Liposomal-Mitoxantrone Overcomes Acquired Resistance and Enhances Venetoclax Synergy in AML-Mechanistic Insights and Preliminary Clinical Evidence

Abstract

Introduction

Liposomal drug delivery has emerged as a promising strategy to improve the efficacy and safety of chemotherapeutic agents, yet whether and how liposomal reformulation fundamentally reshapes therapeutic response and resistance evolution remains unclear. This study compared liposomal mitoxantrone with free mitoxantrone and examined their effects on antileukemic activity, hematologic toxicity, resistance development, and cell-death pathways.

Methods

Liposomal and free mitoxantrone were evaluated in an MLL-AF9–driven acute myeloid leukemia mouse model by survival analysis and peripheral blood measurements. Secondary transplantation experiments were performed to evaluate the sensitivity of residual leukemia cells. Single-cell RNA sequencing was used to characterize transcriptional features of residual leukemia cells. Drug combination with venetoclax and the cell killing pattern of the two drugs were explored based on MOLM13 cell line. Exploratory clinical observations were additionally obtained from four patients who received liposomal mitoxantrone in combination with venetoclax.

Results

Liposomal mitoxantrone significantly prolonged mouse survival compared with free-mitoxantrone and control groups (both p < 0.001), while causing minimal myelosuppression. In contrast, free-mitoxantrone induced marked hematologic toxicity. Secondary transplantation demonstrated that residual cells following free-mitoxantrone treatment developed resistance upon re-exposure (p < 0.001), while those from liposomal-treated mice retained drug sensitivity and conferred sustained survival benefit (both p < 0.01). Single-cell RNA sequencing revealed that free-mitoxantrone induced transcriptional heterogeneity, characterized by enhanced metabolic and immune-related pathways, and monocyte-like features, whereas liposomal-mitoxantrone preserved a proliferative state similar to untreated cells. Consistently, liposomal-mitoxantrone demonstrated stronger synergistic effects with venetoclax.

Discussion

Liposomal reformulation altered the therapeutic, transcriptional, and cell-death pattern of mitoxantrone in the models. These findings suggest that liposomal mitoxantrone may reduce hematologic toxicity and limit the emergence of drug resistance.

Introduction

Liposomal drug delivery systems have provided substantial advances in the clinical translation of chemotherapeutic agents by modulating pharmacokinetics, biodistribution, and toxicity profiles.^1–3 Encapsulation within phospholipid bilayers prolongs systemic circulation, reduces off-target exposure, and enables more controlled drug release compared with free drugs, thereby improving the therapeutic efficiency of cytotoxic agents.^4–6 The success cases of liposomal formulations in clinical practice, such as Doxil and CPX-351 exemplifies the translational feasibility of this platform.^7–10 Pegylated liposomal doxorubicin significantly attenuates anthracycline-associated cardiotoxicity while maintaining antitumor efficacy, whereas CPX-351, by co-encapsulating daunorubicin and cytarabine at a fixed synergistic ratio, improves overall survival in secondary acute myeloid leukemia (AML). These paradigms demonstrate that liposomal reformulation can fundamentally enhance pharmacokinetic characteristics and therapeutic performance rather than merely alter formulation characteristics.

Mitoxantrone exerts potent DNA intercalating and topoisomerase II–inhibitory activity.^11,12 However, its clinical utility is constrained by suboptimal pharmacokinetic stability, cumulative cardiotoxicity, and rapid development of acquired resistance.^13,14 Liposomal encapsulation has been demonstrated that can alter intracellular drug trafficking, leading to delayed nuclear accumulation and a distinct temporal pattern of nuclear drug exposure compared with free drug.¹⁵ Such changes may influence the magnitude and duration of DNA damage, oxidative stress, and downstream cell-death programs. Meanwhile, mitoxantrone resistance has been linked to ATP-binding cassette transporters, particularly ABCG2/BCRP, whereas liposomal encapsulation has been reported to reduce P-glycoprotein-mediated drug extrusion and alter transporter activity in drug-resistant cancer cells.^16,17 These findings suggest that liposomal delivery may partially bypass transporter-mediated resistance and reshape intracellular mitoxantrone exposure. Furthermore, recent studies have shown that amphiphilic PEG moieties incorporated into liposomal bilayers can enhance membrane permeability to reactive oxygen species, including peroxide-derived radicals, thereby facilitating lipid peroxidation and ferroptosis induction in cancer cells.¹⁸ Consistently, emerging evidence indicates that the SLC7A11/GPX4 axis contributes to mitoxantrone resistance in AML cells, supporting a role for ferroptosis in mitoxantrone response.¹⁹ Thus, liposomal formulation may also have the potential to alter the conventional mitoxantrone-induced cell-killing pattern. Liposomal-mitoxantrone has shown improved tolerability and antitumor activity in recurrent or refractory peripheral T-cell lymphoma and extranodal NK/T-cell lymphoma.^20,21 However, whether and how liposomal reformulation can reprogram therapeutic response dynamics or resistance evolution remains insufficiently defined.

Facing the current unmet need, our research presented the characterization of liposomal formation-mediated modulation in the performance of mitoxantrone in the AML model. By integrating single-cell transcriptomic analysis and functional validation, we provided insight into the tumor cell killing pattern, avoidance of drug resistance, and potential combination strategy. These results established a translational and mechanistic framework supporting the clinical application of liposomal-mitoxantrone.

