Model for predicting early chemotherapy failure of DLBCL patient and application thereof

Abstract

The invention relates to the field of tumor molecular diagnosis and precision medicine, in particular to a model for predicting early chemotherapy failure of a DLBCL patient and application of the model. By detecting the ctDNA concentration level of the patient base line and the early treatment period and / or combining clinical characteristic parameters, a multivariable risk assessment algorithm is established, the high-risk patient can be identified in the early treatment period, a reliable basis is provided for individualized treatment, and remarkable clinical and social values are achieved.

Description

Technical Field

[0001] This invention relates to the field of tumor molecular diagnostics and precision medicine, specifically to a model for predicting early chemotherapy failure in DLBCL patients and its application. Background Technology

[0002] Diffuse large B-cell lymphoma (DLBCL) is the most common type of non-Hodgkin lymphoma, and the R-CHOP regimen is its standard first-line treatment. However, approximately 40% of patients experience primary refractory or early relapse after the first round of treatment, i.e., early chemotherapy failure (ECF). ECF patients generally have a very poor prognosis, which has become a core challenge in the current clinical treatment of DLBCL.

[0003] Currently, clinical prognostic assessment mainly relies on traditional clinical indicators such as the International Prognostic Index (IPI). However, these static indicators are difficult to dynamically reflect the true response of tumors to treatment, and their sensitivity in predicting ECF risk is generally insufficient. Circulating tumor DNA (ctDNA), as an emerging non-invasive liquid biopsy biopsy marker, can reflect tumor burden and molecular residual lesion status in real time and with high sensitivity, showing great potential in efficacy assessment and recurrence prediction. However, existing technologies mostly focus on post-treatment monitoring and lack quantitative predictive models that integrate ctDNA levels and clinical characteristics in the early stages of treatment, resulting in the inability to achieve dynamic risk assessment of early efficacy for patients under different treatment regimens.

[0004] Therefore, there is an urgent need in this field to develop a predictive tool that can accurately identify high-risk ECF patients in the early stages of treatment. By constructing a comprehensive predictive model based on ctDNA quantitative analysis and combined with key clinical parameters, it is hoped that early and dynamic stratification of ECF risk can be achieved, thereby providing a reliable basis for clinicians to formulate timely and personalized treatment strategies, which has significant clinical value and social significance. Summary of the Invention

[0005] In view of this, the technical problem to be solved by the present invention is to provide a model for predicting early chemotherapy failure in DLBCL patients and its application. By detecting the patient's baseline and early treatment period ctDNA concentration levels, and / or combining clinical characteristic parameters (such as IPI score, LDH, ECOG score, etc.), a multivariate risk assessment algorithm is established to achieve early identification and stratified management of ECF risk.

[0006] The first aspect of the present invention provides a method for constructing a model to predict early chemotherapy failure in patients with diffuse large B-cell lymphoma, comprising: obtaining the patient's ctDNA level and constructing a comprehensive assessment model for predicting the risk of early chemotherapy failure through multivariate regression analysis.

[0007] In some embodiments, obtaining the patient's ctDNA level includes: obtaining blood samples collected before the start of the first and second treatment cycles, using a tumor-informed ctDNA detection method, identifying tumor-specific mutation sites through targeted capture sequencing, and calculating the ctDNA concentration.

[0008] In some embodiments, the ctDNA detection is referenced to the mutation profile of the patient's tumor tissue.

[0009] In some embodiments, the construction method further includes obtaining clinical characteristic parameters of the patient, including at least one of the following: IPI score, age, Ann Arbor stage, LDH, ECOG score, or extranodal involvement site.

[0010] In some embodiments, the construction step further includes:

[0011] The patient cohort with diffuse large B-cell lymphoma was divided into a training set and a validation set.

[0012] A prediction model is constructed in the training set using the logistic regression algorithm, and the parameters of the prediction model are obtained by screening using multivariate Cox regression.

[0013] The accuracy of the model was verified on the validation set.

[0014] A second aspect of the invention provides a model for predicting early chemotherapy failure in patients with diffuse large B-cell lymphoma.

[0015] In some embodiments, the model is logit(P(ECF=1))=hGE C1 ;

[0016] Alternatively, logit(P(ECF=1))=hGE C2 ;

[0017] Alternatively, logit(P(ECF=1))=hGE C2 ÷hGE C1 ;

[0018] The hGE C1 The hGE concentration is the value of ctDNA in a blood sample collected before the start of the first course of treatment for patients with diffuse large B-cell lymphoma. C2 The value of ctDNA in a blood sample collected before the start of the second course of treatment for a patient with diffuse large B-cell lymphoma.

