NSCLC patient typing and third-generation EGFR-TKI curative effect prediction model based on T790M mutation of plasma metabolite
Through plasma metabolites detection and machine learning algorithms, markers used to predict the efficacy of third-generation EGFR-TKIs in patients with T790M mutation NSCLC were identified, which solved the problem of inaccurate efficacy prediction in the prior art and achieved personalization Treatment guidance and improvement of efficacy.
Patent Information
- Application Number
- CN202411291900.6
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-09-14
- Publication Date
- 2024-12-20
AI Technical Summary
The prior art is difficult to accurately predict the efficacy of NSCLC patients with T790M mutation on third-generation EGFR-TKIs, and the lack of effective typing tools hinders the implementation of personalized treatment.
Through plasma metabolites detection, 28 markers (such as soybean flavin sulfate, N-(2-furoyl)glycine, hexanoylglutamine, etc.) were identified, combined with non-negative matrix factor decomposition and random forest algorithm. Models were established to evaluate the prognosis of third-generation EGFR-TKI treatment, distinguish patient subtypes and predict efficacy.
It has achieved accurate prediction of the classification of patients with T790M mutant NSCLC and the efficacy of third-generation EGFR-TKI, helping clinicians to formulate personalized treatment plans, improve treatment effects and reduce medical costs.
Smart Images
Figure HDA0005044854430000011 
Figure HDA0005044854430000021
Abstract
Description
Technical Field
[0001] The present invention belongs to the field of biomedical detection technology, and specifically relates to a NSCLC patient typing based on T790M mutation of plasma metabolites and a third-generation EGFR-TKI efficacy prediction model. Background Art
[0002] Non-small cell lung cancer (NSCLC) is the most common histological type of lung cancer, and epidermal growth factor receptor-tyrosine kinase inhibitors (EGFR-TKIs) are currently one of the most commonly used drugs for patients with advanced NSCLC who cannot undergo surgery. The first-generation EGFR-TKIs have good efficacy in NSCLC patients with EGFR-sensitive mutations, but most patients tend to develop drug resistance after 9-15 months (Rosell et al., 2012). The most common mechanism of resistance is the EGFR T790M mutation. The third-generation EGFR-TKIs targeting the T790M mutation have shown good efficacy in treating both EGFR-sensitive and T790M-mutated NSCLC patients, bringing new hope for treatment to patients (Ramalingam et al., 2020). Results from several clinical trials have shown that compared with platinum-based chemotherapy, first-generation EGFR-TKIs (e.g., gefitinib and erlotinib) and second-generation EGFR-TKIs (e.g., afatinib and dacomitinib) prolonged progression-free survival (PFS) by 10-14 months, but unfortunately, prolonged overall survival (OS) was not observed (Rosell et al., 2012). The efficacy of third-generation EGFR-TKIs (e.g., osimertinib, ametinib, vometinib, and ruzitotinib) is significantly better than that of first- and second-generation EGFR-TKIs, with a median PFS of 17.8-20.8 months (Shi et al., 2022; Shi et al., 2010; Shi et al., 2023; Soria et al., 2018). More importantly, compared with the first- and second-generation EGFR-TKIs, osimertinib treatment achieved OS benefits (OS: 38.6 months, compared with 31.8 months), making osimertinib a new drug for first-line and later-line treatment of advanced NSCLC with EGFR mutations. Detection of EGFR gene mutations by tissue biopsy or circulating cell-free tumor DNA (ctDNA) is essential for determining treatment strategies before treatment and after disease progression. However, patients with the same mutation are still heterogeneous, and only 70% of patients with T790M mutations respond to third-generation EGFR-TKIs (Mok et al., 2017). In addition, about 50% of patients will inevitably progress after 9-10 months of medication (Mok et al., 2017), and the time of progression is difficult to predict. Relapse and progression can be predicted by monitoring changes in tumor markers in the blood.Studies have shown that carcinoembryonic antigen (CEA), CA125, and ctDNA may be effective tumor biomarkers. However, current research is insufficient, and the detection of these molecules alone is not enough to provide accurate assessments, which hinders the application of tumor marker assessment in clinical practice.
