A method for constructing an acute myeloid leukemia prognosis model based on ferroptosis-related genes

Abstract

This invention discloses a method for constructing a prognostic model for acute myeloid leukemia (AML) based on ferroptosis-related genes. The method involves acquiring gene expression data from AML patients and healthy samples to screen for differentially expressed ferroptosis-related genes (DEGs) associated with AML. Using univariate Cox proportional hazards regression analysis, LASSO regression analysis, and multivariate Cox proportional hazards regression analysis, key genes ACSF2, SLC7A11, DNAJB6, and SOCS1 are selected from these DEGs. A prognostic risk model is constructed based on the expression levels of these key genes and their corresponding multivariate Cox regression coefficients. The risk score formula is: Risk Score = 0.534 × ACSF2 expression value - 0.453 × DNAJB6 expression value + 0.194 × SLC7A11 expression value + 0.308 × SOCS1 expression value. The prognostic model constructed in this invention has high predictive accuracy and reliability, effectively stratifying the risk and assessing the prognosis of AML patients, providing an important reference for the clinical treatment of AML.

Description

Technical Field

[0001] This invention relates to the fields of bioinformatics and medical technology, specifically to a method for constructing a prognostic model for acute myeloid leukemia based on ferroptosis-related genes. Background Technology

[0002] Acute myeloid leukemia (AML) is a highly aggressive hematologic malignancy with high morbidity and mortality. The molecular heterogeneity of AML leads to significant differences in treatment response and prognosis among different subtypes. Current standard treatments such as chemotherapy and hematopoietic stem cell transplantation face challenges including chemotherapy resistance, high relapse rates, and limited donor availability. Therefore, identifying reliable prognostic biomarkers and effective therapeutic targets, and constructing precise prognostic models, are crucial for achieving personalized treatment of AML and improving patient outcomes.

[0003] Ferroprelation is a novel, regulated programmed cell death mechanism characterized by the accumulation of iron-dependent lipid peroxides. Unlike traditional cell death mechanisms such as apoptosis and necrosis, it possesses unique molecular characteristics. Recent studies have shown that ferroptosis plays a crucial role in the occurrence, development, and treatment of tumors. AML cells typically exhibit high iron metabolism levels, making them more sensitive to ferroptosis inducers, thus providing a new therapeutic target for AML. However, current research on the application of ferroptosis-related genes in AML prognostic assessment is insufficient, and reliable prognostic models based on ferroptosis-related genes are lacking. Therefore, it is necessary to develop a method for constructing a prognostic model for acute myeloid leukemia based on ferroptosis-related genes. Summary of the Invention

[0004] The purpose of this invention is to provide a method for constructing a prognostic model for acute myeloid leukemia based on ferroptosis-related genes, so as to solve the problems mentioned in the background art.

[0005] To achieve the above objectives, the present invention provides the following technical solution: a method for constructing a prognostic model for acute myeloid leukemia based on ferroptosis-related genes, comprising the following steps:

[0006] S1. Data Acquisition: Acquire gene expression data, clinical information, and gene expression data of normal samples from patients with acute myeloid leukemia (AML).

[0007] S2. Screening for differentially expressed ferroptosis-related genes (DEGs): Based on known ferroptosis-related genes (FRGs) and combined with the gene expression data obtained in step S1, differential expression analysis is used to screen for differentially expressed ferroptosis-related genes (DEGs) between AML patients and normal samples.

[0008] S3. Screening Key Genes: Univariate Cox proportional hazards regression analysis was performed on the DEGs obtained in step S2 to screen for genes associated with overall survival in AML patients; then, LASSO regression analysis was performed on the screened genes to select genes with non-zero penalty coefficients; finally, multivariate Cox proportional hazards regression analysis was performed on these genes to screen for 4 key genes, namely ACSF2, SLC7A11, DNAJB6, and SOCS1.

[0009] S4. Constructing a prognostic model: Based on the expression levels of the four key genes screened in step S3 and their coefficients in multivariate Cox proportional hazards regression analysis, a prognostic risk model is constructed. The risk score formula of the prognostic risk model is: Risk score = 0.534 × ACSF2 expression value - 0.453 × DNAJB6 expression value + 0.194 × SLC7A11 expression value + 0.308 × SOCS1 expression value.

[0010] Preferably, in step S1, the gene expression data and clinical information of the AML patient are obtained from the TCGA database, and the gene expression data of the normal sample are obtained from the TCGA-GTEx combined database.

[0011] Preferably, in step S2, there are 87 known ferroptosis-related genes (FRGs). By integrating the gene expression data obtained in step S1, 63 FRGs are identified, and then 55 differentially expressed ferroptosis-related genes (DEGs) are screened according to the criteria of P<0.05 and |log2FC|>1.

[0012] Preferably, in step S3, the screening criterion for univariate Cox proportional hazards regression analysis is P<0.05.

[0013] Preferably, the method further includes a step of validating the constructed prognostic model, specifically: obtaining the GSE71014 dataset from the GEO database as a validation cohort, calculating the risk score for each patient in the validation cohort using the prognostic risk model constructed in step S4, dividing patients into low-risk and high-risk groups using the median risk score as the cutoff value, and evaluating the predictive performance of the model through Kaplan-Meier survival analysis and time-dependent ROC curves.

[0014] Preferably, the method further includes the step of constructing a prognostic nomogram, which integrates the risk score, age, and AML risk classification of the prognostic model in step S4 to predict the 1-year, 3-year, and 5-year overall survival of AML patients.

[0015] Preferably, the constructed prognostic model is used to stratify the risk of patients with acute myeloid leukemia. Specifically, the risk score of the patients is calculated, and the patients are divided into low-risk group and high-risk group with the median risk score as the cutoff value.

[0016] Preferably, the prognostic model constructed is used to assess the prognosis of patients with acute myeloid leukemia. Specifically, the patient's risk score is calculated, and the patient's overall survival is assessed according to the risk score. A high risk score corresponds to a poor overall survival, and a low risk score corresponds to a better overall survival.

