Identification of glucuronic acid as a biomarker of poor prognosis in acute myeloid leukemia based on plasma metabolomics

Abstract

Metabolic abnormalities have been identified in various solid tumors and hematologic diseases, with reprogramming of central carbon metabolism occurring to promote disease progression. However, the metabolic profile of central carbon in acute myeloid leukemia (AML) remains unknown. We employed targeted metabolomics to analyze the alterations in central carbon metabolites present in the blood of acute myeloid leukemia (AML) patients. Models constructed using orthogonal partial least squares discriminant analysis (OPLS-DA) were utilized to evaluate intergroup differences in metabolite levels. Furthermore, a public database facilitated the kyoto encyclopedia of genes and genomes (KEGG) pathway enrichment analysis. Additionally, metabolites exhibiting significant differences were selected, and their effects on the proliferation and drug resistance of human myeloid leukemia cell lines were validated in vitro using CCK-8 analysis, MTT assays, and flow cytometry. Our results indicated that 27 targeted metabolites were up-regulated and eight targeted metabolites were down-regulated in the AML group. These metabolites were primarily enriched in pathways related to the biosynthesis of cofactors, glyoxylate and dicarboxylate metabolism, glucagon signaling, 2-oxocarboxylic acid metabolism, biosynthesis of amino acids, the citrate cycle (TCA cycle), and central carbon metabolism in cancer. Notably, significant changes were observed in malic acid, alpha-ketoisovaleric acid, and glucuronic acid. In vitro experiments demonstrated that exogenous glucuronic acid can promote the growth and drug resistance of human AML cells. In conclusion, this study reveals alterations in central carbon metabolites in the blood of AML patients and identifies metabolites that may play a role in AML development and drug resistance.

Introduction

AML is a hematologic disorder characterized by abnormal proliferation, differentiation disorders, and inhibition of apoptosis of myeloid progenitor cells [1, 2]. The accumulation of these primitive cells leads to the suppression of normal hematopoiesis and the infiltration of extramedullary tissues, resulting in tissue damage [3]. AML is the most prevalent form of acute leukemia in adults, predominantly affecting older patients, with a median age at diagnosis of 68 years. [4]. The disease is associated with a high mortality rate, with an average survival of approximately 3 months without specific treatment and a 5-year survival rate ranging from 10 to 35% following treatment [5]. Chemotherapy has traditionally been the primary treatment modality for AML [6]. However, elderly patients, who constitute the majority of those affected, often struggle to tolerate such therapies [7]. Despite significant advancements in recent years—including targeted therapies, allogeneic hematopoietic stem cell transplantation, and chimeric antigen receptor-modified T-cells (CAR-Ts)—the overall prognosis for AML remains poor. This is largely due to the complex etiology of the disease and the high degree of heterogeneity observed in AML, with many patients experiencing drug resistance and relapses [8]. Consequently, there is an ongoing need to elucidate the mechanisms underlying the occurrence and progression of AML, as well as to develop more effective prevention strategies and targeted therapies.Acute myeloid leukemia (AML) represents a malignant hematologic disorder characterized by the rapid proliferation of abnormal white blood cells in the bone marrow, which subsequently impairs normal hematopoiesis [1, 2]. As the most prevalent form of acute leukemia in adults, AML manifests as a complex disease entity involving multiple genetic alterations and epigenetic modifications, leading to diverse phenotypic presentations and substantial therapeutic challenges. Epidemiological data indicate that AML incidence increases with advancing age, with more than half of patients diagnosed beyond 60 years of age. The disease carries a dismal prognosis, exhibiting a high mortality rate and a 5-year survival rate below 30% [9], posing significant threats to human health and life expectancy. Current therapeutic strategies primarily encompass chemotherapy, targeted therapy, and allogeneic hematopoietic stem cell transplantation. Despite remarkable advancements in AML treatment and continuous improvement in therapeutic outcomes in recent years [8, 10], the overall prognosis remains suboptimal due to the emergence of drug resistance. Approximately 50% of patients develop resistance during chemotherapy, resulting in disease progression and relapse [11]. Consequently, elucidating the mechanisms underlying AML pathogenesis, progression, and drug resistance, along with developing more effective preventive strategies and targeted therapies, holds crucial clinical significance for AML management.