Material and Methods

Cell Line

Human leukemia cell line MOLM-13 was originally obtained from American Type Culture Collection (ATCC). The human acute myeloid leukemia cell line MOLM-13 was maintained in our laboratory and authenticated by short tandem repeat (STR) profiling conducted by Beijing Tsingke Biotech Co., Ltd. (Beijing, China). MOLM-13 cells were cultured in RPMI-1640 medium (Thermo Fisher Scientific) supplemented with 10% fetal bovine serum (FBS) and 1% penicillin and streptomycin and maintained at 37 °C in a humidified incubator with 5% CO₂.

Mice

Six- to eight-week-old C57BL/6 mice were purchased from Beijing Vital River Laboratory Animal Technology Co., Ltd and maintained under specific pathogen-free conditions with free access to food and water. At the end of the experiment, all surviving mice were euthanized by cervical dislocation. All procedures are carried out in accordance with institutional guidelines and approved animal care plans. All animal procedures were approved by the Institutional Animal Care and Use Committee of the State Key Laboratory of Experimental Hematology (IHCAMS-DWLL-NSFC2023015-2) and complied with the NIH Guide for the Care and Use of Laboratory Animals. Detailed grouping strategies can be referred to the related section in Figure legends. Mice were randomly assigned to the control, free-mitoxantrone, or liposomal-mitoxantrone group by simple randomization. Treatment was administered separately to each mouse, and all efficacy and toxicity outcomes were measured at the individual-animal level. Therefore, the individual mouse constituted the experimental unit. Each treatment group included six mice. The sample size was determined based on previous studies and our preliminary experiments.^22–24To minimize potential assessment bias, mice and collected samples were assigned coded identifiers that did not disclose treatment allocation. Investigators performing flow-cytometric sample acquisition, data analysis and survival assessment were blinded to group allocation. Treatment assignments were revealed only after the corresponding data collection and analysis had been completed. All animals enrolled were included in the final analysis, no predefined exclusion criteria were applied. Free mitoxantrone and liposomal mitoxantrone were administered according to the experimental design described below. Liposomal and free mitoxantrone were prepared by diluting mitoxantrone in sterile glucose solution immediately before administration. The dose was 3mg/kg. Before injection, liposomal and free mitoxantrone were diluted with sterile glucose to the required concentration. Mice received drugs via tail-vein injection. The injection volume was 200μL per mouse. Control mice received an equivalent volume of glucose solution following the same injection schedule. All formulations were prepared freshly before administration and were gently mixed before injection.

Drugs

For in vitro and vivo studies, ferrostatin-1 (Fer-1), Z-VAD, venetoclax, blank liposome, mitoxantrone, and liposomal-mitoxantrone are respectively from MCE and CSPC Pharmaceutical Group Limited.

Establishment and Monitoring of the Murine MLL-AF9 Leukemia Model

Bone marrow cells were harvested from 6 to 8-week-old wild-type mice. Lineage-negative (Lin-) cells were enriched using a hematopoietic stem/progenitor cell enrichment kit and subsequently transduced with MSCV-MLL-AF9-GFP retrovirus. A total of 5×10⁴ transduced cells were transplanted into C57BL/6 recipient mice via tail vein injection. Leukemia development was monitored by flow cytometric analysis of GFP+ cells in peripheral blood and bone marrow samples. In addition, leukemic cell burden in peripheral blood was quantified using an automated hematology analyzer (Mindray).

In vitro Chemotherapy Drug Treatment and Related Ferroptosis and Apoptosis Test of AML Cell Lines

The acute myeloid leukemia cell line MOLM-13 was cultured via RPMI-1640 medium containing 10% FBS and 1% penicillin and streptomycin. The cells were cultured in a 37 °C, 5% CO ₂ constant temperature incubator. During experimental processing, cells were treated with PBS or the corresponding drug. Cell counting was performed using the trypan blue staining method to eliminate interference from dead cells. Cell ferroptosis and apoptosis detection were performed using the RhnNox-6/CM-H2DCFDA ferroptosis fluorescence detection kit (Cat. No. C1190S, Beyotime Biotechnology) and Annexin V-YSFluor™ 647/7-AAD apoptosis fluorescence detection kit (Cat. No. 40312ES20, Yeasen Biotechnology). Operations were conducted according to the instructions. After staining, positive cells were detected and counted using flow cytometry to evaluate the proportion of ferroptosis and apoptosis in MOLM-13 cells.

Flow Cytometry Analysis

Flow cytometric analysis was performed on a BD FACSAria III flow cytometer using BD FACSDiva Software. For each sample, 10,000 cellular events were acquired. During data analysis, debris was first excluded according to the FSC-A and SSC-A distribution. Doublets were then removed using an FSC-A and FSC-H gate, and subsequent quantification was performed within the single-cell population. For the apoptosis assay, apoptotic cells were identified in the corresponding APC-A-positive gate. The detector voltage settings were FSC-A, 368; FSC-H, 368; SSC-A, 267; and APC-A, 550. For the ferroptosis assay, ferroptotic cells were quantified within the corresponding FITC/BB515-A and PE-A gate after debris and doublet exclusion. The detector voltage settings were FSC-A, 368; FSC-H, 368; SSC-A, 267; FITC/BB515-A, 330; and PE-A, 297. For the ferroptosis and apoptosis assay, FITC/BB515-A and PE-A signals were analyzed using the same gating strategy across all experimental groups.

Single-Cell Sample Preparation

Hindlimb bones were collected from leukemic mice and ground in PBS to release bone marrow cells. The resulting cell suspensions were filtered sequentially through 70-µm and 30-µm cell strainers. GFP+ cells were isolated by fluorescence-activated cell sorting. Sorted cells were resuspended in 1× PBS (Invitrogen) supplemented with 0.04% bovine serum albumin (BSA) at a final concentration of 700–1200 cells/µL and processed using the Chromium Single Cell 3′Kit (10x Genomics; v3.1, PN: 1000268, or v4, PN: 1000691) according to the manufacturer’s instructions at Novogene Bioinformatics Technology Co., Ltd. (Tianjin, China).