[0019] In some embodiments, a high-risk threshold for ctDNA concentration is determined using logistic regression models and ROC curve analysis; preferably, the high-risk threshold is 3 log hGE / mL. Patients with diffuse large B-cell lymphoma and ctDNA concentrations ≥3 log hGE / mL are identified as a group with a high risk of early chemotherapy failure, and these patients are classified as high-risk ECF patients.

[0020] In some specific embodiments,

[0021] The model is logit(P(ECF=1))=hGE C1 Its AUC = 0.826;

[0022] The model is logit(P(ECF=1))=hGE C2 Its AUC = 0.718;

[0023] The model is logit(P(ECF=1))=hGE C2 ÷hGE C1 Its AUC=0.686.

[0024] In some specific embodiments, the danger threshold for ctDNA concentration is set at 3 log hGE / mL, and the model logit(P(ECF=1))=hGE is used. C1 This model predicts the risk of early chemotherapy failure and outputs an early chemotherapy failure risk score and / or early chemotherapy failure risk stratification results based on the risk score. Experimental verification showed that the model achieved a sensitivity of 81.3%, a specificity of 79.4%, and an accuracy of 79.8% in predicting early chemotherapy failure. Under the same experimental conditions, the IPI score had a sensitivity of 62.5%, a specificity of 77.9%, and an accuracy of 75% in predicting early chemotherapy failure. The comparative results indicate that the ctDNA concentration-based prediction model provided by this invention significantly outperforms the traditional IPI score in terms of overall diagnostic performance, particularly in sensitivity, and can more effectively identify patients with a high risk of early chemotherapy failure.

[0025] In some embodiments, the model is logit(P(ECF=1))=α+β1*hGE C1 +β2* IPI+ε;

[0026] Wherein, hGE C1The ctDNA concentration value is the value of a blood sample collected before the start of the first course of treatment for patients with diffuse large B-cell lymphoma; the IPI is the IPI score of patients with diffuse large B-cell lymphoma; α, β1, and β2 are model coefficients, and ε is the random error, which is solved by logistic regression and iterative weighted least squares method; α is -5.0 to -6.0, β1 is 1 to 1.5, β2 is 0.1 to 0.5, and ε is 0 to 0.05. For example, α is -5.1, -5.2, -5.3, -5.4, -5.5, -5.6, -5.7, -5.8, or -5.9; β1 is 1, 1.1, 1.2, 1.3, 1.4, or 1.5; β2 is 0.1, 0.2, 0.3, 0.4, or 0.5; and ε is 0, 0.01, 0.02, 0.03, 0.04, or 0.05. Preferably, α is -5.46, β1 is 1.16, β2 is 0.27, and ε is 0.

[0027] In some embodiments, the model is logit(P(ECF=1))=α+β1*hGE C1 +β2* Hans+β3* AAS +β4* LDH +β5* ECOG +β6* Extranodal +ε;

[0028] Wherein, hGE C1 The values ​​represent: ctDNA concentration in blood samples collected before the start of the first treatment course for patients with diffuse large B-cell lymphoma; Hans classification for diffuse large B-cell lymphoma, with GCB type represented by a value of 0 and non-GCB type by a value of 1; AAS classification for diffuse large B-cell lymphoma, with stages 1 and 2 represented by a value of 0 and stages 3 and 4 by a value of 1; LDH level in serum, with the normal range represented by a value of 0 and elevated LDH levels by a value of 1; ECOG score, with 0 and 1 points representing a value of 0 and ≥2 points representing a value of 1; Extranodal involvement in patients with diffuse large B-cell lymphoma, with 0 or 1 extranodal involvement representing a value of 0 and ≥2 involved sites representing a value of 1.

