Construction and application of RUNX1 rearrangement leukemia prognosis and sensitive drug prediction model

Abstract

The invention discloses a prediction model construction method for RUNX1 rearrangement leukemia prognosis and sensitive drugs, and belongs to the technical field of biomedicine. The invention discloses an objective rule that the RUNX1 fusion gene causes abnormally high expression of the partner gene through a'promoter exchange 'mechanism for the first time, establishes a prognosis layering prediction model based on the partner gene pan-cancer high expression risk ratio (HR), divides the RUNX1 fusion gene into a high risk group and a low risk group, and establishes a prognosis layering prediction model in TARGET-AML (n = 3, 4, 5, 6, 7, 8, 8, 9, 10, 11, 12, 13, 13, 13, 13, 13, 13, 13, 13, 13, 13, 13, 13, 13, 13, 13, 157) and CNpAML (n = 149) queues verify their predicted value for patient lifetime and minimal residual focus level. Meanwhile, a prediction model of RUNX1 fusion gene sensitive drugs based on partner gene drug sensitivity characteristics is developed. The clinical problems that RUNX1 rearrangement leukemia risk stratification is incomplete, evaluation is inaccurate, and individualized treatment medicine lacks are solved.

Description

Technical Field

[0001] The present invention belongs to the field of biomedical technology, and specifically relates to the construction and application of a model for predicting the prognosis and sensitive drugs of RUNX1 rearranged leukemia. The invention is a model construction method that can be used for predicting the prognosis and sensitive drugs of RUNX1 rearranged leukemia. Background Art

[0002] Leukemia is a malignant clonal disease of the hematopoietic system characterized by impaired proliferation and differentiation of abnormal hematopoietic cells in the bone marrow. Leukemia can be categorized as acute myeloid leukemia (AML), acute lymphoblastic leukemia (ALL), chronic myeloid leukemia (CML), and chronic lymphocytic leukemia (CLL). The pathogenesis of leukemia involves a variety of genetic abnormalities, including chromosomal rearrangements and gene mutations.

[0003] Chromosomal rearrangements that generate fusion genes directly drive the development of leukemia. Therefore, fusion genes are listed in clinical guidelines as important indicators for leukemia risk stratification and prognosis assessment. RUNX::RUNX1T1 and CBFB::MYH11 fusion genes are closely associated with a favorable prognosis. Patients with MLL::AF9 fusion genes are classified as intermediate-risk. Patients with NUP98 fusion genes, such as NUP98::NSD1, are classified as high-risk, with poor response to chemotherapy, high relapse rates, and a poor prognosis.

[0004] The RUNX1 gene is a key transcription factor for the development and maintenance of the hematopoietic system. Chromosomal rearrangements are highly likely to occur in AML. Over 50 different RUNX1 fusion genes have been reported, including RUNX1::RUNX1T1, RUNX1::MECOM, RUNX1::CBFA2T2, and RUNX1::GATA2. However, only RUNX1::RUNX1T1 and RUNX1::MECOM have been clearly linked to clinical prognostic risk. RUNX1::RUNX1T1 is generally associated with a poor prognosis, while RUNX1::MECOM is associated with a high prognosis. Other RUNX1 fusion genes are relatively rare, often discovered only in single cases or small sample sizes, making it difficult to characterize their clinical prognostic implications in large-scale clinical cohorts. Therefore, the clinical significance of most RUNX1 fusion genes remains unclear, necessitating the development of prognostic models applicable to different types of RUNX1-rearranged leukemias.

[0005] Currently, clinical treatment for RUNX1-rearranged leukemia primarily relies on cytotoxic chemotherapy, which not only has significant side effects but also makes some patients with RUNX1 fusion gene carriers insensitive to chemotherapy drugs, prone to relapse, difficult to treat, and ultimately fatal. However, because the partners of the RUNX1 fusion gene are complex and most are extremely rare, it is difficult to conduct innovative drug discovery research one by one. Therefore, whether to quickly provide treatment strategies for clinical patients with RUNX1-rearranged leukemia is an urgent problem to be solved. Summary of the Invention

[0006] The purpose of the present invention is to provide a method for constructing a prediction model for the prognosis and sensitive drugs of RUNX1 rearrangement leukemia, which is helpful for the prognosis assessment and individualized medication guidance of leukemia patients carrying the RUNX1 fusion gene.