Single-Cell 3’ RNA Sequencing and Data Analysis

The cell suspension was loaded into 10x Chromium Chip (v3.1 PN:1000120) and barcoded with a 10x Chromium Controller. RNA from the barcoded cells was subsequently reverse-transcribed, amplified and prepared into sequencing libraries with 10x Library Construction Kit (v3.1 PN:1000190) according to the manufacturer’s instructions. Sequencing was performed with Illumina NovaSeq with 150-bp paired-end reads at Novogene Bioinformatics Technology Co., Ltd (Tianjin, China). Raw reads were processed following the steps of Quality Control, Generation and Analysis of Single-Cell Transcriptomes. Details can be referred to Supplementary Materials. To enable cross-species gene mapping, the R package “homologene” was applied to convert mouse gene symbols to their corresponding human orthologs. Pathway activity was evaluated using the R package “AUCell”. Single-cell pseudotime trajectory analysis was performed using the R package “Monocle”.²⁵ Mitochondrial gene percentage was calculated using the mouse mitochondrial gene pattern “^mt-”. Cells with mitochondrial gene content greater than 10% were excluded. Cells were excluded if they had fewer than 200 detected genes, more than 6,000 detected genes, more than 50,000 RNA counts, or mitochondrial gene content greater than 10%. Potential doublets were reduced by excluding cells with abnormally high numbers of detected genes or RNA counts. Clustering was performed with a resolution of 0.1. Highly variable genes were identified using the “vst” method, and 300 variable features were selected. PCA was performed, and the first 10 principal components were used for nearest-neighbor graph construction, clustering, and UMAP visualization.

Bliss Independence Score Calculation

To evaluate the anti-tumor activity of free-mitoxantrone and liposomal-mitoxantrone combination with venetoclax conditions, we referred to the drug combination index calculation method (Bliss independence analysis) in previous literature to evaluate the synergistic effect of free-mitoxantrone and liposomal-mitoxantrone.²⁶

The Bliss independence is to evaluate whether two drugs conduct cellular killing efficacy independently. For each concentration pair, the expected effect was calculated as follows:

Eexp = 1 − (1 − Ea) (1 − Eb)

For drug a and drug b, their single-agent effects are denoted as Ea and Eb, respectively. Eexp represents the expected combined effect of drug combination.

Bliss score = Eobs − Eexp

Where Eobs represents the actual drug combination effect. Then the Bliss score was calculated as the difference between observed and expected effect of drug combination. Therefore, a positive score means that the combination performs better than expected, indicating synergy, whereas a negative score indicates a weaker than expected effect, indicating antagonism. Mean Bliss scores were calculated as the mean value across the concentration-response matrix. The effects of drug combinations were evaluated using a 5×5 checkerboard matrix. Venetoclax was tested at 0–40 μM (0, 5, 10, 20, and 40 μM), in combination with either free mitoxantrone at 0–4 μg/L (0, 0.5, 1, 2, and 4 μg/L) or liposomal mitoxantrone at 0–32 μg/L (0, 4, 8, 16, and 32 μg/L). Each condition was tested in four replicates.

Statistical Analysis

The statistical analysis and data visualization of the study were conducted employing R and GraphPad Prism 10. For the analysis of tail blood routine data at different time points, statistical significance was analyzed using two-way repeated-measures ANOVA. Sphericity was not assumed, and the Geisser–Greenhouse correction was applied to the repeated time factor. Comparisons of WBC among the three groups at each individual time point, ferroptosis and apoptosis cell rates and viable cell counts were performed using Tukey’s multiple-comparisons test. For the analysis of antitumor effects of two drugs on pretreated cells, we employed Dunnett’s multiple comparisons test. The survival status of mice was displayed using Kaplan-Meier curves and compared between groups using log rank tests. Cell viability was assessed after 48 h of treatment with a series of concentrations of free or liposomal mitoxantrone. IC50 values were calculated from GraphPad Prism. p<0.05 is considered statistically significant, and statistical significance is indicated as follows: p <0.05 represents *p <0.01 represents * *p <0.001 represents * * *p <0.0001 represents * * * *. Some of the image materials are sourced from Servier Medical Art (https://smart.servier.com). The use of images follows the Creative Commons Attribution 4.0 International License (CC BY 4.0, https://creativecommons.org/licenses/by/4.0/).

Results

Liposomal-Mitoxantrone Demonstrated a Significantly Better Efficacy and Less Bone Marrow Suppression Compared to Free-Mitoxantrone

To assess the clinical potential of liposomal mitoxantrone and free mitoxantrone, we first evaluated their antileukemic efficacy. After the leukemia model was established, the mice were randomly assigned to receive liposomal mitoxantrone, free mitoxantrone, or no treatment. Untreated leukemia-bearing mice served as the control group (Figure 1A). It demonstrated that mice treated with free-mitoxantrone exhibited significantly prolonged survival compared with control mice (p < 0.001), whereas liposomal-mitoxantrone resulted in a markedly greater survival benefit relative to both control and free-mitoxantrone groups (p < 0.001) (Figure 1B). Median survival was 21 days in the control group and 37 days in the free-mitoxantrone–treated group. In contrast, median survival was not reached in the liposomal-mitoxantrone–treated group, in which only one mouse died at day 44, and all remaining mice survived throughout the observation period (Figure 1B). All dead mice exhibited a marked increase in GFP+ cells and WBC prior to death, confirming leukemia-associated mortality (Figure 1C and Figure S1).