[0017] Preferably, the method further includes a step of preparing a kit for prognostic assessment of acute myeloid leukemia, the kit comprising reagents for detecting the expression levels of ACSF2, SLC7A11, DNAJB6, and SOCS1 genes, the reagents comprising primers for qRT-PCR detection, wherein:

[0018] The primer sequences for ACSF2 are: qACSF2-F: CTGTCTACGTCGGGATGCTG, qACSF2-R: CTCTGGAACTGAGGAAGCGG;

[0019] The primer sequences for SLC7A11 are: qSLC7A11-F: CGCTGTGAAGGAAAAAGCACA, qSLC7A11-R: TGGTGGACACAACAGGCTTT;

[0020] The detection primers for DNAJB6 and SOCS1 were obtained according to conventional design methods.

[0021] Preferably, the method of using the kit includes the following steps:

[0022] (1) Sample processing: Obtain bone marrow or peripheral blood samples from patients with acute myeloid leukemia and extract total RNA from the samples;

[0023] (2) cDNA synthesis: cDNA was synthesized using the PrimeScript RT kit with 1 μg of total RNA extracted in step (1) as a template;

[0024] (3) qRT-PCR detection: qRT-PCR reaction was performed on the StepOnePlus real-time PCR system using SYBR Green PCRMaster Mix. ACTB was used as the internal reference gene to detect the expression levels of ACSF2, SLC7A11, DNAJB6 and SOCS1 genes respectively. The primer sequences for ACTB were ACTB-F: ACCGCGAGAAGATGACCCA and ACTB-R: GGATAGCACAGCCTGGATAGCAA.

[0025] The primer sequences for ACSF2 are qACSF2-F: CTGTCTACGTCGGGATGCTG, qACSF2-R: CTCTGGAACTGAGGAAGCGG;

[0026] The primer sequences for SLC7A11 are qSLC7A11-F: CGCTGTGAAGGAAAAAGCACA, qSLC7A11-R: TGGTGGACACAACAGGCTTT; the detection primers for DNAJB6 and SOCS1 were obtained using conventional design methods.

[0027] (4) Data calculation: The relative expression levels of ACSF2, SLC7A11, DNAJB6 and SOCS1 genes were calculated using the 2^(−ΔΔCt) method;

[0028] (5) Risk score calculation: Substitute the relative expression levels of each gene obtained in step (4) into the risk score formula in step S4 to calculate the patient's risk score;

[0029] (6) Prognostic assessment: The patient’s prognosis is assessed based on the risk score results. A high-risk score indicates a poor overall survival, while a low-risk score indicates a good overall survival.

[0030] Beneficial Effects: This invention screens differentially expressed genes related to ferroptosis associated with AML, uses bioinformatics analysis to identify key genes, and constructs a prognostic model. This model exhibits high predictive accuracy and reliability. External validation and nomogram construction further enhance the model's practicality and clinical application value. This prognostic model can provide important reference for risk stratification and prognostic assessment of AML patients, contributing to personalized AML treatment and improving patient outcomes.

[0031] The above description is merely an overview of the technical solutions of the embodiments of this application. In order to better understand the technical means of the embodiments of this application and to implement them in accordance with the contents of the specification, and to make the above and other objects, features and advantages of the embodiments of this application more apparent and understandable, specific implementation methods of this application are described below. Attached Figure Description

[0032] Figure 1 A flowchart for data collection and analysis; Figure 2 Functional enrichment analysis diagrams for differentially expressed genes related to ferroptosis; (A) GO enrichment analysis bubble diagram; (B) KEGG pathway enrichment analysis bubble diagram.

[0033] Figure 3 The image shows the identification of AML prognostic biomarkers based on ferroptosis-related genes (FRGs); (A) the lambda value that minimizes the error through 1000 cross-validations; (B) the elimination of highly correlated genes by calculating the gene coefficients of the model; and (C) the forest plot of the four FRG characteristic genes obtained from multivariate Cox analysis. Figure 4 The diagram shows the prognostic model construction based on the TCGA-LAML cohort; (A) KM survival analysis of the TCGA-LAML cohort; (B) ROC curves of 1-year, 3-year and 5-year survival; (CD) risk survival status distribution of the TCGA-LAML cohort. Figure 5 The diagram shows the prognostic model construction based on the GSE71014 cohort; (A) KM survival analysis of the GSE71014 cohort; (B) ROC curves of 1-year, 3-year and 5-year survival; (CD) risk survival status distribution of the GSE71014 cohort. Figure 6 This is a schematic diagram illustrating the independent prognostic value of the prognostic model; where (A) is a forest plot of univariate Cox proportional hazards regression; (B) is a forest plot of multivariate Cox regression analysis; and (CE) are ROC curves for risk score, age, and AML hazard stratification. Figure 7 A schematic diagram of the nomogram model construction; wherein, (A) nomograms predicting 1-year, 3-year, and 5-year overall survival rates; (BD) 1-year, 3-year, and 5-year calibration curves; (E) ROC curves evaluating the predictive power of the nomogram; Figure 8 This is a schematic diagram comparing somatic mutations and tumor mutation burden (TMB) of characteristic genomes; (A) somatic mutation waterfall plot for risk groups; (B) correlation analysis between risk score and TMB; (C) comparison of TMB between high- and low-risk groups; (D) mutation frequencies of four characteristic genes (ACSF2, DNAJB6, SLC7A11, SOCS1) in AML patients obtained from the cBioPortal database. Figure 9This diagram illustrates the inhibition of AML cell proliferation by Erastin and RSL3. (A) The CCK-8 assay was used to detect the effects of different concentrations of erastin or RSL3 on the viability of three cell lines (MOLM-13, AML-193, and HL-60). (B) The C11 BODIPY fluorescent probe was used to detect lipid ROS levels: green represents oxidized C11-BODIPY, and red represents non-oxidized C11-BODIPY. Figure 10 This diagram illustrates the synergistic inhibitory effect of ferroptosis inducers combined with ACSF2 / SLC7A11 knockdown on AML cell proliferation. (A) CRISPR-Cas13-mediated ACSF2 and SLC7A11 knockdown efficiency (verified by qPCR at the mRNA level); (B) Effect of ACSF2 or SLC7A11 knockdown on cell viability (CCK-8 assay); (C) Effect of gene knockdown combined with ferroptosis inducers on cell viability (CCK-8 assay). Statistical significance is indicated as: *p < 0.05, **p < 0.01, ****p < 0.0001; "ns" indicates no statistical difference. Figure 11 This diagram illustrates the inhibitory effects of Erastin and RSL3 on HL-60 cell subcutaneous xenografts in nude mice and their influence on major organs. (A) Curves showing the change in HL-60 cell xenograft volume over time in each treatment group (carrier control group - black circles, Erastin group - red triangles, RSL3 group - black triangles, observation period: days 7 to 21), data expressed as tumor volume (mm³) ± error bar; (B) Results of xenograft weight measurements in each group of nude mice, data expressed as mean ± standard error; (C) Curves showing the change in body weight over time in each treatment group of nude mice (control group - black circles, Erastin group - red squares, RSL3 group - red triangles), data expressed as body weight (g) ± error bar; (D) Histopathological analysis of major organs in nude mice from different treatment groups (H&E staining assessment of heart, kidney, liver, lung, and spleen); (E) Serum biochemical parameters (ALT, AST, ALP, BUN, CREA) of nude mice in each group were measured to assess liver and kidney function. Data are expressed as mean ± standard error, and "ns" indicates no statistical significance. Detailed Implementation