[0003] In recent years, many studies have shown that comprehensive metabolite profiling provides new potential biomarkers for precise diagnosis and efficacy testing, and in some aspects is even superior to genomic and proteomics methods. Metabolomics-related research also provides a unique direction for discovering tumor pathophysiological mechanisms. Studies have reported that metabolic reprogramming, such as lipid metabolism reprogramming, increased de novo glutathione synthesis, and abnormal glutamine uptake (Huang et al., 2019; Liu et al., 2021), is involved in the development of resistance to first-generation EGFR-TKIs. However, relevant studies on the circulating metabolic characteristics of NSCLC patients who have a significant response to third-generation EGFR-TKIs are still limited, and there is an urgent need to find new plasma tumor markers that can predict the efficacy of third-generation EGFR-TKIs. Summary of the invention
[0004] Currently, there are no clinical classification tools for NSCLC patients with T790M mutations and tools for predicting the efficacy of third-generation EGFR-TKIs. The present invention studies the classification and efficacy prediction biomarkers of NSCLC patients with T790M mutations, aiming to explore the heterogeneity of this group of people and their association with prognosis based on plasma metabolomics detection of baseline samples, and to establish a model that can well distinguish different groups and predict their efficacy.
[0005] In order to solve the above technical problems, the present invention provides the following technical solutions:
[0006] In a first aspect, the present invention provides a marker for evaluating the prognosis of third-generation EGFR-TKI drug treatment for advanced non-small cell lung cancer with T790M mutation, wherein the marker includes glycitein sulfate, N-(2-furoyl)glycine, N-oleoyltaurine, caproylglutamine, 3-hydroxyoleic acid, myristoleic acid (14:1n5), oleic acid (18:1), palmitic acid (16:0), palmitoleic acid (16:1n7), eicosenoic acid (20:1), docosadienoic acid (22:2n6), heneicosopentaenoic acid (21:5n3), 3-hydroxybutyric acid, 3-hydroxybutyrylcarnitine, caproylcarnitine (C6), 3-hydroxydecanoylcarnitine, 3-hydroxyoleoyl Carnitine, 5-dodecenoylcarnitine (C12:1), decanoylcarnitine (C10), lauroylcarnitine (C12), myristoylcarnitine (C14:1), myristoylcarnitine (C14), octanoylcarnitine (C8), oleoylcarnitine (C18:1), palmitoleoylcarnitine (C16:1), 1-linoleoyl-GPE (18:2), 1,2-dilinoleoyl-GPE (18:2 / 18:2), 1-oleoyl-2-linoleoyl-GPE (18:1 / 18:2), or any combination thereof.
[0007] In a second aspect, the present invention provides a reagent for detecting a marker for use in preparing a kit for evaluating the prognosis of third-generation EGFR-TKI drug treatment for advanced non-small cell lung cancer with T790M mutation, wherein the marker comprises glycitein sulfate, N-(2-furoyl)glycine, N-oleoyltaurine, hexanoylglutamine, 3-hydroxyoleic acid, myristoleic acid (14:1n5), oleic acid (18:1), palmitic acid (16:0), palmitoleic acid (16:1n7), eicosenoic acid (20:1), docosadienoic acid (22:2n6), henicosopentaenoic acid (21:5n3), 3-hydroxy ... -Hydroxybutyric acid, 3-hydroxybutyrylcarnitine, caproylcarnitine (C6), 3-hydroxydecanoylcarnitine, 3-hydroxyoleoylcarnitine, 5-dodecenoylcarnitine (C12:1), decanoylcarnitine (C10), lauroylcarnitine (C12), myristoylcarnitine (C14:1), myristoylcarnitine (C14), octanoylcarnitine (C8), oleoylcarnitine (C18:1), palmitoleoylcarnitine (C16:1), 1-linoleoyl-GPE (18:2), 1,2-dilinoleoyl-GPE (18:2 / 18:2), 1-oleoyl-2-linoleoyl-GPE (18:1 / 18:2), or any combination thereof.