Figure 1.

Liposomal and free-mitoxantrone demonstrated different therapeutic efficacy and leukocyte suppression features. (A) Experimental design for therapeutic evaluation in the mice with MLL-AF9 leukemia. The mice were initially injected with 5×10⁴ MLL-AF9 leukemic cells via the tail vein on day 0. On day 14, mice were injected with 3mg/kg liposomal-mitoxantrone and 3mg/kg free-mitoxantrone via the tail vein or glucose solution as controls. (B) Kaplan–Meier survival analysis of leukemic mice treated with free-mitoxantrone, liposomal-mitoxantrone, or glucose solution as a control. (C) GFP+ leukemic cells in peripheral blood over time, as determined by flow cytometry. (D) Experimental scheme for toxicity evaluation in healthy mice. We administered mice in 3mg/kg liposomal-mitoxantrone, 3mg/kg free-mitoxantrone, and glucose solution as controls via tail vein injection to test the effect of liposomal-mitoxantrone and free-mitoxantrone on bone marrow. (E) WBC in peripheral blood of healthy mice treated with free-mitoxantrone, liposomal-mitoxantrone, or glucose solution as control. Two-way repeated-measures ANOVA: time: p< 0.0001, treatment group: p< 0.0001, time × treatment group interaction p< 0.0001. Comparisons among the three groups at each individual time point were performed using Tukey’s multiple-comparisons test. Free-mito: Free-mitoxantrone; Lip-mito: liposomal-mitoxantrone. p <0.05 represents *, p <0.01 represents * *, p <0.001 represents * * *, p <0.0001 represents * * * *.

To evaluate impacts on hematopoiesis side effect, healthy mice were assigned to the control, liposomal mitoxantrone, or free mitoxantrone group. (Figure 1D). Early after administration (day 3–12), mice treated with free-mitoxantrone exhibited a marked reduction in white blood cell (WBC) counts compared with the control group (Figure 1E). At the later stage (day 18–24), a gradual recovery of WBC levels was observed in mice receiving free-mitoxantrone (Figure 1E). In contrast, WBC counts in the liposomal-mitoxantrone group remained close to those in the control group throughout the monitoring period (Figure 1E).

Cells Treated with Free-Mitoxantrone Developed Resistance, Whereas Those Treated with Liposomal-Mitoxantrone Did Not

Subsequently, to evaluate the acquired resistance to liposomal-mitoxantrone and free-mitoxantrone, mouse models were established using residual tumor cells following treatment with either liposomal-mitoxantrone or free-mitoxantrone (Figure 2A). The corresponding formulations were then re-administered to assess therapeutic response. In AML models initiated from free-mitoxantrone–residual cells, treated mice exhibited significantly shorter survival than the untreated group (median survival: Treated with Free-mito 24 days vs without treatment 27 days; p < 0.001) (Figure 2B and C). GFP+ cells did not increase before death in the treated group, but increased in the control group, which indicated little to no therapeutic efficacy but toxicity of free-mitoxantrone (Figure 2D). The data of WBC counts monitoring at different time points are also presented in Figure S2. In AML models initiated from liposomal-mitoxantrone residual cells, liposomal-mitoxantrone treatment prolonged survival in comparison with untreated controls (median survival: Lip-mito 1 day group treated with Lip-mito 44 days vs Lip-mito 1 day group without treatment 25.5 days, Lip-mito 7 day group treated with Lip-mito 37.5 days vs Lip-mito 7 day group without treatment 25.5 days) (p < 0.01) (Figure 2E and F). GFP+ cells and WBC counts increased in the peripheral blood before death, which suggested that all of them died of leukemia (Figure 2G and S3).

Figure 2.

Cells exposed to free-mitoxantrone developed resistance, but those exposed to liposomal formulation did not. (A) Experimental scheme illustrating the collection of residual leukemic cells. The mice were injected with 5×10⁴ MFF-AF9 leukemic cells via the tail vein on day 0 and administered with 3mg/kg liposomal-mitoxantrone, 3mg/kg free-mitoxantrone on day 14, respectively. On day 14, we sacrificed the mice and collected the bone marrow cells and named them the control group. On day 15, we sacrificed the mice and collected the bone marrow cells and named them as Free-mito 1 day and Lip-mito 1 day. On day 21, mice were sacrificed and bone marrow cells were named as Lip-mito 7 day. (B) Experimental design of cell resistance after exposure to free-mitoxantrone. Secondary recipient mice were transplanted with 5×10⁴ MLL-AF9 leukemic cells. On day 14, mice were administered with 3mg/kg free-mitoxantrone or a glucose solution as the control. (C) Survival of secondary recipient mice transplanted with residual leukemic cells from free-mitoxantrone-treated mice and subsequently treated with free-mitoxantrone or left untreated. (D) Percentage of GFP+ leukemic cells in the peripheral blood of secondary recipient mice transplanted with residual leukemic cells from free-mitoxantrone-treated mice. (E) Experimental design of cell resistance after exposure to 1day and 7day liposomal-mitoxantrone. Secondary recipient mice were transplanted with 5×10⁴ MLL-AF9 Lip-mito 1day and 7day treated residual cells. On day 14, mice were administered with 3mg/kg liposomal-mitoxantrone or a glucose solution as the control. (F) Survival of secondary recipient mice transplanted with residual leukemic cells from Lip-mito 1day and 7day treated mice, with or without subsequent liposomal-mitoxantrone treatment. (G) Percentage of GFP+ leukemic cells in the peripheral blood of secondary recipient mice transplanted with residual cells from Lip-mito 1day and 7day treated cells.