[0034] To make the objectives, technical solutions, and advantages of the embodiments of this application clearer, the technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0035] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application pertains; the terminology used herein in the specification of the application is for the purpose of describing particular embodiments only and is not intended to limit the application; the terms “comprising” and “having”, and any variations thereof, in the specification, claims and drawings of this application are intended to cover non-exclusive inclusion.

[0036] The term "embodiment" as used herein means that a particular feature, structure, or characteristic described in connection with an embodiment may be included in at least one embodiment of this application. The appearance of the phrase "embodiment" in various places throughout the specification does not necessarily refer to the same embodiment, nor is it a separate or alternative embodiment mutually exclusive with other embodiments. It will be explicitly and implicitly understood by those skilled in the art that the embodiments described herein can be combined with other embodiments.

[0037] To enable those skilled in the art to better understand the present application, the technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the accompanying drawings.

[0038] This invention provides the following technical solution: a method for constructing a prognostic model for acute myeloid leukemia based on ferroptosis-related genes, comprising the following steps:

[0039] S1. Data Acquisition: Acquire gene expression data, clinical information, and gene expression data of normal samples from patients with acute myeloid leukemia (AML).

[0040] S2. Screening for differentially expressed ferroptosis-related genes (DEGs): Based on known ferroptosis-related genes (FRGs) and combined with the gene expression data obtained in step S1, differential expression analysis is used to screen for differentially expressed ferroptosis-related genes (DEGs) between AML patients and normal samples.

[0041] S3. Screening Key Genes: Univariate Cox proportional hazards regression analysis was performed on the DEGs obtained in step S2 to screen for genes associated with overall survival in AML patients; then, LASSO regression analysis was performed on the screened genes to select genes with non-zero penalty coefficients; finally, multivariate Cox proportional hazards regression analysis was performed on these genes to screen for 4 key genes, namely ACSF2, SLC7A11, DNAJB6, and SOCS1.

[0042] S4. Constructing a prognostic model: Based on the expression levels of the four key genes screened in step S3 and their coefficients in multivariate Cox proportional hazards regression analysis, a prognostic risk model is constructed. The risk score formula of the prognostic risk model is: Risk score = 0.534 × ACSF2 expression value - 0.453 × DNAJB6 expression value + 0.194 × SLC7A11 expression value + 0.308 × SOCS1 expression value.

[0043] In step S1, the gene expression data and clinical information of the AML patients are obtained from the TCGA database, and the gene expression data of the normal samples are obtained from the TCGA-GTEx combined database.

[0044] In step S2, there are 87 known ferroptosis-related genes (FRGs). By integrating the gene expression data obtained in step S1, 63 FRGs were identified, and then 55 differentially expressed ferroptosis-related genes (DEGs) were screened according to the criteria of P<0.05 and |log2FC|>1.

[0045] In step S3, the screening criterion for univariate Cox proportional hazards regression analysis is P<0.05.

[0046] The present invention also includes a step of validating the constructed prognostic model, specifically: obtaining the GSE71014 dataset from the GEO database as a validation queue, calculating the risk score for each patient in the validation queue using the prognostic risk model constructed in step S4, dividing patients into low-risk and high-risk groups using the median risk score as the cutoff value, and evaluating the predictive performance of the model through Kaplan-Meier survival analysis and time-dependent ROC curves.

[0047] The present invention also includes the step of constructing a prognostic nomogram, which integrates the risk score, age and AML risk classification of the prognostic model in step S4 to predict the 1-year, 3-year and 5-year overall survival of AML patients.

[0048] In this invention, a prognostic model is used to stratify the risk of patients with acute myeloid leukemia. Specifically, the risk score of the patients is calculated, and the patients are divided into low-risk group and high-risk group using the median risk score as the cutoff value.

[0049] In this invention, a prognostic model is used to assess the prognosis of patients with acute myeloid leukemia. Specifically, the risk score of the patient is calculated, and the overall survival of the patient is assessed according to the risk score. A high risk score corresponds to a poor overall survival, and a low risk score corresponds to a better overall survival.

