Risk stratification assessment model, device and construction method for runx1: :runx1t1 positive childhood acute myeloid leukemia

Abstract

This invention discloses a risk stratification assessment model, device, and construction method for RUNX1::RUNX1T1-positive children with acute myeloid leukemia. The construction method includes: acquiring sample clinical data; assessing the correlation between clinical predictors and overall survival (OS) and event-free survival (ORS); for continuous variables, determining the optimal risk cutoff value for predicting poor prognosis; converting the continuous variables into binary variables; performing univariate Cox proportional hazards regression analysis to screen for factors significantly associated with OS and ORS; and incorporating these factors into a multivariate Cox proportional hazards regression model to confirm independent prognostic factors. This invention systematically integrates the clinical variable MRD1 and the percentage of peripheral blood blasts at diagnosis as core predictors, constructing a model capable of accurately assessing ORS. + A prognostic model for pAML risk was developed and rigorously validated. This model demonstrated superior predictive performance, effectively and accurately identifying patients with a high actual risk of relapse and death from the traditionally low-risk patient population.

Description

Technical Field

[0001] This invention belongs to the fields of tumor molecular genetics, bioinformatics and precision medicine, and relates to a risk stratification assessment model, device and construction method for RUNX1::RUNX1T1 positive childhood acute myeloid leukemia. Background Technology

[0002] Childhood acute myeloid leukemia (pAML) is an aggressive hematologic malignancy, and despite recent advancements in treatment, its overall survival (OS) remains only 60%–70%. 10–12% of pAML cases involve a t(8;21)(q22;q22.1) translocation, leading to the formation of the RUNX1::RUNX1T1 (RR) fusion gene. This fusion gene disrupts the function of core binding factor (CBF), thereby affecting hematopoietic differentiation.

[0003] In both adult and pediatric AML, t(8;21) translocations are generally considered to be associated with a favorable prognosis. In pediatric patients, the Pediatric Oncology Cooperative Group (COG) defined a risk group based on the AAML03P1 trial, classifying CBF-positive pAML patients as low-risk; subsequent clinical studies, including the AAML0531 and AAML1031 studies, defined a final risk group that improved predictive accuracy by incorporating immunophenotypic and next-generation sequencing data, also classifying these as low-risk. Recent cytogenetic fusion molecular risk stratification and LncScore models further integrate prognostic factors from both molecular and long non-coding RNA perspectives. However, these models still classify the majority of t(8;21) pAML patients as low-risk.

[0004] Because the RR fusion gene itself is a low-risk prognostic marker, only a very small number of risk prediction tools specifically target RR. + pAML. A recent adult RR + An AML risk prediction tool, combining transcriptomic analysis of 240 genes with clinical characteristics, identified three molecular subtypes. One subtype showed the worst survival outcomes, associated with abnormal expression characteristics of leukemia stem cell and progenitor cell genes. In pediatric patients, a study involving 48 RR... +Patient studies have revealed that extramedullary leukemia (EML) and loss of sex chromosomes (LOS) are independent prognostic factors for overall survival (OS), event-free survival (EFS), and relapse-free survival (RFS) in children with this disease (Yang, J., et al., Prognostic Factors of PediatricAcute Myeloid Leukemia Patients with t(8;21) (q22;q22): A Single-CenterRetrospective Study. Children (Basel), 2024. 11(5).). While these findings enhance the understanding of relapse-free survival (RR)... + While pAML is understood, molecular subtyping may not always be available in clinical practice. Furthermore, minimal residual disease (MRD1) after induction therapy has become a recognized predictor of relapse and poor overall survival, yet up to 30% of MRD-negative patients still experience relapse, suggesting the presence of occult high-risk characteristics in patients.

[0005] In summary, although most RR + Patients with pAML are stratified as low-risk, but a subset still experience poor survival and relapse risk. Therefore, a more accurate and clinically accessible model is urgently needed to improve risk stratification and identify relapse rates (RR). + For high-risk pAML patients, we provide tailored treatment recommendations. Summary of the Invention

[0006] The purpose of this invention is to address the above-mentioned technical problems by providing an accurate and clinically accessible model that can improve risk stratification and identify risk responses (RR). + For high-risk pAML patients, risk assessment can be conducted to provide treatment recommendations.

[0007] To achieve the above-mentioned objectives, the present invention provides the following technical solution:

[0008] In a first aspect, the present invention provides a method for constructing a risk stratification assessment model for RUNX1::RUNX1T1-positive childhood acute myeloid leukemia, comprising the following steps:

[0009] S1. Obtain clinical data of positive samples carrying the RR fusion gene and randomly divide them into a training cohort and an internal validation cohort.