Residual Cells After Liposomal and Free-Liposomal Treatments Exhibited Heterogeneous Transcriptome Features

In light of the marked differences in resistance between liposomal and free-mitoxantrone-treated residual cells, we applied single-cell RNA sequencing to investigate the potential underlying mechanisms. After the quality control, 8855, 11,521, 10145, and 10133 cells from free-mitoxantrone, liposomal-mitoxantrone 1day, liposomal-mitoxantrone 7day, and the control group were finally submitted to analysis, respectively. We found that cells from two liposomal-treated groups and the control group predominantly clustered together, whereas cells from the free-mitoxantrone-treated group did not cluster with these populations on the UMAP plot (Figure 3A and B). These results suggested that liposomal-mitoxantrone treatment induced minimal transcriptomic alterations in residual cells, whereas free-mitoxantrone treatment induced marked transcriptional changes in residual cells. Next, we analyzed the proportional contributions of residual cells from each treatment group across the identified single-cell clusters (Figure 3C and D). Clusters 0 and 1 were predominantly composed of cells derived from the 1day and 7day liposomal-mitoxantrone treated groups as well as the control group, whereas clusters 2 and 4 were largely enriched for cells from the free-mitoxantrone treated group (Figure 3D and E). To further characterize the biological features of each cell cluster, we identified cluster-specific marker genes and performed enrichment analysis (Figure 3F and Table S1). Clusters 0 and 1, predominantly composed of 1day and 7day liposomal-mitoxantrone-treated cells and control cells, were enriched in chromatin organization, transcriptional regulation, and mitotic cell cycle pathways, indicating preservation of proliferative capacity. In contrast, clusters 2 and 4, enriched in free-mitoxantrone-treated cells, were associated with immune and inflammatory responses, lipid metabolic reprogramming, and growth regulation, suggesting stress-induced transcriptional remodeling.

Figure 3.

Single-cell transcriptomic landscape of AML cells after liposomal and free-mitoxantrone treatments. (A) UMAP projection of AML cells from control, Free-mito, and Lip-mito 1-day and 7-day groups. (B) UMAP visualization showing the distribution of cells from each treatment group plotted, respectively. (C) UMAP embedding colored by identified cell clusters. (D) Bar plot showing the proportion of each cell cluster across different treatment groups. (E) Split UMAP plots displaying cluster distribution within each treatment group, respectively. (F) Heatmap showing the top marker genes for each AML cluster and their associated enriched biological processes and signaling pathways.

Liposomal-Mitoxantrone Showed Stronger Synergy with Venetoclax Through Distinct Gene Expression Profiling After Treatments

Next, pseudotime trajectory analysis was performed to delineate the developmental landscape of residual AML cells. The reconstructed trajectory revealed a continuous differentiation progression across all cells (Figure 4A). Free-mitoxantrone-treated cells preferentially occupied a distinct region of the inferred trajectory, whereas untreated and liposomal-mitoxantrone-treated cells showed greater overlap in other regions (Figure 4B–D). This distribution was consistent with the greater transcriptional remodeling observed after free-mitoxantrone treatment. Pseudotime-dependent monocyte feature pattern and UMAP visualization demonstrated an enrichment of monocyte-like cells in the free-mitoxantrone group, whereas liposomal-mitoxantrone-treated cells had lower monocyte score (Figure 4E and F). Given that monocyte-like states have been linked to venetoclax resistance, we therefore hypothesized that liposomal-mitoxantrone may represent a more effective agent combined with venetoclax. In MOLM-13 cells, Bliss independence analysis demonstrated synergistic cytotoxicity for mitoxantrone and venetoclax combinations; notably, venetoclax plus liposomal-mitoxantrone showed significantly stronger synergy than venetoclax plus free-mitoxantrone (Figure 4G and H).

Figure 4.

Single-cell profiling of residual AML cells following free and liposomal-mitoxantrone treatment revealed divergent differentiation states and distinct venetoclax responses. (A) Pseudotime trajectory reconstruction of residual AML cells after treatment illustrated differentiation progression across all groups. (B) Distribution of differentiation phases among control, free-mitoxantrone, and 1day and 7day liposomal-mitoxantrone-treated residual cells. (C) Group-wise visualization of residual cells distributed along the pseudotime trajectory. (D) Density distribution of residual cells along pseudotime. (E) Monocyte signature scores change along pseudotime. (F) UMAP visualization of monocyte signature scores in all cells and in each treatment group. (G and H) Drug matrix heatmaps of venetoclax combined with free-mitoxantrone (G) and liposomal-mitoxantrone (H). Positive Bliss scores indicated synergy and were displayed on the down-sided heatmaps.

Based on in vitro evidence suggesting that liposomal-mitoxantrone may be a promising partner for venetoclax-based regimens, we retrospectively explored the therapeutic potential of this combination in four clinical cases. In this case series, four patients with newly diagnosed acute myeloid leukemia (AML), aged 43 to 62 years, were treated with a combination of venetoclax, azacitidine, and liposomal-mitoxantrone (VAM). All patients were classified as Adverse-risk according to the ELN 2022 criteria. Detailed clinical characteristics of the four patients are summarized in Table 1 and Figure 5. Following induction therapy with VAM, three of the four patients achieved complete remission (CR) with negativity for measurable residual disease (MRD). Patient 1 achieved molecular complete remission with a negative MLL-AF10 fusion gene. Notably, among the three responding patients, neutrophil recovery occurred between 21 and 32 days, and platelet recovery between 0 and 14 days; two of these patients never experienced platelet counts below 20×10⁹/L. Red blood cell transfusion requirements in these three patients ranged from 5 to 6 units, and platelet transfusion requirements from 0 to 4 units. All patients developed febrile neutropenia. Patients 1, 2, and 4 presented with pulmonary infections at admission, while Patient 3 developed oral ulcers, and all infections resolved following antimicrobial therapy. Adverse events of AML patients receiving VAM regimen can be reffered to Table S2.