[0029] α, β1, β2, β3, β4, β5 and β6 are model coefficients, and ε is the random error, which is solved by logistic regression and iterative weighted least squares method;

[0030] The values ​​of α are -5.0 to -6.0, β1 is 1 to 1.5, β2 is 0.1 to 0.5, β3 is 0.2 to 0.7, β4 is -0.7 to -0.3, β5 is 1 to 1.5, β6 is 0.1 to 0.5, and ε is 0 to 0.05. For example, α is -5.1, -5.2, -5.3, -5.4, -5.5, -5.6, -5.7, -5.8, or -5.9; β1 is 1, 1.1, 1.2, 1.3, 1.4, or 1.5; β2 is 0.1, 0.2, 0.3, 0.4, or 0.5; β3 is 0.2, 0.3, 0.4, 0.5, 0.6, or 0.7; β4 is -0.7, -0.6, -0.5, -0.4, or -0.3; β5 is 1, 1.1, 1.2, 1.3, 1.4, or 1.5; β6 is 0.1, 0.2, 0.3, 0.4, or 0.5; and ε is 0, 0.01, 0.02, 0.03, 0.04, or 0.05. Preferably, α is -5.6, β1 is 1.312, β2 is 0.282, β3 is 0.451, β4 is -0.501, β5 is 1.19, β6 is 0.145, and ε is 0.

[0031] In some specific embodiments,

[0032] The model is logit(P(ECF=1))=α+β1*hGE C1 +β2* IPI+ε, its AUC=0.837;

[0033] The model is logit(P(ECF=1))=α+β1*hGE C1 +β2* Hans+β3* AAS +β4* LDH +β5*ECOG +β6* Extranodal +ε, its AUC=0.843;

[0034] Under the same experimental conditions, the IPI score predicted the risk of early chemotherapy failure, with an AUC of 0.746.

[0035] The comparative results show that, in predicting early chemotherapy failure in patients with diffuse large B-cell lymphoma, the predictive model constructed in this invention, which integrates ctDNA concentration and clinical characteristic parameters, significantly outperforms the traditional IPI score. Specifically, the model based on ctDNA combined with clinical parameters has higher sensitivity than the IPI score in predicting early chemotherapy failure; the overall predictive accuracy and stability of different ctDNA-related models are also superior to the single IPI score. The model provided by this invention can more accurately identify high-risk patients, providing a more reliable basis for effective risk stratification in clinical practice.

[0036] In some embodiments, the model described above is used to predict the risk of early chemotherapy failure, and the early chemotherapy failure risk score and / or the early chemotherapy failure risk stratification result based on the risk score are output. If the risk score is greater than or equal to a preset threshold, the patient is classified as a high-risk group for early chemotherapy failure; otherwise, the patient is classified as a low-risk group for early chemotherapy failure.

[0037] In some specific embodiments,

[0038] The model used is logit(P(ECF=1))=α+β1*hGE C1 +β2* IPI+ε predicts the risk of early chemotherapy failure, with a preset threshold of 0.10; if the risk score is greater than or equal to 0.10, the patient is classified as a high-risk group for early chemotherapy failure, and if the risk score is less than 0.10, the patient is classified as a low-risk group for early chemotherapy failure.

[0039] The model used is logit(P(ECF=1))=α+β1*hGE C1 +β2* Hans+β3* AAS +β4* LDH +β5* ECOG+β6* Extranodal +ε predicts the risk of early chemotherapy failure, with a preset threshold of 0.10; if the risk score is greater than or equal to 0.10, the patient is classified as a high-risk group for early chemotherapy failure, and if the risk score is less than 0.10, the patient is classified as a low-risk group for early chemotherapy failure.

[0040] Under the same experimental conditions, the IPI score predicts the risk of early chemotherapy failure, with a preset threshold of 2.50; if the risk score is greater than or equal to 2.50, the patient is classified as a high-risk group for early chemotherapy failure, and if the risk score is less than 2.50, the patient is classified as a low-risk group for early chemotherapy failure.

[0041] Based on the relative risk stratification results indicated by this risk score, high-risk patients who fail early chemotherapy can be given priority for intensive treatment (such as CAR-T therapy, bispecific antibody therapy, etc.) and / or more intensive disease monitoring, thereby providing an objective basis for clinical decision-making.

[0042] A third aspect of the invention provides the use of a model constructed by the method described above, or a model described above, in the preparation of a product for predicting the risk of early chemotherapy failure in patients with diffuse large B-cell lymphoma.

[0043] In some embodiments, the product is an in vitro diagnostic product, which includes diagnostic kits, diagnostic systems, or diagnostic devices.

[0044] A fourth aspect of the present invention provides a diagnostic system or device for predicting early chemotherapy failure in patients with diffuse large B-cell lymphoma, characterized in that it comprises:

[0045] The data acquisition module is used to acquire combined indicator data of the patients to be tested, including the patients' ctDNA levels and / or clinical characteristic parameters.