[0010] S2. Using Kaplan-Meier survival analysis combined with log-rank test, the correlation between multiple potential clinical predictors and overall survival and event-free survival was evaluated one by one in the training cohort.

[0011] S3. For continuous variables, receiver operating characteristic (ROC) curve analysis is used, and the optimal risk cutoff value for predicting poor prognosis is determined according to the Youden index principle. The optimal cutoff value is determined by plotting the ROC curve and calculating the Youden index corresponding to each point on the curve, and taking the diagnostic threshold corresponding to the maximum value of the index.

[0012] S4. Transform the above continuous variables into binary variables, namely high risk and low risk, according to their optimal cutoff values.

[0013] S5. Perform univariate Cox proportional hazards regression analysis to screen out factors that are significantly associated with overall survival and event-free survival; then incorporate these significant factors into the multivariate Cox proportional hazards regression model to correct for the mutual influence between factors and finally identify independent prognostic factors.

[0014] S6. Determine minimal residual disease after the first induction therapy and the percentage of peripheral blood blasts at diagnosis as prognostic factors.

[0015] As a preferred implementation, the thresholds for minimal residual disease after the first induction therapy and the percentage of peripheral blood blasts at diagnosis are calculated using optimal cutoff values. The optimal risk cutoff value for minimal residual disease after the first induction therapy is 0.015, and the optimal cutoff value for the percentage of peripheral blood blasts at diagnosis is 67.5%.

[0016] As a preferred implementation, an evaluation is RR when at least one of the following conditions is met. + High risk: Minimal residual disease ≥0.015 after the first induction therapy, and peripheral blood blast percentage ≥67.5% at diagnosis; if neither of the above two conditions is met, the risk is assessed as recurrent respiratory failure (RR). + Low risk.

[0017] In a preferred embodiment, the method further includes an internal verification step.

[0018] In a preferred embodiment, the method further includes an external verification step.

[0019] Secondly, the present invention provides a risk stratification assessment model for RUNX1::RUNX1T1 positive children with acute myeloid leukemia, which is constructed according to the above method.

[0020] Thirdly, the present invention provides a risk stratification assessment device for RUNX1::RUNX1T1-positive children with acute myeloid leukemia, comprising:

[0021] The data acquisition unit is used to acquire prognostic factor data of the subjects;

[0022] An analysis and evaluation unit is used to assess the prognostic risk stratification status of the subject based on the prognostic factor data;

[0023] The results output unit is used to output the risk stratification assessment results.

[0024] In a preferred embodiment, the prognostic factor data includes minimal residual disease after the first induction therapy and the percentage of peripheral blood blasts at diagnosis.

[0025] As a preferred implementation, an evaluation is RR when at least one of the following conditions is met. + High risk: Minimal residual disease ≥0.015 after the first induction therapy, and peripheral blood blast percentage ≥67.5% at diagnosis; if neither of the above two conditions is met, the risk is assessed as recurrent respiratory failure (RR). + Low risk.

[0026] This invention systematically and effectively integrates the clinical variables of minimal residual disease (MRD1) after the first induction therapy and the percentage of peripheral blood blasts at diagnosis as core predictors, constructing a system capable of accurately assessing relapse rate (RR). + A concise and powerful clinical prognostic model for pAML risk has been developed and rigorously validated in independent cohorts. This model, rigorously validated in multiple independent cohorts, demonstrates superior predictive performance, effectively and accurately identifying patients with high actual risk of relapse and death from those traditionally considered "good prognostic" low-risk patients. This model has potential value in guiding treatment decisions, providing evidence-based support for intensive or novel treatment strategies for high-risk patients, and contributing to relapse and relapse prevention (RR). + This invention provides a practical tool for individualized risk assessment and precision treatment decision-making in pAML, enabling integrated precision intervention from risk prediction to treatment guidance. Furthermore, it reveals specific biological characteristics of high-risk patients, such as specific immune checkpoint gene expression profiles and chemotherapy drug sensitivity patterns, and points to potential targeted therapies such as dasatinib for high-risk patients carrying KIT mutations, while exploring the possibility of other intervention strategies such as immunotherapy for high-risk patients without KIT mutations. Attached Figure Description

[0027] Figure 1 This demonstrates the technical process for developing and validating the RR pAML model.