Table 1.

Clinical Information of the Series of Patients Received Therapy of Venetoclax, Azacitidine, and Liposomal Mitoxantrone

	Patient 1	Patient 2	Patient 3	Patient 4
Date of admission	2025.12.23	2025.12.02	2025.12.26	2025.12.29
Age	49	50	43	62
Gender	Female	Male	Female	Male
Complete blood count	WBC 5.36×10^9 /L	WBC 15.40×10^9 /L	WBC 1.65×10^9 /L	WBC 8.76×10^9 /L
	Hb 78g/L	Hb 68g/L	Hb 79g/L	Hb 73g/L
	PLT 170×10^9 /L	PLT 18×10^9 /L	PLT 45×10^9 /L	PLT 63×10^9 /L
Cytogenetics	46,XX,ins(10;11)(p13;q23q13)[17]/46,XX[3]	46,XY,add(17)(p11.1)[9]/45,XY,der(10;17)(q10;q10)[6]/46,XY[5]	46,XX[10]	46,XY[20]
Molecular genetics	MLL-AF10,SBDS; SETD2	MLL-PTD; ASXL1; U2AF1; KRAS; TP53	DDX41;ASXL1	MLL-PTD; DDX41; CEBPA;ASXL1
ELN 2022 risk classification	Adverse	Adverse	Adverse	Adverse
Neutrophil recovery time	21days	35days	23days	32days
Platelet recovery time	0days (Not lower than 20×10^9 /L)	25days	0days (Not lower than 20×10^9 /L)	14days
Therapy response	CR after one course of chemotherapy	NR	CR after one course of chemotherapy	CR after one course of chemotherapy
Amount of red blood cell transfusions	5.5U	23.5U	5U	6U
Amount of platelet cell transfusions	0	11	1	4
Treatment regimen	Venetoclax 100mg -d2, 200mg -d1, 400mg d1-7; Azacitidine 128mg d1-7; Liposomal-mitoxantrone 40mg d1	Venetoclax 100mg d1, 200mg d2, 400mg d3-9; Azacitidine 150mg:d3-6,100mg:d7-9; Liposomal-mitoxantrone 40mg d2	Venetoclax100mg d1, 200mg d2,400mg d3-9; Azacitidine 130mg d1-7; Liposomal-mitoxantrone 40mg d2	Venetoclax 50mg d1,2; Azacitidine 150mg d1-7; Liposomal-mitoxantrone 50mg d1

Figure 5.

Dynamic changes of patient blood routine with treatment time in a clinic series. Dynamic changes of WBC (A), RBC (B), HB (C), PLT (D), and NEU (E), with treatment time in the clinic series.

Liposomal-Mitoxantrone Predominantly Triggered Ferroptosis in AML Cells, Whereas Free-Mitoxantrone Induced Apoptosis

Single-cell RNA sequencing revealed pronounced heterogeneity among residual leukemic cells, potentially reflecting distinct tumor-killing mechanisms induced by the two formulations. To clarify the programmed cell death (PCD) modalities involved, we analyzed signature scores for multiple PCD pathways. Free-mitoxantrone-treated residual cells exhibited elevated expression of genes associated with apoptosis, autophagy, and ferroptosis resistance (Figure 6A). In contrast, short-term and long-term liposomal-mitoxantrone-treated residual cells did not show upregulation of ferroptosis resistance signatures and instead displayed profiles comparable to control cells (Figure 6A). These findings suggest that residual cells surviving free-mitoxantrone treatment may acquire adaptive resistance to ferroptosis, whereas ferroptosis may remain functionally active and contribute to AML cell elimination following liposomal-mitoxantrone treatment. To validate these findings, apoptosis and ferroptosis were quantified in MOLM-13 cells. At 48 h, the IC50 values were 1.16 µg/L for free-mitoxantrone and 8.153 µg/L for liposomal-mitoxantrone. Accordingly, 1 μg/L free mitoxantrone and 8 μg/L liposomal mitoxantrone, which approximate their respective 48-h IC50 values, were selected to provide approximately IC50-equivalent cytotoxic exposures for subsequent analyses. Based on these results, MOLM-13 cells were treated with 1 µg/L free-mitoxantrone, 1 µg/L liposomal-mitoxantrone, or 8 µg/L liposomal-mitoxantrone for subsequent analyses. To further investigate the involvement of ferroptosis and apoptosis in drug-induced cell death, cells in each treatment group were co-treated with either 5 µM Fer-1, or 50 µM Z-VAD. After 48 h of incubation, ferroptotic cell death, apoptotic cell death, and viable cell counts were assessed. According to the results, ferroptosis was most pronounced in the 8 µg/L liposomal-mitoxantrone group, followed by 1 µg/L liposomal-mitoxantrone, 1 µg/L free-mitoxantrone, and control (Figure 6B). Moreover, inhibition of apoptosis by Z-VAD further increased ferroptosis in all three mitoxantrone-treated groups, suggesting a compensatory shift toward ferroptotic cell death when apoptosis was blocked. (Figure 6B). In contrast, apoptosis was significantly more prominent in the 1 µg/L free mitoxantrone group than in either of the liposomal mitoxantrone groups (Figure 6C). Inhibition of ferroptosis by Fer-1 further enhanced apoptosis in the free mitoxantrone group, indicating that blockade of ferroptosis may redirect cell death toward apoptosis under free mitoxantrone treatment. Consistent with these findings, cell count analysis showed that Fer-1 partially rescued cell viability in both liposomal mitoxantrone groups, whereas it had no significant effect on the cytotoxicity of free mitoxantrone (Figure 6D). Conversely, apotosis inhibitor Z-VAD attenuated the cytotoxic effects of 8 µg/L liposomal mitoxantrone and 1 µg/L free mitoxantrone, but did not significantly affect the cytotoxicity of 1 µg/L liposomal mitoxantrone (Figure 6D). Taken together, these results suggest that liposomal mitoxantrone predominantly induces ferroptosis in MOLM-13 cells, whereas free mitoxantrone primarily triggers apoptosis. Pharmacological inhibition experiments further support the distinct cell death mechanisms induced by the two formulations: inhibition of ferroptosis attenuated the cytotoxicity of liposomal mitoxantrone, while inhibition of apoptosis reduced the cytotoxicity of free mitoxantrone and high-dose liposomal mitoxantrone.

