs hematopoietic cells/AML essential.

Of the 170 genes from the DepMap set, 55 were pan-essentials, and 115 were AML essentials. Of these 115 AML essentials, 47 were a hit in 21 of 23 AML cell lines evaluated, suggesting that these are true positives. Thirty-five of the 115 AML essentials were called in only 1 of the 23 screens, suggesting that these could be false positives (supplemental Figure 1A; supplemental Table 1). Of the 264 genes from the literature-based screen data, 70 were pan-essentials, and 21% of the hits (n = 56/264) occurred in 14 of the 22 screens, suggesting that these are likely true positives (supplemental Figure 1B; supplemental Table 1). We reasoned that targeting AML-specific CYCLOPS genes would minimize any negative systemic effects. There was a significant overlap between the genes identified in the 2 independent data sets, DepMap and literature-based screen analyses, yielding 127 essential genes, which we prioritized for downstream analysis (Figure 1A, step 5; Figure 1B).

CYCLOPS genes are essential genes with reduced expression in cancer cells, making them uniquely vulnerable to further suppression due to their reliance on limited gene products for survival. Conversely, our analysis revealed a complementary vulnerability in a distinct class of genes we refer to as ARGuS (Amplification-Related Gain Of Sensitivity) genes, inspired by the many-eyed giant Argus Panoptes of Greek mythology. In this case, cells with amplification or overexpression of essential genes become paradoxically more dependent on their continued activity. To classify essential chromosome 7 genes in these 2 categories, we correlated the expression levels of each candidate gene from CCLE to the dependency score for that gene across cell lines using DepMap data (Figure 1A, step 6; supplemental Table 2). Thirty of the 127 essential genes fit CYCLOPS criteria, where the gene expression level is significantly and negatively correlated with the gene essentiality score (Figure 1A, step 7; Figure 1C, blue). Among these 30 genes was PSMC2, a previously identified CYCLOPS gene on chromosome 7, validating our approach.¹² We identified a significant positive correlation between gene expression and dependency for 21 genes, thus fitting the ARGuS criteria (Figure 1A, step 8; Figure 1C, red). CDK6 was one of the most significant ARGuS genes, demonstrating that higher gene expression correlates with increased dependency, as shown in a prior study.²² Our data show that integrating CRISPR-Cas9 screen data with gene expression reveals cancer cell–specific liabilities from chromosomal loss.

We then investigated whether US Food and Drug Administration (FDA)–approved drugs or small-molecule inhibitors exist that could decrease the activity of any of the 30 identified CYCLOPS genes using data from the National Institutes of Health Illuminating the Druggable Genome project (supplemental Table 3).²³ Illuminating the druggable genome classifies each gene into 1 of 4 druggability levels, ranging from targets with unknown biological functions to targets with FDA-approved drugs. Although NDUFA5 comes up as a hit with an approved FDA drug, metformin, it is not directly bound and inhibited by the drug. Therefore, we eliminated it from the downstream analysis. Most genes (25/30) had no compounds targeting their protein functions. Only one of the CYCLOPS proteins, NAMPT, had a direct small-molecule inhibitor, which was pursued in downstream analyses. (supplemental Figure 2A; supplemental Table 3).

First, we confirmed NAMPT as a CYCLOPS gene by analyzing its correlation with DNA copy number and transcript levels across DepMap cell lines (Figure 2A; supplemental Figure 2B-C). NAMPT expression and copy number were significantly correlated, particularly in hematopoietic and myeloid cells. Cell lines with low NAMPT expression (bottom 25% of mRNA expression across all cancers) were more dependent on it than those with high expression (top 25%; Figure 2B). CRISPR dependency scores showed NAMPT is selectively essential in hematopoietic cancer cell lines (supplemental Figure 2D), likely due to lower transcript levels than solid tumors (Figure 2C).

