Diagnostic kit based on acute leukemia prognosis and drug sensitivity prediction

Abstract

The invention relates to the technical field of biomedicine, in particular to a diagnostic kit based on acute leukemia prognosis and drug sensitivity prediction, which comprises a prediction module I and a prediction module II. The prediction module I is used for predicting the clinical survival rate of a patient and takes one or more of 32 molecules such as CCND3, FERMT3 and PLD4 as prognostic markers; the prediction module II is used for predicting the sensitivity of the body to drugs, and one or more molecules of BMP8B, IGF1R, OTULINL, SLC22A15, CERS1 and PDE4A are used as drug sensitivity markers. By constructing a dual prediction model, the limitation of prediction efficiency of a single biomarker is broken through, functional coupling of survival prognosis and multi-drug sensitivity prediction is realized, and a clinically convertible integrated tool is provided for accurate treatment of AML (acute myeloid leukemia).

Description

TECHNICAL FIELD

[0001] The present application relates to the field of biomedical technology, and in particular to a diagnostic kit based on acute leukemia prognosis and drug sensitivity prediction. BACKGROUND

[0002] Acute leukemia is a malignant clonal disease of hematopoietic stem / progenitor cells, mainly divided into acute lymphoblastic leukemia (ALL) and acute myeloid leukemia (AML). AML is a highly heterogeneous hematopoietic malignancy, with significant differences in molecular characteristics, treatment response and prognosis among different patients. The significant difference in patient prognosis is mainly due to the complexity of genomic and molecular lineage. FMS-like tyrosine kinase 3 (FLT3) is a receptor tyrosine kinase located on the surface of hematopoietic stem cells, which regulates cell proliferation, differentiation and survival in normal hematopoiesis. In AML, FLT3 gene mutation is one of the most common driver mutations, which is not only closely related to the occurrence and development of the disease, but also significantly affects the prognosis of patients and the selection of treatment strategies.

[0003] FLT3 inhibitors are key targeted drugs for FLT3 mutations in AML. Although FLT3 inhibitors (such as midostaurin, gilteritinib, and quizartinib) have become an important breakthrough in targeted therapy, there are still two major challenges in clinical practice: first, the initial response rate of FLT3 mutant patients to inhibitors is low, and most patients with initial effectiveness are prone to acquired drug resistance; second, the current risk stratification system based on the European Leukemia Network (ELN) standard mainly relies on static genetic markers, which is difficult to dynamically capture the clonal evolution trajectory of the disease and the molecular mechanism of drug resistance during treatment.

[0004] Therefore, it is urgent to deeply analyze the biological basis of the heterogeneity of FLT3 inhibitor treatment response in the clinical management of AML in order to achieve precise stratification of patient survival prognosis. Although there are currently biomarker explorations based on a single omics level, such as using FLT3 internal tandem duplication (ITD) mutation status or kinase domain (TKD) mutation in genomics data for prognosis evaluation and drug selection, such strategies have significant limitations: Firstly, the multidimensional analysis of drug resistance mechanisms has not been systematized, especially lacking in-depth integration of protein function regulation. Although existing research has established drug resistance / prognosis models or drug sensitivity models using genomic or transcriptomic data, and can partially explain the phenotype of drug resistance, the research is usually limited to these transcriptomic markers / models, and fails to conduct systematic integration analysis at the functional level (for example, exploring the dynamic association, covariation rule and potential causal mechanism between drug resistance related protein expression changes and transcriptomic model predicted scores / molecular subtypes). However, a large amount of evidence shows that the core driving signal of acquired drug resistance often occurs at the protein function regulation level. The current limited analysis mode may lead to insufficient identification of key drug resistance driving targets, and it is difficult to effectively distinguish between driving functional changes and accompanying biological variations, limiting the depth and convertibility of mechanism analysis.

[0005] Secondly, the existing marker system has redundancy and clinical translation barriers. Survival prediction models usually rely on multi-gene combination risk scores, which have high detection costs and are difficult to standardize; and FLT3 inhibitor response prediction markers show drug-specific dispersion phenomenon (such as independent markers required for gilteritinib and midostaurin), making it difficult to build a unified cross-drug universal prediction model. This marker fragmentation phenomenon forces clinical detection to design independent schemes for each drug, significantly increasing the complexity and cost of operation. So far, a universal efficacy prediction model compatible with different FLT3 inhibitors is still blank, which cannot meet the urgent need of clinical dynamic drug selection for individual patients. SUMMARY

[0006] The present application aims to at least solve one of the technical problems in the related art. To this end, the object of the present application is to provide a diagnostic kit based on acute leukemia prognosis and drug sensitivity prediction.