[0028] Figure 2 The development and validation of the RR pAML model are shown. (A) An RR pAML model was constructed using a training cohort, classifying patients into RR groups. + Low-risk and high-risk groups. Results showed that RR... +(A) Overall survival (OS) and event-free survival (EFS) were significantly reduced in the high-risk group. (B) Area under the ROC curve (AUC) of OS and EFS at 1, 3, 5, and 7 years in the training cohort. (C) Forest plot showing the percentage of peripheral blood blasts and the contribution of MRD1 to the model. (D) Internal validation of the RR pAML model. (E) AUC of OS and EFS at 1, 3, 5, and 7 years in the internal validation cohort. (F) Calibration curves of OS and EFS in the internal validation cohort. (GH) Validation of the RR pAML model in external cohorts in Southwest and Northeast China.

[0029] Figure 3 The results show that the RR pAML model exhibits superior prognostic performance compared to existing risk stratification systems. (AB) In both the training and internal validation cohorts, the C-index of the RR pAML model is higher than that of the study-defined risk group and the final risk group for both OS and EFS. (CD) In ​​both the training and internal validation cohorts, the 3-year AUC values ​​of the RR pAML model for both OS and EFS are higher than those of the study-defined risk group and the final risk group. (EF) Decision curve analysis (DCA) at 3-year time points for OS and EFS in both the training and internal validation cohorts. (GH) In the external validation cohort, compared to the Chinese AML stratification tool, the RR pAML model shows higher 3-year AUC values ​​for both OS and EFS. (IJ) RR + The reclassification of the cohorts showed that more than 30% of patients were reclassified as high-risk in the training, internal and external validation cohorts.

[0030] Figure 4 The effects of KIT mutation and dasatinib treatment on overall survival (OS) and emergency survival (EFS) are shown. (A) The mutation waterfall plot shows that KIT mutation was the most common mutation in all cohorts. (B) Kaplan-Meier survival curve analysis shows that KIT mutation significantly reduced OS (p = 0.00021) and EFS (p < 0.0001) in all cohorts. (CD) Dasatinib treatment in KIT mutation-positive patients showed positive response rate (RR). + In the high-risk group, the treatment efficacy was more significant for OS (p = 0.025) and EFS (p = 0.018). (E) Dasatinib treatment in KIT mutation-positive RR + In both the low-risk and high-risk groups, there were no significant differences in overall survival (OS) (p = 0.33) and emergency response time (EFS) (p = 0.73), indicating that treatment benefit was not affected by risk stratification. (F) KIT mutation-positive RR in patients not receiving dasatinib treatment + High-risk patients consistently had lower overall survival (OS) (p = 0.00047) and free risk of failure (EFS) (p < 0.0001).

[0031] Figure 5 This demonstrates transcriptomic analysis based on immune checkpoint genes and prediction of chemotherapeutic drug sensitivity. (A) Transcriptomic analysis of immune checkpoint genes (ICGs), comparing healthy individuals and KIT mutation-negative RR. + Low-risk and high-risk groups. Results showed differences in immune checkpoint gene expression between the groups. (B) KEGG analysis suggested that these genes may be involved in the immunosuppressive process. (C) Based on IC50 values ​​(half the inhibitory concentration) in KIT mutation-negative RR... + The differences between the low-risk and high-risk groups identified four chemotherapy drugs that may provide a reference for individualized chemotherapy regimens for patients at different risks. Detailed Implementation

[0032] It should be noted that the following detailed descriptions are exemplary and intended to provide further illustration of the invention. Unless otherwise specified, all technical and scientific terms used in this invention have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains. It is important to note that the terminology used herein is for the purpose of describing particular embodiments only and is not intended to limit the exemplary embodiments according to the present invention. As used herein, the singular form is intended to include the plural form as well, unless the context clearly indicates otherwise. Furthermore, it should be understood that when the terms “comprising” or “including” are used in this specification, they indicate the presence of features, steps, operations, devices, components, or combinations thereof. Experimental methods described in the following detailed descriptions, unless specific conditions are specified, are generally performed according to conventional methods and conditions in molecular biology within the art, which are fully explained in the literature. See, for example, the techniques and conditions described in Sambrook et al., *Molecular Cloning: A Laboratory Manual*, or according to the conditions recommended by the manufacturer.

[0033] The present invention will be further illustrated with specific examples. These examples are for illustrative purposes only and do not limit the scope of the invention. Unless otherwise specified, experimental conditions not explicitly stated in the examples are generally performed under conventional conditions or as recommended by the selling company. Materials and reagents used in the examples, unless otherwise specified, are commercially available.

[0034] definition

[0035] Minimal residual disease after induction therapy (MRD): The state in which a small number of leukemia cells remain in the body of a leukemia patient after complete remission following induction chemotherapy (or after bone marrow transplantation).

[0036] Minimal residual disease after first induction therapy (MRD1): The state of a small number of leukemia cells remaining in the body of a leukemia patient after complete remission of the first course of induction chemotherapy (or after bone marrow transplantation).