[0050] The present invention also includes the step of preparing a kit for prognostic assessment of acute myeloid leukemia, said kit comprising reagents for detecting the expression levels of ACSF2, SLC7A11, DNAJB6, and SOCS1 genes, said reagents comprising primers for qRT-PCR detection, wherein:

[0051] The primer sequences for ACSF2 are: qACSF2-F: CTGTCTACGTCGGGATGCTG, qACSF2-R: CTCTGGAACTGAGGAAGCGG;

[0052] The primer sequences for SLC7A11 are: qSLC7A11-F: CGCTGTGAAGGAAAAAGCACA, qSLC7A11-R: TGGTGGACACAACAGGCTTT;

[0053] The detection primers for DNAJB6 and SOCS1 were obtained according to conventional design methods.

[0054] In this invention, the method of using the reagent kit includes the following steps:

[0055] (1) Sample processing: Obtain bone marrow or peripheral blood samples from patients with acute myeloid leukemia and extract total RNA from the samples;

[0056] (2) cDNA synthesis: cDNA was synthesized using the PrimeScript RT kit with 1 μg of total RNA extracted in step (1) as a template;

[0057] (3) qRT-PCR detection: qRT-PCR reaction was performed on the StepOnePlus real-time PCR system using SYBR Green PCRMaster Mix. ACTB was used as the internal reference gene to detect the expression levels of ACSF2, SLC7A11, DNAJB6 and SOCS1 genes respectively. The primer sequences for ACTB were ACTB-F: ACCGCGAGAAGATGACCCA and ACTB-R: GGATAGCACAGCCTGGATAGCAA.

[0058] The primer sequences for ACSF2 are qACSF2-F: CTGTCTACGTCGGGATGCTG, qACSF2-R: CTCTGGAACTGAGGAAGCGG;

[0059] The primer sequences for SLC7A11 are qSLC7A11-F: CGCTGTGAAGGAAAAAGCACA, qSLC7A11-R: TGGTGGACACAACAGGCTTT; the detection primers for DNAJB6 and SOCS1 were obtained using conventional design methods.

[0060] (5) Data calculation: The relative expression levels of ACSF2, SLC7A11, DNAJB6 and SOCS1 genes were calculated using the 2^(−ΔΔCt) method;

[0061] (5) Risk score calculation: Substitute the relative expression levels of each gene obtained in step (4) into the risk score formula in step S4 to calculate the patient's risk score;

[0062] (6) Prognostic assessment: The patient’s prognosis is assessed based on the risk score results. A high-risk score indicates a poor overall survival, while a low-risk score indicates a good overall survival.

[0063] Example:

[0064] Differential analysis and enrichment analysis of FRGs in AML

[0065] Figure 1 This demonstrates the overall workflow of our study. We identified 55 differentially expressed FRGs, of which 24 were upregulated and 32 were downregulated in AML. GO analysis ( Figure 2 A) shows that these genes are mainly involved in biological processes such as oxidative stress, chemical stress, and iron homeostasis (BP). At the cellular component (CC) level, the products of these genes are mainly located in membrane structures such as the outer mitochondrial membrane, organelle outer membranes, and lysosomes. In terms of molecular function (MF), these genes are significantly enriched in RNA polymerase II-specific DNA-binding transcription factor activity, long-chain fatty acid-CoA ligase activity, and oxidoreductase activity. KEGG pathway annotation analysis ( Figure 2 B) Further evidence shows that differentially expressed genes are significantly enriched in key signaling pathways such as ferroptosis regulation and microRNA biosynthesis.

[0066] FRG-based prognostic feature identification of AML

[0067] To establish a robust multigene expression profile for predicting AML prognosis, we progressively screened and modeled the expression data of 55 candidate genes. Univariate Cox regression analysis (Table 1) identified 28 genes associated with overall survival, categorized as protective genes (HR<1) and risk genes (HR>1) (p < 0.05). To avoid overfitting, LASSO regression analysis was performed using the "glmnet" package. Figure 3(A-3B) (Table 2) was used to screen out 15 genes most associated with survival outcomes. Subsequently, multivariate Cox proportional hazards analysis was used to model and identify 4 FRGs for developing the optimal prognostic features for patient OS. Coefficients are shown in Table 3. Multivariate analysis of forest plots ( Figure 3 C) shows that ACSF2, SLC7A11, and SOCS1 are adverse factors affecting clinical outcomes in AML patients, while DNAJB6 is a protective factor.

[0068] Construction and validation of a risk prognostic model for AML patients

[0069] The risk score is calculated using the expression levels of four characteristic genes and their corresponding coefficients, according to the following formula:

[0070] Risk Score

[0071] =(0.534*ACSF2)+(-0.453*DNAJB6)+(0.194*SLC7A11)+(0.308*SOCS1)

[0072] Patients in the TCGA dataset (n=130) were divided into low-risk and high-risk groups based on the median value. Kaplan-Meier analysis ( Figure 4 A) showed that the overall survival of AML patients in the low-risk group was significantly better than that in the high-risk group (p < 0.001). To assess the predictive efficiency of the model, a time-dependent ROC analysis was performed. The results showed that the AUC values ​​for 1-year, 3-year, and 5-year survival predictions were 0.852, 0.82, and 0.818, respectively. Figure 4 B) indicates that the model has high predictive accuracy. Furthermore, the distribution of risk scores shows that patients with high risk scores have a significantly increased risk of death (B). Figure 4 C-4D).

[0073] To verify the reliability of the risk scoring model, we performed external validation using the GSE71014 dataset. The same formula was applied to quantify the test set samples, and they were grouped with the same cutoff value. Consistent with the results from the TCGA cohort, Kaplan-Meier analysis (…) Figure 5 A) showed that a high-risk score was significantly associated with poor overall survival (OS) (p < 0.001). The AUCs for 1-year, 3-year, and 5-year OS were 0.756, 0.729, and 0.72, respectively. Figure 5 B). The distribution of risk scores is similar to that of the training set. Figure 5 (C-5D), indicating that the model performs well in predicting the survival rate of AML patients.