[0046] The scoring module is used to input the joint indicator data into at least one model as described above and calculate the risk score.

[0047] The judgment module compares the risk score with a set threshold. When the risk score is greater than or equal to the set threshold, the patient is classified as a high-risk patient with early chemotherapy failure of diffuse large B-cell lymphoma; when the risk score is less than the set threshold, the patient is classified as a non-high-risk patient.

[0048] A fifth aspect of the present invention provides a computer program product including computer-readable instructions that, when executed on an electronic device, cause the electronic device to implement the model building method as described above, or to implement the model prediction function as described above, and output the early chemotherapy failure risk score and / or the early chemotherapy failure risk stratification result based on the risk score.

[0049] A sixth aspect of the present invention provides an electronic device comprising at least one processor and a memory connected to said processor, wherein:

[0050] The memory is used to store computer programs;

[0051] The processor is used to execute the computer program to enable the electronic device to implement the model building method as described above, or to implement the model prediction function as described above, and to output the early chemotherapy failure risk score and / or the early chemotherapy failure risk stratification result based on the risk score.

[0052] A seventh aspect of the present invention provides a computer storage medium carrying one or more computer programs that, when executed by an electronic device, enable the electronic device to implement the model building method as described above, or to implement the model prediction function as described above, and output the early chemotherapy failure risk score and / or the early chemotherapy failure risk stratification result based on the risk score.

[0053] This invention relates to a model for predicting early chemotherapy failure in DLBCL patients and its application. A multivariate risk assessment algorithm is established by detecting the baseline and early treatment-phase ctDNA concentration levels of patients, and / or combining clinical characteristic parameters (such as IPI score, LDH, ECOG score, etc.). Results show that patients with ctDNA levels ≥3 log hGE / mL are considered high-risk, with an AUC of 0.821 predicting early chemotherapy failure, significantly better than the traditional indicator IPI's AUC of 0.746. This invention enables early stratification and individualized treatment decisions for DLBCL patients, providing a basis for precision medicine and showing significant clinical application potential.

[0054] Compared with the prior art, the present invention has the following significant advantages:

[0055] It significantly improved the predictive sensitivity for early chemotherapy failure;

[0056] It achieved multidimensional data fusion and quantitative assessment of ctDNA levels and clinical characteristics;

[0057] It supports dynamic monitoring and real-time risk assessment of patient efficacy under different treatment regimens. Attached Figure Description

[0058] Figure 1 This is a flowchart of the model construction and verification process for Example 2;

[0059] Figure 2 Example 2 compares the ROC curves of the ctDNA model for all cases in the training set (84 cases) and the validation set (A, ROC curve comparison of ctDNA before the first treatment cycle, ctDNA before the second treatment cycle, and the ratio of ctDNA before and after the two ctDNA cycles; B, Comparison of the predictive performance of ctDNA before the first treatment cycle in the training set with IPI score; C, Comparison of the predictive performance of ctDNA before the first treatment cycle in the validation set with IPI score).

[0060] Figure 3 For Example 2, a risk stratified survival curve comparison was performed on all cases in the training set (84 cases) and the validation set (61 cases) (A, training set; B, validation set).

[0061] Figure 4 For Example 3, tumor tissue DNA, oral mucosal DNA, and blood samples collected before the start of the first and second treatment cycles from 115 patients with diffuse large B-cell lymphoma were used for ctDNA detection. The ROC curves of the training set model constructed in conjunction with other clinical parameters were compared. The C1_hGE curve represents the model logit(P(ECF=1))=hGE. C1The IPI curve represents the IPI score prediction; the hGE+IPI curve represents the model logit(P(ECF=1))=α+β1*hGE C1 +β2* IPI+ε; hGE+5_factors represents the model logit(P(ECF=1))=α+β1* hGE C1 +β2* Hans+β3* AAS +β4* LDH +β5*ECOG +β6* Extranodal +ε;

[0062] Figure 5 Example 3 used tumor tissue DNA, oral mucosal DNA, and blood samples collected before the start of the first and second treatment cycles from 115 patients with diffuse large B-cell lymphoma for ctDNA detection. The ROC curves of a validation set model constructed using these samples, along with other clinical parameters, were compared. The C1_hGE curve represents the model logit(P(ECF=1))=hGE. C1 The IPI curve represents the IPI score prediction; the hGE+IPI curve represents the model logit(P(ECF=1))=α+β1*hGE C1 +β2* IPI+ε; hGE+5_factors represents the model logit(P(ECF=1))=α+β1* hGE C1 +β2* Hans+β3* AAS +β4* LDH +β5*ECOG +β6* Extranodal +ε;