[0007] The prediction model construction method is achieved by the following steps: (1) Analyze the RUNX1 fusion gene type and obtain the specific partner gene by high-throughput sequencing technology combined with transcriptome analysis algorithms such as STAR-fusion; (2) Based on the pan-cancer gene expression and patient survival data from the TCGA database, the Cox proportional hazards regression model was used to calculate the hazard ratio (HR) between high and low expression of wild-type partner genes in pan-cancer. (3) The RUNX1 fusion gene risk stratification was assessed based on the HR value. If the HR was > 1, the patient was classified into the high-risk group; if the HR was < 1, the patient was classified into the low-risk group. (4) Based on the gene expression and drug sensitivity data of leukemia cell lines in the GDSC database, the cell lines were divided into high-expressing cell lines and low-expressing cell lines according to the median expression level of the fusion partner gene in the cell lines, and the average inhibitory activity (Z-score) of different drugs on high-expressing and low-expressing cells was calculated respectively; (5) Based on the differential inhibitory activity of cells with high expression of the wild-type partner gene compared with cells with low expression, drugs with large differences and strong inhibitory activity were selected to predict potential sensitive drugs for the corresponding partner's RUNX1 fusion gene.

[0008] The RUNX1 fusion genes include but are not limited to RUNX1::MECOM, RUNX1::ZFPM2, RUNX1::STK3, RUNX1::USP42, RUNX1::FBXO43, RUNX1::TRPM2, RUNX1::BCL11B, RUNX1::QSER1, RUNX1::RUNX1T1, RUNX1::TIAM1, RUNX1::GATA2, RUNX1::AFF3, RUNX1::CBFA2T3, RUNX1::CBFA2T2 and other RUNX1 fusion genes located at the N-terminus.

[0009] The second object of the present invention is to provide the application of the above-constructed model in the prediction of prognosis and sensitive drugs of RUNX1 rearranged leukemia.

[0010] The application of the method for constructing a predictive model for prognosis and sensitive drugs described in the present invention is to use the pan-cancer high expression risk ratio of the wild-type partner gene and the drug sensitivity characteristics to predict the prognosis and sensitive drugs of the corresponding partner RUNX1 fusion gene. The present invention reveals for the first time the objective law that the RUNX1 fusion gene drives the abnormally high expression of the partner gene by exchanging a strong promoter, and on this basis uses the Cox proportional hazard regression model to calculate the hazard ratio (HR) of pan-cancer high expression of the partner gene, which is used to quickly predict the risk stratification and prognosis of the RUNX1 fusion gene. The above-mentioned prediction model construction method was applied in the public TARGET-AML (n=3,157) database and the self-built CNpAML (n=149) cohort, respectively, and the feasibility of the prediction method was verified, which has good predictive accuracy for patient survival and minimal residual disease (MRD) levels.

[0011] Furthermore, based on the common feature of abnormally high expression of partner genes in RUNX1 fusion genes, the present invention proposes for the first time that the drug sensitivity characteristics of wild-type partner genes can be used to predict the sensitive drugs of RUNX1 fusion genes. The gene expression of 25 leukemia cell lines and their sensitivity test results for 205 drugs were analyzed using the GDSC database ( https: / / www.cancerrxgene.org / , Genomics of Drug Sensitivity in Cancer), and obtained a specific sensitive drug for AML cells with high expression of wild-type partner genes, which can become a candidate therapeutic drug for patients with RUNX1 fusion genes. The application of the above-constructed model was verified using RUNX1::QSER1 as an example. Based on the GDSC database, the present invention found that compared with acute myeloid leukemia cells with low QSER1 expression, leukemia cells with high QSER1 expression were highly sensitive to sorafenib. We then recommended sorafenib to a patient with relapsed and refractory leukemia carrying the RUNX1::QSER1 fusion gene. The patient achieved complete remission with sorafenib treatment (the original cells were reduced from 70% to 0% after sorafenib treatment).

