A biomarker combination, kit and system for efficacy evaluation of lung adenocarcinoma targeted therapy

Abstract

The application discloses a biomarker combination, a kit and a system for evaluating the curative effect of lung adenocarcinoma targeted therapy. The biomarker combination is composed of sphingosine base metabolites, 3-methyl-2-piperidone and 4-hydroxydodecanedioyl carnitine. Through non-targeted metabolomics combined with LASSO regression algorithm (lasso regression), the application accurately locks three key markers highly related to EGFR gene mutation lung adenocarcinoma targeted therapy from numerous differential metabolites, and constructs an auxiliary evaluation system. Verification shows that the combination and system have very high sensitivity and specificity in distinguishing disease progression PD and partial remission PR, and the AUC value is greater than 0.95. The application has the advantages of minimally invasive, accurate and objective based on ex vivo serum detection, and provides an important kit and system for dynamic monitoring of lung adenocarcinoma targeted therapy.

Description

Technical Field

[0001] This invention belongs to the field of medical testing and bioinformatics technology, specifically relating to a combination of biomarkers, reagent kits and systems for evaluating the efficacy of targeted therapy for lung adenocarcinoma. Background Technology

[0002] In China, approximately one million new cases of lung cancer are diagnosed annually, with about 733,000 deaths, accounting for nearly half of the global total. Both incidence and mortality rates are the highest in the country. In Asian populations, lung adenocarcinoma is the main pathological type of lung cancer. EGFR gene mutations refer to alterations in the EGFR gene sequence, with a mutation rate as high as 30%-60%. Third-generation EGFR-TKIs (epidermal growth factor receptor tyrosine kinase inhibitors), represented by osimertinib, serve as the first-line standard treatment for EGFR-mutant advanced lung adenocarcinoma, significantly prolonging patient survival. However, acquired drug resistance remains a significant and unavoidable challenge in clinical practice, becoming a major obstacle limiting long-term patient benefits. Currently, the efficacy evaluation of targeted therapy for patients mainly relies on imaging examinations. Specifically, regular monitoring of dynamic changes in serological markers may enable the earlier detection of drug-resistant clones before lesion enlargement is observed on imaging, potentially providing guidance for personalized precision treatment of lung cancer.

[0003] Tumor cells may develop drug resistance using reversible regulatory mechanisms, including epigenetic, transcriptional, and metabolic reprogramming, evading early drug killing and maintaining survival. This may be the main cause of acquired drug resistance. Metabolomics plays an increasingly important role in discovering tumor biomarkers, assessing overall metabolic changes, and studying pathogenesis from the perspective of small molecule metabolites. Previous studies have shown that metabolomics has shown good potential in the early diagnosis of lung cancer. For example, some studies have constructed a discriminant model for differentiating non-small cell lung cancer tissues based on four differentially expressed metabolites, while others have used combinations of 19 metabolites for lung cancer screening and diagnosis. However, the efficacy evaluation of targeted therapy for patients with EGFR-mutant lung adenocarcinoma has not yet been addressed. Currently, metabolite combination models for evaluating the efficacy of targeted therapy in such patients still need to be established. To this end, this invention constructs a targeted therapy efficacy evaluation system based on serum differentially expressed metabolites. By detecting the serum metabolite levels of patients, it enables regular monitoring of the efficacy of targeted drug therapy, which helps clinicians to identify drug-resistant patients in a timely manner and adjust treatment plans, thereby potentially improving treatment benefits. Summary of the Invention

[0004] This invention provides a biomarker combination for evaluating the efficacy of targeted therapy for lung adenocarcinoma, the biomarker combination consisting of sphingosine base metabolite, 3-methyl-2-piperidinone, and 4-hydroxydodecanoic acid carnitine.

[0005] Furthermore, it includes reagents for detecting the levels of three biomarkers in isolated serum samples, the three biomarkers being composed of sphingosine base metabolites, 3-methyl-2-piperidone, and 4-hydroxydodecanoic acid carnitine.