[0063] Figure 6 This is a schematic diagram of the structure of the diagnostic system or device provided by the present invention;

[0064] Figure 7 This is a schematic diagram of the composition structure of an electronic device provided by the present invention. Detailed Implementation

[0065] This invention provides a model for predicting early chemotherapy failure in DLBCL patients and its application. Those skilled in the art can refer to the content of this document and appropriately modify the process parameters to achieve the desired result. It should be particularly noted that all similar substitutions and modifications are obvious to those skilled in the art and are considered to be included in this invention. The methods and applications of this invention have been described through preferred embodiments. Those skilled in the art can clearly modify or appropriately change and combine the methods and applications described herein without departing from the content, spirit, and scope of this invention to implement and apply the technology of this invention.

[0066] Unless otherwise defined in this invention, scientific and technical terms related to this invention shall have the meanings understood by one of ordinary skill in the art.

[0067] The terms “comprising,” “including,” and “having” are used interchangeably to indicate the inclusiveness of a scheme, meaning that the scheme may contain elements other than those listed. It should also be understood that the use of “comprising,” “including,” and “having” herein also provides for schemes “consisting of…”.

[0068] When used herein, the term “and / or” includes the meaning of “and,” “or,” and “all or any other combination of elements linked by the term.”

[0069] The term "at least one" means one or more, and "more than one" means two or more. "At least one of the following" or similar expressions refer to any combination of these items, including any combination of single or multiple items.

[0070] The term "first course of treatment" refers to the first course of a specific treatment regimen for patients with diffuse large B-cell lymphoma who begin standard treatment, with the first-line treatment being the R-CHOP regimen, consisting of 6 or 8 courses of treatment or other treatment regimens. Generally, one course of treatment lasts 21 days.

[0071] The term "second course of treatment" refers to the second course of treatment for patients with diffuse large B-cell lymphoma who have started standard treatment, with the first-line treatment being the R-CHOP regimen, consisting of 6 or 8 courses of treatment or other specific treatment regimens. Generally, one course of treatment lasts 21 days.

[0072] It should be understood that in the various embodiments of this application, the order of the above processes does not imply the order of execution. Some or all steps may be executed in parallel or sequentially. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of this application.

[0073] The numerical ranges and parameters involved in this invention have been presented as precisely as possible in the specific embodiments. However, any numerical value inevitably contains standard deviations due to individual test methods. Therefore, unless otherwise expressly stated, it should be understood that all numerical ranges or specific data used in this disclosure may have a reasonable deviation within a certain range, such as ±10%, ±5%, ±1%, or ±0.5%.

[0074] Early chemotherapy failure (ECF) is defined as: patients with diffuse large B-cell lymphoma who do not achieve complete remission (CR) or partial remission (PR) after first-line treatment, or who relapse / progress within 12 months of initial diagnosis.

[0075] Although the first-line treatment regimen R-CHOP (rituximab combined with cyclophosphamide, doxorubicin, vincristine, and prednisone) has significantly improved patient outcomes, 30%–40% of patients with diffuse large B-cell lymphoma still face the risk of early chemotherapy failure. These patients have an extremely poor prognosis, with a five-year survival rate of less than 20%. However, emerging therapies such as CAR-T cell therapy, bispecific T-cell connectives (BiTEs), and antibody-drug conjugates (ADCs) offer new hope for patients who have failed early chemotherapy. Therefore, accurately identifying high-risk patients and adjusting treatment regimens in a timely manner remains a key challenge for improving prognosis.

[0076] The test materials used in this invention are all common commercial products and can be purchased on the market.

[0077] The present invention will be further illustrated below with reference to the embodiments:

[0078] Example 1: Construction of a risk prediction model based on ctDNA levels

[0079] 1. Sample collection and ctDNA detection

[0080] Eighty-four newly diagnosed DLBCL patients were selected as the training set (see attached). Figure 1 ECF (extracorporeal blood failure) is defined as relapse or non-response within 12 months of chemotherapy. Peripheral blood samples were collected from patients before the first cycle of chemotherapy (C1) and after the first cycle of chemotherapy (C2).