[0007] In order to achieve the above-mentioned object, the technical solution adopted by the present application is as follows: The diagnostic kit based on acute leukemia prognosis and drug sensitivity prediction comprises a prediction module I and a prediction module II, the prediction module I is used to predict the clinical survival rate of patients, which can guide the stratified treatment of AML patients, and intensive or combined therapy is needed for high-risk patients; the prediction module II is used to predict the sensitivity of the body to drugs, to support the decision-making of clinical drug use; The prediction module I takes one or more molecules of CCND3, FERMT3, PLD4, TOP1MT, NRGN, RCAN1, ABCD1, ALOX5AP, CCL5, CEBPB, CTSZ, FLOT1, HCK, IL4I1, IL6R, ITGA7, ITGAM, MCOLN2, NEDD9, PEA15, PECAM1, POU2F2, PSMB9, RNPEP, RRAS, SRGAP2, SYK, THEMIS2, TNFAIP2, UNC13D, VDR and ZNF385A as a prognostic marker; The prediction module II takes one or more molecules of BMP8B, IGF1R, OTULINL, SLC22A15, CERS1 and PDE4A as a drug sensitivity marker.

[0008] Preferably, the prognostic marker includes six molecules of CCND3, FERMT3, PLD4, TOP1MT, NRGN and RCAN1, and the prognostic risk score formula is as follows: Risk score = weight coefficient I × A + weight coefficient II × B + weight coefficient III × C + weight coefficient IV × D + weight coefficient V × E + weight coefficient VI × F; Wherein, the cutoff value of risk score is 1-1.05; A represents the gene expression amount of CCND3, B represents the gene expression amount of FERMT3, C represents the gene expression amount of PLD4, D represents the gene expression amount of TOP1MT, E represents the gene expression amount of NRGN, and F represents the gene expression amount of RCAN1.

[0009] Preferably, the weight coefficient I is 0.39, the weight coefficient II is 0.225, the weight coefficient III is 0.122, the weight coefficient IV is 0.087, the weight coefficient V is 0.082, and the weight coefficient VI is 0.075.

[0010] Preferably, the drug is selected from FLT3 inhibitors.

[0011] Preferably, the FLT3 inhibitor is selected from one or more of quizartinib, crenolanib, gilteritinib, midostaurin and sorafenib.

[0012] Preferably, the normalized AUC is used to evaluate the sensitivity of the body to the drug; Wherein, AUC is the area under the drug dose-effect curve.

[0013] Preferably, the prognostic markers are screened by using single factor Cox regression analysis based on proteomic data of FLT3 inhibitor-resistant cell models, combined with TCGA-LAML and Target-AML clinical cohort data; Wherein, TCGA-LAML is adult acute myeloid leukemia patients, and Target-AML is children and adolescents with acute myeloid leukemia.

[0014] Preferably, the drug sensitivity markers are screened by using LASSO-Logistic regression machine learning method.

[0015] Preferably, the drug sensitivity includes the sensitivity of the body to the drug before and after taking the drug.

[0016] Preferably, the acute leukemia is acute myeloid leukemia.

[0017] The above one or more technical solutions in the embodiments of the present application have at least one of the following technical effects: The diagnostic kit for predicting the prognosis and drug sensitivity of acute leukemia provided by the present application comprises prediction module I and prediction module II. Prediction module I is used for predicting the clinical survival rate of patients, and one or more molecules of CCND3, FERMT3, PLD4, TOP1MT, NRGN, RCAN1, ABCD1, ALOX5AP, CCL5, CEBPB, CTSZ, FLOT1, HCK, IL4I1, IL6R, ITGA7, ITGAM, MCOLN2, NEDD9, PEA15, PECAM1, POU2F2, PSMB9, RNPEP, RRAS, SRGAP2, SYK, THEMIS2, TNFAIP2, UNC13D, VDR and ZNF385A are used as prognostic markers; prediction module II is used for predicting the sensitivity of the body to the drug, and one or more molecules of BMP8B, IGF1R, OTULINL, SLC22A15, CERS1 and PDE4A are used as drug sensitivity markers. The present application breaks through the limitation of single biomarker prediction efficiency by constructing a double prediction model, realizes the functional coupling of survival prognosis and multiple drug sensitivity prediction, and provides an integrated tool for clinical transformation for AML precise treatment. At the same time, the proteomic driven marker dimension reduction strategy is adopted to overcome the redundancy problem of the traditional marker detection system, simplify the operation process, reduce the clinical detection cost and improve the efficiency.

