Efficacy prediction biomarker screening method, efficacy prediction method and apparatus, storage medium, and program product

Abstract

An efficacy prediction biomarker screening method, an efficacy prediction method and apparatus, a storage medium, and a program product. The efficacy prediction biomarker screening method comprises: preprocessing sample data; performing screening from the preprocessed sample data to obtain differentially expressed genes related to a treatment response; constructing a classifier model on the basis of gene features of the differentially expressed genes; and evaluating the performance of the constructed classifier model, and selecting an efficacy prediction biomarker on the basis of an evaluation result.

Description

Methods for screening predictive biomarkers of therapeutic efficacy, methods and devices for predicting therapeutic efficacy, storage media and procedures. Technical Field

[0001] This disclosure relates to, but is not limited to, the field of biotechnology, and particularly to a method for screening biomarkers for predicting therapeutic efficacy, a method and apparatus for predicting therapeutic efficacy, a storage medium, and a program product. Background Technology

[0002] Neoadjuvant therapy plays a crucial role in breast cancer treatment. Its primary goal is to provide systemic therapy to breast cancer patients before surgery, reducing tumor size, improving disease stage, and thus increasing the chances of surgical resection. This treatment strategy can help some breast cancer patients who were previously inoperable to meet the surgical criteria, and provide the possibility of breast-conserving surgery for patients who were not initially suitable. Furthermore, neoadjuvant therapy can assess tumor sensitivity to drugs, providing guidance for subsequent treatment. Complete pathological response (pCR) has become an important research endpoint in neoadjuvant therapy clinical trials, and whether or not pCR is achieved influences the choice of subsequent treatment regimens. Therefore, pathological evaluation of breast specimens after neoadjuvant therapy is crucial for assessing treatment efficacy and predicting patient prognosis. However, the pathological characteristics of the tumor may change after neoadjuvant therapy, posing challenges to pathological evaluation. Therefore, accurate assessment of the treatment efficacy of neoadjuvant therapy is essential.

[0003] In China, the Miller & Payne (MP) system is commonly used in pathology to assess treatment efficacy. This system compares core needle biopsy specimens before treatment with surgical specimens after treatment, primarily assessing the cellular abundance of residual tumor cells in the primary breast lesion after neoadjuvant therapy. It is classified into five grades: Grade 1 (G1): No change in invasive cancer cells or only a few cancer cells showing change; the overall number of cancer cells has not decreased. Grade 2 (G2): Slight reduction in invasive cancer cells, but the total number remains high; the reduction in cancer cells does not exceed 30%. Grade 3 (G3): Reduction in invasive cancer cells between 30% and 90%. Grade 4 (G4): Significant reduction in invasive cancer cells exceeding 90%, with only scattered small clusters of cancer cells or single cancer cells remaining. Grade 5 (G5): No invasive cancer cells remain in the original tumor bed, but ductal carcinoma in situ may be present. Although the MP system is commonly used, its application in international clinical trials is relatively limited due to its limitations. For example, this system only assesses the primary breast lesion and not the axillary lymph nodes; its application is also challenging when tumor cell density is heterogeneous after chemotherapy. Furthermore, due to the limited availability of core needle biopsies, the abundance of cells within the specimen sometimes does not represent the overall cell density of the tumor.

[0004] The Residual Cancer Burden (RCB) system recommended by the International Breast Cancer Consortium assesses the extent of residual tumor in the primary breast lesion (mm × mm), residual tumor cell density (%), percentage of carcinoma in situ (%), number of positive lymph nodes, and maximum diameter of residual metastatic lymph nodes (mm). Inputting these five pathological parameters into an online calculator (www.mdanderson.org / breastcancer_RCB) yields the RCB index and corresponding RCB grades (RCB-0: pCR; RCB-I: minimal residual tumor; RCB-II: moderate residual tumor; RCB-III: extensive residual tumor). Patients with RCB-0 and RCB-I after neoadjuvant therapy have a better prognosis and a low risk of distant recurrence. The RCB system provides reproducible assessments and is suitable for pathological evaluation after treatment of different subtypes of breast cancer. However, the RCB system requires meticulous specimen collection and microscopic evaluation after neoadjuvant therapy, and manual interpretation can be labor-intensive.

[0005] Summary of the Invention

[0006] The following is an overview of the subject matter described in detail herein. This overview is not intended to limit the scope of the claims.

[0007] This disclosure provides a method for screening predictive biomarkers for therapeutic efficacy, including:

[0008] Preprocess the sample data;

[0009] Differentially expressed genes associated with treatment response were screened from the preprocessed sample data;

[0010] A classifier model is constructed based on the gene characteristics of the differentially expressed genes;

[0011] The performance of the constructed classifier model was evaluated, and efficacy prediction biomarkers were selected based on the evaluation results.

[0012] This disclosure also provides an apparatus for screening predictive biomarkers of therapeutic efficacy, including a memory and a processor connected to the memory, the memory being used to store instructions, and the processor being configured to execute the steps of the predictive biomarker screening method of any embodiment of this disclosure based on the instructions stored in the memory.

[0013] This disclosure also provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the efficacy prediction biomarker screening method described in any embodiment of this disclosure.

[0014] This disclosure also provides a program product including instructions that, when executed by a computer, perform the efficacy predictive biomarker screening method as described in any embodiment of this disclosure.

[0015] This disclosure also provides a device for screening biomarkers for predicting therapeutic efficacy, including a preprocessing module, a screening module, a model building module, and an evaluation and selection module, wherein:

[0016] The preprocessing module is configured to preprocess transcriptome expression data;

[0017] The screening module is configured to screen differentially expressed genes related to treatment response from preprocessed transcriptome expression data;

[0018] The model building module is configured to build a classifier model based on the gene characteristics of the differentially expressed genes;

[0019] The evaluation selection module is configured to evaluate the performance of the constructed classifier model and select efficacy prediction biomarkers based on the evaluation results.