[0081] Cell-free DNA from plasma, DNA from tumor tissue, and DNA from the oral mucosa were extracted. Tumor-informed ctDNA detection was used to identify tumor-specific mutation sites through targeted capture sequencing.

[0082] Calculate ctDNA concentration (expressed as haploid genome equivalents / mL) and variant allele frequency (VAF).

[0083] The level of ctDNA after log-transformation (log hGE / mL) was used as the model input variable.

[0084] For the calculation of ctDNA concentration, please refer to the patent application number CN2024 / 11678397.X.

[0085] 2. Model Building

[0086] Normalize the ctDNA data;

[0087] The optimal threshold was determined using logistic regression and ROC curve analysis.

[0088] Baseline ctDNA concentration (hGE) C1≥ 3 log hGE / mL is considered a high-risk threshold;

[0089] The analysis showed that the baseline ctDNA concentration in ECF patients was significantly higher than that in the non-ECF group (P<0.001). When the baseline ctDNA concentration was ≥ 3 log hGE / mL, the sensitivity, specificity, and accuracy for predicting ECF were 81.3%, 79.4%, and 79.8%, respectively. The sensitivity, specificity, and accuracy for predicting ECF using the IPI score were 62.5%, 77.9%, and 75%, respectively. The analytical performance of baseline ctDNA concentration in predicting ECF was superior to that of the IPI score. (Appendix) Figure 2 A, 2B)

[0090] As attached Figure 2 As shown in Figure C, baseline ctDNA concentration exhibits superior analytical performance and stability in predicting ECF compared to IPI score. (The AUC for ctDNA concentration in predicting ECF was 0.815, with a sensitivity of 69.2%, a specificity of 93.8%, and an accuracy of 88.5%; the AUC for IPI score in predicting ECF was 0.679, with a sensitivity of 46.2%, a specificity of 89.6%, and an accuracy of 80.3%).

[0091] 3. Risk score calculation and clinical application

[0092] When the risk score is ≥ a set threshold (e.g., 3.0), the patient is identified as a high-risk ECF patient;

[0093] High-risk patients may be given priority for intensive treatment (such as CAR-T or dual anti-inflammatory therapy) and intensive monitoring.

[0094] Example 2: Validation of a risk prediction model based on ctDNA levels

[0095] Sixty-one newly diagnosed DLBCL patients were selected as the validation set (see attached). Figure 1 Peripheral blood samples were collected and ctDNA levels were measured. ECF was defined as relapse or no response within 12 months of chemotherapy.

[0096] The analysis showed that baseline ctDNA concentrations in ECF patients were significantly higher than those in the non-ECF group (P<0.001). When baseline ctDNA ≥ 3 log hGE / mL, the sensitivity and specificity for predicting ECF were 69.2% and 93.8%, respectively. The sensitivity and specificity for predicting ECF using the IPI score were 46.2% and 89.6%, respectively. The analytical performance of baseline ctDNA concentration in predicting ECF was superior to that of the IPI score. (Appendix) Figure 2 A, 2C)

[0097] The logistic regression model based on ctDNA levels achieved a prediction accuracy of 79.8%. In the independent validation cohort, the AUC was 0.815, validating the model's stability. (Appendix) Figure 2 B, 2C)

[0098] Risk stratification revealed that the 2-year progression-free survival rate was significantly lower in the high-risk group than in the low-risk group (P<0.01). (Appendix) Figure 3 )

[0099] Example 3: Construction and Validation of a Risk Prediction Model Based on ctDNA Levels and Clinical Data

[0100] ctDNA was detected using tumor tissue, oral mucosa, and blood samples collected before the start of the first treatment cycle (C1) and the start of the second treatment cycle (C2) from 115 DLBCL patients. Following the detection procedure described below, the ctDNA concentration in the blood samples of each patient before the start of the first treatment cycle (baseline) and the start of the second treatment cycle was obtained. The calculation of ctDNA concentration is detailed in patent application number CN2024 / 11678397.X.