To determine the frequency of NAMPT deletions in human −7/del(7q) myeloid diseases, we assessed the samples from patients with −7/del(7q) AML within the The Cancer Genome Atlas (TCGA) Single Nucleotide Polymorphism (SNP) array data.²⁴ (https://www.cancer.gov/tcga). One allele of NAMPT was deleted in 81.5% of −7/del(7q) cases (n = 22) in the TCGA patient cohort (Figure 2D). The normalized z score of NAMPT mRNA expression was significantly lower (51%) in samples with NAMPT deletions than in patients with diploid chromosome 7 status (Figure 2E). Together, these data suggest that a reduced NAMPT DNA copy number, which correlates with decreased NAMPT transcripts, represents a potential collaterally lethal liability in most −7/del(7q) patients.

We next investigated the relationship between NAMPT expression and response to NAMPT inhibitors in AML. First, we used the Genomics of Drug Sensitivity in Cancer data set 2, which contains drug sensitivity data for 349 cancer cell lines treated with the NAMPT inhibitor, FK-866.²⁵ A total of 255 of 349 cell lines also had transcriptome profiling in the CCLE database.²¹ We found a significant positive correlation between NAMPT expression and FK-866 area under the curve in 256 pan-cancer (supplementary Figure 3A) and 43 hematopoietic cancer cell lines (Figure 3A), indicating that cells with reduced NAMPT are more sensitive to NAMPT inhibition.

To test the relationship between NAMPT levels and sensitivity to NAMPT inhibitors, we genetically modified human AML cell lines to reduce NAMPT DNA copy number. We used OCI-AML3 and MOLM-13, which are triploid and diploid, respectively, based on CCLE data for NAMPT. Via CRISPR-Cas9 editing, we inactivated 2 (OCI-AML3) or 1 (MOLM-13) NAMPT alleles and confirmed that these edits partially reduced NAMPT levels compared to isogenic controls (Figure 3B; supplemental Figure 3B-C). We generated a clonal control cell line using a guide targeting intron 2 of HPRT. We assessed the viability of these cell lines treated with increasing doses of KPT-9274, a NAMPT inhibitor currently in a phase 1 clinical trial (ClinicalTrials.gov identifier: NCT04914845). Half-maximal inhibitory concentration (IC₅₀) values were significantly lower in the NAMPT-edited clones than the control line (Figure 3C-D; supplemental Figure 3D). KPT-9274 inhibits both NAMPT and p21-activated kinase 4 (PAK4).²⁶ To exclude off-target effects, we used a NAMPT-specific inhibitor, FK-866. Similar results were observed with FK-866 in the OCI-AML3 cell line (supplemental Figure 3E). These findings using genetic tools reveal the on-target efficacy of NAMPT inhibitors in AML cells with reduced NAMPT expression.

The depletion of NAD via NAMPT inhibition leads to cell death, and the pathways driving this process include apoptosis, autophagy, and oncosis.^27-29 Flow cytometry following annexin V and 7-AAD staining showed that KPT-9274 treatment resulted in a significant increase in the percentage of annexin V–positive and 7-AAD–negative (apoptotic) and annexin V and 7-aminoactinomycin D (7-AAD) dual positive (necrotic) staining in NAMPT-deficient cells compared to control cells (Figure 3E). Next, we used a pan-caspase inhibitor (Z-VAD-FMK), necroptosis inhibitor (necrostatin-1 [Nec-1]), and ferroptosis inhibitor (ferrostatin-1) to determine the mode of death mediated by NAMPT inhibition in our study. As expected, Z-VAD-FMK completely rescued caspase-dependent cell death (Figure 3F, left) in NAMPT-deficient cells but only partially rescued total cell death (Figure 3F, right). Ferroptosis inhibition did not alter NAMPT-mediated cytotoxicity. The significant increase in cell death we observe in response to the Nec-1 inhibitor (Figure 3F, left) might be due to Nec-1’s role in enhancing other forms of cell death while inhibiting necroptosis, as reported previously.^30,31 Together, these data suggest that NAMPT inhibition results in higher cytotoxic effects via both caspase-dependent and independent cell death pathways on cells expressing lower levels of NAMPT.