[0006] Furthermore, the reagents include: an extractant for extracting serum metabolites, a mobile phase for liquid chromatography separation, and the three biomarkers consisting of sphingosine base metabolites, 3-methyl-2-piperidone, and 4-hydroxydodecanoic acid carnitine, with isotope-labeled internal standards for the three biomarkers.

[0007] This invention also provides a system for evaluating the efficacy of targeted therapy for lung adenocarcinoma, comprising: a data acquisition module for acquiring relative abundance detection data of three biomarkers in an ex vivo serum sample to be tested, wherein the three biomarkers are composed of sphingosine base metabolites, 3-methyl-2-piperidone, and 4-hydroxydodecanoic acid carnitine; a model processing module for substituting the relative abundance detection data into a preset logistic regression model to calculate the efficacy prediction probability value; and a result output module for outputting an auxiliary evaluation conclusion on the efficacy of targeted therapy based on the prediction probability value.

[0008] Further, the process includes the following steps: S1, collecting and processing serum samples: serum samples were collected from patients with disease progression (PD) and partial remission (PR) after targeted therapy for lung adenocarcinoma, and protein was removed; S2, detection: non-targeted metabolomics analysis was performed on the processed samples using liquid chromatography-tandem mass spectrometry (LC-MS / MS) to obtain raw mass spectrometry data; S3, data preprocessing: after converting the raw mass spectrometry data format, peak extraction, alignment, retention time correction, and missing values ​​were filtered and filled to obtain a two-dimensional matrix containing metabolite information; S4, multidimensional statistical analysis: principal component analysis (PCA) and orthogonal partial least squares discriminant analysis (OPLS-DA) were performed on the two-dimensional matrix sequentially. S5, Differential metabolite screening: Differential metabolites are screened based on the variable importance projection value (VIP), corrected false discovery rate (FDR), and fold difference (FC) of the OPLS-DA model; S6, Key feature screening: The differential metabolites are screened using the LASSO regression algorithm (lasso regression) to identify the three biomarkers mentioned above as key differential metabolites, namely, sphingosine base metabolites, 3-methyl-2-piperidinone, and 4-hydroxydodecanoic acid carnitine; S7, Model construction and evaluation: A logistic regression evaluation model is constructed based on the screened key differential metabolites, and its robustness is evaluated according to the model performance indicators.

[0009] Furthermore, in step S3, the raw data is converted into a common format using ProteoWizard (a free and open-source data processing toolkit for raw mass spectrometry data in proteomics and metabolomics research), and peak extraction and alignment are performed using the R package XCMS (short for eXtracted Chromatogram Mass Spectrometry, the most classic open-source data processing software in the field of metabolomics).

[0010] Furthermore, in step S5, the screening criteria for differential metabolites are: VIP value greater than 1, FDR less than 0.05, and FC greater than 1.5 or FC less than 0.667.

[0011] The present invention also provides the application of the above-described biomarker combination, consisting of sphingosine base metabolites, 3-methyl-2-piperidinone and 4-hydroxydodecanoic acid carnitine, in the preparation of an in vitro detection kit for assisting in the evaluation of the efficacy of targeted therapy for lung adenocarcinoma.

[0012] Compared with the prior art, the beneficial effects of this invention are as follows:

[0013] 1. High sensitivity and specificity in detection performance;

[0014] This invention successfully screened three key differentially expressed metabolites: sphingosine base metabolite, 3-methyl-2-piperidinone, and 4-hydroxydodecanoic acid carnitine. Based on these, a logistic regression model was constructed to achieve early differentiation between PD (progressive disease) and PR (partial remission) patients undergoing targeted therapy for EGFR-mutant lung adenocarcinoma. The model demonstrated stability in 10-fold cross-validation, with accuracy, sensitivity, specificity, and AUC (area under the curve) all greater than 0.95. This metabolite combination is suitable for LC-MS / MS (liquid chromatography-mass spectrometry) platforms and can be further developed into a diagnostic kit for evaluating the efficacy of targeted therapy for PD and PR patients in EGFR-mutant lung adenocarcinoma.