[0020] This disclosure also provides a method for predicting therapeutic efficacy, including:

[0021] Determine efficacy predictive biomarkers, which are obtained by the method as described in any embodiment of this disclosure;

[0022] Determine the size of efficacy predictive biomarkers in the test samples, and predict efficacy based on the determined efficacy predictive biomarker sizes.

[0023] This disclosure also provides an efficacy prediction device, including a memory; and a processor connected to the memory, the memory being used to store instructions, the processor being configured to perform the steps of the efficacy prediction method as described in any embodiment of this disclosure based on the instructions stored in the memory.

[0024] This disclosure also provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the efficacy prediction method as described in any embodiment of this disclosure.

[0025] This disclosure also provides a computer program product including instructions that, when executed by a computer, perform the efficacy prediction method as described in any embodiment of this disclosure.

[0026] The efficacy prediction biomarker screening method, efficacy prediction method and device, storage medium and program product of this disclosure screen differentially expressed genes related to treatment response from preprocessed sample data, construct a classifier model based on the gene characteristics of the differentially expressed genes, evaluate the performance of the constructed classifier model, and select efficacy prediction biomarkers based on the evaluation results. This can accurately predict and evaluate the efficacy of disease treatment, and the method is simple to use, reducing the workload of manual judgment.

[0027] Other features and advantages of this disclosure will be set forth in the following description, and will be apparent in part from the description, or may be learned by practicing the disclosure. Other advantages of this disclosure may be realized and obtained by means of the methods described in the description and the accompanying drawings. Attached Figure Description

[0028] The accompanying drawings are used to provide an understanding of the technical solutions of this disclosure and form part of the specification. They are used together with the embodiments of this disclosure to explain the technical solutions of this disclosure and do not constitute a limitation on the technical solutions of this disclosure.

[0029] Figure 1 is a flowchart illustrating an exemplary embodiment of the present disclosure of a method for screening biomarkers for predicting therapeutic efficacy.

[0030] Figure 2A is a schematic diagram of differentially expressed genes related to treatment response in all samples of an exemplary embodiment of this disclosure;

[0031] Figure 2B is a schematic diagram of the gene expression levels of differentially expressed genes for all samples corresponding to exemplary embodiments of this disclosure;

[0032] Figure 3A is a schematic diagram of differentially expressed genes related to treatment response in a pre-treatment sample of an exemplary embodiment of the present disclosure;

[0033] Figure 3B is a schematic diagram of the gene expression levels of differentially expressed genes corresponding to pre-treatment samples in an exemplary embodiment of this disclosure;

[0034] Figure 4A is a schematic diagram of differentially expressed genes related to treatment response in post-treatment samples of an exemplary embodiment of this disclosure;

[0035] Figure 4B is a schematic diagram of the gene expression levels of differentially expressed genes corresponding to post-treatment samples in an exemplary embodiment of this disclosure.

[0036] Figures 5A to 5D are schematic diagrams illustrating the model performance of classifier models constructed using the four gene features of all samples according to an exemplary embodiment of this disclosure.

[0037] Figures 6A to 6D are schematic diagrams illustrating the model performance of classifier models constructed using four gene features of pre-treatment samples according to exemplary embodiments of the present disclosure.

[0038] Figures 7A to 7D are schematic diagrams illustrating the model performance of classifier models constructed using four gene features of post-treatment samples according to exemplary embodiments of the present disclosure.

[0039] Figure 8A is a schematic diagram of the classification threshold results of efficacy prediction biomarkers for all samples in an exemplary embodiment of this disclosure;

[0040] Figure 8B is a schematic diagram of the specificity and sensitivity of the efficacy prediction biomarker classification thresholds for all samples in an exemplary embodiment of this disclosure;

[0041] Figure 9A is a schematic diagram of the classification threshold division results of efficacy prediction biomarkers in pre-treatment samples according to an exemplary embodiment of the present disclosure;

[0042] Figure 9B is a schematic diagram of the specificity and sensitivity of the classification threshold of efficacy prediction biomarkers for pre-treatment samples in an exemplary embodiment of the present disclosure.

[0043] Figure 10A is a schematic diagram of the classification threshold division results of efficacy prediction biomarkers for post-treatment samples in an exemplary embodiment of this disclosure;

[0044] Figure 10B is a schematic diagram of the specificity and sensitivity of the classification threshold of efficacy prediction biomarkers for post-treatment samples in an exemplary embodiment of this disclosure.

[0045] Figure 11 is a schematic diagram of a therapeutic efficacy prediction biomarker screening device according to an exemplary embodiment of the present disclosure;

[0046] Figure 12 is a schematic diagram of another therapeutic efficacy prediction biomarker screening device according to an exemplary embodiment of the present disclosure;

[0047] Figure 13 is a flowchart illustrating an exemplary embodiment of the present disclosure of a method for predicting therapeutic efficacy;

[0048] Figure 14 is a schematic diagram of another therapeutic effect prediction device according to an exemplary embodiment of the present disclosure. Detailed Implementation

[0049] This disclosure describes several embodiments, but these descriptions are exemplary and not limiting, and it will be apparent to those skilled in the art that many more embodiments and implementations are possible within the scope of the embodiments described herein. Although many possible combinations of features are shown in the drawings and discussed in the detailed description, many other combinations of the disclosed features are also possible. Unless specifically limited, any feature or element of any embodiment may be used in combination with, or may replace, any feature or element of any other embodiment.