[0018] Additional aspects and advantages of the application will be set forth in part in the description which follows, and in part will become apparent to those skilled in the art upon examination of the following and the accompanying drawings. BRIEF DESCRIPTION OF DRAWINGS

[0019] Figure 1 is a differential protein cluster heat map of the drug-resistant cell line provided by the protein group detection of embodiment 1 of the present application.

[0020] Figure 2 is a process diagram of screening of 32 risk proteins related to drug resistance and patient survival provided by embodiment 1 of the present application.

[0021] Figure 3 is a graph of the relationship between the marker combination and the area under the model curve of the survival model construction in the TCGA data provided by embodiment 1 of the present application.

[0022] Figure 4 is an AUC curve diagram corresponding to the optimal marker combination provided by embodiment 1 of the present application.

[0023] Figure 5 is a survival curve diagram of the optimal marker combination model provided by embodiment 1 of the present application.

[0024] Figure 6 is the performance of the optimal marker combination model in the Target AML validation set provided by embodiment 1 of the present application.

[0025] Figure 7 is the optimal marker combination of the five FLT3 inhibitors for establishing a prediction drug sensitivity model respectively provided by embodiment 1 of the present application.

[0026] Figure 8 is an AUC curve diagram corresponding to the optimal marker combination of the five FLT3 inhibitors provided by embodiment 1 of the present application.

[0027] Figure 9 is an AUC curve diagram corresponding to the six optimal marker combinations using the data corresponding to gilteritinib as the test set provided by embodiment 1 of the present application.

[0028] Figure 10 is an AUC curve diagram corresponding to the six optimal marker combinations using the data corresponding to quizaritinib, crenolanib, midostaurin and sorafenib respectively as the validation set provided by embodiment 1 of the present application. DETAILED DESCRIPTION

[0029] In order to make the purpose, technical scheme and advantages of the present application clearer, the technical scheme in the present application will be described clearly and completely below in combination with specific embodiments. Obviously, the described embodiments are part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor fall within the scope of protection of the present application. The following embodiments are used to illustrate the present application, but cannot be used to limit the scope of the present application.

[0030] Example 1 I. Prognostic risk model construction To elucidate the clinical relevant proteomic dynamics in drug resistance evolution, the present application performed K-means clustering on the differentially expressed proteins, and identified five clusters with distinct functional features, as shown in FIG. 1. From the figure, it can be seen that C2 and C5 modules are highly expressed in wild type, and the expression decreases with the development of drug resistance (early stage→late stage); C1, C3 and C4 modules are lowly expressed in wild type, and the expression increases with the development of drug resistance. Figure 1

[0031] Among them, the number of proteins in cluster C1 is 111, and the function is: innate immune response, carboxylic acid metabolism, lymphocyte activation, type II interferon response, positive regulation of cell-matrix adhesion; The expression trend of cluster C1 is rising, which indicates that the immune response and cell adhesion are enhanced, which may help drug-resistant cells escape immune surveillance or remodel the microenvironment.

[0032] The number of proteins in cluster C2 is 71, and the function is: regulation of reactive oxygen metabolism, positive regulation of cysteine-type endopeptidase activity, mitochondrial assembly, negative regulation of cell-cell adhesion, alcohol biosynthesis; The expression trend of cluster C2 is decreasing, which indicates that the functions related to mitochondrial assembly and reactive oxygen metabolism are inhibited with the development of drug resistance, which may reflect the adaptive changes of cell energy metabolism.

[0033] The number of proteins in cluster C3 is 370, and the function is: innate immune response, inflammatory response, endoplasmic reticulum stress response, glycoprotein metabolism, oxidative stress response; The expression trend of cluster C3 is rising, which indicates that inflammation and endoplasmic reticulum stress are activated, reflecting the compensatory mechanism of drug-resistant cells in response to stress (such as drug stimulation).

