Tissue metabolism fingerprint detection method based on solid-phase mass spectrum, application of tissue metabolism fingerprint detection method in gastric cancer diagnosis and prognosis model and metabolite combination

Abstract

The invention discloses a tissue metabolism fingerprint detection method based on solid-phase mass spectrometry, application of the tissue metabolism fingerprint detection method in a gastric cancer diagnosis prognosis model and a metabolite combination, and relates to the field of molecular detection.Metabolite is extracted through methyl alcohol, dewaxed through heating, in-situ captured through nano-particles, sample application is conducted through a sample priority strategy, and the tissue metabolism fingerprint detection method based on solid-phase mass spectrometry is obtained. The invention develops an FFPE tissue metabolism fingerprint rapid analysis technology based on nano-particle enhanced laser desorption ionization mass spectrometry (NPELDI-MS). On the basis of the optimized tissue metabolism fingerprint extraction method in combination with a machine learning model, tissue metabolism fingerprint combined machine learning feature screening and diagnosis, typing and prognosis model construction are realized, and a high-efficiency, low-sample-consumption and low-cost metabolism fingerprint analysis technology is provided; and integrated application of cancer diagnosis, subtype differentiation and prognosis prediction is realized.

Description

Technical Field

[0001] The present invention relates to the field of molecular detection, and in particular to a tissue metabolic fingerprint detection method based on solid-phase mass spectrometry, its application in a gastric cancer diagnosis and prognosis model, and a metabolite combination. Background Art

[0002] Histopathological assessment, the gold standard for cancer diagnosis, identifies cancerous changes by observing tissue morphological features under a microscope. However, its results are highly dependent on the pathologist's experience and are subject to significant inter-observer variability, which may lead to the risk of missed or misdiagnosis. In addition, molecular diagnostic methods such as immunohistochemistry (IHC) and fluorescence in situ hybridization (FISH) provide a basis for treatment strategies by quantitatively analyzing specific biomarkers such as HER2. However, these methods rely on complex labeling and staining procedures, which are cumbersome and time-consuming. Although metabolites (molecular weight <1000Da) can directly reflect pathological status and can be detected through label-free methods, existing metabolic analysis technologies still face technical bottlenecks.

[0003] Mass spectrometry (MS) is a core tool for metabolite analysis, which can achieve high-resolution (ppm level) analysis by detecting the mass-to-charge ratio (m / z) of metabolites. However, traditional MS requires the combination of chromatographic techniques (such as liquid chromatography or gas chromatography) for metabolite enrichment and purification, resulting in a single sample analysis time of several hours and a large sample consumption (e.g., approximately 2.36 mm per test). 3 Although nanoparticle-enhanced laser desorption ionization mass spectrometry (NPELDI-MS) selectively captures metabolites through nanoparticle solid phase, simplifying the pre-processing process and reducing sample requirements, its application in directly extracting tissue metabolic fingerprints (TMFs) is still immature, limiting its promotion in clinical pathology-assisted diagnosis.

[0004] In existing technologies, the development of metabolic markers is mostly based on plasma samples. Although they have the advantage of being non-invasive, they cannot directly reflect the local metabolic characteristics of the tumor microenvironment. In addition, traditional pathological diagnosis and molecular typing rely on multi-step detection (such as morphological evaluation combined with IHC / FISH). The process is complicated and cannot meet the clinical needs of fast and low sample consumption. Especially for institutions with limited medical resources, there is an urgent need for an integrated technology that can simultaneously achieve accurate diagnosis, molecular typing and prognosis evaluation based on a small amount of FFPE tissue (routine clinical preservation samples) while avoiding subjective interpretation errors.

[0005] Therefore, those skilled in the art are committed to establishing an efficient, low-sample-consumption, and low-cost metabolic fingerprint analysis technology to directly capture the metabolic characteristics in FFPE tissues and realize the integrated application of cancer diagnosis, subtype differentiation, and prognosis prediction. Summary of the Invention

[0006] In view of the above-mentioned defects of the prior art, the technical problem to be solved by the present invention is to provide a high-efficiency, low-sample-consumption, and low-cost metabolic fingerprint analysis technology, and to realize the integrated application of cancer diagnosis, subtype differentiation, and prognosis prediction.

[0007] To achieve the above object, the present invention provides a method for detecting tissue metabolic fingerprints based on solid-phase mass spectrometry, which is characterized by including the following steps:

[0008] Step 1: Place the tissue sample in a methanol solution to release metabolites;

[0009] Step 2: Heat the sample to dewax;

[0010] Step 3: Ice-bath the sample;

[0011] Step 4: Centrifuge the sample to remove wax;

[0012] Step 5: Collect the supernatant as a sample;

[0013] Step 6: Perform sample loading using the sample-first spotting method;

[0014] Step 7: Perform mass spectrometry detection to obtain the metabolic fingerprint of the tissue sample.