Metabolic reprogramming, a defining feature of malignant tumors, supports cancer cell survival, proliferation, and metastasis under hostile conditions, playing a pivotal role in tumor development and therapeutic resistance [12]. These metabolic shifts include up-regulated glucose metabolism, enhanced glutaminolysis, and altered amino acid and lipid metabolism, particularly the activation of the mevalonate pathway for cholesterol biosynthesis, which is commonly observed in various cancers, including AML [13]. AML patients demonstrate widespread metabolic dysregulation across multiple pathways. For instance, a study investigating AML metabolic signatures identified significant alterations in glycolysis, the TCA cycle, and the pentose phosphate pathway in FLT3-ITD AML cells treated with ATO/ATRA [14]. Additionally, targeting glutamine metabolism has been shown to inhibit leukemia cell growth, improve the therapeutic efficacy of cytarabine, and mitigate drug resistance [15]. These observations highlight the central role of metabolic reprogramming in AML progression and treatment resistance. Metabolomics, a comprehensive analytical tool for monitoring metabolite dynamics, has become instrumental in characterizing metabolic profiles. While the application of metabolomics has spurred considerable interest in AML metabolism, systematic studies exploring its implications for drug resistance are still lacking.

In this study, we employed LC–MS/MS technology to systematically compare the targeted metabolic alterations of 56 central carbon metabolites in blood samples between healthy controls and AML patients. By analyzing differential metabolites and associated pathways, we aimed to identify key factors influencing AML pathogenesis, progression, and prognosis, thereby providing valuable insights for clinical management. Furthermore, we selected the most significantly altered metabolite for in vitro validation to investigate its role in AML cell proliferation and drug resistance. This approach seeks to elucidate the functional significance of this metabolite in AML and potentially identify novel therapeutic targets for AML treatment.

The metabolome refers to the complete set of small molecule metabolites present in a tissue or cell of an organism at a specific physiological time point or under pathological conditions. Metabolomics encompasses the qualitative and quantitative analysis of these metabolomes, aimed at elucidating the dynamic patterns of metabolomic changes in response to interventions or disease states. In recent years, metabolomics has gained significant traction in the study of AML. By analyzing metabolite alterations in the blood of healthy individuals compared to AML patients, researchers have identified distinctive metabolic profiles associated with AML [16], as well as biomarkers of clinical relevance for detection and prognosis [17–19]. Central carbon metabolism (CCM) traditionally includes the Embden-Meyerhof-Parnas pathway (EMP), the Pentose Phosphate Pathway (PPP), and the Tricarboxylic Acid Cycle (TCA). CCM serves as the primary source of energy for organisms and provides essential precursors for other metabolic processes within the body. Furthermore, various cancer cell types have been reported to undergo reprogramming of central carbon metabolism to facilitate disease progression [20, 21]. Consequently, investigating the expression of central carbon metabolites is crucial for the development of targeted therapies aimed at inhibiting cancer cell metabolism.

In this study, we investigated the targeted metabolic changes of 56 central carbon metabolites in the blood of healthy controls and patients with AML using LC–MS/MS technology. Additionally, we selected metabolites that were significantly up-regulated in AML patients to validate their effects on malignant progression and drug resistance in human acute myeloid leukemia cells in vitro. This study elucidates the metabolic profile during the onset and progression of AML, providing new insights for the identification of effective therapeutic targets.

Materials and methods

Clinical samples

This study included a total of 20 clinical samples, with 10 participants in the control group and 10 in the AML group. The inclusion criteria for the AML group required that acute myeloid leukemia (AML) be confirmed through blood tests, and participants must not have had serious cardiac, hepatic, or renal dysfunction, nor have used medications that could influence metabolism in the recent past. For the control group, inclusion criteria mandated the exclusion of serious complications in other organs, as well as severe cardiac, hepatic, or renal dysfunction, and the absence of other malignancies. All participants fasted for 8 to 12 h prior to the collection of fasting blood samples, which were taken in the morning and subsequently stored at -80°C until metabolomics testing could be conducted. The study received approval from the Ethics Committee of the First Affiliated Hospital of Kunming Medical University, and all patients provided written informed consent before treatment in accordance with the Helsinki Declaration.