[0037] Percentage of blast cells in peripheral blood at diagnosis: In the diagnosis of acute leukemia, the percentage of blast cells (including blast granulocytes, blast monocytes and blast megakaryocytes) in the total number of leukocytes in peripheral blood.

[0038] RUNX1::RUNX1T1 positive childhood acute myeloid leukemia: occurs in children and adolescents, is a malignant clonal disease originating from bone marrow hematopoietic stem / progenitor cells, where myeloid blast cells proliferate and accumulate uncontrollably in the bone marrow, peripheral blood or other tissues, suppressing hematopoiesis, leading to anemia, bleeding and infection, and is detected to carry the characteristic genetic marker RUNX1::RUNX1T1 fusion gene.

[0039] Technical solution

[0040] (I) Overall Technical Approach

[0041] This invention first constructs a binary risk stratification model (RR pAML model) based on MRD1 and the percentage of peripheral blood blasts at diagnosis from large-scale multicenter data. After rigorous internal and external validation and comparison with existing models, its effectiveness in assessing RR is confirmed. + The model enhances the accuracy and superiority of prognostic assessment for pAML patients. Furthermore, it delves into the biological characteristics of high-risk individuals identified by the model, integrating KIT mutation analysis, immune checkpoint gene expression profiling, and computational drug sensitivity prediction to form a complete closed loop from risk identification to mechanism exploration and treatment strategy recommendations.

[0042] Figure 1 This demonstrates the technical process for developing and validating the RR pAML model.

[0043] (II) Constructing a prognostic model based on two core clinical variables

[0044] First, clinical data from a total of 2009 pAML patients under the age of 28 years were systematically collected from the public data portal of the National Cancer Institute's TARGET database (Therapeutically Applicable Research to Generate Effective Treatments) (https: / / portal.gdc.cancer.gov / ). Through bioinformatics analysis, 284 positive samples carrying the RR fusion gene were selected from these 2009 samples, forming the core RR... + Patient cohorts. Using computer-generated random numbers, patients were randomly divided into two independent sub-cohorts in a 7:3 ratio: a training cohort (70% of the total population, i.e., N = 199 cases) and an internal validation cohort (30% of the total population, i.e., N = 85 cases).

[0045] In the training cohort (N=199), the model building process was initiated. First, using Kaplan-Meier survival analysis combined with log-rank tests, the correlation between multiple potential clinical predictors (including but not limited to: sex, race, age, white blood cell count at diagnosis, percentage of bone marrow blasts at diagnosis, percentage of peripheral blood blasts at diagnosis, presence of central nervous system disease, FAB classification, chromosomal genetic abnormalities, mutations, MRD1, MRD2, etc.) and overall survival (OS) and event-free survival (EFS) was evaluated in the training cohort. For continuous variables, particularly the core variables MRD1 level and percentage of peripheral blood blasts at diagnosis, receiver operating characteristic (ROC) curve analysis was used, and the optimal risk cutoff value for predicting poor prognosis (such as death or relapse) was determined based on the Youden Index principle. The optimal cutoff value was determined by plotting ROC curves and calculating the Youden Index (sensitivity + specificity - 1) for each point on the curve, taking the diagnostic threshold corresponding to the maximum value of this index.

[0046] Then, the continuous variables were transformed into binary variables (high risk, low risk) based on their optimal cutoff values. Subsequently, univariate Cox proportional hazards regression analysis was performed to screen for factors significantly associated with OS / EFS (white blood cell count at diagnosis, MRD1 level, and percentage of peripheral blood blasts at diagnosis). These significant factors were then incorporated into a multivariate Cox proportional hazards regression model to correct for interactions between factors, ultimately confirming independent prognostic factors. In this implementation scheme, after the above steps, the MRD1 level after induction chemotherapy (binary) and the percentage of peripheral blood blasts at diagnosis (binary) were identified as two independent and potent prognostic factors. The thresholds for MRD1 and peripheral blood were calculated using optimal cutoff values. The optimal risk cutoff value for MRD1 was 0.015. The optimal cutoff value for the percentage of peripheral blood blasts was 67.5%. Based on this, a concise binary risk stratification model, namely the "RR pAML model," was constructed. The model's rules are defined as follows: if a patient meets either "MRD1 level is higher than or equal to the cutoff value" or "percentage of peripheral blood blasts at diagnosis is higher than or equal to the cutoff value" or both, they are classified as a high-risk group; if both are lower than the cutoff value, they are classified as a low-risk group.