Although −7/del(7q) is an independent high-risk factor, it often co-occurs with other cytogenetic abnormalities and somatic mutations. TP53 abnormalities occur in 43% to 56% of −7/del(7q) patients.^7,8 In the −7/del(7q) TCGA cohort, we found that TP53 is mutated or deleted in 12 of 22 cases (54.5%) of −7/del(7q) with partial NAMPT loss (Figure 4A). Therefore, we next tested NAMPT inhibition in TP53- and NAMPT-deficient cell lines.

We assessed NAMPT dependency by comparing Chronos scores between TP53 wild-type and mutant cell lines and observed no significant difference (supplemental Figure 4A). Furthermore, GDSC2 drug response and CCLE mutational data showed no significant difference in FK-866 response between wild-type and mutant TP53 cell lines (supplemental Figure 4B).

To experimentally validate our data mining, we used CRISPR-Cas9 to knockout TP53 in OCI-AML3 and MOLM-13 cell lines with and without NAMPT deficiency (Figure 4B). NAMPT-deficient cell lines with wild-type TP53 had comparable IC₅₀ values to those without TP53 when treated with increasing doses of KPT-9274 (Figure 4C). Next, we examined cell death induced by KPT-9274 in TP53 null and NAMPT-deficient cell lines. Surprisingly, annexin V/7-AAD staining revealed that the absence of TP53 protein did not diminish the effects of NAMPT inhibition on the death of cell lines with reduced NAMPT (Figure 4D). Edited OCI-AML3 cells treated with KPT-9274 showed no cleaved caspase 3 activation at 24 or 48 hours, unlike staurosporine-treated cells (Figure 4E), suggesting NAMPT inhibitor cytotoxicity may occur via a TP53- and caspase-3–independent cell death process. These data align with previous findings that NAMPT inhibition-induced NAD depletion causes cell death via oncosis, independent of TP53 and caspase-3.³² However, all cell death mechanisms in our study remain unexplored. In summary, our data indicate that the cytotoxic effects of KPT-9274 in cancer cells with reduced NAMPT levels persist despite the TP53 mutational status, highlighting the potential to exploit NAMPT inhibitors for the treatment of −7/del7q patients with TP53 deficiency.

To establish the clinical relevance of our findings, we assessed the effects of NAMPT pharmacological inhibition in primary patient samples (supplemental Table 4). We first compared the relative viability of samples from patients with AML with and without −7 after treatment with KPT-9274 in liquid cultures. Samples from patients with −7 AML were significantly more responsive to NAMPT inhibition than AML samples diploid for chromosome 7 (Figure 5A). In addition, −7 patient cells with P53 inactivation were equally responsive to NAMPT inhibition (Figure 5A). There was no significant viability difference among vehicle-treated groups at 0 and 48 hours, indicating that observed differences are not due to varying culture conditions for different genetic subtypes (supplemental Figure 5A). Next, we determined the effects of KPT-9274 on normal hematopoietic cells in liquid cultures. Up to 500 nM of KPT-9274 showed no significant cytotoxicity in these cells at 24 and 48 hours, indicating selective killing of −7 cells while sparing normal CD34⁺ hematopoietic cells (supplemental Figure 5B; Figure 5B), consistent with previous studies.^33-35 

Previous studies have validated in vitro colony-forming unit (CFU) assays as predictors of in vivo drug toxicity in normal cells and efficacy on leukemic cells in preclinical models.^36-38 Therefore, we used CFU assays to determine the effectiveness of pharmacological NAMPT inhibition in samples from patients with −7 AML with and without TP53 inactivation. Samples from patients with −7 AML treated with KPT-9274 formed significantly fewer colonies than vehicle-treated samples. Remarkably, TP53-deficient −7 samples showed the same sensitivity to KPT-9274 as −7 samples alone (Figure 5C). Notably, −7 samples with TP53 inactivation were also CK (supplemental Table 4). These findings confirm that NAMPT haploinsufficiency in −7 and some del(7q) patients, with or without TP53 inactivation or CK, presents a targetable therapeutic vulnerability.