Metabolite extraction

The samples were thawed in an ice-water bath, and 100 μL was pipetted and transferred to a clean 1.5 mL EP tube. Next, 30 μL of methanol, pre-cooled to −40 °C, was added to the sample, which was then vortexed and mixed for 30 s. Following this, the sample was sonicated in the ice-water bath for 15 min. The sample was allowed to stand at 40 °C for 1 h, after which it was centrifuged at 4 °C for 15 min at 12,000 rpm. Subsequently, 320 μL of the supernatant was aspirated into a clean EP tube, vacuum spun dry, and then reconstituted in 160 μL of purified water. Finally, the supernatant was transferred into the UPLC-MS/MS injection bottle for HPIC-MS/MS analysis.

Metabolomics analysis

In this project, a Thermo Scientific Dionex ICS-6000 high-performance ion chromatography (HPIC) system (Thermo Scientific, USA) was utilized. The guard column employed was an AG11-HC RFIC with dimensions of 2 × 50 mm, while the separation column was an AS11-HC RFIC measuring 2 × 250 mm. The A-phase of the ion chromatography consisted of a 100 mM NaOH aqueous solution, and the D-phase was pure water. Additionally, methanol containing 2 mM acetic acid was introduced post-column at a flow rate of 0.15 ml/min. The column oven temperature was maintained at 30 °C, the sample tray was set to 4 °C, and the injection volume was 5 μL. For mass spectrometry analysis, a 6500 QTrap + mass spectrometer was employed in multiple reaction monitoring (MRM) mode. The ion source parameters were configured as follows: capillary voltage = -4500 V, source temperature = 450 °C, gas1 = 45 psi, gas2 = 45 psi, and CUR = 30 psi.

Preprocessing of data

We extracted 56 central carbon metabolites from 20 experimental samples, of which 35 metabolites were retained after filtering individual metabolites and modeling missing values in the raw data.

Cell culture and treatment

Human acute myeloid leukemia cells (THP-1 and KG-1) were purchased from Procell (Wuhan, China). All cells were cultured in 1640 RPMI medium (Gibco, USA) supplemented with 10% FBS and 1% penicillin/streptomycin (Gibco, USA) at 37 °C in a humidified environment with 5% CO2. The treatment concentration of glucuronic acid (Sigma, USA) was set at 1.0 mM, while that of cytarabine (Sigma, USA) was 0.5 μM.

Cell Counting Kit-8 (CCK-8) assay

The proliferative capacity of the cells was assessed using the CCK-8 kit. Briefly, THP-1 and KG-1 cells were inoculated into 96-well plates and incubated in medium supplemented with metabolites for 24 h. Subsequently, CCK-8 reagent was added dropwise to each well, and after an additional 2 h of incubation, the absorbance was measured at 450 nm.

Methylthiazolyldiphenyl-tetrazolium bromide (MTT) assay

Suspension cells were inoculated into 96-well plates and cultured in incubators for 12 h. Following this incubation period, drugs with varying concentration gradients were added. After an additional 24 h in the incubator, MTT solution was added, and the culture was continued for 4 h. Subsequently, the samples were centrifuged, and the supernatant was carefully removed. Then, 100 μL of dimethyl sulfoxide was added to each well, and the mixture was shaken at low speed for 10 min on a shaker. Finally, the absorbance at 490 nm was measured using an enzyme-linked immunoassay.

Flow cytometry

The detection of apoptosis was performed using the Annexin V/PI double staining assay. Cell suspensions were inoculated into well plates and treated according to experimental requirements, followed by incubation in a cell incubator for 24 h. Subsequently, cells were collected by centrifugation and washed twice with pre-cooled PBS to obtain 1 to 5 × 10^5 cells. To resuspend the cells, 500 μL of Binding Buffer was added. Next, 5 μL of Annexin V with a fluorescent dye was added and mixed thoroughly, followed by the addition of 5 μL of propidium iodide, which was also mixed well. The cells were then incubated for 20 min at room temperature in the absence of light, after which they were observed and detected.