[0012] The present invention has the following beneficial effects: Based on an in-depth analysis of different RUNX1 fusion genes, the present invention reveals for the first time the objective rule that RUNX1 fusion genes directly lead to high expression of partner genes through "promoter exchange." Based on this characteristic, on the one hand, the prognosis of RUNX1 fusion genes was successfully predicted by integrating the pan-cancer high expression risk ratio of partner genes. On the other hand, based on the drug sensitivity of cells with high expression of partner genes, potential sensitive drugs for RUNX1 fusion genes were predicted. The technical solution of the present invention solves the clinical problems of the existing technology such as imperfect risk stratification, inaccurate assessment, and lack of personalized treatment drugs for RUNX1 rearrangement leukemia. BRIEF DESCRIPTION OF THE DRAWINGS

[0013] Figure 1 The expression of RUNX1 gene and its partner genes in RUNX1 rearranged and non-rearranged leukemias; A is the expression heat map of RUNX1 gene and its partner genes, and B is the expression of RUNX1 gene in pan-cancer and the schematic diagram of promoter exchange mechanism.

[0014] Figure 2 This is a risk stratification and prognostic analysis chart for the partner gene of the RUNX1 fusion gene in the TARGET-AML database; A is the classification of high / low-risk partner genes based on the pan-cancer high expression risk ratio, and B is the comparison of survival curves for patients with high / low-risk RUNX1 fusion genes.

[0015] Figure 3 Figure 2 is a validation diagram of the risk stratification and minimal residual disease (MRD) level of RUNX1 fusion partner genes in the CNpAML cohort; A is the classification of high / low-risk partner genes based on the pan-cancer high expression risk ratio, and B is the comparison of MRD levels in patients with high / low-risk RUNX1 fusion gene carriers.

[0016] Figure 4It is the prediction and clinical application results of sensitive drugs for the RUNX1::QSER1 fusion gene based on high expression of the partner gene; including A is a scatter plot of the sensitivity of AML cells with high / low expression of QSER1 in the GDSC database to various drugs, and B is the effect of sorafenib on the clearance of leukemia cells in refractory patients carrying RUNX1::QSER1 fusion. DETAILED DESCRIPTION

[0017] The present invention will be further described in detail below with reference to the accompanying drawings and examples. The following examples are intended to illustrate the present invention only and are not intended to limit the scope of the invention. Experimental methods in the examples where specific conditions are not specified are generally carried out under conventional conditions or as recommended by the manufacturer.

[0018] Example 1: RNA-seq analysis of 3,157 samples from the TARGET-AML database successfully identified 10 RUNX1 fusion genes using the STAR-Fusion algorithm. Comparison of gene expression between fusion-positive and -negative samples revealed significant upregulation of partner genes in RUNX1 fusion-positive samples. Further analysis combined with the TCGA database revealed significantly elevated expression of wild-type RUNX1 in LAML compared with other cancer types, suggesting that the RUNX1 promoter is strongly activated in AML. In the case of RUNX1 fusions, the partner gene promoter is swapped with the strong RUNX1 promoter, leading to abnormally high gene expression. These results reveal for the first time the objective mechanism by which RUNX1 fusions lead to increased expression of partner genes through a "promoter swap" mechanism.

[0019] Example 2: To evaluate the clinical significance of different RUNX1 fusion genes, a pan-cancer hazard ratio (HR) for overexpression across 33 cancer types was integrated from the TCGA database. A Cox proportional hazards regression model was used to calculate the HR for overexpression of partner genes in the pan-cancer setting, stratifying RUNX1 fusion genes into high- and low-risk groups (HR > 1, high risk; HR < 1, low risk). The analysis showed that the known RUNX1::RUNX1T1 (HR = 0.8) was classified as low-risk, while RUNX1::MECOM (HR = 1.4) was classified as high-risk, indicating that this model is consistent with current clinical consensus. Subsequently, survival analysis of the TARGET-AML cohort confirmed that patients with high-risk RUNX1 fusion genes had significantly shorter median survival than those with low-risk RUNX1 fusion genes. This finding provides a rapid predictive model for risk stratification and prognostic assessment of patients with different RUNX1-rearranged leukemias.