PARP1 plays a critical role in chromatin organization and DNA repair. PARP1 requires oxidized NAD (NAD⁺) as a substrate for adenosine 5′-diphosphate ribosylation in DNA damage repair, and NAMPT inhibition reduces NAD⁺ available for PARP, leading to increased DNA damage.³⁹ Recent studies have shown that NAMPT inhibitors synergize with PARP inhibition in Ewing sarcoma.⁴⁰ We and others have shown that genes involved in DNA damage recognition and repair pathways, such as CUX1,^41,EZH2,^42,KMT2C,⁴³ and KMT2E,⁴⁴ are frequently deleted in −7 and del(7q) conditions,⁹ suggesting that DNA repair factor inhibitors could be a potential treatment in these cancers. Moreover, NAD is also generated via Preiss-Handler and Kynurenine (de novo) synthesis pathways in addition to the NAMPT-mediated salvage pathway. An increase in expression or function of the key players of those pathways, such as nicotinic acid phosphoribosyl transferase and quinolinate phosphoribosyl transferase, can replace NAD, resulting in NAMPT inhibitor resistance.^45,46 For these reasons, we explored whether combining PARP inhibitors with NAMPT inhibitors could enhance the cytotoxic effects in NAMPT haploinsufficient AML conditions and be used to overcome potential resistance. Incremental doses of KPT-9274 and the PARP inhibitor niraparib were administered based on the IC₅₀ value of each drug in NAMPT-deficient OCI-AML3 cell lines with or without P53 inactivation. We assessed drug synergy using the SynergyFinder application.⁴⁷ We observed significant synergy between KPT-9274 and niraparib in inhibiting cell viability in NAMPT-deficient AML cells (mean Zero interaction potency [ZIP] score, 10.24) and in NAMPT plus TP53-deficient AML cells, which displayed greater synergy (mean ZIP score, 17.10; Figure 6A-B). Remarkably, TP53-null cell lines also exhibited synergy when treated with the combination therapy (supplemental Figure 6; mean ZIP score, 10.99), suggesting that molecular pathways other than NAD synthesis might play a role in synergy. These results provide supporting evidence for PARP and NAMPT inhibitor combination therapy in −7/del(7q) and TP53-mutant patients with AML.

Despite molecular and genetic insights, no targeted therapies exist for −7/del(7q) myeloid malignancies. Previous studies have identified cancer-specific vulnerabilities using RNA interference (RNAi) and CRISPR screens.⁴⁸ In this study, we used DepMap genome-wide CRISPR-Cas9 screen, gene expression, and copy number data to identify NAMPT as a targetable, codeleted, and essential passenger gene in −7/del(7q) myeloid neoplasms (Figure 6C). Our study provides a data-driven framework for uncovering collateral lethal genes in other cancers with recurrent chromosomal arm–level deletions using CRISPR screens and gene expression data. In addition to identifying known CYCLOPS dependencies, our study uncovered several ARGuS genes. These findings suggest that the inhibition of ARGuS genes could present a novel therapeutic strategy, particularly in tumors in which these genes are amplified or aberrantly expressed at high levels. By exploiting the addictive dependence that some cancer cells develop on overexpressed genes, ARGuS vulnerabilities may expand the landscape of precision oncology targets, especially in cancers lacking targetable mutations but characterized by gene amplification or transcriptional upregulation.^22,49 