Figure 6.

Liposomal-mitoxantrone and free-mitoxantrone induce distinct cell death programs in MOLM-13 cells. (A) Ridge plots of single-cell gene signature scores for apoptosis, pyroptosis, autophagy, necroptosis, ferroptosis (markers and suppressors), and cuproptosis-related pathways (promoting and inhibitory signatures) in control, free-mitoxantrone, and 1-day and 7-day liposomal-mitoxantrone treated leukemic cells, based on single-cell RNA sequencing. Each ridge represents the density distribution of pathway signature scores across individual cells within one treatment condition. The x-axis indicates the calculated pathway activity score. A rightward shift of the density curve indicates an increase in the relative activity of the corresponding pathway, while overlapping curves indicate comparable pathway activity among groups. (B and C) The proportion of ferroptosis (B) and apoptosis (C) cells in different drug-treated cell groups. (D) Viable cell counts of different drug-treated cell groups. Comparisons of ferroptosis and apoptosis cell rates and viable cell counts were performed using Tukey’s multiple-comparisons test. p <0.05 represents *, p <0.01 represents * *, p <0.001 represents * * *, p <0.0001 represents * * * *.

Additionally, to determine whether the drug-resistant phenotype persisted after drug withdrawal and to minimize the potential influence of residual drug, MOLM-13 cells were pretreated with liposomal mitoxantrone or free mitoxantrone for 48 h. After drug removal, the pretreated cells were cultured in drug-free medium and passaged twice. The cells were then seeded at 1×10⁵ cells per well and re-challenged with either liposomal mitoxantrone or free mitoxantrone. Cells previously exposed to liposomal mitoxantrone remained sensitive to both formulations while cells pretreated with free mitoxantrone showed reduced sensitivity to free mitoxantrone upon re-challenge (Figure S4), suggesting that free mitoxantrone induced a more persistent resistance phenotype.

Discussion

Our study suggested that liposomal-mitoxantrone had a different therapeutic effect from free-mitoxantrone, characterized by different tumor-killing mechanisms, which result in decreased acquired drug resistance, enhanced anti-leukemic efficacy, synergistic cytotoxicity with venetoclax, and reduced hematologic toxicity. In addition, we provided preliminary clinical evidence of the combination regimen of venetoclax, azacitidine, and liposomal-mitoxantrone in AML patients.

Liposomal delivery systems can influence not only the plasma pharmacokinetics of anticancer agents but also their intracellular disposition kinetics.¹⁵ This feature may be particularly relevant for drugs whose primary site of action is the nucleus, such as mitoxantrone. Since liposomal formulations can generate a distinct temporal pattern of nuclear drug exposure compared with the free drug, thereby potentially altering both the duration and intensity of pharmacological activity.¹⁵ In addition, liposomal encapsulation may modulate cellular drug efflux by reducing P-glycoprotein-mediated drug extrusion and altering the activity of drug transporters in resistant cancer cells.¹⁶ Notably, amphiphilic polyethylene glycol moieties incorporated into liposomal bilayers may also enhance membrane permeability to reactive oxygen species, thereby promoting lipid peroxidation and facilitating ferroptotic cell death.¹⁸ Given the emerging evidence linking ferroptosis dysregulation to mitoxantrone resistance in AML, these mechanisms highlight the therapeutic potential of liposomal mitoxantrone.¹⁹ Beyond improving systemic pharmacokinetic features, liposomal formulation may also have potential in influencing nuclear drug exposure, attenuating transporter-mediated drug efflux, and increasing susceptibility to ferroptosis, thereby improving the therapeutic efficacy of mitoxantrone in AML.

Liposomal encapsulation has been widely applied in clinical oncology and has achieved significant therapeutic success, as demonstrated by several approved liposomal agents. For example, CPX-351 (liposomal cytarabine/daunorubicin) delivers both agents at a fixed synergistic ratio and exhibits pharmacokinetic and pharmacodynamic properties distinct from conventional formulations.^27–30 Similarly, pegylated liposomal doxorubicin shows markedly altered tissue distribution and reduced cardiotoxicity compared with free doxorubicin.^31–34 In our case of mitoxantrone, liposomal formulation likely contributes to decreased acquired resistance, reduced hematologic toxicity, and synergy with venetoclax. Meanwhile, liposomal delivery may enable the redevelopment of combination regimens previously limited by toxicity or resistance, particularly in high-risk and relapsed or refractory disease.