[0050] This disclosure includes and contemplates combinations of features and elements known to those skilled in the art. The embodiments, features, and elements disclosed in this disclosure may also be combined with any conventional features or elements to form a unique inventive scheme as defined by the claims. Any feature or element of any embodiment may also be combined with features or elements from other inventive schemes to form another unique inventive scheme as defined by the claims. Therefore, it should be understood that any feature shown and / or discussed in this disclosure may be implemented individually or in any suitable combination. Therefore, the embodiments are not limited except by the limitations imposed by the appended claims and their equivalents. Furthermore, various modifications and changes may be made within the scope of the appended claims.

[0051] Furthermore, in describing representative embodiments, the specification may have presented methods and / or processes as a specific sequence of steps. However, the method or process should not be limited to the specific order of steps described herein, to the extent that the method or process does not depend on the specific order of steps described herein. As will be understood by those skilled in the art, other sequences of steps are also possible. Therefore, the specific order of steps set forth in the specification should not be construed as a limitation of the claims. Moreover, the claims relating to the method and / or process should not be limited to the steps performed in the order written, and those skilled in the art will readily understand that these orders can be varied and still remain within the spirit and scope of the embodiments disclosed herein.

[0052] As shown in Figure 1, this disclosure provides a method for screening predictive biomarkers for therapeutic efficacy, including:

[0053] Step 101: Preprocess the sample data;

[0054] Step 102: From the preprocessed sample data, screen out differentially expressed genes related to treatment response;

[0055] Step 103: Construct a classifier model based on the gene characteristics of differentially expressed genes;

[0056] Step 104: Evaluate the performance of the constructed classifier model and select efficacy prediction biomarkers based on the evaluation results.

[0057] The efficacy prediction biomarker screening method of this disclosure screens differentially expressed genes related to treatment response from preprocessed sample data, constructs a classifier model based on the gene characteristics of the differentially expressed genes, evaluates the performance of the constructed classifier model, and selects efficacy prediction biomarkers based on the evaluation results. This method can accurately predict and evaluate the efficacy of disease treatment, and is simple to use, reducing the workload of manual judgment.

[0058] In some exemplary embodiments, in step 101, the sample data can be a breast cancer sample dataset. However, this disclosure is not limiting in this regard, and the sample data can also be any other type of disease dataset.

[0059] In some exemplary embodiments, the sample data includes gene expression data and corresponding clinical information.

[0060] In this embodiment of the disclosure, the sample data can be obtained by downloading from a public database. For example, gene expression data and corresponding clinical information, including survival time and survival status, can be obtained from the Gene Expression Omnibus (GEO) database or the TCGA database.

[0061] For example, the sample information is shown in Table 1. This embodiment of the disclosure selected a total of 87 breast cancer samples, including 47 pre-treatment (baseline) samples and 40 post-treatment (postoperative) samples. In Table 1, Ki-67, also known as the proliferation index, represents the proliferation index of tumor cells. ER is the estrogen receptor, PR is the progesterone receptor, and the human epidermal growth factor receptor family (HER2) index is divided into four levels: -, 1+, 2+, and 3+. P63 is a marker for myoepithelial cells. Pathological classification indicates pathological classification. BRCA represents breast invasive carcinoma, DCIS represents ductal carcinoma in situ, MP (Miller-Payne) grading includes five levels from G1 to G5, and residual tumor burden (RCB) grading includes four levels: 0, I, II, and III. RCB=0 indicates complete remission (pCR). ALN metastasis indicates axillary lymph node metastasis, and bilateral lesion indicates bilateral lesions.

[0062] Table 1

[0063] As shown in Table 1, in all, baseline, and postoperative samples, clinical characteristics such as age, P63 protein expression, and axillary lymph node metastasis were significantly associated with treatment response (p<0.05).

[0064] In some exemplary embodiments, step 101 involves preprocessing the sample data, including:

[0065] Perform quality control filtering on the sample data.

[0066] For example, quality control filtering includes removing low-expression genes and retaining high-expression genes. However, this disclosure is not limited thereto, and preprocessing may also include other processing operations.

[0067] In this disclosure, the screening criteria for low-expression and high-expression genes are not unique and depend on the data. For example, a high-expression gene can be a gene whose expression level (counts) is greater than X in at least N% of the sample size. In some exemplary embodiments, N can be between 10 and 30, and X can be between 1 and 2; however, this disclosure does not limit this. For example, N can be 20, and X can be 1.

[0068] In some exemplary embodiments, in step 102, the screening criteria for differentially expressed genes are: the statistical significance p-value of the treatment response is less than or equal to a preset p-value threshold and |log2(FoldChange)| is greater than or equal to a preset second numerical threshold, where FoldChange is the fold change.

[0069] In this embodiment of the disclosure, FoldChange can be calculated according to the following formula:

[0070] In this embodiment of the disclosure, the T-test or F-test method can be used to calculate the statistical significance p-value of each gene in the treatment response, and m genes with p-value ≤ a preset p-value threshold and |log2(FoldChange)| ≥ a preset second numerical threshold are selected as differentially expressed genes, where m is the number of genes after selection.

[0071] In some exemplary implementations, the preset p-value threshold can be between 0.01 and 0.05, for example, the preset p-value threshold can be set to 0.05.

[0072] In some exemplary embodiments, the preset second numerical threshold is greater than or equal to 0.3. For example, the preset second numerical threshold may be set to 0.3.

[0073] In some exemplary embodiments, in step 102, differentially expressed genes related to treatment response are screened from the preprocessed sample data, including:

[0074] Select m1 upregulated genes and m2 downregulated genes that meet the criteria of statistical significance regarding treatment response (p-value less than or equal to a preset p-value threshold) and |log2(FoldChange)| greater than or equal to a preset second numerical threshold. m1 and m2 are natural numbers, with m1 greater than or equal to 1 and m2 greater than or equal to 1.