[0015] 2. Efficient screening methods;

[0016] To address the challenges of high dimensionality and high noise in metabolomics data, this invention first uses an orthogonal partial least squares discriminant analysis (OPLS-DA) model to select metabolites with a variable importance projection value greater than 1 for subsequent analysis. Then, univariate hypothesis testing is performed on these metabolites, and the Benjamini-Hochberg method (false discovery rate correction method) is used to correct the obtained p-values ​​to control the false discovery rate (FDR < 0.05). This is combined with the criteria of fold change (FC) > 1.5 or FC < 0.667 to screen differentially expressed metabolites. Then, using a LASSO model, three key differentially expressed metabolites are identified, significantly reducing model complexity. The logistic regression model constructed based on these three key differentially expressed metabolites demonstrates excellent efficacy assessment performance: an AUC of 0.996, accuracy of 0.985, sensitivity of 1.000, and specificity of 0.968.

[0017] In summary, this invention utilizes a logistic regression model constructed from three key differentially expressed serum metabolites to effectively distinguish between serum samples from PD (progressive disease) and PR (partial response) patients undergoing targeted therapy for lung adenocarcinoma with EGFR gene mutations. This model demonstrates high sensitivity, specificity, and accuracy. Furthermore, by employing liquid biopsy technology, this model can evaluate the efficacy of targeted therapy for lung adenocarcinoma patients, providing a basis for timely adjustments to treatment plans and potentially improving clinical outcomes and patient prognosis. Attached Figure Description

[0018] Figure 1 PCA score plot (principal component analysis) of serum sample mass spectrometry data of PD and PR patients with EGFR gene mutation lung adenocarcinoma targeted therapy: (A) positive ion mode (B) negative ion mode.

[0019] Figure 2 OPLS-DA score plot (orthogonal partial least squares discriminant analysis score plot) of serum sample mass spectrometry data of PD and PR patients with EGFR gene mutation lung adenocarcinoma targeted therapy: (A) positive ion mode; (B) negative ion mode.

[0020] Figure 3 , 3 The relative abundance of key differentially expressed metabolites in the serum of PD patients and PR patients undergoing targeted therapy for EGFR gene-mutant lung adenocarcinoma.

[0021] Figure 4 ROC curves of a model constructed based on a combination of biomarkers from three key metabolites. Detailed Implementation

[0022] The present invention will now be described in further detail with reference to the accompanying drawings and embodiments.

[0023] Example:

[0024] 1. Subject information and sample collection:

[0025] All volunteers included in the study obtained written informed consent before serum sample collection. Serum samples were collected from 33 patients with progressive disease (PD) and 33 patients with partial disease (PR) undergoing targeted therapy for EGFR-mutant lung adenocarcinoma at a single center (Hospital 1). All serum samples were collected in the morning on an empty stomach. All collected serum samples were stored at -80 degrees Celsius to avoid repeated freeze-thaw cycles.

[0026] 1.1 Inclusion criteria:

[0027] a) Age ≥ 18 years; b) Biopsy specimen pathology type is NSCLC (non-small cell lung cancer); c) Tumor tissue or blood gene test results at initial diagnosis show EGFR gene mutation; d) Previous oral treatment with a third-generation epidermal growth factor receptor tyrosine kinase inhibitor EGFR-TKI for ≥ 3 weeks, including osimertinib, vormetinib, and amitinib; e) No previous treatment with a first- or second-generation EGFR-TKI; f) Previous treatment with or without radical surgery; g) No previous treatment with chemotherapy, radiotherapy, immunotherapy, traditional Chinese medicine, or other treatments before imaging assessment of disease progression; h) After developing resistance to third-generation EGFR-TKIs, received / did not receive chemotherapy, radiotherapy, immunotherapy, traditional Chinese medicine, or other treatments; i) Complete clinical medical record information, including but not limited to gender, age, pathological type, gene mutation type, previous treatment methods, and treatment methods after developing resistance to third-generation EGFR-TKIs.