Venetoclax-based therapy has substantially advanced AML treatment, particularly for aged patients or patients unfit for intensive chemotherapy.^35–37 In this population, the combination of venetoclax and azacitidine has demonstrated superior response rates and survival compared with hypomethylating agents alone.^38–41 However, the limited response durability and the non-negligible relapse rate highlighted the unmet need of exploring novel combination strategies.^39,42 In this context, our research data demonstrated stronger synergy between venetoclax and liposomal-mitoxantrone compared with free-mitoxantrone, potentially attributed to the unique tumor-killing pattern and reduced hematologic toxicity of liposomal-mitoxantrone. Consistently, our clinical case series demonstrated the potential that the VAM regimen may achieve high therapeutic efficacy while alleviating the treatment-related myelosuppression in clinical practice. Recent research demonstrated that AML cells with monocyte differentiation characteristics, often downregulate BCL-2 and rely on MCL-1 to maintain mitochondrial oxidative phosphorylation and cell survival, thus gain the therapeutic resistance resistant to venetoclax.⁴³ Notably, our single-cell RNA-sequencing analysis revealed a significant enrichment of monocytic-like features among residual AML cells following treatment with free mitoxantrone, whereas such enrichment was not observed after liposomal mitoxantrone treatment. These findings suggest that the enhanced efficacy of the venetoclax–liposomal mitoxantrone combination has the high potential of being attributed to the absence of monocytic-like features in residual cells from liposomal mitoxantrone treatment. Despite the limited sample size, this small case series provides the preliminary clinical evidence supporting the combination efficacy of liposomal-mitoxantrone and venetoclax in patients with AML. Together, these findings suggest that liposomal-mitoxantrone may potentially represent a promising partner for venetoclax and warrant further clinical evaluation in AML.

There were limitations to this study. First, the limited sample size of single-cell sequencing analysis may constrain our conclusions. Future studies incorporating larger cohorts and single-cell multi-omics approaches are warranted to comprehensively characterize the biological features of residual cells following treatment with liposomal and free mitoxantrone from the perspective of the entire transcriptional regulatory landscape. Second, the clinical evidence supporting the combination of venetoclax and liposomal mitoxantrone was derived from a small case series of only four patients. Larger patient cohorts and well-designed prospective randomized controlled trials will therefore be required to provide more definitive evidence regarding the efficacy and safety of this therapeutic strategy. Third, patient-derived xenograft and primary-cell AML models are essential for bridging mechanistic findings from the laboratory to the clinic, whereas the in vivo experiments in the present study were primarily based on an MLL-AF9 AML model. Further validation in PDX models and primary AML samples will be important to establish the translational relevance of our findings. Despite these limitations, our study demonstrates that liposomal reformulation alters the resistance feature and drug-combination characteristics of mitoxantrone, highlighting liposomal mitoxantrone’s therapeutic potential in AML.

Funding Statement

Beijing Xisike Clinical Oncology Research Foundation (Y-SYBLD2022ZD-0031), National Natural Science Foundation of China (82370183), Tian Jin Natural Science Foundation (23JCZXJC00310), CAMS Innovation Fund for Medical Sciences (2023-I2M-C&T-A-012), Tianjin Municipal Science and Technology Commission Grant (25ZXRKSY00050).

Declaration of generative AI and AI-assisted technologies in the manuscript preparation process

During the preparation of this work, the authors used ChatGPT (OpenAI) (version 5.3 and 5.4)for language polishing and improving the readability of the manuscript. The authors thoroughly reviewed and edited the generated content and take full responsibility for the final version of the manuscript.

Abbreviations

AML, Acute myeloid leukemia; ANOVA, Analysis of variance; ATP, Adenosine triphosphate; CR, Complete remission; ELN, European LeukemiaNet; Fer-1, Ferrostatin-1; Free-mito, Free-mitoxantrone; GFP, Green fluorescent protein; Hb, Hemoglobin; IC50, Half-maximal inhibitory concentration; Lip-mito, Liposomal-mitoxantrone; MRD, Measurable residual disease; NEU, Neutrophil; NR, No response; PBS, Phosphate-buffered saline; PCD, Programmed cell death; PDX, Patient-derived xenograft; PEG, Polyethylene glycol; PLT, Platelet; RBC, Red blood cell; ROS, Reactive oxygen species; WBC, White blood cell.

Data Sharing Statement

All related data can be acquired upon the reasonable request by contacting corresponding author.

Ethics Approval

Institutional ethics committee of Blood Diseases Hospital Ethics Committee have approved the research. Ethic number is IIT2020024-EC-1. All animal procedures were approved by the Institutional Animal Care and Use Committee of the State Key Laboratory of Experimental Hematology (IHCAMS-DWLL-NSFC2023015-2) and complied with the NIH Guide for the Care and Use of Laboratory Animals.

The overall design and conduction of the research was fully accordance with the Declaration of Helsinki. For the patients enrolled in the clinic series in the research, written informed consent was collected.

Author Contributions

All authors made a significant contribution to the work reported, whether that is in the conception, study design, execution, acquisition of data, analysis and interpretation, or in all these areas; took part in drafting, revising or critically reviewing the article; gave final approval of the version to be published; have agreed on the journal to which the article has been submitted; and agree to be accountable for all aspects of the work. All authors agreed to submit the manuscript. Everyone has no discrepancy for the research content.

Disclosure

The authors declare that the research was conducted in the absence of any potential competing interests.