[0075] The upregulated gene with the smallest p-value was selected as the most upregulated gene, and the downregulated gene with the smallest p-value was selected as the most downregulated gene.

[0076] The most upregulated and the most downregulated genes were selected as the differentially expressed genes in the final screening.

[0077] In this embodiment of the disclosure, the m genes include m1 upregulated genes and m2 downregulated genes. Among the m1 upregulated genes, the upregulated gene with the smallest p-value is selected as the most upregulated gene; among the m2 downregulated genes, the downregulated gene with the smallest p-value is selected as the most downregulated gene. The most upregulated gene and the most downregulated gene are selected as the differentially expressed genes in the final screening.

[0078] For example, for all breast cancer samples, as shown in Figures 2A and 2B, differentially expressed genes associated with treatment response (pCR) were screened based on the significance p-value and FoldChange, namely OR4Q2 and GCNT1 genes, which were significantly downregulated and significantly upregulated, respectively, in all pCR samples.

[0079] For example, for pre-treatment samples, as shown in Figures 3A and 3B, differentially expressed genes associated with treatment response (pCR) were identified as ITIH1 and MMP11 genes based on the significance p-value and FoldChange, which were significantly downregulated and significantly upregulated, respectively, in all pCR samples.

[0080] For example, for post-treatment samples, as shown in Figures 4A and 4B, differentially expressed genes associated with treatment response (pCR) were identified as HMGCS2 and RGS1 genes based on the significance p-value and FoldChange, which were significantly downregulated and significantly upregulated, respectively, in all pCR samples.

[0081] In some exemplary embodiments, in step 103, the genetic characteristics of the differentially expressed gene include any one of the following:

[0082] Gene expression level of a single differentially expressed gene, the difference in gene expression levels of multiple differentially expressed genes, and the ratio of gene expression levels of multiple differentially expressed genes.

[0083] For example, for all breast cancer samples, the gene characteristics of differentially expressed genes can be any one of the following: OR4Q2 gene expression level, GCNT1 gene expression level, the difference between OR4Q2 gene expression level and GCNT1 gene expression level, or the ratio of OR4Q2 gene expression level to GCNT1 gene expression level.

[0084] For example, for pre-treatment samples, the gene characteristics of differentially expressed genes can be any one of the following: ITIH1 gene expression level, MMP11 gene expression level, the difference between ITIH1 gene expression level and MMP11 gene expression level, or the ratio of ITIH1 gene expression level to MMP11 gene expression level.

[0085] For example, for post-treatment samples, the gene characteristics of differentially expressed genes can be any one of the following: HMGCS2 gene expression level, RGS1 gene expression level, the difference between HMGCS2 gene expression level and RGS1 gene expression level, or the ratio of HMGCS2 gene expression level to RGS1 gene expression level.

[0086] In some exemplary embodiments, in step 103, the classifier model is any one of the following: Logistic Regression (LR) model, Support Vector Machine (SVM) model, and Naive Bayes model.

[0087] Logistic regression, also known as the log-odds model, is widely used in machine learning, deep learning, recommender systems, advertising prediction, intelligent marketing, biology, sociology, and other fields due to its simplicity, strong interpretability, and ease of implementation. Support vector machines (SVMs) are linear classifier models, essentially defined as linear classifiers that maximize the margin in the feature space; fundamentally, they are a problem of solving a convex quadratic programming problem. Naive Bayes models are based on Bayesian quantification and include the condition (independence between features). Naive Bayes models are relatively simple in principle and widely used.

[0088] In some exemplary embodiments, step 104, evaluating the performance of the constructed classifier model, includes:

[0089] The classifier model is trained using K-fold cross-validation.

[0090] Calculate the Area Under the Receiver Operating Characteristic (AUC) value of the trained classifier model, and evaluate the performance of the classifier model based on the AUC value.

[0091] K-fold cross-validation is a commonly used model evaluation technique for assessing the performance of machine learning models on training data. It divides the original dataset into K subsets, called folds. The model is then trained and validated K times on K distinct training sets. In each training iteration, one fold is used as the validation set, while the remaining K-1 folds are used as the training set.

[0092] In some exemplary embodiments, the number of folds K in the K-fold cross-validation method is greater than or equal to 3, for example, K = 5 or 10; however, this disclosure does not limit it.

[0093] The Area Under the Curve (AUC), also known as the area under the ROC (Receiver Operating Characteristic) curve, is used to evaluate the performance of a binary classifier. A UC value closer to 1 indicates better classifier performance. During K-fold cross-validation, the AUC value is calculated at each iteration. Finally, by averaging or otherwise combining the AUC values ​​from all iterations, a comprehensive AUC value is obtained, which serves as an important metric for evaluating model performance.

[0094] In some exemplary embodiments, in step 104, selecting a predictive biomarker for efficacy based on the evaluation results includes:

[0095] Choose the classifier model with the highest AUC value as the optimal classifier model;

[0096] The classification threshold of the optimal classifier model is determined based on specificity and / or sensitivity.

[0097] For example, still using the aforementioned all, baseline, and postoperative samples, classifier models are constructed using the gene features of the selected differentially expressed genes. Assuming a logistic regression model is chosen as the classifier model and five-fold cross-validation is used to train it, the corresponding model performance is shown in Figures 5A to 5D, 6A to 6D, and 7A to 7D. As can be seen from Figures 5A to 5D, 6A to 6D, and 7A to 7D, in this embodiment of the disclosure, the AUC values ​​of models using gene expression ratios as features are all higher than the AUC values ​​of models using the gene expression levels of individual genes or the difference in gene expression levels as features.