[0074] To assess the independent predictive ability of the risk scoring model, univariate Cox regression analysis was first performed. The results showed that age, AML risk classification, and risk score were all significantly associated with patient prognosis (p < 0.05). Figure 6 A). Subsequent multivariate Cox regression analysis further confirmed that the risk score was an independent prognostic factor for AML patients. Figure 6 B). The ROC values ​​of the risk score at 1 year, 3 years, and 5 years were 0.851, 0.836, and 0.815, respectively, all higher than those of AML risk classification (1-year ROC=0.648, 3-year=0.684, 5-year=0.681) and age (1-year ROC=0.68, 3-year=0.733, 5-year=0.744). Figure 6 (C-6E), indicating that the prognostic model has high specificity and sensitivity. Based on this, we believe that the prognostic model is an important independent prognostic factor.

[0075] To build a more practical personalized prediction tool, a nodal plot was constructed based on age, AML risk classification, and risk score. Figure 7 A). The final nomogram score for each patient, obtained by integrating various indicators, can be used to predict 1-year, 3-year, and 5-year survival rates. Furthermore, we established calibration curves to validate the effectiveness of the nomogram model in predicting 1-year, 3-year, and 5-year overall survival (OS). The results showed that the calibration curves demonstrated good agreement between predicted and actual survival. Figure 7 B-7D). Nominal plot performance evaluation shows that the AUCs for 1-year, 3-year, and 5-year survival are 0.695, 0.749, and 0.761, respectively. Figure 7 E).

[0076] Comparative analysis of somatic mutations and TMB in different AML risk score groups

[0077] The development of tumors is often accompanied by the accumulation of gene mutations. Figure 8 A shows the simple nucleotide variants in the risk score groups of AML patients, revealing that the 20 genes with the highest mutation rates in AML are NPM1, TP53, ASXL1, DNMT3A, RUNX1, IDH2, KRAS, TTN, BCORL1, FLT3, KIT, MUC16, FAT2, IDH1, IQCN, PAN2, PRPF8, DNAH7, HIVEP3, and KMT2D. In the TCGA cohort, risk score was not significantly correlated with TMB. Figure 8 B), there was no significant difference in TMB between the high-risk group and the low-risk group. Figure 8 C). Furthermore, the mutation frequencies of the four characteristic genes (ACSF2, SLC7A11, SOCS1, and DNAJB6) are low. Figure 8 D).

[0078] Ferrocyte-inducing agents combined with ACSF2 / SLC7A11 knockout synergistically inhibit AML cell proliferation.

[0079] To evaluate the potential anticancer effects of ferroptosis inducers on AML cells, Erastin and RSL3 were used, and the viability of MOLM-13, AML-193, and HL-60 cells was detected by the CCK-8 assay. The results showed that Erastin and RSL3 treatment significantly reduced AML cell viability in a dose-dependent manner. With increasing Erastin and RSL3 concentrations, the relative survival rate of AML cells significantly decreased; this phenomenon was observed in all three AML cell lines (MOLM-13, AML-193, and HL-60), indicating that Erastin and RSL3 significantly inhibited AML cell proliferation. Figure 9 A).

[0080] To further investigate the mechanism of ferroptosis, we assessed the oxidative status of AML-193 cells under different treatment conditions using fluorescence microscopy. Compared with the control group, the Erastin treatment group showed stronger green fluorescence, indicating an increased intracellular oxidation level; while the control group mainly showed red fluorescence, reflecting a lower oxidation level. Figure 9 B). These findings are consistent with the mechanism of ferroptosis, namely the accumulation of lipid peroxides. The increased oxidative stress in cells treated with Erastin further supports the induction of ferroptosis in AML cells.

[0081] Based on these findings, we used the CRISPR-Cas13 system to selectively knock out the key regulatory genes ACSF2 and SLC7A11 to further explore the role of FRGs in AML treatment. Two gRNA targets (gACSF2-2 and gACSF2-3) were designed for ACSF2, and three gRNA targets (gSLC7A11-1, gSLC7A11-2, and gSLC7A11-3) were designed for SLC7A11. By optimizing transfection conditions, a stable gene knockout model was successfully established. qRT-PCR results confirmed that the mRNA expression of ACSF2 and SLC7A11 in AML cells was significantly reduced. Figure 10 A), validating the effectiveness of the gene editing system. Functional experiments showed that knockout of ACSF2 or SLC7A11 alone significantly inhibited the proliferation of AML-193 cells (A). Figure 10 B). Notably, ACSF2 knockout combined with Erastin or RSL3, and SLC7A11 knockout combined with RSL3, both exhibited strong synergistic pro-cell death effects. Figure 10C). However, no significant synergistic effect was observed in the combination of SLC7A11 knockout and Erastin, which is consistent with the mechanism by which Erastin induces ferroptosis by inhibiting SLC7A11 activity (C). Figure 10 (C) These results indicate that ACSF2 and SLC7A11 play important roles in AML cell survival and ferroptosis resistance. The combination of gene editing and drug therapy enhanced the inhibition of AML cell proliferation, suggesting a potential therapeutic strategy.

[0082] Ferrocyte inducers showed potential to inhibit tumor growth in the HL-60 xenograft mouse model.

[0083] To evaluate the inhibitory effects of the ferroptosis inducers Erastin and RSL3 on in vivo tumor growth in AML, we established a HL-60 cell xenograft model. After subcutaneous implantation of HL-60 cells into 5-week-old male nude mice, the mice were randomly divided into a control group (Vehicle), an Erastin (40 mg / kg) group, and an RSL3 (40 mg / kg) group according to tumor formation, and administered intraperitoneal injections daily. Tumor volume was dynamically monitored. Figure 11 A) showed that Erastin and RSL3 treatment significantly delayed tumor progression ( Figure 11 (B) Specifically, there was no significant difference between the treatment group and the control group in the early stages (d7-d13); however, from day 17 (d17), the tumor growth rate in the Erastin and RSL3 groups was significantly lower than that in the continuously growing control group, and the inhibitory effect was particularly significant in the later stages (d17-d21). These results indicate that ferroptosis inducers exert a delayed anti-tumor effect in AML models by targeting cell death pathways, providing crucial experimental support for the clinical translation of ferroptosis therapy.