[0008] In a third aspect, the present invention provides a kit for evaluating the prognosis of third-generation EGFR-TKI drug treatment for advanced non-small cell lung cancer with T790M mutation, the kit comprising reagents for detecting markers, the markers comprising glycitein sulfate, N-(2-furoyl)glycine, N-oleoyltaurine, hexanoylglutamine, 3-hydroxyoleic acid, myristoleic acid (14:1n5), oleic acid (18:1), palmitic acid (16:0), palmitoleic acid (16:1n7), eicosenoic acid (20:1), docosadienoic acid (22:2n6), henicosopentaenoic acid (21:5n3) , 3-hydroxybutyric acid, 3-hydroxybutyrylcarnitine, caproylcarnitine (C6), 3-hydroxydecanoylcarnitine, 3-hydroxyoleoylcarnitine, 5-dodecenoylcarnitine (C12:1), decanoylcarnitine (C10), lauroylcarnitine (C12), myristoylcarnitine (C14:1), myristoylcarnitine (C14), octanoylcarnitine (C8), oleoylcarnitine (C18:1), palmitoleoylcarnitine (C16:1), 1-linoleoyl-GPE (18:2), 1,2-dilinoleoyl-GPE (18:2 / 18:2), 1-oleoyl-2-linoleoyl-GPE (18:1 / 18:2), or any combination thereof.
[0009] In a fourth aspect, the present invention provides a method for evaluating the prognosis of third-generation EGFR-TKI drug treatment for advanced non-small cell lung cancer with T790M mutation, the method comprising detecting markers, the markers comprising glycitein sulfate, N-(2-furoyl)glycine, N-oleoyltaurine, hexanoylglutamine, 3-hydroxyoleic acid, myristoleic acid (14:1n5), oleic acid (18:1), palmitic acid (16:0), palmitoleic acid (16:1n7), eicosenoic acid (20:1), docosadienoic acid (22:2n6), henicopentaenoic acid (21:5n3), 3-hydroxy 1-linoleoyl-GPE (18:2), 1,2-dilinoleoyl-GPE (18:2 / 18:2), 1-oleoyl-2-linoleoyl-GPE (18:1 / 18:2), or any combination thereof.
[0010] In some embodiments, the marker is from the plasma of the subject. In some embodiments, the marker is from the plasma of the subject before receiving the drug.
[0011] In some embodiments, the markers include glycitein sulfate, N-(2-furoyl)glycine, N-oleoyltaurine, caproylglutamine, 3-hydroxyoleic acid, myristoleic acid (14:1n5), oleic acid (18:1), palmitic acid (16:0), palmitoleic acid (16:1n7), eicosenoic acid (20:1), docosadienoic acid (22:2n6), einosapentaenoic acid (21:5n3), 3-hydroxybutyric acid, 3-hydroxybutyrylcarnitine, caproylcarnitine (C6), 3-hydroxydecanoylcarnitine The invention relates to a combination of alkali, 3-hydroxyoleoylcarnitine, 5-dodecenoylcarnitine (C12:1), decanoylcarnitine (C10), lauroylcarnitine (C12), myristoylcarnitine (C14:1), myristoylcarnitine (C14), octanoylcarnitine (C8), oleoylcarnitine (C18:1), palmitoleoylcarnitine (C16:1), 1-linoleoyl-GPE (18:2), 1,2-dilinoleoyl-GPE (18:2 / 18:2), and 1-oleoyl-2-linoleoyl-GPE (18:1 / 18:2).
[0012] In some embodiments, the marker is glycitein sulfate, N-(2-furoyl)glycine, N-oleoyltaurine, caproylglutamine, 3-hydroxyoleic acid, myristoleic acid (14:1n5), oleic acid (18:1), palmitic acid (16:0), palmitoleic acid (16:1n7), eicosenoic acid (20:1), docosadienoic acid (22:2n6), einosapentaenoic acid (21:5n3), 3-hydroxybutyric acid, 3-hydroxybutyrylcarnitine, caproylcarnitine (C6), 3-hydroxydecanoylcarnitine The invention relates to a combination of alkali, 3-hydroxyoleoylcarnitine, 5-dodecenoylcarnitine (C12:1), decanoylcarnitine (C10), lauroylcarnitine (C12), myristoylcarnitine (C14:1), myristoylcarnitine (C14), octanoylcarnitine (C8), oleoylcarnitine (C18:1), palmitoleoylcarnitine (C16:1), 1-linoleoyl-GPE (18:2), 1,2-dilinoleoyl-GPE (18:2 / 18:2), and 1-oleoyl-2-linoleoyl-GPE (18:1 / 18:2).
[0013] In some embodiments, the third generation EGFR-TKI drug includes Osimertinib, Almonertinib, Furmonertinib, Rezivertinib, or a combination thereof.