[0047] Figure 2 The AC results demonstrate the development of the RR pAML model. The RR pAML model was constructed using a training cohort, classifying patients into RR groups. + Low-risk and high-risk groups. Results showed that RR... + Both overall survival (OS) and event-free survival (EFS) were significantly reduced in the high-risk group. Figure 2A). Area under the ROC curve (AUC) of OS and EFS in the training cohort at 1, 3, 5, and 7 years. Figure 2 (B). Forest plots show the percentage of peripheral blood blasts and the contribution of MRD1 to the model ( Figure 2 (C).

[0048] (III) Internal and external validation of the prognostic model

[0049] The initially constructed RR pAML model was validated using an independent internal validation cohort (N=85). Kaplan-Meier survival curves were plotted for the high-risk and low-risk groups, and the log-rank test was used to compare whether there were significant differences in OS and EFS between the two groups, in order to visually demonstrate the model's discriminative ability.

[0050] Figure 2 The DH shows the validation of the RR pAML model: internal validation of the RR pAML model ( Figure 2 (D); AUC values ​​of OS and EFS in the internal validation queue at 1 year, 3 years, 5 years, and 7 years ( Figure 2 The calibration curves of OS and EFS in the internal validation queue (E). Figure 2 (F).

[0051] In addition, an external validation cohort from three independent clinical centers was introduced, totaling 294 RRs. + Patients with pAML whose initial diagnosis spanned from March 2013 to June 2022.

[0052] The 294 patients included: 196 in the Northeast cohort, of which 128 remained after feature screening; and 98 in the Southwest cohort, of which 86 remained after feature screening.

[0053] Screening requirements include:

[0054] (1) The inclusion criteria were: diagnosed with AML; belonging to the RUNX1::RUNX1T1 subtype; and no key data missing.

[0055] (2) Exclusion criteria: M3 subtype; diagnosed after June 2022.

[0056] The external validation cohorts specifically included: (a) the Northeast China cohort (N=128); and (b) the Southwest China cohort (N=86). In these internal and external validation cohorts, patients were grouped by risk using established cutoff values ​​and stratification rules, and survival analyses were repeated to test the reproducibility and predictive power of the RR pAML model across different populations.

[0057] Figure 2The GH data shows the validation of the RR pAML model in external cohorts in Southwest and Northeast China. The validation results demonstrate that the model of this invention has good repeatability and accurate predictive power.

[0058] (iv) Quantitative evaluation and comparison of model performance

[0059] Multiple statistical indicators were used to rigorously quantify the performance of the RR pAML model. First, Harrell's C-index (the concordance index between the model's predicted risk score and the patient's actual survival time) was calculated; a value closer to 1 indicates higher predictive accuracy. Second, time-dependent receiver operating characteristic (ROC) curves were plotted, and the area under the curve (AUC) was calculated to assess the model's discriminative power in predicting 1-year, 3-year, and 5-year survival rates. Furthermore, decision curve analysis (DCA) was performed to quantify the net clinical benefit of the model at different threshold probabilities compared to the "all treatment" or "no treatment" strategies, thereby evaluating its clinical applicability. To demonstrate the superiority of this model, it was compared head-to-head with other widely used prognostic stratification tools, including "study-defined risk groups," "improved final risk models," and "Chinese AML risk stratification tools." In the same validation cohort, the C-index, AUC, and other indicators of these competing models were calculated and compared in parallel to highlight the RR pAML model's superiority in predicting childhood RR. + The model exhibits superior prognostic performance in pAML. To further evaluate the model's robustness, the entire RR was also analyzed. + Five-fold cross-validation is performed on the queue (N=284), which means that the queue is randomly divided into 5 subsets of similar size, and the model is built by taking turns using 4 of the subsets as the training set and validating on the remaining 1 subset. This process is repeated 5 times, and finally the results of the 5 validations are combined to obtain a stable performance estimate.

[0060] The study defined risk groups: Based on the risk model developed by the Children's Oncology Group (COG) based on the AAML03P1 trial, CBF-positive patients were classified as low-risk, distinguishing them from high-risk patients with high-risk cytogenetic features (such as monosomy 7, 5q deletion, or high allele ratio FLT3-ITD mutation) (Pollard, JA, et al., Gemtuzumab Ozogamicin Improves Event-Free Survival and Reduces Relapse in Pediatric KMT2A-Rearranged AML: Results From the Phase III Children's Oncology Group Trial AAML0531. J Clin Oncol, 2021. 39(28): p.3149-3160.).