Statistical analysis

Metabolome data were analyzed using multivariate statistical techniques to identify differential metabolites. OPLS-DA was employed to evaluate intergroup differences in metabolite levels. Pathway enrichment analysis was conducted utilizing the KEGG database. Receiver operating characteristic (ROC) curves were implemented to assess the predictive capacity of metabolites for disease. All data were statistically analyzed using SPSS 20.0 and GraphPad 9.0, and results are presented as mean ± SD. Significance in cell proliferation and apoptosis detection experiments was determined using one-way ANOVA, with p < 0.05 considered statistically significant.

Results

Model construction

We constructed OPLS-DA supervised models for 35 targeted metabolites using the original data and conducted 200 randomized test examinations. The score scatter plot clearly demonstrates that the control and AML groups can be significantly distinguished, with a Q2 value of 0.659 (Fig. 1A). The intercepts of Q2 and R2 obtained from the permutation test further indicate that the OPLS-DA model is reliable and not overfitted (Fig. 1B). Additionally, the p < 0.05 result from the replacement test suggests that the model developed in this study is robust (Fig. 1C).

Fig. 1.

Model Construction A Scatterplot depicting the OPLS-DA scores for blood samples, with the control and AML groups distinguished by different colors; each point represents an individual sample. B Replacement test result chart. C Histogram of replacement test results. HJ represents the control group and WHJ represents the AML group

Analysis of differential metabolites

A multivariate approach was employed to quantify the expression of 35 metabolites in both the control and acute myeloid leukemia (AML) groups, with the results of the statistical analysis visualized as volcano plots. Among these metabolites, 28 were found to be up-regulated and seven down-regulated in the AML group (Fig. 2A). Of these, glucuronic acid was the metabolite with the most statistically significant changes. To provide a clearer visual representation of metabolite expression in each group, 35 metabolites were clustered and analyzed (Fig. 2B). These results suggest a tendency for clustering of serum metabolites between the control and AML groups. Normalization of metabolites across different samples was performed, and based on the relative content of metabolites, 10 significantly up-regulated and seven significantly down-regulated metabolites were selected for Z-score plotting. Figure 2C illustrates that the contents of these metabolites exhibited notable differences between the two groups. We then drew a boxplot to assess the difference in the distribution of these 17 metabolites between the two groups (Fig. 2D). The results indicated that the differences in alpha-Ketoglutaric acid, alpha-Ketoisovaleric, fumaric acid, glucuronic acid, isocitric acid, malic acid, and succinic acid were statistically significant (p < 0.05).

Fig. 2.

Analysis of differential metabolites A Volcano plots of targeted metabolite expression in control and AML groups. The horizontal axis indicates the magnitude of change for each metabolite, while the vertical axis represents the P-value. The size of the scatter points corresponds to the VIP value derived from the OPLS-DA model. B Cluster analysis plot of metabolites. C Z-score plots for metabolites, the Z-score calculated as z = (x-μ)/σ: where x is a specific score, μ is the mean, and σ is the standard deviation. D Boxplots of metabolite expression in the two groups. p < 0.05 was considered statistically significant

Annotation of differential metabolites

We utilized the KEGG database to annotate the pathways associated with differential metabolites and discovered that these metabolites were predominantly enriched in the following pathways: Biosynthesis of cofactors, Glyoxylate and dicarboxylate metabolism, glucagon signaling pathway, 2-Oxocarboxylic acid metabolism, biosynthesis of amino acids, the citrate cycle (TCA cycle), and central carbon metabolism in cancer (Fig. 3A). Notably, the TCA cycle pathway exhibited the highest level of enrichment (Fig. 3B). The abundance scores were determined by calculating the ratio of the difference between the number of up-regulated and down-regulated differential metabolites annotated to these pathways, relative to the total number of metabolites within the pathway. We observed that all metabolites enriched in the GABAergic synapse pathway among the annotated pathways displayed a trend toward upregulation in the AML group (Fig. 3C). Subsequently, we calculated the sum of the relative expression of each metabolite within the annotated pathways and represented this data as a heat map (Fig. 3D). Finally, we conducted a comprehensive analysis, which included enrichment and topology analyses, of the pathways containing the differential metabolites and identified glyoxylate and dicarboxylate metabolism, as well as the TCA cycle pathway, as key pathways exhibiting the highest correlation with the differential metabolites (Figs. 3E and F). Based on the differential metabolites obtained from the previous analysis, we constructed a regulatory network that encompasses metabolic pathways, modules, enzymes, reactions, and metabolites. The results illustrate the intersections among metabolic pathways and highlight potential target enzymes and metabolites under specific study conditions (Fig. 3G).