[0014] In some embodiments, the third-generation EGFR-TKI drug is rizitinib.
[0015] In some embodiments, the advanced non-small cell lung cancer is stage III and stage IV non-small cell lung cancer as defined by the TNM staging system.
[0016] In some embodiments, reagents for detecting a marker include reagents for qualitatively or quantitatively detecting the marker.
[0017] In some embodiments, reagents for detecting a marker include marker standards used in an ultra-performance liquid chromatography / tandem mass spectrometry (UPLC-MS / MS) method to identify the marker.
[0018] In some embodiments, the marker standards include corresponding standards of the following markers: glycitein sulfate, N-(2-furoyl)glycine, N-oleoyltaurine, caproylglutamine, 3-hydroxyoleic acid, myristoleic acid (14:1n5), oleic acid (18:1), palmitic acid (16:0), palmitoleic acid (16:1n7), eicosenoic acid (20:1), docosadienoic acid (22:2n6), einicosopentaenoic acid (21:5n3), 3-hydroxybutyric acid, 3-hydroxybutyrylcarnitine, caproylcarnitine (C6), 3 -hydroxydecanoylcarnitine, 3-hydroxyoleoylcarnitine, 5-dodecenoylcarnitine (C12: 1), decanoylcarnitine (C10), lauroylcarnitine (C12), myristoylcarnitine (C14: 1), myristoylcarnitine (C14), octanoylcarnitine (C8), oleoylcarnitine (C18: 1), palmitoleoylcarnitine (C16: 1), 1-linoleoyl-GPE (18: 2), 1,2-dilinoleoyl-GPE (18: 2 / 18: 2), 1-oleoyl-2-linoleoyl-GPE (18: 1 / 18: 2) or any combination thereof. In some preferred embodiments, the marker standard is the corresponding standard of the above 28 markers.
[0019] In some embodiments, the detection method of the marker includes an ultra-high performance liquid chromatography / tandem mass spectrometry (UPLC-MS / MS) method. In some embodiments, marker identification is performed based on marker standards by an ultra-high performance liquid chromatography / tandem mass spectrometry (UPLC-MS / MS) method. In some embodiments, marker quantification is performed by calculating the area under the chromatographic peak curve.
[0020] In some embodiments, the heterogeneity of circulating metabolomics of non-small cell lung cancer is analyzed by non-negative matrix factorization (NMF) method, and patients are divided into two subtypes with significantly different metabolic characteristics.
[0021] In some embodiments, the non-negative matrix factorization (NMF) method divides patients into two subtypes with distinct metabolic characteristics. Subtype I is characterized by arginine and proline metabolism, glycerophospholipid metabolism, purine metabolism, and pyrimidine metabolism, and has a good prognosis for drug treatment; subtype II is characterized by elevated biosynthesis of unsaturated fatty acids, and has a poor prognosis for drug treatment.
[0022] In some embodiments, the prognosis of third-generation EGFR-TKI drug treatment for advanced non-small cell lung cancer is assessed by a random forest algorithm based on marker quantification (e.g., area under the curve obtained by ultra-performance liquid chromatography / tandem mass spectrometry (UPLC-MS / MS)).
[0023] In some embodiments, when the good prognosis probability obtained by the random forest algorithm is greater than the poor prognosis probability (ie, when the probability that the patient is subtype I is greater than 50%), the subject has a good prognosis for drug treatment.
[0024] In some embodiments, when the good prognosis probability obtained by the random forest algorithm is less than the poor prognosis probability (ie, when the probability that the patient is subtype I is less than 50%), the subject has a poor prognosis for drug treatment.
[0025] In some embodiments, the probability that the 28 markers predict a good prognosis (i.e., predicting that the patient is subtype I) is higher than 50% (i.e., the probability of poor prognosis is less than the probability of good prognosis) in subjects with advanced non-small cell lung cancer with T790M mutation is suitable for treatment with third-generation EGFR-TKI drugs.
[0026] In some embodiments, non-negative matrix factorization (NMF) and random forest algorithms are performed using R package "NMF" (version 0.20.6) and R package "randomForest" (version 4.7-1.1) of R language software (version 4.2.2), respectively. In some specific embodiments, non-negative matrix factorization (NMF) is performed using the following parameters: Brunet algorithm, nrun = 200. In some specific embodiments, the random forest algorithm is performed using the following parameters: interaction.depth = 6, n.trees = 300, shrinkage = 0.001, n.minobsinnode = 10.