[0028] 1.2 Exclusion criteria:

[0029] a) Age <18 years; b) Pregnant or breastfeeding women; c) Previous oral treatment with third-generation EGFR-TKI for ≤3 weeks; d) Combined with other treatments during third-generation EGFR-TKI treatment; e) Patients with missing clinical medical records.

[0030] 1.3 Efficacy evaluation and observation indicators:

[0031] Patient efficacy evaluation was based on the 2009 WHO Response Evaluation Criteria in Solid Tumors (RECIST, version 1.1): Partial Response (PR): The sum of the long diameter of all target lesions and the short diameter of lymph nodes decreased by ≥30% from baseline. Progressive Disease (PD): The sum of the long diameter of all target lesions and the short diameter of lymph nodes increased by at least 20% from the recorded minimum, and the absolute value was greater than 5 mm, or at least one new lesion appeared, or there was clear progression of non-target lesions.

[0032] 2. LC-MS / MS detection:

[0033] Metabolite analysis was performed using an ExionLC liquid chromatography system (SCIEX) coupled with a SCIEX Zeno TOF 7600 high-resolution mass spectrometer to obtain raw mass spectrometry data.

[0034] 2.1 Chromatographic conditions:

[0035] Chromatographic separation was performed on a Waters ACQUITY UPLC HSS T3 column (1.8 μm, 2.1 × 100 mm) at a column temperature maintained at 40°C. The mobile phase consisted of an aqueous solution containing 0.1% formic acid (phase A) and an acetonitrile solution containing 0.1% formic acid (phase B), with a flow rate of 0.3 mL / min and an injection volume of 2 μL. A gradient elution program was used: 0–1.0 min, maintaining 2% B; 1.0–8.0 min, linearly increasing phase B from 2% to 98%; 8.0–10.0 min, maintaining 98% B; then recovering to 2% B within 0.1 min and equilibrating to 12.0 min.

[0036] 2.2 Mass spectrometry conditions:

[0037] Mass spectrometry detection employed an electrospray ionization source, performing alternating full scans and information-dependent acquisition in positive and negative ion modes. Key parameters were as follows: ion source temperature 500°C; nebulizer gas and auxiliary heating gas both 45 psi; curtain gas 35 psi. The capillary voltage was +5500 V in positive ion mode and −4500 V in negative ion mode. After triggering IDA (information-dependent acquisition), time-of-flight mass spectrometry (TOF-MS) scans were performed in both positive (+) and negative (−) ion modes, i.e., the TOF(+) / TOF(−) MS scan range was 20–1200 m / z, with collision energies of +35 V (positive mode) and −35 V (negative mode), respectively.

[0038] 2.3 Mass spectrometry data processing:

[0039] Mass spectrometry data were converted to mzXML or mzML format using ProteoWizard (a free and open-source data processing toolkit for raw mass spectrometry data in proteomics and metabolomics research). Peak extraction, retention time correction, and peak alignment were then performed using XCMS (eXtracted Chromatogram Mass Spectrometry, a classic open-source data processing software in the field of metabolomics) to generate a raw peak list. Data preprocessing mainly included missing value filtering, imputation, and normalization. Specifically, firstly, characteristic peaks with a non-zero value ratio exceeding 75% in all samples were retained; then, the remaining missing values ​​were imputed using half the minimum value of the characteristic peaks. Further, based on quality control samples, characteristic peaks with a relative standard deviation greater than 30% were filtered out to improve data quality. Finally, the samples were standardized using the total intensity normalization method to obtain a data matrix usable for subsequent statistical analysis.

[0040] 3. Metabolite identification:

[0041] The preprocessed primary and secondary mass spectrometry data were compared with public metabolic databases such as HMDB (Human Metabolome Database) and laboratory-built standard spectral libraries to complete metabolite identification. At the same time, with the help of tools such as MetDNA, the identities of potential metabolites were inferred based on metabolic reaction networks, thereby achieving systematic identification of differential metabolites.