[0084] Experimental endpoint tumor weight analysis ( Figure 11 (C) showed that, compared with the control group, both the Erastin and RSL3 groups exhibited a decreasing trend, but the difference was not statistically significant (RSL3 vs. Vehicle: P = 0.2415; Erastin vs. Vehicle: P = 0.3303). Notably, this trend was consistent with the results of dynamic tumor volume monitoring. Figure 11 (B) suggests that the two ferroptosis inducers may work by delaying tumor growth.

[0085] Safety assessments showed that the weight change curves in mice treated with Erastin and RSL3 were in high agreement with those in the control group. Figure 11 D), indicating no significant effect on systemic metabolism. Histopathological analysis showed no significant pathological changes in major organs (heart, liver, kidney, lung, spleen). Figure 11E), serum liver function (ALT, AST, ALP) and kidney function (BUN, CREA) parameters were all within the normal physiological range. Figure 11 F) confirmed that the short-term intervention had no acute toxicity.

[0086] Despite advancements in treatment, AML remains a significant clinical challenge due to its high relapse rate and chemotherapy resistance, resulting in poor prognosis. Therefore, developing optimized prognostic models and treatment regimens based on the biological characteristics of AML is crucial for improving patient survival. Ferropreservation is an iron-dependent, regulated form of cell death triggered by loss of glutathione peroxidase activity and iron-dependent lipid peroxidation. This process leads to the accumulation of lipid reactive oxygen species (ROS), exceeding the cell's antioxidant capacity, causing lipid oxidative damage to the cell membrane, and ultimately resulting in cell death. Biomarker profiling based on ferropreservation-related genes may more accurately segment AML patients into different risk groups than traditional factors. Further research into the genes regulating ferropreservation and their prognostic effects will help optimize risk stratification and provide more personalized treatment strategies for AML patients.

[0087] This study first screened differentially expressed genes related to ferroptosis using gene expression profiles from AML patients and normal samples. Subsequently, GO and KEGG analyses were used to annotate and enrich these genes functionally, exploring their potential biological functions and metabolic pathways. GO bioprocess analysis revealed that these genes are involved in regulating redox reactions and responses to signals that may cause oxidative damage in AML cells. Cell component enrichment analysis showed that these genes are associated with the outer membrane of organelles, which is rich in polyunsaturated fatty acids and susceptible to ROS-induced oxidative damage, representing a key site for ferroptosis. Differentially expressed genes located in organelle membranes may influence ferroptosis by regulating iron and lipid metabolism. Molecular function (MF) analysis showed that these genes mainly regulate DNA-binding transcription factors and RNA polymerase II-specific DNA-binding transcription factors, playing a role in gene expression regulation and signal transduction in AML cells, affecting cell proliferation, differentiation, and apoptosis. KEGG analysis indicated that ferroptosis-related differentially expressed genes are mainly involved in ferroptosis-related processes and microRNA-related processes, revealing the potential molecular mechanisms and regulatory networks of ferroptosis in AML cells, as well as the role of microRNAs in these processes. Ferroprelation-related processes include signaling pathways, glutathione metabolism, lipid metabolism, and ROS clearance, which determine the sensitivity or resistance of AML cells to ferroptosis. MicroRNA-related processes encompass the biosynthesis, maturation, transport, and function of microRNAs, as well as cancer-related pathways such as the p53 pathway, Wnt pathway, and PI3K-Akt pathway, which are crucial for the survival, proliferation, apoptosis, invasion, and metastasis of AML cells.

[0088] We developed a risk-based prognostic model based on four characteristic genes (ACSF2, DNAJB6, SLC7A11, and SOCS1) to categorize AML patients into high- and low-risk groups using a risk score. This model not only provides a new tool for risk stratification of AML patients but also demonstrates its predictive efficacy through Kaplan-Meier and ROC curve analyses. Notably, patients in the high-risk group had significantly worse prognoses, a finding further validated in the GEO dataset. Univariate and multivariate Cox regression analyses confirmed that the risk score is an independent prognostic factor for AML patients. Furthermore, our developed nomogram model integrates risk scores and clinical characteristics, providing a more accurate tool for personalized prognostic assessment of AML patients. The calibration plot shows high predictive efficiency, indicating its potential clinical application value. These findings not only provide a new perspective on AML prognostic assessment but also lay the foundation for future development of molecular biomarker-based personalized treatment strategies. Among the four characteristic genes, ACSF2, SLC7A11, and SOCS1 were identified as poor prognostic biomarkers, while DNAJB6 was associated with a favorable prognosis. As a key enzyme in fatty acid metabolism, ACSF2 participates in fatty acid activation by catalyzing the binding of fatty acids to coenzyme A to form acyl-CoA. Studies have shown that ACSF2 not only regulates fatty acid metabolism but may also promote AML progression by influencing lipid peroxidation and ferroptosis sensitivity, although the specific mechanisms require further investigation. SLC7A11 encodes a cysteine / glutamate antitransporter that protects cells from ferroptosis by maintaining glutathione (GSH) levels and preventing oxidative stress. Inhibition of SLC7A11 leads to GSH depletion and increases ferroptosis in AML cells, especially when combined with low-dose demethylating drugs via the MAGEA6-AMPK-SLC7A11-GPX4 signaling pathway. SOCS1 affects ferroptosis sensitivity by regulating the JAK / STAT pathway and participates in iron metabolism and inflammatory signaling. Conversely, DNAJB6 is a molecular chaperone protein that, when overexpressed, promotes ferroptosis by regulating GPX4 levels and protein homeostasis, exhibiting anticancer properties. The low expression of DNAJB6 in tumor cells and its ability to induce lipid peroxidation through GPX4 regulation make it a promising prognostic marker. The unique role of these genes in the regulation of ferroptosis and their prognostic significance provide new insights into potential therapeutic strategies for AML, especially through the regulation of the ferroptosis pathway.