[0098] Specifically, in the efficacy prediction of all samples, the ratio of OR4Q2 gene expression to GCNT1 gene expression can be used as a efficacy prediction biomarker (AUC = 0.85); in the efficacy prediction of pre-treatment (baseline) samples, the ratio of ITIH1 gene expression to MMP11 gene expression can be used as a efficacy prediction biomarker (AUC = 0.98); and in the efficacy prediction of post-treatment (postoperative) samples, the ratio of HMGCS2 gene expression to RGS1 gene expression can be used as a efficacy prediction biomarker (AUC = 0.98).

[0099] In this embodiment of the disclosure, when the sample data is other types of disease datasets, the corresponding efficacy prediction biomarkers can also be screened out according to the above-described efficacy prediction biomarker screening method.

[0100] In some exemplary embodiments, determining the classification threshold of the optimal classifier model based on specificity and / or sensitivity includes:

[0101] Obtain the preset initial classification threshold, step size, and maximum classification threshold;

[0102] Set the classification threshold starting from the initial classification threshold and gradually increase it to the maximum classification threshold by step size. Calculate the specificity and / or sensitivity of the optimal classifier model at each classification threshold.

[0103] The classification threshold that satisfies the preset specificity threshold and / or sensitivity threshold is used as the classification threshold of the final determined optimal classifier model.

[0104] This disclosure determines the classification threshold of the optimal classifier model by setting a specificity threshold and / or a sensitivity threshold. In different business scenarios, if it is necessary to ensure that specificity or sensitivity is as high as possible, an initial classification threshold of R can be set for the optimal classifier model on the training set, and then increased by a step size S until the specificity Spe / sensitivity Sen is greater than or equal to the specificity threshold / sensitivity threshold, at which point the corresponding classification threshold R+S is retained. n and sensitivity Sen / specificity Spe, S n This is the step size added at the end of the increment process; when the classification threshold is incremented to the preset maximum classification threshold, for example, the preset maximum classification threshold is 0.99, assuming that when the classification threshold is incremented to 0.99, the above condition of until the specificity Spe / sensitivity Sen is greater than or equal to the specificity threshold / sensitivity threshold is still not met, then the highest specificity Spe and the corresponding sensitivity Sen are output.

[0105] In this embodiment, the model's specificity or sensitivity can be adjusted according to the specific scenario to meet the performance requirements of different scenarios. In scenarios emphasizing specificity, the specificity threshold can be set to 99%; in scenarios emphasizing sensitivity, the sensitivity threshold can be set to 99%.

[0106] In an exemplary embodiment, the specificity threshold is set to 99%, the initial classification threshold is set to 0.5, the step size is set to 0.001, and the maximum classification threshold is set to 0.99. Assuming a set of training samples has already generated a set of probability values ​​under the model's prediction, at the current classification threshold, by comparing the probability value of each sample with the classification threshold, it is determined whether each sample is classified as positive or negative, and the specificity corresponding to this classification threshold is then calculated. If the specificity does not reach the specificity threshold, the step size is incremented by 0.001, and each sample is reclassified as positive or negative, and the specificity corresponding to this classification threshold is then calculated. The larger the classification threshold, the stricter the condition for a sample to be classified as positive, until the specificity is greater than or equal to 99%, at which point the corresponding classification threshold and sensitivity are retained.

[0107] Taking the sample data in Table 1 as an example, as shown in Figures 8A and 8B, in the screening of efficacy prediction biomarkers for all samples, the efficacy prediction biomarker was the ratio of OR4Q2 gene expression level to GCNT1 gene expression level. The final classification threshold was 0.786. That is, when the OR4Q2 gene expression level / GCNT1 gene expression level in the sample to be tested is <0.786, the sample to be tested is judged as a pCR sample; when the OR4Q2 gene expression level / GCNT1 gene expression level in the sample to be tested is ≥0.786, the sample to be tested is judged as a nonPCR sample. Under this classification threshold, the model's sensitivity is 0.848 and the model's specificity is 0.741.

[0108] As shown in Figures 9A and 9B, during the screening of efficacy prediction biomarkers for pre-treatment (baseline) samples, the efficacy prediction biomarker was the ratio of ITIH1 gene expression level to MMP11 gene expression level. The final classification threshold was determined to be 0.592. That is, when the ITIH1 gene expression level / MMP11 gene expression level in the sample to be tested is <0.592, the sample to be tested is judged as a pCR sample; when the ITIH1 gene expression level / MMP11 gene expression level in the sample to be tested is ≥0.592, the sample to be tested is judged as a nonPCR sample. Under this classification threshold, the model sensitivity = 1 and the model specificity = 0.786.

[0109] As shown in Figures 10A and 10B, in the screening of efficacy prediction biomarkers for postoperative samples, the efficacy prediction biomarker was the ratio of HMGCS2 gene expression to RGS1 gene expression. The final classification threshold was determined to be 0.583. That is, when the HMGCS2 gene expression ratio / RGS1 gene expression in the sample to be tested is <0.583, the sample to be tested is judged as a pCR sample; when the HMGCS2 gene expression ratio / RGS1 gene expression in the sample to be tested is ≥0.583, the sample to be tested is judged as a nonPCR sample. Under this classification threshold, the model sensitivity = 1 and the model specificity = 0.923.

[0110] In summary, the efficacy prediction biomarker screening method of this disclosure identifies up- and down-regulated genes related to treatment response by analyzing differential gene expression data of transcriptomes before and after treatment under treatment response conditions. Ultimately, it provides the expression ratio of ITIH1 gene to MMP11 gene before treatment and the expression ratio of HMGCS2 gene to RGS1 gene after treatment as biomarkers for predicting the efficacy of breast cancer treatment. It can predict and evaluate the efficacy of neoadjuvant therapy for breast cancer with high sensitivity and high specificity, and the method is simple to use, reducing the workload of manual judgment.