Although we found 30 essential genes on chromosome 7 that met the criteria for a CYCLOPS gene, NAMPT was the only gene with existing small-molecule inhibitors directly targeting it. We confirmed that cellular NAMPT levels determine sensitivity to pharmacological inhibition by using genetically modified cellular models of NAMPT haploinsufficiency. Congruently, primary samples from patients with −7 AML were more sensitive to NAMPT inhibition than other AML subtypes. This aligns with recent findings by Eldfors et al⁵⁰ that −7/del(7q) samples were more responsive to NAMPT inhibitors in an AML drug screen. Although Eldfors et al⁵⁰ demonstrated that primary −7/del(7q) samples were sensitive to NAMPT inhibitors, the relationship between NAMPT expression and sensitivity to NAMPT inhibitors was not established. In addition, we used CRISPR-Cas9 genetic editing to reduce NAMPT levels in AML cell lines and establish NAMPT inhibitor on-target effects. Moreover, we performed CFU assays using patient samples, a clinically relevant tool proven to effectively evaluate the toxic effects of drugs, to show that NAMPT inhibitors were effective in primary −7/del(7q) patient samples. Our work described herein firmly establishes NAMPT as a therapeutic vulnerability in −7/del(7q) myeloid neoplasms. Furthermore, it provides a strong rationale for designing clinical trials to target NAMPT-deficient cancers with NAMPT inhibitors.

We also demonstrated that the cytotoxic effects of the dual inhibitor KPT-9274 in −7 AML are not attributable to off-target PAK4 inhibition. In AML, inhibiting PAK4 has been associated with maintaining leukemic stem cell quiescence.⁵¹ However, functional studies in AML cell lines show that altering PAK4 expression does not affect proliferation or sensitivity to KPT-9274.³³ Using CRISPR-edited NAMPT-deficient cells and the NAMPT-specific inhibitor FK866, we confirmed that NAMPT is the primary driver of KPT-9274–associated cytotoxicity in AML. Moreover, PAK4 is not differentially expressed in −7 AML compared to other karyotypes in our patient cohort (data not shown).⁵² Thus, we identify NAMPT, rather than PAK4, as the primary driver of KPT-9274–associated cytotoxicity in −7 AML.

TP53 is the most frequently mutated gene in cancer, and more than half of 7/del(7q) patients display P53 inactivation, resulting in worse patient survival outcomes.^7,8 Despite advancements in cancer therapeutics, P53 inactivation remains a common cause of drug resistance, leading to poor prognoses.¹⁰ We observed that TP53 status did not affect NAMPT inhibitor sensitivity in cell lines and primary patient samples. We believe the P53-independent nature of NAMPT inhibitor–induced cell death widens the clinical applicability of NAMPT inhibitors. Moreover, 50% of patients with therapy-related myeloid neoplasm (t-MN) with −7/del7q, often combined with TP53 loss, face poor outcomes.⁷ Successful application of NAMPT inhibitors in the clinic has the potential for a substantial impact on the growing number of cancer survivors at risk of t-MN.

Studies have shown that NAMPT inhibitors mediate cell death via apoptosis, autophagy, and oncosis.^27-29 Although we found that NAMPT inhibitor treatment leads to P53-independent cell death, possibly via oncosis, we also saw an increase in annexin V–positive/7-AAD–negative staining, commonly used as a marker for early apoptosis. However, reports have shown that annexin V staining is not specific to apoptotic cell death. Indeed, oncosis also results in the exposure of phosphatidylserine residues on the outside layer of the plasma membrane, detected by annexin V,⁵³ which may explain our observed results of annexin V–positive cells without caspase 3 activation at 48 hours. However, because we observed partial rescue of cell death upon using a pan-caspase inhibitor, we predict that other caspase-dependent, nonapoptotic cell death pathways are at play. The exact cell death mechanisms in high-risk −7 and P53-inactive AML upon NAMPT inhibition will require systematic future investigations.