[0034] The number of proteins in cluster C4 is 118, and the function is: acute phase response, positive regulation of cell adhesion, regulation of inflammatory response, regulation of MAPK cascade, regulation of interleukin-6 (IL-6) production; The expression trend of cluster C4 is rising, which indicates that inflammatory signals (such as MAPK and IL-6) and cell adhesion are up-regulated, which may promote drug resistance through pro-inflammatory microenvironment or signal pathway reprogramming.

[0035] The number of proteins in cluster C5 is 350, and the function is: oxidative phosphorylation, mitochondrial electron transport (ubiquinone→cytochrome C; cytochrome C→oxygen), mitochondrial transport, oxidative stress response; The expression trend of cluster C5 is decreasing, which indicates that the core function of mitochondrial respiratory chain (oxidative phosphorylation) is significantly down-regulated, suggesting that drug-resistant cells may rely on alternative metabolism such as glycolysis.

[0036] ​Wild type in the figure represents the initial drug sensitive, drug resistant, but with FLT3-ITD mutation; early drug resistance represents drug resistance after 3-6 months of drug use and combined with FLT3-TKD mutation on the basis of FLT3-ITD mutation; late drug resistance represents drug resistance after 6 months of drug use, which is combined with NRAS mutation on the basis of early drug resistance mutation.

[0037] As shown in Figure 2 The analysis process from protein clustering to prognosis risk protein screening is as follows: I. The proteins are divided into two groups by K-means clustering algorithm: C3 and C4 clusters: a total of 488 proteins, and the arrow pointing up indicates that the proteins in this group are positively correlated with the "drug resistance" phenotype (drug resistance proteins); C2 and C5 clusters: a total of 421 proteins, and the arrow pointing down indicates that the proteins in this group are negatively correlated with the "drug resistance" phenotype (drug sensitive proteins).

[0038] II. Univariate Cox regression analysis: RNA expression in two AML data sets (TCGA-LAML, Target-AML) was analyzed by univariate Cox proportional hazards model to screen proteins related to survival prognosis: 32 "risk proteins" were screened (hazard ratio > 1, indicating that high expression is associated with poor prognosis); only one "beneficial protein" (hazard ratio < 1, indicating that high expression is associated with good prognosis).

[0039] III. Locking core risk proteins through a Venn diagram: The intersection of three groups of protein data is shown by a Venn diagram, and the three groups of data are drug resistance proteins of C3 and C4 clusters, risk transcripts of Target AML data, and risk transcripts of TCGA-LAML data; Central intersection (32): molecules belonging to the above three groups at the same time, representing the most stable and core prognosis risk molecules (both belonging to the drug resistance cluster in clustering and being verified as risk factors in two independent data sets).

[0040] The above analysis process from "clustering grouping" discovers phenotype differences (high / low risk clusters), then screens prognosis related proteins through "survival analysis", and finally locks the core risk proteins through "multi-dataset intersection", ensuring the reliability and generalizability of the results (cross-dataset verification).

[0041] To verify the clinical relevance of the differential proteins screened by drug sensitivity model, the transcriptome and survival data of TCGA-LAML (n=151) and Target-AML (n=1074) independent clinical cohorts were integrated: 32 risk proteins significantly negatively correlated with overall survival of patients were screened out by single factor Cox proportional hazards regression model analysis (hazard ratio >1, p<0.05), as shown in Table 1 and Table 2.

[0042] To overcome overfitting of high-dimensional data, LASSO regression algorithm was applied for feature compression: 6 core prognostic markers were screened out by 10-fold cross-validation to optimize the regularization parameter λ: CCND3, TOP1MT, RCAN1, NRGN, PLD4 and FERMT3, and the weight coefficients of the six prognostic markers were 0.39, 0.255, 0.122, 0.087, 0.082 and 0.075 respectively, and a prognostic risk score formula was established as follows: Risk score = 0.39 × CCND3 gene expression + 0.255 × TOP1MT gene expression + 0.122 × RCAN1 gene expression + 0.087 × NRGN gene expression + 0.082 × PLD4 gene expression + 0.075 × FERMT3 gene expression; In the TCGA-AML training set, the high-risk group (n=66) and the low-risk group (n=66) were divided by the median value of the risk score (risk score = 1.027056) as the cut-off value, the 5-year survival rate of the high-risk group was only 9.85%, while the 5-year survival rate of the low-risk group could reach 35.83% (hazard ratio = 2.380, 95% CI: 1.512-3.746, p<0.001); This result was repeated in the Target-AML validation set (hazard ratio = 1.360, 95% CI: 1.204-1.536, p<0.001).