[0111] As shown in Figure 11, this embodiment of the present disclosure also provides a device for screening predictive biomarkers for therapeutic efficacy, including a preprocessing module 1110, a screening module 1120, a model building module 1130, and an evaluation and selection module 1140, wherein:

[0112] Preprocessing module 1110 is configured to preprocess sample data;

[0113] The screening module 1120 is configured to screen differentially expressed genes related to treatment response from preprocessed sample data;

[0114] The model building module 1130 is configured to build a classifier model based on the gene characteristics of the differentially expressed genes;

[0115] The evaluation selection module 1140 is configured to evaluate the performance of the constructed classifier model and select efficacy prediction biomarkers based on the evaluation results.

[0116] In some exemplary embodiments, the screening criteria for the differentially expressed genes are: the statistical significance p-value of the treatment response is less than or equal to a preset p-value threshold and |log2(FoldChange)| is greater than or equal to a preset second numerical threshold, where FoldChange is the fold change.

[0117] In some exemplary embodiments, the screening module 1120 filters differentially expressed genes related to treatment response from preprocessed sample data, including:

[0118] Select m1 upregulated genes and m2 downregulated genes that meet the statistical significance of the treatment response p-value being less than or equal to a preset p-value threshold and |log2(FoldChange)| being greater than or equal to a preset second numerical threshold. FoldChange is the fold change, and m1 and m2 are both natural numbers, with m1 being greater than or equal to 1 and m2 being greater than or equal to 1.

[0119] The upregulated gene with the smallest p-value was selected as the most upregulated gene, and the downregulated gene with the smallest p-value was selected as the most downregulated gene.

[0120] The most upregulated and least upregulated genes were selected as the differentially expressed genes in the final screening.

[0121] In some exemplary embodiments, the gene characteristics of differentially expressed genes include any one of the following: the gene expression level of a single differentially expressed gene, the difference in gene expression levels of multiple differentially expressed genes, and the ratio of gene expression levels of multiple differentially expressed genes.

[0122] In some exemplary implementations, the classifier model is any one of the following: a logistic regression model, a support vector machine model, and a Naive Bayes model.

[0123] In some exemplary embodiments, the evaluation selection module 1140 evaluates the performance of the constructed classifier model, including:

[0124] The classifier model is trained using K-fold cross-validation, where K is greater than or equal to 3.

[0125] Calculate the Area Under the Receiver Operating Characteristic (AUC) value of the trained classifier model, and evaluate the performance of the classifier model based on the AUC value.

[0126] In some exemplary embodiments, the evaluation selection module 1140 selects efficacy predictive biomarkers based on the evaluation results, including:

[0127] Choose the classifier model with the highest AUC value as the optimal classifier model;

[0128] The classification threshold of the optimal classifier model is determined based on specificity and / or sensitivity.

[0129] In some exemplary embodiments, the evaluation selection module 1140 determines the classification threshold of the optimal classifier model based on specificity and / or sensitivity, including:

[0130] Obtain the preset initial classification threshold, step size, and maximum classification threshold;

[0131] The classification threshold is set starting from the initial classification threshold and gradually increased to the maximum classification threshold according to the step size. The specificity and / or sensitivity of the optimal classifier model at each classification threshold are calculated.

[0132] The classification threshold that satisfies the preset specificity threshold and / or sensitivity threshold is used as the classification threshold of the final determined optimal classifier model.

[0133] This disclosure also provides an apparatus for screening predictive biomarkers of therapeutic efficacy, including a memory; and a processor connected to the memory, the memory being used to store instructions, the processor being configured to perform the steps of the predictive biomarker screening method as described in any embodiment of this disclosure based on the instructions stored in the memory.

[0134] As shown in Figure 12, in one example, the efficacy prediction biomarker screening device may include: a first processor 1210, a first memory 1220, a first bus system 1230, and a first transceiver 1240. The first processor 1210, the first memory 1220, and the first transceiver 1240 are connected via the first bus system 1230. The first memory 1220 stores instructions, and the first processor 1210 executes the instructions stored in the first memory 1220 to control the first transceiver 1240 to transmit and receive signals. Specifically, the first transceiver 1240 can acquire sample data under the control of the first processor 1210. The first processor 1210 preprocesses the sample data; from the preprocessed sample data, differentially expressed genes related to treatment response are screened out; a classifier model is constructed based on the gene characteristics of the differentially expressed genes; the performance of the constructed classifier model is evaluated, and efficacy prediction biomarkers are selected based on the evaluation results.

[0135] It should be understood that the first processor 1210 can be a central processing unit (CPU), or it can be other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), off-the-shelf programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor can be a microprocessor or any conventional processor, etc.

[0136] The first memory 1220 may include read-only memory and random access memory, and provides instructions and data to the first processor 1210. A portion of the first memory 1220 may also include non-volatile random access memory. For example, the first memory 1220 may also store device type information.

[0137] In addition to the data bus, the first bus system 1230 may also include a power bus, a control bus, and a status signal bus. However, for clarity, all buses are labeled as the first bus system 1230 in Figure 12.

[0138] In implementation, the processing performed by the processing device can be accomplished through integrated logic circuits in the hardware of the first processor 1210 or through software instructions. That is, the method steps of this embodiment can be executed by the hardware processor, or by a combination of hardware and software modules within the processor. The software modules can reside in random access memory, flash memory, read-only memory, programmable read-only memory, electrically erasable programmable memory, registers, or other storage media. This storage medium is located in the first memory 1220. The first processor 1210 reads information from the first memory 1220 and, in conjunction with its hardware, completes the steps of the above method. To avoid repetition, further details are omitted here.

[0139] This disclosure also provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the efficacy predictive biomarker screening method as described in any embodiment of this disclosure. The efficacy predictive biomarker screening method driven by executing executable instructions is essentially the same as the efficacy predictive biomarker screening method provided in the above embodiments of this disclosure, and will not be described in detail here.