[0027] In some embodiments, the drug treatment prognosis includes the prognosis of third-generation EGFR-TKI drug treatment for advanced non-small cell lung cancer, such as efficacy (good (complete remission (CR), partial remission (PR)) and poor (stable disease (SD), progressive disease (PD)), progression-free survival (PFS) and overall survival (OS).
[0028] Beneficial effects of the present invention:
[0029] The present invention discloses the application of 28 plasma metabolic markers in the classification of patients with advanced non-small cell lung cancer with T790M mutation and the prediction of the prognosis and efficacy of the third-generation EGFR-TKI drug treatment of advanced non-small cell lung cancer. According to the above application, by detecting glycitein sulfate, N-(2-furoyl)glycine, N-oleoyltaurine, caproylglutamine, 3-hydroxyoleic acid, myristoleic acid (14:1n5), oleic acid (18:1), palmitic acid (16:0), palmitoleic acid (16:1n7), eicosenoic acid (20:1), docosadienoic acid (22:2n6), heneicosopentaenoic acid (21:5n3), 3-hydroxybutyric acid, 3-hydroxybutyrylcarnitine, caproylcarnitine (C6), 3-hydroxydecanoylcarnitine, 3-hydroxyoleoylcarnitine, 5-dodecenoylcarnitine (C12:1), decanoylcarnitine (C10), lauroylcarnitine ( The contents of myristoylcarnitine (C12), myristoylcarnitine (C14:1), myristoylcarnitine (C14), octanoylcarnitine (C8), oleoylcarnitine (C18:1), palmitoleoylcarnitine (C16:1), 1-linoleoyl-GPE (18:2), 1,2-dilinoleoyl-GPE (18:2 / 18:2), and 1-oleoyl-2-linoleoyl-GPE (18:1 / 18:2) combined with the random forest model can determine the prognosis of the subjects, predict the efficacy of the treatment regimen, assist in the selection and evaluation of the treatment regimen, and thus guide clinical physicians to adopt targeted treatment regimens for the subjects, which is beneficial to improve the prognosis of the subjects and reduce medical expenses. BRIEF DESCRIPTION OF THE DRAWINGS
[0030] Figure 1 Figure 3 shows the plasma metabolome revealing heterogeneity in patients with EGFR-mutated non-small cell lung cancer: (A) Metabolic benchmark clustering of non-small cell lung cancer samples (n=139) based on 150 metabolites, with samples represented in columns, metabolites represented in rows, and protein signal density Z-scored between individuals; (B) Unsaturated fatty acids were significantly increased in subtype II compared with subtype I, using a two-sided Wilcoxon's test, the middle line of the box plot is the median, the top and bottom edges represent the upper and lower quartiles, and the whiskers represent the minimum and maximum values; (C) Kaplan-Meier curves of OS for the two subtypes in the discovery cohort (n=139); (D) Kaplan-Meier curves of OS for the two subtypes in patients with brain metastases in the discovery cohort; (E) Overall survival curves according to metabolic subtypes in patients with EGFR exon 19 deletion and L858R mutation in the discovery cohort (log-rank test).
[0031] Figure 2The clinical characteristics of the two subtypes and the subtype prediction model based on the random forest algorithm are shown: (A) The random forest algorithm sorts the variables by the average ranking of Gini decrease and selects the variables with the average reduced Gini value greater than 0.5; (B) AUC of the subtype model in the training dataset (n=97) and the test dataset (n=42); (C) and (D) Kaplan-Meier analysis of OS of the two subtypes predicted by the model in cohort 1 (n=139) and the validation cohort (n=47); (E) Box plot of fatty acid levels of R and NR patients in the external validation cohort. DETAILED DESCRIPTION
[0032] The present invention is further described below in conjunction with the accompanying drawings and specific examples. These examples are only used to illustrate the present invention and are not intended to limit the scope of the present invention. The experimental methods in the following examples where specific conditions are not specified are usually performed according to conventional conditions in the art or according to conditions recommended by the manufacturer. Unless otherwise specified, all conventional methods are used. Unless otherwise defined, the professional and scientific terms used in the text have the same meanings as those familiar to those skilled in the art.