[0043] The six prognostic markers, prognostic model (CCND3, FERMT3, PLD4, TOP1MT, NRGN, RCAN1), as shown in Figure 3 ; The prediction results in the TCGA LAML test set are shown in Figure 4 and Figure 5 , which shows that the six prognostic markers have excellent prediction performance; The prediction results in the Target AML validation set are shown in Figure 6As shown in the figure, the results indicate that these six prognostic markers can still maintain high prediction efficiency (AUC=0.836).

[0044] In summary, the prognostic risk score formula established based on the six prognostic markers demonstrated superior stratification capabilities in both the training set (TCGA-LAML) and the validation set (Target-AML), dividing patients into high-risk and low-risk groups, with significantly better predictive efficacy than the traditional ELN classification. Therefore, using these six prognostic markers as a predictive model can serve as risk stratification biomarkers to identify patients who may benefit from resistance proteomics-guided combination therapy.

[0045] 2. Construction of drug sensitivity model.

[0046] This study employed a dose-escalation approach to establish early-stage resistance (ER) and late-stage resistance (LR) cell models for mechanistic investigation and development of reversal strategies. Compared to parental (WT) cells, ER cells exhibited a >10-fold increase in IC50 values ​​and were associated with FLT3-TKD mutations, while LR cells exhibited over 100-fold resistance due to secondary NRAS mutations.

[0047] To predict FLT3 inhibitor sensitivity, we integrated pharmacogenomic data from primary AML samples (n=671) from the BeatAML database. The area under the dose-effect curve (AUC) for five inhibitors, quizartinib, crenoclab, gilteritinib, midostaurin, and sorafenib, was collected as a sensitivity indicator (lower AUCs indicate higher sensitivity). For each inhibitor, the AUC values ​​were normalized between 0 and 1 using the formula 1-AUC / 300 (higher normalized AUCs indicate higher sensitivity). Patients were then grouped based on the normalized AUC values, with the upper third designated as the sensitive group and the lower third as the resistant group. The differences between the sensitive and resistant groups for each inhibitor were compared to identify differentially expressed molecules. Differential molecules were used to independently construct a prediction model for each inhibitor: using iterative LASSO-logistic regression (L1 regularization coefficient α = 1, 10-fold cross-validation to optimize λ), 22 to 31 marker combinations were screened for each drug, and the prediction efficiency AUC of each model was > 0.85 (e.g., AUC = 0.87 for the gilteritinib model). Figure 7 shown.

[0048] The predictive power of strong predictive markers identified by molecular profiling of each FLT3 inhibitor sensitive and resistant cohort, such as Figure 8 As shown (AUC=0.82~0.90).

[0049] The intersection of the markers of each drug was taken, and six common core markers were identified as drug sensitivity markers. These six drug sensitivity markers are BMP8B, IGF1R, OTULINL, SLC22A15, CERS1, and PDE4A.

[0050] The corresponding data of gilteritinib is used as a test set to test the model, and the result is shown in Figure 9 The result shows that the prediction performance can reach 0.733; Quizaritinib, crenolanib, midostaurin and sorafenib are used as a verification set, and the result is shown in Figure 10 (AUC 0.73-0.81), which can still maintain stable prediction performance.

[0051] The diagnostic kit provided by the application is based on acute leukemia prognosis and drug sensitivity prediction, and a double prediction module based on drug resistance proteome-clinical transcriptome cross verification: by establishing an AML acquired drug resistance cell model, the drug resistance related protein target is screened by using deep proteomics analysis; the clinical transcriptome data of AML patients in public databases such as TCGA are integrated at the same time, and the survival correlation verification (Cox regression analysis) of the above target is carried out, so as to ensure that the marker has drug resistance function explanation and patient overall survival (OS) prediction value. This breaks through the traditional single-omics limitation and realizes the direct mapping of functional protein mechanism and clinical phenotype. A survival risk score model is established based on a simplified marker set (≤10 core target points): a quantitative risk assessment system is generated by using multivariate Cox regression; a cross-inhibitor response prediction model: the drug sensitivity data (AUC value) of five FLT3 inhibitors (quizaritinib, crenolanib, gilteritinib, midostaurin and sorafenib) in beatAML database are called, and a unified prediction framework is trained by machine learning algorithm, so as to realize the synchronous evaluation of multi-drug efficacy. The system uses proteome-driven marker dimension reduction strategy to overcome the traditional marker redundancy problem; at the same time, through the universality of the core target of drug resistance mechanism, it realizes the functional coupling of survival prognosis and multi-drug response prediction for the first time, and provides an integrated tool for clinical transformation for AML precise treatment.