[0140] In some possible implementations, various aspects of the efficacy predictive biomarker screening method provided in this disclosure may also be implemented in the form of a program product, which includes program code that, when the program product is run on a computer device, causes the computer device to perform the steps in the efficacy predictive biomarker screening method according to various exemplary embodiments of this disclosure as described above. For example, the computer device may perform the efficacy predictive biomarker screening method described in the embodiments of this disclosure.

[0141] As shown in Figure 13, this embodiment of the present disclosure also provides a method for predicting therapeutic efficacy, including:

[0142] Step 1301: Determine efficacy prediction biomarkers, which are obtained by the efficacy prediction biomarker screening method as described in any embodiment of this disclosure;

[0143] Step 1302: Determine the size of the efficacy prediction biomarker in the sample to be tested, and predict the efficacy based on the determined efficacy prediction biomarker size.

[0144] In some exemplary embodiments, when the sample to be tested is a pre-treatment breast cancer sample, the ratio of the gene expression levels of the ITIH1 gene to the MMP11 gene is used as a predictive biomarker for efficacy; when the sample to be tested is a post-treatment breast cancer sample, the ratio of the gene expression levels of the HMGCS2 gene to the RGS1 gene is used as a predictive biomarker for efficacy.

[0145] This disclosure also provides an efficacy prediction device, including a memory; and a processor connected to the memory, the memory being used to store instructions, the processor being configured to perform the steps of the efficacy prediction method as described in any embodiment of this disclosure based on the instructions stored in the memory.

[0146] As shown in Figure 14, in one example, the efficacy prediction device may include: a second processor 1410, a second memory 1420, a second bus system 1430, and a second transceiver 1440. The second processor 1410, the second memory 1420, and the second transceiver 1440 are connected via the second bus system 1430. The second memory 1420 stores instructions, and the second processor 1410 executes the instructions stored in the second memory 1420 to control the second transceiver 1440 to transmit and receive signals. Specifically, the second transceiver 1440 can acquire a sample to be tested under the control of the second processor 1410. The second processor 1410 determines efficacy prediction biomarkers, which are obtained through the efficacy prediction biomarker screening method as described in any embodiment of this disclosure. It also determines the size of the efficacy prediction biomarkers in the sample to be tested and predicts the efficacy based on the determined size of the efficacy prediction biomarkers.

[0147] It should be understood that the second processor 1410 can be a central processing unit (CPU), or it can be other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), off-the-shelf programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor can be a microprocessor or any conventional processor.

[0148] The second memory 1420 may include read-only memory and random access memory, and provides instructions and data to the second processor 1410. A portion of the second memory 1420 may also include non-volatile random access memory. For example, the second memory 1420 may also store device type information.

[0149] In addition to the data bus, the second bus system 1430 may also include a power bus, a control bus, and a status signal bus. However, for clarity, all buses are labeled as the second bus system 1430 in Figure 12.

[0150] In implementation, the processing performed by the processing device can be accomplished through integrated logic circuits in the hardware of the second processor 1410 or through software instructions. That is, the method steps of this embodiment can be executed by a hardware processor, or by a combination of hardware and software modules within the processor. The software modules can reside in random access memory, flash memory, read-only memory, programmable read-only memory, electrically erasable programmable memory, registers, or other storage media. This storage medium is located in the second memory 1420. The second processor 1410 reads information from the second memory 1420 and, in conjunction with its hardware, completes the steps of the aforementioned method. To avoid repetition, further details are omitted here.

[0151] This disclosure also provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the efficacy prediction method as described in any embodiment of this disclosure. The efficacy prediction method driven by executing executable instructions is essentially the same as the efficacy prediction method provided in the above embodiments of this disclosure, and will not be described in detail here.

[0152] In some possible implementations, various aspects of the efficacy prediction method provided in this disclosure may also be implemented as a program product comprising program code that, when run on a computer device, causes the computer device to perform the steps in the efficacy prediction method according to various exemplary embodiments of this disclosure as described above. For example, the computer device may perform the efficacy prediction method described in the embodiments of this disclosure.

[0153] The program product may employ any combination of one or more readable media. A readable medium may be a readable signal medium or a readable storage medium. A readable storage medium may be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination thereof. More specific examples of readable storage media (a non-exhaustive list) include: an electrical connection having one or more wires, a portable disk, a hard disk, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination thereof.

[0154] It will be understood by those skilled in the art that all or some of the steps, systems, or apparatuses disclosed above, and their functional modules / units, can be implemented as software, firmware, hardware, or suitable combinations thereof. In hardware implementations, the division between functional modules / units mentioned above does not necessarily correspond to the division of physical components; for example, a physical component may have multiple functions, or a function or step may be performed collaboratively by several physical components. Some or all components may be implemented as software executed by a processor, such as a digital signal processor or microprocessor, or as hardware, or as an integrated circuit, such as an application-specific integrated circuit (ASIC). Such software may be distributed on a computer-readable medium, which may include computer storage media (or non-transitory media) and communication media (or transient media). As is known to those skilled in the art, the term computer storage media includes volatile and non-volatile, removable and non-removable media implemented in any method or technology for storing information (such as computer-readable instructions, data structures, program modules, or other data). Computer storage media include, but are not limited to, RAM, ROM, EEPROM, flash memory or other memory technologies, CD-ROM, digital versatile disc (DVD) or other optical disc storage, magnetic cartridges, magnetic tape, disk storage or other magnetic storage devices, or any other medium that can be used to store desired information and can be accessed by a computer. Furthermore, it is well known to those skilled in the art that communication media typically contain computer-readable instructions, data structures, program modules, or other data in modulated data signals such as carrier waves or other transmission mechanisms, and may include any information delivery medium.