[0020] Example 3: In a cohort of 149 AML patients at Zhejiang University Children's Hospital, four novel RUNX1 fusion genes (RUNX1::FBXO43, RUNX1::TRPM2, RUNX1::BCL11B, and RUNX1::QSER1) were identified. Expression analysis showed that the hazard ratios for high expression of the gene partners of these fusion genes were all greater than 1, suggesting that they may be high-risk RUNX1 fusion genes. Further clinical data demonstrated that the MRD levels of these patients after the first induction chemotherapy were indeed significantly higher than those in the RUNX1::RUNX1T1 group, confirming their high-risk nature. This result further validated the reliability of our proposed RUNX1 fusion gene risk stratification prediction model from the perspective of minimal residual disease.

[0021] Example 4: To address the difficulty in choosing medication for patients with RUNX1 fusion, this paper analyzes the gene expression of 25 leukemia cell lines and their sensitivity to 205 drugs based on the GDSC database ( https: / / www.cancerrxgene.org / , Genomics of Drug Sensitivity in Cancer). The specific method involves dividing cell lines into high- and low-expression groups based on the median expression of the target fusion partner gene. The average inhibitory activity (Z-score) of each group against each drug is calculated. Drugs with large differentials and strong inhibitory activity are selected to predict potential sensitive drugs for the corresponding partner's RUNX1 fusion gene. The following application uses the RUNX1 / QSER1 as an example. Clinically, a patient with AML harboring the RUNX1::QSER1 fusion gene experienced relapse and refractory disease 16 months after transplantation. For this patient with the RUNX1::QSER1 fusion gene, GDSC database analysis revealed that QSER1-overexpressing AML cells were resistant to 74% of chemotherapy drugs but were specifically sensitive to sorafenib and venetoclax (the average Z-score of the QSER1-overexpressing cell lines was lower than that of the low-expressing cells). Considering the patient's high resistance to clinical chemotherapy after relapse (after treatment with HAG + epirubicin, the tumor percentage only decreased from 90% to 70%), we recommended sorafenib treatment for this refractory patient. The results showed that after sorafenib monotherapy, the patient's tumor cell percentage decreased from 70% to 0% after the RUNX1::QSER1 fusion gene carrier was treated with sorafenib alone, achieving complete remission. This case demonstrates that drug sensitivity characteristics based on partner genes can quickly predict personalized treatment drugs for patients with RUNX1-rearranged leukemia.

Claims

1. A method for constructing a model for predicting the prognosis and drug sensitivity of RUNX1-rearranged leukemia, characterized in that: This is achieved by following these steps: (1) Analyze the RUNX1 fusion gene type and obtain the specific partner gene by high-throughput sequencing technology combined with transcriptome analysis algorithms such as STAR-fusion; (2) Based on the pan-cancer gene expression and patient survival data from the TCGA database, the Cox proportional hazards regression model was used to calculate the hazard ratio (HR) between high and low expression of wild-type partner genes in pan-cancer. (3) The RUNX1 fusion gene risk stratification was assessed based on the HR value. If the HR was > 1, the patient was classified into the high-risk group; if the HR was < 1, the patient was classified into the low-risk group. (4) Based on the gene expression and drug sensitivity data of leukemia cell lines in the GDSC database, the cell lines were divided into high-expressing cell lines and low-expressing cell lines according to the median expression level of the fusion partner gene in the cell lines, and the average inhibitory activity (Z-score) of different drugs on high-expressing and low-expressing cells was calculated respectively; (5) Based on the differential inhibitory activity of cells with high expression of the wild-type partner gene compared with cells with low expression, drugs with large differences and strong inhibitory activity were selected to predict potential sensitive drugs for the corresponding partner's RUNX1 fusion gene.

2. The RUNX1 fusion gene according to claim 1, characterized in that Including but not limited to RUNX1::MECOM, RUNX1::ZFPM2, RUNX1::STK3, RUNX1::USP42, RUNX1::FBXO43, RUNX1::TRPM2, RUNX1::BCL11B, RUNX1::QSER1, RUNX1::RUNX1T1, RUNX1::TIAM1, RUNX1::GATA2, RUNX1::AFF3, RUNX1::CBFA2T3, RUNX1::CBFA2T2 and other fusion genes with RUNX1 located at the N-terminus.

3. Application of the model constructed according to the method of claim 1 in predicting drug sensitivity in RUNX1 rearrangement leukemia.

Images