[0042] 4. Screening of differentially expressed metabolites and model establishment:

[0043] 4.1 Screening for differentially expressed serum metabolites between PD and PR patients after targeted therapy for EGFR gene-mutant lung adenocarcinoma;

[0044] Unsupervised principal component analysis (PCA) was used to examine the overall distribution of the samples. A significant separation trend in the serum metabolic profiles of PD and PR patients was observed in both positive and negative ion modes. Figure 1 ).

[0045] Then, supervised orthogonal partial least squares discriminant analysis (OPLS-DA) was used to distinguish the differences in metabolic profiles between PD patients and PR patients after targeted therapy; such as Figure 2 As shown, significant metabolic differences and intergroup classification trends were observed between the PD and PR groups under both positive and negative ion modes. Under the positive ion mode, the model's R²X = 0.403, R²Y = 0.978, and Q²Y = 0.778; under the negative ion mode, the model's R²X = 0.144, R²Y = 0.893, and Q²Y = 0.665.

[0046] Based on the orthogonal partial least squares discriminant analysis (OPLS-DA) model, metabolites with a variable importance projection value greater than 1 were selected for subsequent analysis. Next, univariate hypothesis testing was performed on these metabolites, and the obtained p-values ​​were corrected using the Benjamini-Hochberg method to control for false detection rate. The fold change between groups was also calculated. Finally, the differentially expressed metabolites selected had to simultaneously meet the following conditions: VIP > 1, FDR < 0.05, and FC > 1.5 or FC < 0.667; 35 differentially expressed metabolites were obtained.

[0047] 4.2 Key differentially expressed metabolites were screened using the LASSO algorithm:

[0048] Based on the abundance matrix of 35 differentially expressed metabolites obtained in step 4.1, this invention uses LASSO regression to screen variables, with a feature importance score greater than 0.5 calculated from the regression coefficient as the threshold. Ultimately, three key differentially expressed metabolites were identified: sphingosine base metabolites, 3-methyl-2-piperidinone, and 4-hydroxydodecanoic acid carnitine. See Table 1 and... Figure 3 As shown, compared with the PR group, sphingosine base metabolites and 3-methyl-2-piperidinone were significantly elevated in the serum of PD patients, while 4-hydroxydodecanoic acid carnitine was significantly decreased.

[0049] 4.3 Establishing a efficacy assessment model based on key differential metabolite combinations:

[0050] Based on the three key differentially expressed metabolites listed in Table 1, and using PD and PR as response variables, a discriminant model for evaluating the efficacy of targeted therapy for EGFR-mutant lung adenocarcinoma was constructed using a logistic regression model. Multivariate ROC curve analysis based on the three key differentially expressed metabolites showed that AUC = 0.996 (…). Figure 4 The results showed that the combination of biomarkers was highly effective in evaluating the efficacy of targeted therapy, with an accuracy of 0.985, sensitivity of 1.000, and specificity of 0.968, which can provide important support for clinical practice.

[0051] High sensitivity ensures the capture of true positive cases (reducing the rate of missed diagnoses), while high specificity effectively eliminates false positives (controlling the risk of misdiagnosis). The key biomarker combination discovered in this invention demonstrates outstanding performance in both dimensions, suggesting its high reliability in assisting clinical assessment of disease progression after targeted therapy for EGFR-mutant lung adenocarcinoma. Based on this advantage, this combination has good application potential in screening for and early identifying drug resistance risks in patients undergoing targeted therapy for EGFR-mutant lung adenocarcinoma, and is expected to contribute to achieving more precise individualized treatment.

[0052] Table 1. Three key differentially expressed metabolites

[0053] Serial Number Metabolites Ion mode ESI+ / - Retention time RT Mass-to-charge ratio Difference factor FC False discovery rate (FDR) LASSO rating 1 Sphingosine base metabolites ESI+ 5.92 230.2473 0.55 2.79E-07 0.70 2 4-Hydroxydodecanoylcarnitine ESI+ 5.90 390.2546 1.57 1.37E-06 1.21 3 3-Methyl-2-piperidinone ESI+ 5.11 114.0910 0.66 1.90E-12 2.09

[0054] In the embodiments of the present invention, the preprocessing of non-targeted metabolomics data is calibrated using the Total Intensity Normalization method to eliminate systematic errors caused by fluctuations in injection volume and mass spectrometry detection between samples, thereby screening out differentially significant metabolites.