[0101] The concentration values ​​of ctDNA (hGE) in two blood samples collected from each patient were used. C1 and hGE C2 Combining patient clinical information including IPI score, age, Ann Arbor stage, LDH, ECOG score, and extranodal involvement site, different regression models were constructed to predict the likelihood of patients developing ECF status after treatment. The models are as follows:

[0102] Model 1:

[0103] logit(P(ECF=1))=α+β1*hGE C1 +β2* IPI+ε

[0104] Model 2:

[0105] logit(P(ECF=1))=α+β1*hGE C1 +β2* Hans+β3* AAS +β4* LDH +β5* ECOG +β6*Extranodal +ε

[0106] Based on the ECF status of 115 samples, regression analysis was performed on Model 1 and Model 2 respectively. The logistic regression model was used for fitting, and the iterative weighted least squares algorithm was used to solve the model, resulting in the following parameter estimation results:

[0107] Model 1:

[0108] logit(P(ECF=1)) = 1.16 * hGE C1 +0.27 * IPI - 5.46

[0109] Model 2:

[0110] logit(P(ECF=1))=1.312*hGE C1 +0.282* Hans+0.451* AAS -0.501* LDH +1.19*ECOG +0.145* Extranodal -5.6

[0111] As attached Figure 4 As shown, the logistic regression model based on ctDNA levels and clinical parameters predicts the following results:

[0112] The model 1 predicted the risk of early chemotherapy failure with an AUC value of 0.837.

[0113] Model 2 predicted the risk of early chemotherapy failure with an AUC value of 0.843.

[0114] Under the same experimental conditions, the IPI score predicted the risk of early chemotherapy failure, with an AUC of 0.746.

[0115] The comparative results show that, compared with the traditional IPI score, the prediction model constructed in this invention, which integrates ctDNA concentration and clinical characteristic parameters, demonstrates superior overall performance in predicting early chemotherapy failure in patients with diffuse large B-cell lymphoma. Its prediction accuracy, sensitivity, and specificity are all significantly improved, and it has a better risk stratification discrimination ability.

[0116] As attached Figure 5 As shown, in the independent validation queue, Model 1 predicted an AUC of 0.856, and Model 2 predicted an AUC of 0.845. Both AUCs are not lower than 0.8, which verifies the stability of the models.

[0117] Compared with the prior art, the present invention has the following advantages:

[0118] 1. High sensitivity: The AUC of baseline ctDNA in predicting ECF reached 0.867, which is better than the IPI score (AUC 0.758); the sensitivity of predicting ECF based on ctDNA level and clinical parameters is significantly higher than that of predicting ECF based on IPI score, which improves the positive detection rate of patients who have failed early chemotherapy. The ECF prediction model based on ctDNA and clinical parameters is better than the IPI score.

[0119] 2. Early identification: High-risk groups can be identified before the first cycle of chemotherapy, enabling early intervention and providing auxiliary reference information for clinicians in addition to the IPI score;

[0120] 3. High clinical operability: Based on routine peripheral blood samples, the testing process is simple and suitable for widespread application;

[0121] 4. Multi-center validation: The model has been validated by an external validation cohort to demonstrate stable predictive performance;

[0122] 5. Guiding treatment: Providing risk stratification basis for CAR-T, dual anti-inflammatory and gene subtype-guided therapy.

[0123] This invention provides a model for predicting early chemotherapy failure in DLBCL patients and its application, involving tumor gene detection, biostatistical modeling, and clinical risk stratification. By detecting the patient's baseline and early treatment ctDNA concentration levels, and / or combining clinical characteristic parameters (such as IPI score, LDH, ECOG score, etc.), a multivariate risk assessment algorithm is established to achieve early identification and stratified management of ECF risk.

[0124] The above are merely preferred embodiments of the present invention. It should be noted that those skilled in the art can make various improvements and modifications without departing from the principle of the present invention, and these improvements and modifications should also be considered within the scope of protection of the present invention.

Claims

1. A method for constructing a model to predict early chemotherapy failure in patients with diffuse large B-cell lymphoma, characterized in that, include: We obtained patients' ctDNA levels and constructed a comprehensive assessment model to predict the risk of early chemotherapy failure through multivariate regression analysis.

2. The construction method according to claim 1, characterized in that, It also includes obtaining the patient's clinical characteristic parameters, which include at least one of the following: IPI score, age, Ann Arbor stage, LDH, ECOG score, or extranodal involvement site.

3. The construction method according to claim 1, characterized in that, The construction steps also include: The patient cohort with diffuse large B-cell lymphoma was divided into a training set and a validation set. A prediction model is constructed in the training set using the logistic regression algorithm, and the parameters of the prediction model are obtained by screening using multivariate Cox regression. The accuracy of the model was verified on the validation set.