[0033] Example
[0034] In this study, baseline samples of 186 patients with advanced NSCLC with T790M mutation who participated in the third-generation EGFR-TKI ruzitotinib (BPI-7711) clinical trial were collected for non-targeted metabolomics study to characterize the metabolic characteristics of patients with different efficacy and prognosis of third-generation EGFR-TKI. The clinical characteristics of the two cohorts are shown in Table 1. Cohort 1 was divided into a discovery cohort (n = 97) and a validation cohort (n = 42). A random forest model was established based on the metabolic data of 139 patients and validated in an external validation cohort, cohort 2 (n = 47).
[0035] Table 1. Clinical characteristics of NSCLC patients included in the study
[0036] variable Queue 1, N=139 Queue 2, N=47 Age, mean ± SD 59(51,66) 59(54,66) Gender, n (%) female 90(65％) 32(68％) male 49(35％) 15(32％) CNS metastasis, n (%) N 73(53％) 32(68％) Y 66(47％) 15(32％) Tissue type, n (%) Adenocarcinoma (NOS) 71(51％) 23(49％) Adenocarcinoma: Acinar 59(42％) 22(47％) Adenocarcinoma: Nipple 3(2.2％) 2(4.3％) Adenosquamous carcinoma 2(1.4％) \ other 3(2.2％) \ Squamous cell carcinoma 1(0.7％) \ BMI.group Low (<18.5) 17(12％) 6(13％) Normal (18.5-24.9) 69(50％) 26(55％) Overweight (25.0–29.9) 53(38％) 15(32％)
[0037] Experimental methods
[0038] 1. Queue characteristics
[0039] Clinical data and samples of the expanded cohort of the clinical trial of the third-generation EGFR-TKI targeted anti-tumor drug BPI-7711 with T790M mutation were collected, and the human genetic resource registration for sample use rights has been approved. Finally, non-targeted metabolomics testing was performed on baseline plasma samples of 186 patients. According to the definition of objective response rate, the good efficacy (R) group was defined as complete remission or partial remission after treatment, and the poor efficacy (NR) group was defined as stable disease or progressive disease. The good efficacy and poor efficacy groups were relatively matched in clinical information such as age, gender, tissue type, presence or absence of brain metastasis, and most test indicators.
[0040] 2.UPLC-MS / MS detection
[0041] Non-targeted metabolomics testing was performed by the CalOmics metabolomics laboratory of Dian Diagnostics using its CalOmics metabolomics platform. Methanol was added to the samples in a ratio of 1:4. After oscillation for 3 minutes to mix, insoluble matter was precipitated by centrifugation at 4,000 × g for 10 minutes at 20°C. After centrifugation, 100 μL of supernatant was taken from each sample and transferred to 4 sample plates, dried with nitrogen, and then added to the reconstitution solution for subsequent UPLC-MS / MS detection. The four different UPLC-MS / MS methods were all performed using ACQUITY2D UPLC (Waters, Milford, MA, USA) ultra-high performance liquid chromatography combined with Q Exactive (QE) high-resolution mass spectrometry (Thermo Fisher Scientific, San Jose, USA). The QE mass spectrometry parameters were a scan resolution of 35,000 and a scan range of 70-1000 m / z. The specific parameters of the four UPLC-MS / MS methods are as follows: Method 1: QE was detected in positive ion electrospray ionization (ESI) mode, and the liquid phase was separated using a C18 column (UPLC BEH C18, 2.1×100 mm, 1.7 μm; Waters), and the mobile phase was water (A) and methanol (B) containing 0.05% PFPA and 0.1% FA; Method 2: QE was detected in negative ion electrospray ionization (ESI) mode, and the liquid phase was separated using a C18 column (UPLC BEH C18, 2.1×100 mm, 1.7 μm; Waters), and the mobile phase was water (A) and methanol (B) containing 6.5 mM ammonium bicarbonate; Method 3: QE was detected in positive ion electrospray ionization (ESI) mode, and the liquid phase was separated using a C18 column (UPLC BEH C18, 2.1×100 mm, 1.7 μm; Waters) for separation, and the mobile phases were water (A) and methanol / acetonitrile / water (B) containing 0.05% PFPA and 0.1% FA; Method 4: QE was detected in negative ion electrospray ionization (ESI) mode, and the liquid phase was separated using a HILIC column (UPLC BEH Amide, 2.1×150 mm, 1.7 μm; Waters), and the mobile phases were water (A) and acetonitrile (B) containing 10 mM ammonium formate.