Previous studies have investigated NAMPT inhibitors as potential cancer therapeutics. Despite promising preclinical results, first-generation NAMPT inhibitors (FK866 and CHS828) failed in human clinical trials because of their limited efficacy and severe side effects.^54,55 In preclinical assessments, second-generation NAMPT inhibitors KPT-9274 and OT-82 have shown more promise in hematologic malignancies^33-35,56 and are undergoing phase 1 clinical trials (ClinicalTrials.gov identifiers: NCT04914845 and NCT03921879). OT-82 lacked cardiac, neurological, or retinal toxicities seen in other NAMPT inhibitors and did not affect mouse aging or longevity.⁵⁶ Our findings show that −7/del(7q) AML samples with NAMPT haploinsufficiency are highly sensitive to NAMPT inhibition. Low, nontoxic doses of KPT-9274 and OT-82 could be effective in −7/del(7q) patients while sparing normal hematopoietic cells. These lower doses are safe for normal hematopoietic cells, as shown in our study and others.^33,34 

Although NAMPT inhibitors are effective as single agents at low concentrations in −7 AML, resistance may limit their long-term efficacy. Previous studies have shown enhanced activity when combined with venetoclax or stearoyl-Coenzyme A desaturase inhibitors (CAY10566; MF-438).^33-35,50 Combining NAMPT inhibitors with the FDA-approved PARP inhibitor, niraparib, enhanced cytotoxicity in NAMPT-deficient AML cells, regardless of TP53 status, broadening therapeutic potential. Notably, PARP inhibitors alone show limited efficacy in patients with −7/del(7q) abnormalities,⁵⁰ making the observed sensitivity in TP53-null cells particularly compelling. Given that both PARP and sirtuins rely on NAD, NAMPT inhibition is expected to affect their function. Supporting this, prior work showed synergy between NAMPT and SIRT6 inhibition in AML.⁵⁷ These findings highlight the potential of NAMPT-based combination therapies in hematologic malignancies with NAMPT or DNA repair pathway deficiencies.

The authors acknowledge Amittha Wickrema for the gift of CD34⁺ primary human hematopoietic cells and Angela Stoddart for her feedback on this manuscript. The authors thank the University of Pennsylvania Stem Cell and Xenograft core (RRID: SCR_010035) staff for sample acquisition and shipment, the University of Chicago’s DNA Sequencing Facility staff for assistance with Sanger sequencing, and the Cytometry and Antibody Technology Facility staff for assistance with cell sorting. The University of Chicago DNA Sequencing Facility (RRID: SCR_019196) and Cytometry and Antibody Technology (RRID: SCR_017760) Facilities receive financial support from the Cancer Center Support Grant P30CA014599 from the National Institutes of Health (NIH)/National Cancer Institute (NCI).

This work was supported by grants from the NIH/NCI; R01CA231880 (M.E.M.); T32CA009594 (H.Z.B.); NIH/National Heart, Lung, and Blood Institute (R01HL142782 and R01HL166184 [M.E.M.]; F32 HL156600 [M.D.S.]); the V Foundation for Cancer Research Pediatric Cancer Research V Scholar All-Star Award (M.E.M.); and the Cancer Research Foundation Fletcher Scholar Award (M.E.M.). M.E.M. is a Scholar of the Leukemia & Lymphoma Society. M.P.C. is supported by U54CA283759 from the NCI and Veterans Merit Award (I01BX000918).

Contribution: M.D.S. designed and performed the experiments, analyzed and interpreted data, and wrote the manuscript; H.Z.B. designed and performed CD34 viability assays and cell death inhibitor experiments, analyzed and interpreted data, and contributed to writing methods and editing of the manuscript; S.K. provided technical assistance for cell culture, DNA sequencing, and immunoblotting; S.J.S. and M.P.C. provided primary material from patients with AML, protocols for primary human AML colony-forming unit assays, and feedback on the manuscript; and M.E.M. conceived the study, assisted in experimental design and data interpretation, supervised the study, edited the manuscript, and provided research infrastructure.

Conflict-of-interest disclosure: The authors declare no competing financial interests.

Correspondence: Megan E. McNerney, Department of Pathology, Knapp Center for Biomedical Discovery, The University of Chicago, 5-128 900 East 57th St, Chicago, IL 60637; email: megan.mcnerney@bsd.uchicago.edu.