4. A model for predicting early chemotherapy failure in patients with diffuse large B-cell lymphoma, characterized in that, It is: logit(P(ECF=1))=hGE C1 ; Alternatively, logit(P(ECF=1))=hGE C2 ; Alternatively, logit(P(ECF=1))=hGE C2 ÷hGE C1 ; hGE C1 The hGE concentration is the value of ctDNA in a blood sample collected before the start of the first course of treatment for patients with diffuse large B-cell lymphoma. C2 The value of ctDNA in a blood sample collected before the start of the second course of treatment for a patient with diffuse large B-cell lymphoma.

5. A model for predicting early chemotherapy failure in patients with diffuse large B-cell lymphoma, characterized in that, The model is as follows: logit(P(ECF=1))=α+β1* hGE C1 +β2* IPI+ε; Wherein, the hGE C1 The ctDNA concentration value is the blood sample collected before the start of the first course of treatment for patients with diffuse large B-cell lymphoma; the IPI is the IPI score of patients with diffuse large B-cell lymphoma; α, β1, and β2 are model coefficients, and ε is the random error, which is solved by logistic regression and iterative weighted least squares method; α is -5.0 to -6.0, β1 is 1 to 1.5, β2 is 0.1 to 0.5, and ε is 0 to 0.05; preferably, α is -5.46, β1 is 1.16, β2 is 0.27, and ε is 0.

6. A model for predicting early chemotherapy failure in patients with diffuse large B-cell lymphoma, characterized in that, The model is as follows: logit(P(ECF=1))=α+β1* hGE C1 +β2* Hans+β3* AAS +β4* LDH +β5* ECOG +β6*Extranodal +ε; Wherein, the hGE C1 The values ​​represent: ctDNA concentration in blood samples collected before the start of the first treatment course for patients with diffuse large B-cell lymphoma; Hans classification for diffuse large B-cell lymphoma, with GCB type represented by a value of 0 and non-GCB type by a value of 1; AAS classification for diffuse large B-cell lymphoma, with stages 1 and 2 represented by a value of 0 and stages 3 and 4 by a value of 1; LDH level in serum, with the normal range represented by a value of 0 and elevated LDH levels by a value of 1; ECOG score, with 0 and 1 points representing a value of 0 and ≥2 points representing a value of 1; Extranodal involvement in patients with diffuse large B-cell lymphoma, with 0 or 1 extranodal involvement representing a value of 0 and ≥2 involved sites representing a value of 1. α, β1, β2, β3, β4, β5 and β6 are model coefficients, and ε is the random error, which is solved by logistic regression and iterative weighted least squares method; The values ​​of α are -5.0 to -6.0, β1 is 1 to 1.5, β2 is 0.1 to 0.5, β3 is 0.2 to 0.7, β4 is -0.7 to -0.3, β5 is 1 to 1.5, β6 is 0.1 to 0.5, and ε is 0 to 0.05; preferably, the values ​​of α are -5.6, β1 is 1.312, β2 is 0.282, β3 is 0.451, β4 is -0.501, β5 is 1.19, β6 is 0.145, and ε is 0.

7. The model constructed by the method according to any one of claims 1 to 3, or the model according to any one of claims 4 to 6, is used in the preparation of a product for predicting the risk of early chemotherapy failure in patients with diffuse large B-cell lymphoma.

8. A computer program product, characterized in that, The device includes computer-readable instructions that, when executed on an electronic device, cause the electronic device to implement the model building method as described in any one of claims 1 to 3, or to implement the model prediction function as described in any one of claims 4 to 6, and output the early chemotherapy failure risk score and / or the early chemotherapy failure risk stratification result based on the risk score.

9. An electronic device, characterized in that, It includes at least one processor and a memory connected to the processor, wherein: The memory is used to store computer programs; The processor is used to execute the computer program so that the electronic device can implement the model building method as described in any one of claims 1 to 3, or implement the model prediction function as described in any one of claims 4 to 6, and output the early chemotherapy failure risk score and / or the early chemotherapy failure risk stratification result based on the risk score.

10. A computer storage medium, characterized in that, The storage medium carries one or more computer programs, which, when executed by an electronic device, enable the electronic device to implement the model building method as described in any one of claims 1 to 3, or the model prediction function as described in any one of claims 4 to 6, and output the early chemotherapy failure risk score and / or the early chemotherapy failure risk stratification result based on the risk score.

Images