Fig. 3.

Annotation of differential metabolites A KEGG classification map of differential metabolites. B KEGG enrichment map of differential metabolites. C Differential abundance score plot for pathways associated with differential metabolites; a score of 1 indicates a trend toward upregulation of the expression of all annotated differential metabolites in the pathway, while a score of -1 indicates a trend toward downregulation. D KEGG Heat Map, the color blocks on the heat map represent the relative expression of all the differential metabolites annotated in the pathway at the corresponding location, with darker red colors indicating higher levels of all the differential metabolites annotated in the pathway, and darker blue colors indicating lower levels of all the differential metabolites annotated in the pathway. E Bubble map of the metabolic pathway. F Enriched rectangular tree diagrams for pathway synthesis analysis. G Regulatory network analysis of differential metabolites

Glucuronic acid promotes malignant progression and drug resistance of human acute myeloid leukemia cells

Based on the existing studies reported, we selected glucuronic acid among the statistically significant differential metabolites to explore its relationship with the onset and progression of acute myeloid leukemia in vitro. First, we plotted the ROC curve for glucuronic acid with an AUC value of 0.76, showing high diagnostic accuracy (Fig. 4A). As demonstrated in Fig. 4B-E, glucuronic acid significantly promoted the proliferation of THP-1 and KG-1 cells while inhibiting their apoptosis. Furthermore, glucuronic acid was found to reverse the inhibitory effects of cytarabine on the proliferation of human acute myeloid leukemia cells. Additionally, glucuronic acid significantly increased the IC50 of cytarabine (Fig. 4F and G), indicating that the resistance of THP-1 and KG-1 cells to cytarabine was enhanced in the presence of glucuronic acid.

Fig. 4.

Glucuronic acid promotes malignant progression and drug resistance of human acute myeloid leukemia cells A ROC curve for glucuronic acid. B and C CCK-8 measured the proliferation capacity of THP-1 and KG-1 in each treatment group. D and E Flow cytometry was used to detect the apoptosis rate of cells in each treatment group. F and G IC50 curves of cytarabine treatment of THP-1 and KG-1. * p < 0.05, ** p < 0.01, *** p < 0.001

Discussion

AML is a highly heterogeneous disease characterized by a poor prognosis, attributed to its complex etiology, genetic features, and variability in treatment responses. Despite recent advancements in AML treatment, the rates of complete remission and long-term survival for patients remain low, with challenges related to relapse and drug resistance proving difficult to overcome. Currently, metabolic abnormalities have been identified across various tumor types, and targeting tumor metabolism has emerged as a promising therapeutic strategy. The field of metabolomics enables the quantification of metabolites under pathological conditions, facilitating a deeper understanding of the relationship between metabolism-related pathways, regulatory mechanisms, and disease progression. In this study, we employed metabolomics to investigate the characteristics of central carbon metabolism within the AML cohort, aiming to identify potential therapeutic targets. Our findings demonstrated that the OPLS-DA model effectively distinguished between the control and AML groups, indicating that the levels of central carbon metabolites are altered during the onset and progression of AML, thereby suggesting a potential association between AML and abnormal metabolite levels.