[0089] Mutational characterization in AML patients revealed that while high-frequency mutated genes (such as NPM1, TP53, and FLT3) play important roles in disease progression, their mutation frequencies did not differ significantly between high- and low-risk groups, and risk scores were not significantly correlated with total liver muscle mass (TMB). In contrast, the low mutation frequencies of ferroptosis-related genes (ACSF2, SLC7A11, SOCS1, and DNAJB6) suggest that their expression levels, rather than genetic variations, dominate prognosis, and are likely more influenced by epigenetic or post-transcriptional regulation. The lack of a significant correlation between risk scores and TMB highlights the independence of driver gene mutations and the regulation of ferroptosis-related gene expression in AML. This finding provides a theoretical basis for therapeutic strategies that simultaneously target driver gene mutations and the ferroptosis pathway.

[0090] This study systematically evaluated the antitumor effects and molecular mechanisms of ferroptosis inducers (Erastin and RSL3) in AML through in vitro and in vivo experiments, and revealed the key roles of ACSF2 and SLC7A11 in the regulation of AML cell proliferation. In vitro experiments showed that Erastin and RSL3 significantly reduced AML cell viability in a dose-dependent manner. Further investigation revealed that Erastin treatment significantly increased intracellular oxidation levels compared to the control group, confirming its ability to trigger ferroptosis by inducing oxidative stress. This observation is consistent with the typical characteristics of ferroptosis—the pathological accumulation of reactive oxygen species (ROS) and lipid peroxides ultimately leading to cell death. The significant oxidative stress response induced by Erastin not only confirms its ferroptosis-inducing effect in AML cells but also reinforces the crucial role of ferroptosis in the pathogenesis of AML, providing solid experimental evidence for its therapeutic targeting.

[0091] To investigate the function of ferroptosis-related genes, we used the CRISPR-Cas13 system to specifically knock out ACSF2 and SLC7A11 in AML cells. The results showed that knocking out either ACSF2 or SLC7A11 alone significantly inhibited cell proliferation. When combined with ferroptosis inducers, ACSF2 knockout showed significant synergistic inhibitory effects with Erastin and RSL3, while SLC7A11 knockout only synergized with RSL3 and had no synergistic effect with Erastin. This difference may stem from Erastin's unique mechanism—it induces ferroptosis by directly inhibiting SLC7A11 activity. Specifically, Erastin inhibits SLC7A11 activity, blocking cystine entry into the cell, leading to GSH depletion. GSH, along with GPX4, synergistically scavenge lipid peroxides, protecting the cell membrane from damage and playing a crucial role in maintaining intracellular lipid stability. When GSH levels are depleted, cells lose their ability to effectively scavenge lipid peroxides, leading to their accumulation, mitochondrial membrane rupture, and ultimately ferroptosis. These findings suggest that ACSF2 and SLC7A11 play key roles in AML cell survival and ferroptosis resistance, and that targeting these genes may enhance the therapeutic effects of ferroptosis inducers.

[0092] In in vivo experiments, Erastin and RSL3 significantly delayed tumor growth in the HL-60 xenograft model, particularly exhibiting significant anti-tumor effects in the later stages of the experiment (days 17-21). Although the reduction in tumor weight was not statistically significant, the decreasing trend was consistent with the dynamic changes in tumor volume, suggesting that ferroptosis inducers may exert a therapeutic effect by slowing tumor growth. The current results may lack statistical power due to sample size limitations or the need for optimization of the dosing regimen, requiring further validation by increasing the sample size or adjusting the dosage. Safety assessments showed that Erastin and RSL3 treatment had no significant effects on mouse body weight, histopathology of major organs, and serum biochemical markers, indicating good short-term safety. However, further studies are needed to evaluate long-term toxicity and pharmacokinetic characteristics to determine their therapeutic window and clinical potential.

[0093] This study has some limitations. First, although the model was validated using the TCGA and GEO datasets, the sample size and diversity of these datasets may not fully encompass the heterogeneity of AML patients. Second, while we demonstrated the crucial roles of SLC7A11 and ACSF2 in AML ferroptosis, the potential mechanisms of other ferroptosis-related genes and their impact on AML prognosis remain to be explored. Third, although the in vitro and in vivo results are encouraging, these findings require validation in larger-scale preclinical studies and clinical trials to confirm their safety and feasibility for clinical application.

[0094] Nevertheless, our findings offer new perspectives on AML treatment, particularly enhancing the efficacy of existing therapies by targeting the ferroptosis pathway. ACSF2 and SLC7A11, as key regulators of ferroptosis, may serve as important biomarkers for predicting AML prognosis and treatment response. Furthermore, the combination of ferroptosis inducers and gene editing technology exhibits significant synergistic anti-tumor effects, laying the foundation for precision medicine in AML. Future research should focus on elucidating the specific regulatory mechanisms of ACSF2 and SLC7A11 in AML, and assessing their expression patterns and treatment responses in different AML subtypes. Integrating multi-omics analyses and optimizing preclinical models will further advance the clinical translation of ferroptosis-targeted therapies in AML.

[0095]

[0096]

[0097]

[0098] In summary, this invention screens differentially expressed genes related to ferroptosis associated with AML, utilizes bioinformatics analysis to identify key genes, and constructs a prognostic model. This model exhibits high predictive accuracy and reliability. External validation and nomogram construction further enhance the model's practicality and clinical application value. This prognostic model can provide important reference for risk stratification and prognostic assessment of AML patients, contributing to personalized AML treatment and improving patient outcomes.

[0099] The above-described embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit them. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of this application.