[0061] Improved final risk model: Based on the final risk model established in the AAML0531 and AAML1031 studies, the accuracy of prognostic prediction was further optimized by incorporating immunophenotypic features and next-generation sequencing (NGS) data (Cooper, TM, et al., Revised Risk Stratification Criteria for Children with Newly Diagnosed Acute Myeloid Leukemia: A Report from the Children's Oncology Group. Blood, 2017. 130 (Supplement 1): p. 407.)

[0062] The Chinese AML Risk Stratification Tool is based on a combination of the patient's cytogenetic and molecular genetic characteristics and the results of MRD testing after treatment. The low-risk group is established based on the presence of a clearly favorable genetic abnormality, including t(8;21) translocation, inv(16) or t(16;16), CEBPA gene mutation (without FLT3-ITD mutation), and NPM1 or IDH gene mutation (also without FLT3-ITD mutation), with a negative MRD1 result. For the high-risk group, a patient is classified as high-risk if they have any of the following: FLT3-ITD mutation, monosomy 5 or 7, t(6;9) translocation, t(8;16) translocation, rare t(6;21) translocation, inv(3) or t(3;3) translocation, or a complex karyotype. The intermediate-risk group is for patients who do not possess either low-risk or high-risk genetic characteristics.

[0063] Figure 3 The RR pAML model demonstrates superior prognostic performance compared to existing risk stratification systems. In both the training and internal validation cohorts, the C-index of the RR pAML model is higher than that of both the study-defined risk group and the final risk group for both OS and EFS. Figure 3 The RR pAML model achieved higher 3-year AUC values ​​for both OS and EFS in the training and internal validation cohorts than the study-defined risk group and the final risk group. Figure 3 (CD). Decision curve analysis (DCA) of OS and EFS at 3-year time points in the training and internal validation queues is as follows: Figure 3 As shown in the EF diagram. In the external validation queue, compared with the Chinese AML layering tool, the RR pAML model showed a higher 3-year AUC value on both OS and EFS. Figure 3 (GH). RR + The reclassification of the cohorts showed that over 30% of patients were reclassified as high-risk in the training, internal, and external validation cohorts. Figure 3 (IJ).

[0064] (v) Linking Risk Stratification with Personalized Treatment

[0065] Building upon the successful differentiation between high-risk and low-risk groups using the RR pAML model, further translational medicine research is being conducted to link risk stratification with personalized treatment decisions. First, the molecular genetic characteristics of high-risk patients are analyzed, with a focus on detecting the mutation status of the KIT gene (GeneID: 3815). Next-generation sequencing technology is used to sequence tumor samples from high-risk patients, analyzing the presence of activating mutations in the exon regions of the KIT gene (especially exons 8, 11, and 17). If high-risk patients carrying KIT mutations are identified, tyrosine kinase inhibitors (such as dasatinib) are recommended or evaluated as potential targeted therapy options, thus providing precise treatment guidance for specific high-risk subgroups identified by the model.

[0066] Figure 4 This chart shows the impact of KIT mutations and dasatinib treatment on overall survival (OS) and emergency survival (EFS). The mutation waterfall plot shows that KIT mutations were the most common mutation across all cohorts. Figure 4 Kaplan-Meier survival curve analysis showed that the KIT mutation significantly reduced overall survival (OS) (p = 0.00021) and overall survival (EFS) (p < 0.0001) in all cohorts. Figure 4 (B). Dasatinib treatment in KIT mutation-positive RR + In the high-risk group, the treatment effects on OS (p = 0.025) and EFS (p = 0.018) were more significant. Figure 4(CD). Dasatinib treatment in KIT mutation-positive RR + There were no significant differences in overall survival (OS) (p = 0.33) and emergency response time (EFS) between the low-risk and high-risk groups, indicating that treatment benefit was not affected by risk stratification. Figure 4 (E). KIT mutation-positive RR in patients not receiving dasatinib treatment. + High-risk patients consistently had lower overall survival (OS) (p = 0.00047) and effective survival (EFS) (p < 0.0001). Figure 4 (F).