[0155] It should be noted that the above embodiments or implementation methods are merely exemplary and not restrictive. Therefore, this disclosure is not limited to the content specifically shown and described herein. Various modifications, substitutions, or omissions can be made to the form and details of the implementations without departing from the scope of this disclosure.

Claims

1. A method for screening predictive biomarkers for therapeutic efficacy, comprising: Preprocess the sample data; Differentially expressed genes associated with treatment response were screened from the preprocessed sample data; A classifier model is constructed based on the gene characteristics of the differentially expressed genes; The performance of the constructed classifier model was evaluated, and efficacy prediction biomarkers were selected based on the evaluation results.

2. The method for screening predictive biomarkers for therapeutic efficacy according to claim 1, wherein, The screening criteria for differentially expressed genes are: the statistical significance p-value of the treatment response is less than or equal to a preset p-value threshold and |log2(FoldChange)| is greater than or equal to a preset second numerical threshold, where FoldChange is the fold change.

3. The method for screening predictive biomarkers for therapeutic efficacy according to claim 1, wherein, The process of screening differentially expressed genes related to treatment response from preprocessed sample data includes: Select m1 upregulated genes and m2 downregulated genes that meet the statistical significance of the treatment response p-value being less than or equal to a preset p-value threshold and |log2(FoldChange)| being greater than or equal to a preset second numerical threshold. FoldChange is the fold change, and m1 and m2 are both natural numbers, with m1 being greater than or equal to 1 and m2 being greater than or equal to 1. The upregulated gene with the smallest p-value was selected as the most upregulated gene, and the downregulated gene with the smallest p-value was selected as the most downregulated gene. The most upregulated and the most downregulated genes were selected as the differentially expressed genes in the final screening.

4. The method for screening predictive biomarkers for therapeutic efficacy according to claim 1, wherein, The gene characteristics of the differentially expressed genes include any one of the following: the gene expression level of a single differentially expressed gene, the difference in gene expression levels of multiple differentially expressed genes, and the ratio of gene expression levels of multiple differentially expressed genes.

5. The method for screening predictive biomarkers for therapeutic efficacy according to claim 1, wherein, The classifier model can be any one of the following: logistic regression model, support vector machine model, and Naive Bayes model.

6. The method for screening predictive biomarkers for therapeutic efficacy according to claim 1, wherein, The evaluation of the performance of the constructed classifier model includes: The classifier model is trained using K-fold cross-validation, where K is greater than or equal to 3. Calculate the Area Under the Receiver Operating Characteristic (AUC) value of the trained classifier model, and evaluate the performance of the classifier model based on the AUC value.

7. The method for screening predictive biomarkers for therapeutic efficacy according to claim 6, wherein, The selection of efficacy predictive biomarkers based on the assessment results includes: Choose the classifier model with the highest AUC value as the optimal classifier model; The classification threshold of the optimal classifier model is determined based on specificity and / or sensitivity.

8. The method for screening predictive biomarkers for therapeutic efficacy according to claim 7, wherein, The step of determining the classification threshold of the optimal classifier model based on specificity and / or sensitivity includes: Obtain the preset initial classification threshold, step size, and maximum classification threshold; The classification threshold is set starting from the initial classification threshold and gradually increased to the maximum classification threshold according to the step size. The specificity and / or sensitivity of the optimal classifier model at each classification threshold are calculated. The classification threshold that satisfies the preset specificity threshold and / or sensitivity threshold is used as the classification threshold of the final determined optimal classifier model.

9. An apparatus for screening predictive biomarkers of therapeutic efficacy, comprising a memory; and a processor connected to the memory, the memory for storing instructions, the processor being configured to perform the steps of the method for screening predictive biomarkers of therapeutic efficacy as described in any one of claims 1 to 8 based on the instructions stored in the memory.

10. A computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the method for screening efficacy predictive biomarkers as described in any one of claims 1 to 8.

11. A computer program product comprising instructions that, when executed by a computer, perform the efficacy predictive biomarker screening method as described in any one of claims 1 to 8.

12. A device for screening predictive biomarkers for therapeutic efficacy, comprising a preprocessing module, a screening module, a model building module, and an evaluation and selection module, wherein: The preprocessing module is configured to preprocess the sample data; The screening module is configured to screen differentially expressed genes related to treatment response from preprocessed sample data; The model building module is configured to build a classifier model based on the gene characteristics of the differentially expressed genes; The evaluation selection module is configured to evaluate the performance of the constructed classifier model and select efficacy prediction biomarkers based on the evaluation results.

13. A method for predicting therapeutic efficacy, comprising: Determine efficacy prediction biomarkers, wherein the efficacy prediction biomarkers are obtained by the efficacy prediction biomarker screening method as described in any one of claims 1 to 8; Determine the size of efficacy predictive biomarkers in the test samples, and predict efficacy based on the determined efficacy predictive biomarker sizes.

14. The efficacy prediction method according to claim 13, wherein, When the sample to be tested is a pre-treatment breast cancer sample, the ratio of ITIH1 gene expression to MMP11 gene expression is used as a predictive biomarker for efficacy; when the sample to be tested is a post-treatment breast cancer sample, the ratio of HMGCS2 gene expression to RGS1 gene expression is used as a predictive biomarker for efficacy.

15. An efficacy prediction device, comprising a memory; and a processor connected to the memory, the memory for storing instructions, the processor being configured to perform the steps of the efficacy prediction method as claimed in any one of claims 13 to 14 based on the instructions stored in the memory.

16. A computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the efficacy prediction method as described in any one of claims 13 to 14.

17. A computer program product comprising instructions that, when executed by a computer, perform the efficacy prediction method as described in any one of claims 13 to 14.

Images