[0042] 3. Substance identification and quantification
[0043] First, the test data is preprocessed, and after confirming that it meets the quality control requirements, the chromatographic peak is extracted using the self-developed data processing system. Then, by comparing the self-built metabolite database based on the standard, the metabolite identification is carried out based on the strict matching of the three indicators of retention index, primary accurate mass number and secondary fragment spectrum. Metabolite quantification is completed by calculating the area under the chromatographic peak curve to obtain the original peak area of each metabolite.
[0044] 4. Data Standardization
[0045] After obtaining the raw peak area of each metabolite, it was standardized for subsequent statistical and bioinformatics analysis. First, the raw peak area of each metabolite was logarithmically transformed (log2) with a base of 2 to reduce the skewed distribution of the overall value and make the data close to a normal distribution, then normalized using the median, and finally the minimum value of all samples was used to fill the missing values.
[0046] 5. Statistical Analysis
[0047] Metabolites with missing data in more than 80% of the samples will be removed from the data set. Missing values are filled with 1 / 5 of the minimum value of the metabolite in all samples. Metabolic data are log10 transformed. The original peak area of each metabolite is logarithmically transformed (log2) and normalized by the median. The R package "NMF" (version 0.20.6) was used for NSCLC plasma typing, and the R package "randomForest" (version 4.7-1.1) was used for feature selection and model building. In the random forest analysis, one hundred and eighty trees were used, and the ten-fold cross validation was repeated 3 times. The R package "pROC" (version 1.18.0) calculated the AUC to evaluate the model performance.
[0048] Experimental Results
[0049] Metabolomic profile of NSCLC patients in response to third-generation EGFR-TKI
[0050] A total of 186 patients were included in this study. Cohort 1 (n = 139) was divided into a discovery cohort (n = 97) and a validation cohort (n = 42), of which the discovery cohort (n = 97) was used to build the model, and the validation cohort (n = 42) and external validation cohort (n = 47) were used as validation sets. All three cohorts used ultra-performance liquid chromatography / tandem mass spectrometry (UPLC-MS / MS) non-targeted metabolomics methods, and plasma samples were analyzed by stable isotope-labeled metabolomics strategy TMTpro (16plex) (Li et al., 2020).
[0051] First, we performed omics analysis on cohort 1, and applied the non-negative matrix factorization (NMF) method to analyze the heterogeneity of circulating metabolomics in non-small cell lung cancer, and divided the patients into two subtypes with significantly different metabolic characteristics. Subtype I was characterized by arginine and proline metabolism, glycerophospholipid metabolism, purine metabolism, and pyrimidine metabolism ( Figure 1 A), subtype II is characterized by elevated biosynthesis of unsaturated fatty acids ( Figure 1 A and 1B). Metabolic group subtype II had a worse OS than subtype I (P = 0.033, Figure 1C). Meanwhile, in NSCLC patients with central nervous system metastasis (n=66), the survival of subtype II patients was significantly shorter than that of subtype I patients (P=0.0051, Figure 1 D). In patients with EGFR exon 19 deletion or EGFR L858R mutation, subtype II patients also had worse OS than subtype I patients (P = 0.069, P = 0.096; Figure 1 E).
[0052] Then, a model involving a random forest algorithm was built to simplify metabolomics subtypes. Cohort 1 was divided into a discovery cohort (n = 97) and a validation cohort (n = 42). Based on the mean reduced Gini value selected in the discovery cohort, the model was constructed using metabolites with a mean reduced Gini value greater than 0.5 ( Figure 2 A), the area under the curve (AUC) of the random forest model in the training data set for predicting subtypes I and II is 1.000, and the AUC of the model in the test data set for predicting subtypes I and II is 0.999 ( Figure 2 B), indicating that the random forest model can effectively predict metabolic subtypes. In cohort 1, the OS of subtype II patients predicted by the model was significantly shorter than that of subtype I patients (P = 0.035, Figure 2 C). The efficacy prediction effect of the random forest model on metabolomics subtypes was further verified in the external validation cohort (cohort 2, n = 47). The OS of the two subtypes was still significantly different (P = 0.032, Figure 2 D). At the same time, in the external validation cohort, it was found that fatty acids in subtype II were still upregulated compared with subtype I ( Figure 2 E).