[0067] Secondly, we performed an expression profile analysis of immune checkpoint-related genes (ICGs). We obtained 76 experimentally validated immune checkpoint targets and regulators from the Cancer Knowledge Transcriptome Tumor Database (CKTTD). Using RR... + Existing RNA sequencing data from the cohort were standardized using the TMM method (trimmed mean of M-values, a commonly used gene expression normalization method designed to eliminate non-biological biases introduced by experimental conditions, thereby improving the accuracy and reproducibility of the analysis). The DESeq2 software package was used to analyze data from healthy individuals and RR. + Low-risk patients and RR + Differences in ICG expression among high-risk patients were investigated. Heatmaps visualized the differences in ICG expression profiles among the three groups, and pathway enrichment analysis was performed on the differentially expressed ICG sets to reveal the potential characteristics of the tumor immune microenvironment in high-risk patients, providing a theoretical basis for exploring the application prospects of immune checkpoint inhibitors in this high-risk population. Finally, chemotherapy drug sensitivity prediction analysis was performed using the R language package oncoPredict, based on RR... + Tumor gene expression profiling data from the cohort were used to predict the in vitro half-maximal inhibitory concentration (IC50) of 198 commonly used chemotherapeutic drugs for each patient using a ridge regression model pre-trained on the Genomics of Cancer Drug Sensitivity (GDSC) database. Outliers in the predicted IC50 values ​​were identified and processed using a ROUT method (with a Q value of 5%). By comparing the distribution differences in predicted IC50 for various chemotherapeutic drugs (such as cytarabine, daunorubicin, and mitoxantrone) between high-risk and low-risk patients, drugs that may be more sensitive or more resistant in the high-risk group were identified, thus providing a computational biology-level reference for optimizing chemotherapy regimens for high-risk patients.

[0068] Figure 5 This study demonstrates transcriptomic analysis based on immune checkpoint genes and prediction of chemotherapeutic drug sensitivity. Transcriptional analysis of immune checkpoint genes (ICGs) was performed, comparing the response rates (RR) of healthy individuals and those with KIT mutation-negative responses. +Low-risk and high-risk groups. Results showed differences in immune checkpoint gene expression among the different groups. Figure 5 A). KEGG analysis suggests that these genes may be involved in the immunosuppressive process (A). Figure 5 B). Based on the IC50 value (half the inhibitory concentration) in KIT mutation-negative RR. + The differences between the low-risk and high-risk groups identified four chemotherapy drugs that may provide a reference for individualized chemotherapy regimens for patients at different risks. Figure 5 (C).

[0069] RR pAML prognostic risk stratification assessment device

[0070] This invention provides a risk stratification assessment device for RUNX1::RUNX1T1-positive children with acute myeloid leukemia, comprising:

[0071] The data acquisition unit is used to acquire prognostic factor data of the subject, including minimal residual disease after the first induction therapy and the percentage of peripheral blood blasts at the time of diagnosis.

[0072] An analysis and evaluation unit is used to assess the prognostic risk stratification status of the subject based on the prognostic factor data;

[0073] The results output unit is used to output the risk stratification assessment results.

[0074] An assessment is RR if at least one of the following conditions is met. + High risk: Minimal residual disease ≥0.015 after the first induction therapy, and peripheral blood blast percentage ≥67.5% at diagnosis; if neither of the above two conditions is met, the risk is assessed as recurrent respiratory failure (RR). + Low risk.

[0075] Example

[0076] This example uses a confirmed RR patient. + Using pAML patients as subjects, this paper explains how to utilize the RR pAML prognostic model described in this invention to achieve risk stratification during the disease diagnosis period, and to formulate individualized precision treatment plans based on the stratification results, ultimately improving patient survival prognosis.

[0077] The specific steps are as follows:

[0078] (a) Collection of basic patient information and initial diagnosis data

[0079] The following indicators were collected: percentage of peripheral blood blasts, minimal residual disease (MRD1) in bone marrow after standard induction chemotherapy (based on COG AAML1031 regimen), and molecular genetic testing (KIT gene exon mutation).

[0080] (ii) Applying the model of this invention for risk stratification

[0081] Input the above indicators into the prognostic risk stratification system of this invention:

[0082] According to the model definition, patients who meet one or both of the following criteria are classified as “RR”: ① Peripheral blood blast percentage ≥67.5% at diagnosis; ② MRD1 ≥0.015 after the first induction therapy. + The "high-risk" group. If neither of the above two factors is met, it is classified as "RR". + "Low-risk" group.

[0083] (III) Clinical interpretation and decision support of risk stratification results

[0084] 1. If the patient has "RR" + The "high-risk" group had a significantly lower expected 5-year overall survival rate than the low-risk group (approximately 73.0% vs. 91.0%), with a significantly increased risk of relapse.

[0085] (1) If a KIT exon mutation is detected

[0086] Targeted therapy: Systemic therapy with the addition of the tyrosine kinase inhibitor dasatinib during consolidation chemotherapy is the preferred recommendation. Clinical data show that this regimen can significantly improve the survival of these patients (both OS and EFS are significantly improved, p<0.05).

[0087] Intensive treatment: It is recommended to use high-risk group intensive chemotherapy regimens and to assess the timing and feasibility of hematopoietic stem cell transplantation (HSCT) at an early stage.

[0088] (2) No KIT exon mutations were detected.