Claims

1. A method for constructing a prognostic model for acute myeloid leukemia based on ferroptosis-related genes, characterized in that: Includes the following steps: S1. Data Acquisition: Acquire gene expression data, clinical information, and gene expression data of normal samples from patients with acute myeloid leukemia (AML). S2. Screening for differentially expressed ferroptosis-related genes (DEGs): Based on known ferroptosis-related genes (FRGs) and combined with the gene expression data obtained in step S1, differential expression analysis is used to screen for differentially expressed ferroptosis-related genes (DEGs) between AML patients and normal samples. S3. Screening key genes: Perform univariate Cox proportional hazards regression analysis on the DEGs obtained in step S2 to screen for genes related to overall survival of AML patients; Next, LASSO regression analysis was performed on the screened genes to select genes with non-zero penalty coefficients; finally, multivariate Cox proportional hazards regression analysis was performed on these genes to screen out four key genes, namely ACSF2, SLC7A11, DNAJB6 and SOCS1. S4. Constructing a prognostic model: Based on the expression levels of the four key genes screened in step S3 and their coefficients in multivariate Cox proportional hazards regression analysis, a prognostic risk model is constructed. The risk score formula of the prognostic risk model is: Risk score = 0.534 × ACSF2 expression value - 0.453 × DNAJB6 expression value + 0.194 × SLC7A11 expression value + 0.308 × SOCS1 expression value.

2. The method for constructing a prognostic model for acute myeloid leukemia based on ferroptosis-related genes according to claim 1, characterized in that: In step S1, the gene expression data and clinical information of the AML patients are obtained from the TCGA database, and the gene expression data of the normal samples are obtained from the TCGA-GTEx combined database.

3. The method for constructing a prognostic model for acute myeloid leukemia based on ferroptosis-related genes according to claim 1, characterized in that: In step S2, there are 87 known ferroptosis-related genes (FRGs). By integrating the gene expression data obtained in step S1, 63 FRGs were identified, and then 55 differentially expressed ferroptosis-related genes (DEGs) were screened according to the criteria of P<0.05 and |log2FC|>1.

4. The method for constructing a prognostic model for acute myeloid leukemia based on ferroptosis-related genes according to claim 1, characterized in that: In step S3, the screening criterion for univariate Cox proportional hazards regression analysis is P<0.

05.

5. The method for constructing a prognostic model for acute myeloid leukemia based on ferroptosis-related genes according to claim 1, characterized in that: It also includes a step to validate the constructed prognostic model, specifically: obtaining the GSE71014 dataset from the GEO database as a validation cohort, calculating the risk score for each patient in the validation cohort using the prognostic risk model constructed in step S4, dividing patients into low-risk and high-risk groups using the median risk score as the cutoff value, and evaluating the predictive performance of the model through Kaplan-Meier survival analysis and time-dependent ROC curves.

6. The method for constructing a prognostic model for acute myeloid leukemia based on ferroptosis-related genes according to claim 1, characterized in that: It also includes the step of constructing a prognostic nomogram, which integrates the risk score, age, and AML risk classification of the prognostic model in step S4 to predict the 1-, 3-, and 5-year overall survival of AML patients.

7. The method for constructing a prognostic model for acute myeloid leukemia based on ferroptosis-related genes according to claim 1, characterized in that: The constructed prognostic model was used to stratify the risk of patients with acute myeloid leukemia. Specifically, the risk score of the patients was calculated, and the patients were divided into low-risk group and high-risk group with the median risk score as the cutoff value.

8. The method for constructing a prognostic model for acute myeloid leukemia based on ferroptosis-related genes according to claim 1, characterized in that: The prognostic model was used to assess the prognosis of patients with acute myeloid leukemia. Specifically, the risk score of the patient was calculated, and the overall survival of the patient was assessed according to the risk score. A high risk score corresponds to a poor overall survival, and a low risk score corresponds to a better overall survival.

9. The method for constructing a prognostic model for acute myeloid leukemia based on ferroptosis-related genes according to claim 1, characterized in that: It also includes steps for preparing a kit for prognostic assessment of acute myeloid leukemia, the kit comprising reagents for detecting the expression levels of the ACSF2, SLC7A11, DNAJB6, and SOCS1 genes, the reagents comprising primers for qRT-PCR detection, wherein: The primer sequences for ACSF2 are: qACSF2-F: CTGTCTACGTCGGGATGCTG, qACSF2-R: CTCTGGAACTGAGGAAGCGG; The primer sequences for SLC7A11 are: qSLC7A11-F: CGCTGTGAAGGAAAAAGCACA, qSLC7A11-R: TGGTGGACACAACAGGCTTT; The detection primers for DNAJB6 and SOCS1 were obtained according to conventional design methods.

10. The method for constructing a prognostic model for acute myeloid leukemia based on ferroptosis-related genes according to claim 9, characterized in that, The method of using the kit includes the following steps: (1) Sample processing: Obtain bone marrow or peripheral blood samples from patients with acute myeloid leukemia and extract total RNA from the samples; (2) cDNA synthesis: cDNA was synthesized using the PrimeScript RT kit with 1 μg of total RNA extracted in step (1) as a template; (3) qRT-PCR detection: qRT-PCR reaction was performed on the StepOnePlus real-time PCR system using SYBR Green PCR MasterMix. ACTB was used as the internal reference gene to detect the expression levels of ACSF2, SLC7A11, DNAJB6 and SOCS1 genes respectively. The primer sequences for ACTB were ACTB-F: ACCGCGAGAAGATGACCCA and ACTB-R: GGATAGCACAGCCTGGATAGCAA. The primer sequences for ACSF2 are qACSF2-F: CTGTCTACGTCGGGATGCTG, qACSF2-R: CTCTGGAACTGAGGAAGCGG; The primer sequences for SLC7A11 are qSLC7A11-F: CGCTGTGAAGGAAAAAGCACA, qSLC7A11-R: TGGTGGACACAACAGGCTTT; the detection primers for DNAJB6 and SOCS1 were obtained using conventional design methods. (4) Data calculation: The relative expression levels of ACSF2, SLC7A11, DNAJB6 and SOCS1 genes were calculated using the 2^(−ΔΔCt) method; (5) Risk score calculation: Substitute the relative expression levels of each gene obtained in step (4) into the risk score formula in step S4 to calculate the patient's risk score; (6) Prognostic assessment: The patient’s prognosis is assessed based on the risk score results. A high-risk score indicates a poor overall survival, while a low-risk score indicates a good overall survival.

Images