[0052] Finally, it should be noted that: the above examples are only used to illustrate the technical solutions of the present application, and not to limit them; although the present application has been described in detail with reference to the foregoing examples, those skilled in the art should understand that: it can still modify the technical solutions recorded in the foregoing examples, or make equivalent replacement for part of the technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present application.

Claims

1. A diagnostic kit based on the prediction of acute leukemia prognosis and drug sensitivity, characterized in that: It includes a prediction module I and a prediction module II, wherein the prediction module I is used to predict the patient's clinical survival rate, and the prediction module II is used to predict the body's sensitivity to drugs; The prediction module I uses one or more molecules among CCND3, FERMT3, PLD4, TOP1MT, NRGN, RCAN1, ABCD1, ALOX5AP, CCL5, CEBPB, CTSZ, FLOT1, HCK, IL4I1, IL6R, ITGA7, ITGAM, MCOLN2, NEDD9, PEA15, PECAM1, POU2F2, PSMB9, RNPEP, RRAS, SRGAP2, SYK, THEMIS2, TNFAIP2, UNC13D, VDR and ZNF385A as prognostic markers; The prediction module II uses one or more molecules among BMP8B, IGF1R, OTULINL, SLC22A15, CERS1 and PDE4A as drug sensitivity markers.

2. The diagnostic kit for acute leukemia prognosis and drug sensitivity prediction according to claim 1, wherein: The prognostic markers include six molecules: CCND3, FERMT3, PLD4, TOP1MT, NRGN, and RCAN1, and the prognostic risk score formula is as follows: Risk score = weight coefficient I × A + weight coefficient II × B + weight coefficient III × C + weight coefficient IV × D + weight coefficient V × E + weight coefficient VI × F; Among them, the cutoff value of the risk score is 1 to 1.05; A represents the gene expression level of CCND3, B represents the gene expression level of FERMT3, C represents the gene expression level of PLD4, D represents the gene expression level of TOP1MT, E represents the gene expression level of NRGN, and F represents the gene expression level of RCAN1.

3. The diagnostic kit based on acute leukemia prognosis and drug sensitivity prediction according to claim 2, characterized in that: The weight coefficient I is 0.39, the weight coefficient II is 0.225, the weight coefficient III is 0.122, the weight coefficient IV is 0.087, the weight coefficient V is 0.082, and the weight coefficient VI is 0.

075.

4. The diagnostic kit based on acute leukemia prognosis and drug sensitivity prediction according to claim 1, characterized in that: The drug is selected from FLT3 inhibitors.

5. The diagnostic kit based on acute leukemia prognosis and drug sensitivity prediction according to claim 4, characterized in that: The FLT3 inhibitor is selected from one or more of quizartinib, crenoclab, gilteritinib, midostaurin and sorafenib.

6. The diagnostic kit for predicting the prognosis and drug sensitivity of acute leukemia according to any one of claims 1 to 5, wherein: The normalized AUC is used to evaluate the body's sensitivity to the drug; AUC is the area under the drug dose-effect curve.

7. The diagnostic kit based on acute leukemia prognosis and drug sensitivity prediction according to claim 5, characterized in that: The prognostic markers were screened based on the proteomic data of FLT3 inhibitor-resistant cell models, combined with TCGA-LAML and Target-AML clinical cohort data, using univariate Cox regression analysis; Among them, TCGA-LAML is for adult patients with acute myeloid leukemia, and Target-AML is for children and adolescents with acute myeloid leukemia.

8. The diagnostic kit based on acute leukemia prognosis and drug sensitivity prediction according to claim 1, characterized in that: The drug sensitivity markers were screened using the LASSO-logistic regression machine learning method.

9. The diagnostic kit based on acute leukemia prognosis and drug sensitivity prediction according to claim 1, characterized in that: The drug sensitivity includes the body's sensitivity to the drug before and after taking the drug.

10. The diagnostic kit based on acute leukemia prognosis and drug sensitivity prediction according to claim 1, characterized in that: The acute leukemia is acute myeloid leukemia.

Images