Among the 28 metabolites that differed between the control and AML groups, we selected 10 differential metabolites that were higher in terms of their relative content to analyze the distribution of the differences. The results showed that glucuronic acid, α-ketoglutaric acid, α-ketoisoglutaric acid, fumaric acid, isocitric acid, malic acid, and succinic acid were statistically different between the two groups. Several of these metabolites, except glucuronic acid, are metabolites or intermediates of the TCA cycle, which many studies have now reported to be strongly associated with AML. For example, recent studies have shown that α-ketoglutarate dehydrogenase (α-KGDH), a key enzyme in the TCA cycle, primarily catalyzes the formation of succinyl coenzyme A, carbon dioxide, and NADH from α-KG, NAD + , and coenzyme A. An α-KGDH inhibitor (ONC-213) inhibits the enzymatic activity of α-KGDH, which induces a mitochondrial stress response through the activation of ATF4 and promotes apoptosis of ANL cells, thereby reducing AML treatment resistance [22]. Zeng’s team showed that wild-type isocitrate dehydrogenase-2 (Wt-IDH2) is an important molecule for AML cell survival and proliferation and that the inability to convert α-KG to isocitrate in the TCA cycle through inhibition of Wt-IDH2 resulted in reduced lipid synthesis and c-Myc expression in AML cells significantly reduced and inhibited AML cell proliferative activity, suggesting that Wt-IDH2 may be a potential therapeutic target for AML[23]. These studies suggest that the TCA cycle is an important metabolic process in the development and progression of AML, and targeting key metabolites or enzyme activities in the TCA cycle is an effective strategy to inhibit AML progression. However, the link between glucuronic acid and AML has not been reported, and the pentose phosphate pathway in which it resides has been reported only rarely in AML progression. In addition, the results analyzed in this study showed that among the metabolites with statistically significant differences, glucuronic acid showed the most significant changes, suggesting that glucuronic acid is likely to be a key metabolite of the pentose phosphate pathway in promoting AML progression. Therefore, we selected glucuronic acid for subsequent molecular experimental validation. Glucuronic acid is formed by the oxidation of the C-6 hydroxyl group of glucose to a carboxyl group and is produced through the hydrolysis of UDP-glucuronic acid in the glucuronic acid pathway. UDP-glucuronic acid serves as an active donor of glucuronic acid and transfers it to the nucleophilic site of drugs through the action of UDP-glucuronosyltransferase (UGT), which catalyzes the removal of UDP [24, 25]. The glycosylation of this drug can lead to the development of resistance in AML patients when treated with verin or cytarabine. Additionally, following the removal of UDP, glucuronic acid can generate D-xylulose 5-phosphate, which enters the pentose phosphate pathway (PPP) under the action of various enzymes. The PPP has been reported to play a crucial role in fulfilling the anabolic needs of cancer cells and is involved in malignant behaviors such as growth, migration, and invasion in AML [26–28].

The upregulation of these metabolites in the blood of AML patients suggests a potential role in facilitating the onset and progression of AML. In cellular experiments, we added glucuronic acid to human AML cells, which resulted in enhanced cell growth and reduced apoptosis, as anticipated. Furthermore, the addition of exogenous glucuronic acid increased the survival rate of cells treated with cytarabine and enhanced their resistance to this chemotherapeutic agent.

Our study provides potential targets and rationale for the clinical treatment of AML drug resistance. The development of direct or indirect inhibitors of glucuronic acid and synthases by targeting glucuronic acid is a potential strategy for future clinical treatment and may become a new therapeutic option for AML therapy. However, there are several limitations to this study. First, we only tested the correlation between glucuronide and poor AML prognosis in an in vitro cell model, which should be further validated in an in vivo model. Second, the effect of glucuronic acid synthesis via the pentose phosphate pathway on AML resistance should be further investigated under conditions that inhibit glucuronic acid synthesis.

Conclusion

In conclusion, this study elucidates the evolving characteristics of central carbon metabolites in plasma, as revealed by targeted metabolomic sequencing under the pathological conditions of acute myeloid leukemia (AML). Through in vitro cellular experiments, we determined that the upregulation of glucuronic acid in the plasma of AML patients promotes malignant progression and drug resistance in human AML cells, thereby offering new potential therapeutic targets for the treatment of AML.

Author contribution

J. Y. and Z.W. performed the conceptualization, investigation, and writing the original draft; K. W. performed the validation; J. R. and B. N. performed the formal analysis; Q. Z., L.L., L. L., and F. Z. performed the visualization; M. S. and Y. Z. performed the supervision and editing the original draft.

Funding

This study is supported by the Yunnan Province High-Level Talent Training Support Program “Famous Medical Special Project”(No. RLMY20200020).

Data availability

The data that support the findings of this study are available from the corresponding author upon reasonable request. No datasets were generated or analyzed during the current study.

Declarations

Conflict of interests

The authors declare no competing interests.