[0015] In a preferred embodiment of the present invention, 80% methanol is used in Step 1.

[0016] In another preferred embodiment of the present invention, in Step 2, the sample is specifically heated at 70 °C.

[0017] In another preferred embodiment of the present invention, in Step 6, inorganic nanoparticles are specifically prepared into a nanoparticle matrix solution with a concentration of 1 mg / mL using deionized water. First, each extracted sample is spotted on a mass spectrometry target plate and dried at room temperature; then, the nanoparticle matrix solution is spotted on the dried sample and dried at room temperature.

[0018] The present invention also provides the application of the above method in constructing a gastric cancer diagnosis and typing model, which is characterized in that first, the metabolic fingerprints of the tissues of gastric cancer patients are preprocessed; then, the collected metabolic fingerprint data is divided into a training set and a test set, and cross-validation is used to optimize the parameters and train the model for the machine learning algorithm on the training set to obtain the performance of the algorithm on the training set, and the trained model is used to make predictions on the test set to obtain the performance of the algorithm on the test set, thereby obtaining a gastric cancer diagnosis and typing model.

[0019] In a preferred embodiment of the present invention, the tissue samples used for collecting metabolic fingerprints are paired cancer tissues of gastric cancer patients, namely GCT, and adjacent normal tissues, namely ANT.

[0020] In another preferred embodiment of the present invention, the preprocessing includes data resampling, spectral line smoothing, baseline correction, spectral peak alignment, and missing value filling to obtain m / z signals.

[0021] In another preferred embodiment of the present invention, 5-fold cross-validation with 20 repetitions is specifically used to optimize the parameters and train the model of the machine learning algorithm on the training set.

[0022] The present invention also provides an application of the above method in constructing a prognostic prediction model for gastric cancer metabolites, which is characterized in that gastric cancer patient samples with complete survival information are randomly assigned to a discovery group and a validation group. Univariate Cox regression analysis is performed on the metabolic fingerprint data set of the discovery group to screen metabolic features significantly related to the overall survival of the patients. Then, through multivariate Cox regression analysis, metabolites independently related to the overall survival are selected, and a prognostic prediction TMP scoring model for gastric cancer is constructed using the selected metabolites.

[0023] The present invention also provides a metabolite combination in the above prognostic prediction model for gastric cancer metabolites, which is characterized in that the metabolite combination consists of malondialdehyde, a microbial co-metabolite benzoic acid, an indoleacetic acid derivative of tryptophan, hexadecanamide ethanol, and stearic acid, which are oxidative stress markers.

[0024] Technical effects

[0025] 1. The present invention has developed a rapid analysis technology for FFPE tissue metabolic fingerprints based on nanoparticle-enhanced laser desorption ionization mass spectrometry (NPELDI-MS).

[0026] Metabolites are captured by nanoparticles in solid phase: Fe3O4 nanoparticles (with a pore size of 3.5 nm) are designed to selectively enrich small molecule metabolites (<1000 Da), and are directly ionized by laser desorption without chromatographic separation.

[0027] Standardization of core components: Large-scale synthesis of nanoparticles: Fe3O4 nanoparticles are synthesized by the solvothermal method, with mature technology, controllable cost, high crystallinity (verified by XRD) and uniform morphology (verified by SEM), suitable for mass production.

[0028] Automation of chip microarray: A 384-sample / chip design is adopted, supporting high-throughput detection (3 seconds / sample), and compatible with the upgrade of existing mass spectrometry equipment.

[0029] Optimized workflow: Wax removal by heating at 70°C (TIC increased by 60%), methanol / water (80 / 20) extractant (extraction efficiency increased by 30%), and sample priority spotting method are adopted to achieve high repeatability (CV < 15%) and rapid detection (3 s / sample). Label-free and stain-free, the single detection time is shortened to 40 minutes (traditional IHC takes 3 days).

[0030] Sample requirement is extremely low: only 0.016 mm is needed for single detection 3 of FFPE tissue (2.36 mm is needed for traditional mass spectrometry 3 ), meeting the analysis requirements of clinical micro-samples (such as needle biopsy).

[0031] Rapid detection: The complete workflow (dewaxing, extraction, detection) takes ~40 minutes / sample, with significantly improved efficiency compared to traditional IHC (3 days) and chromatography-mass spectrometry (several hours).

[0032] Low reagent consumption: No high-cost consumables such as chromatographic columns and labeled antibodies are required.

[0033] Low equipment modification cost: Based on the upgrade of commercial laser desorption mass spectrometers, there is no need to purchase new equipment.