[0055] It should be noted that the kit and system of the present invention are calibrated by adding isotope-labeled internal standards of the three biomarkers. It is understood that the isotope internal standard method and the normalization method in the examples have a high degree of technical consistency in assessing the trend of relative abundance changes of metabolites. Using the isotope internal standard method can further eliminate matrix effects, achieving more accurate quantitative analysis of target biomarkers, thereby further improving the robustness, sensitivity, and specificity of the logistic regression assessment model in clinical applications. Therefore, the assessment model constructed based on the normalized detection data in the examples is also applicable to detection data obtained through the internal standard method, and the technical effects of both in assisting the evaluation of the efficacy of targeted therapy for lung adenocarcinoma are consistent.

[0056] The above-described embodiments are merely preferred embodiments of the present invention. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A combination of biomarkers for evaluating the efficacy of targeted therapy for lung adenocarcinoma, characterized in that, The biomarker combination consists of sphingosine base metabolite, 3-methyl-2-piperidone, and 4-hydroxydodecanoic acid carnitine.

2. A kit for evaluating the efficacy of targeted therapy for lung adenocarcinoma, characterized in that, The reagent includes a method for detecting the levels of three biomarkers in an isolated serum sample, wherein the three biomarkers are a combination as described in claim 1.

3. The reagent kit according to claim 2, characterized in that, The reagents include: an extractant for extracting serum metabolites, a mobile phase for liquid chromatography separation, and isotope-labeled internal standards for the three biomarkers.

4. A system for evaluating the efficacy of targeted therapy for lung adenocarcinoma, characterized in that, include: The data acquisition module is used to acquire the relative abundance detection data of three biomarkers in the isolated serum sample to be tested, wherein the three biomarkers are the combination described in claim 1; The model processing module is used to substitute the relative abundance detection data into a preset logistic regression model to calculate the efficacy prediction probability value; the result output module is used to output the auxiliary evaluation conclusion of the targeted therapy efficacy based on the prediction probability value.

5. A method for constructing the system for evaluating the efficacy of targeted therapy for lung adenocarcinoma as described in claim 4, characterized in that, The process includes the following steps: S1, collecting and processing serum samples: serum samples were collected from patients with disease progression and those in partial remission undergoing targeted therapy for lung adenocarcinoma, and protein was removed; S2, detection: non-targeted metabolomics analysis was performed on the processed samples using liquid chromatography-tandem mass spectrometry (LC-MS / MS) to obtain raw mass spectrometry data; S3, data preprocessing: after converting the raw mass spectrometry data format, peak extraction, alignment, retention time correction, and missing values ​​were filtered and filled to obtain a two-dimensional matrix containing metabolite information; S4, multidimensional statistical analysis: principal component analysis and... S5. Orthogonal-partial least squares discriminant analysis to establish a supervised multivariate statistical model; S6. Differential metabolite screening: Based on the variable importance projection values, corrected false discovery rate, and difference fold of the orthogonal-partial least squares discriminant analysis model, differential metabolites are screened; S7. Key feature screening: The LASSO regression algorithm is used to screen the differential metabolites, and the three biomarkers are identified as key differential metabolites; S8. Model construction and evaluation: Based on the screened key differential metabolites, a logistic regression evaluation model is constructed, and its robustness is evaluated according to the model performance indicators.

6. The method according to claim 5, characterized in that, In step S3, the raw data is converted into a common format using data processing tools, and peak extraction and alignment are performed using data processing software.

7. The method according to claim 5, characterized in that, In step S5, the screening criteria for differential metabolites are: variable importance projection value greater than 1, corrected false discovery rate less than 0.05, and difference fold greater than 1.5 or less than 0.

667.

8. The use of the biomarker combination of claim 1 in the preparation of an in vitro detection kit for assisting in the evaluation of the efficacy of targeted therapy for lung adenocarcinoma.

Images