Claims
1. A marker for evaluating the prognosis of third-generation EGFR-TKI drug treatment for advanced non-small cell lung cancer with T790M mutation, the marker comprising glycitein sulfate, N-(2-furoyl)glycine, N-oleoyltaurine, hexanoylglutamine, 3-hydroxyoleic acid, myristoleic acid (14:1n5), oleic acid (18:1), palmitic acid (16:0), palmitoleic acid (16:1n7), eicosenoic acid (20:1), docosadienoic acid (22:2n6), henicosopentaenoic acid (21:5n3), 3-hydroxybutyric acid, 3-hydroxybutyryl myristoleic acid Carnitine, caproylcarnitine (C6), 3-hydroxydecanoylcarnitine, 3-hydroxyoleoylcarnitine, 5-dodecenoylcarnitine (C12:1), decanoylcarnitine (C10), lauroylcarnitine (C12), myristoylcarnitine (C14:1), myristoylcarnitine (C14), octanoylcarnitine (C8), oleoylcarnitine (C18:1), palmitoleoylcarnitine (C16:1), 1-linoleoyl-GPE (18:2), 1,2-dilinoleoyl-GPE (18:2 / 18:2), 1-oleoyl-2-linoleoyl-GPE (18:1 / 18:2), or any combination thereof.
2. Use of a reagent for detecting the marker according to claim 1 in the preparation of a kit for evaluating the prognosis of third-generation EGFR-TKI drug treatment for advanced non-small cell lung cancer with T790M mutation.
3. A kit for evaluating the prognosis of third-generation EGFR-TKI drug treatment for advanced non-small cell lung cancer with T790M mutation, the kit comprising a reagent for detecting the marker according to claim 1.
4. The marker according to claim 1 or the use according to claim 2 or the kit according to claim 3, wherein the marker is derived from the plasma of a subject before receiving the drug.
5. The marker according to claim 1 or the use according to claim 2 or the kit according to claim 3, wherein the marker comprises or is: glycitein sulfate, N-(2-furoyl)glycine, N-oleoyltaurine, hexanoylglutamine, 3-hydroxyoleic acid, myristoleic acid (14:1n5), oleic acid (18:1), palmitic acid (16:0), palmitoleic acid (16:1n7), eicosenoic acid (20:1), docosadienoic acid (22:2n6), henicopentaenoic acid (21:5n3), 3-hydroxybutyric acid, 3-hydroxy The invention relates to a combination of butylcarnitine, caproylcarnitine (C6), 3-hydroxydecanoylcarnitine, 3-hydroxyoleoylcarnitine, 5-dodecenoylcarnitine (C12:1), decanoylcarnitine (C10), lauroylcarnitine (C12), myristoylcarnitine (C14:1), myristoylcarnitine (C14), octanoylcarnitine (C8), oleoylcarnitine (C18:1), palmitoleoylcarnitine (C16:1), 1-linoleoyl-GPE (18:2), 1,2-dilinoleoyl-GPE (18:2 / 18:2), and 1-oleoyl-2-linoleoyl-GPE (18:1 / 18:2).
6. The marker according to claim 1 or the use according to claim 2 or the kit according to claim 3, wherein the third-generation EGFR-TKI drug comprises osimertinib, almonertinib, furmonertinib, rezivertinib or a combination thereof.
7. The use according to claim 2 or the kit according to claim 3, wherein the kit is used for marker quantification to evaluate the prognosis of third-generation EGFR-TKI drug treatment for advanced non-small cell lung cancer with T790M mutation.
8. The use or kit according to claim 7, wherein the marker identification is performed based on the marker standard by ultra-performance liquid chromatography / tandem mass spectrometry (UPLC-MS / MS) method, and the marker quantification is performed by calculating the area under the chromatographic peak curve.
9. The use or kit according to claim 8, wherein the prognosis of third-generation EGFR-TKI drug treatment for advanced non-small cell lung cancer with T790M mutation is evaluated based on marker quantification by random forest algorithm.
10. The use or kit according to claim 9, wherein when the probability of poor prognosis obtained by the random forest algorithm is greater than the probability of good prognosis, the subject has a poor prognosis for drug treatment; or when the probability of poor prognosis obtained by the random forest algorithm is less than the probability of good prognosis, the subject has a good prognosis for drug treatment.