[0089] Make full use of the extended information provided by the model (immune checkpoint expression, drug sensitivity prediction) and actively seek opportunities for cross-indication drug use or enrollment in relevant clinical trials. Plan hematopoietic stem cell transplantation (HSCT) as a core curative treatment and use the above-mentioned targeted / immunotherapy as important tools for reducing tumor burden before transplantation and maintaining / preventing recurrence after transplantation.

[0090] 2. If the patient is "RR" + Low-risk group: only standard dose chemotherapy was used.

[0091] (iv) Clinical implementation and follow-up results

[0092] Clinicians adopted the model-recommended treatment pathway: combining targeted therapy with subsequent consolidation chemotherapy. After completing chemotherapy, the patient maintained molecular remission. At 36 months of follow-up, the patient was relapse-free and survived without experiencing serious treatment-related toxicities.

[0093] This embodiment demonstrates that the RR pAML prognostic model of the present invention can achieve early and accurate stratification, and can identify "occult high-risk" patients in the traditional classification after diagnosis and induction therapy; it provides clear clinical decision-making basis, automatically generates individualized treatment recommendations based on risk stratification and molecular characteristics (KIT mutation); it improves patient prognosis, significantly reduces the risk of recurrence and improves long-term survival rate by guiding high-risk patients to receive intensive and targeted therapy.

[0094] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention in any way. Therefore, any simple modifications, equivalent changes, and alterations made to the above embodiments based on the technical essence of the present invention without departing from the scope of the present invention shall still fall within the scope of the present invention.

Claims

1. A method for constructing a risk stratification assessment model for RUNX1::RUNX1T1-positive childhood acute myeloid leukemia, characterized in that, Includes the following steps: S1. Obtain clinical data of positive samples carrying the RR fusion gene and randomly divide them into a training cohort and an internal validation cohort. S2. Using Kaplan-Meier survival analysis combined with log-rank test, the correlation between multiple potential clinical predictors and overall survival and event-free survival was evaluated one by one in the training cohort. S3. For continuous variables, receiver operating characteristic (ROC) curve analysis is used, and the optimal risk cutoff value for predicting poor prognosis is determined according to the Youden index principle. The optimal cutoff value is determined by plotting the ROC curve and calculating the Youden index corresponding to each point on the curve, and taking the diagnostic threshold corresponding to the maximum value of the index. S4. Transform the above continuous variables into binary variables, namely high risk and low risk, according to their optimal cutoff values. S5. Perform univariate Cox proportional hazards regression analysis to screen out factors that are significantly associated with overall survival and event-free survival; then incorporate these significant factors into the multivariate Cox proportional hazards regression model to correct for the mutual influence between factors and finally identify independent prognostic factors. S6. Determine minimal residual disease after the first induction therapy and the percentage of peripheral blood blasts at diagnosis as prognostic factors.

2. The method according to claim 1, characterized in that, The thresholds for minimal residual disease after the first induction therapy and the percentage of peripheral blood blasts at diagnosis were calculated using the optimal cutoff values. The optimal risk cutoff value for minimal residual disease after the first induction therapy was 0.015, and the optimal cutoff value for the percentage of peripheral blood blasts at diagnosis was 67.5%.

3. The method according to claim 2, characterized in that, An assessment is RR if at least one of the following conditions is met. + High risk: Minimal residual disease ≥0.015 after the first induction therapy, and peripheral blood blast percentage ≥67.5% at diagnosis; if neither of these conditions is met, the risk is assessed as recurrent respiratory failure (RR). + Low risk.

4. The method according to claim 1, characterized in that, The method also includes an internal verification step.

5. The method according to claim 1, characterized in that, The method also includes an external verification step.

6. A risk stratification assessment model for RUNX1::RUNX1T1 positive children with acute myeloid leukemia, which is constructed according to the method of any one of claims 1 to 5.

7. A risk stratification assessment device for RUNX1::RUNX1T1-positive children with acute myeloid leukemia, characterized in that... include: The data acquisition unit is used to acquire prognostic factor data of the subjects; An analysis and evaluation unit is used to assess the prognostic risk stratification status of the subject based on the prognostic factor data; The results output unit is used to output the risk stratification assessment results. In a preferred embodiment, the prognostic factor data includes minimal residual disease after the first induction therapy and the percentage of peripheral blood blasts at diagnosis.

8. The risk stratification assessment device according to claim 7, characterized in that, An assessment is RR if at least one of the following conditions is met. + High risk: Minimal residual disease ≥0.015 after the first induction therapy, and peripheral blood blast percentage ≥67.5% at diagnosis; if neither of these conditions is met, the risk is assessed as recurrent respiratory failure (RR). + Low risk.

Images