[0034] Strong compatibility: Directly analyze FFPE tissue (clinical routine sample), with a similarity of 0.94 (Pearson correlation) to the metabolic profile of frozen tissue, avoiding the need for cryopreservation.

[0035] Diagnosis, molecular typing (HER2), and prognosis assessment are achieved synchronously in a single detection, replacing traditional multi-step detections (morphology + IHC / FISH + TNM staging), reducing the comprehensive cost.

[0036] 2. Combining the above optimized tissue metabolic fingerprint extraction method with a machine learning model to achieve synchronous completion of the diagnosis and typing of pathological tissues in a single detection. Tissue metabolic fingerprint combined with machine learning feature screening and model construction. The AUC for gastric cancer diagnosis reaches 0.979 (95% CI: 0.921 - 0.998), and the sample consumption is reduced to 0.016 mm 3 / time (1 / 150 of the traditional method). The AUC for distinguishing HER2 positive / negative in gastric cancer reaches 0.963 (95% CI: 0.917 - 1.000).

[0037] Convenient data integration: The machine learning model can be encapsulated as a software module and embedded in the hospital LIS (Laboratory Information System) or third-party detection platform to achieve automated analysis of "sample in - result out", seamlessly connecting with the clinical process.

[0038] 3. Develop a tissue metabolic prognosis (TMP) scoring system to predict the survival risk of patients based on the Cox regression model of metabolic markers. Screen metabolites significantly related to overall survival (such as downregulated hydroxy pyruvate) and construct the TMP scoring formula. Through Kaplan-Meier analysis, the median survival time in the high TMP group is significantly lower than that in the low TMP group (p < 0.05).

[0039] 4. Industrial application prospects

[0040] Clinical diagnosis market

[0041] Primary healthcare promotion: The technology is independent of pathologists' interpretation and is applicable to resource - scarce areas.

[0042] Precision medicine scenario: HER2 typing guides targeted therapy (such as trastuzumab), and the prognostic scoring system optimizes chemotherapy regimens, promoting individualized treatment of gastric cancer.

[0043] Research and drug development

[0044] Mechanism research tools: Combining metabolomic and proteomic data (such as abnormal PPP pathway) to accelerate the analysis of the mechanism of cancer metabolic reprogramming and assist in target discovery.

[0045] Efficacy monitoring: Dynamically monitor the changes in the metabolic profile before and after treatment to evaluate drug response and drug resistance.

[0046] Third - party testing services

[0047] Commercial testing packages: Develop an integrated testing kit for "gastric cancer diagnosis - typing - prognosis", targeting hospitals, physical examination centers and research institutions, with great market potential.

[0048] Industry trend support

[0049] AI + medical integration: Combining metabolic fingerprints with machine learning, in line with the trend of the industrialization of medical AI and easily favored by capital.

[0050] The following will further illustrate the concept, specific structure and technical effects of the present invention in conjunction with the accompanying drawings to fully understand the purpose, features and effects of the present invention. Brief description of the drawings

[0051] Figure 1 It is a schematic diagram of the sample fingerprint detection step of a preferred embodiment of the present invention;

[0052] Figure 2 It is a schematic diagram of the similarity of the metabolic profiles between FFPE tissues (clinical routine samples) and frozen tissues in a preferred embodiment of the present invention;

[0053] Figure 3 It is a schematic diagram of the sample fingerprint map of a preferred embodiment of the present invention;

[0054] Figure 4 It is a schematic diagram of the diagnostic AUC results of cancer and adjacent tissues based on the metabolic fingerprint of gastric cancer tissues in a preferred embodiment of the present invention;

[0055] Figure 5 It is a schematic diagram of the AUC results of HER2 discrimination based on the metabolic fingerprint of gastric cancer tissues in a preferred embodiment of the present invention;

[0056] Figure 6Schematic diagram of the scoring results of the metabolite prognosis prediction (TMP) scoring system for gastric cancer prognosis prediction in a preferred embodiment of the present invention. Detailed implementation manners

[0057] The following introduces multiple preferred embodiments of the present invention with reference to the accompanying drawings of the specification to make its technical content clearer and easier to understand. The present invention can be embodied in many different forms of embodiments, and the protection scope of the present invention is not limited to the embodiments mentioned in the text.

[0058] As Figure 1 shown, the tissue metabolic fingerprint detection method based on solid-phase mass spectrometry of the present invention

[0059] Step 1: Place the tissue sample in 80% methanol to release metabolites;

[0060] Step 2: Heat the sample at 70°C for 20 minutes to dewax;

[0061] Step 3: Ice-bath the sample for 10 minutes;

[0062] Step 4: Centrifuge the sample at 4°C and 10,000 rpm for 10 minutes to remove wax;

[0063] Step 5: Collect the supernatant as a sample;

[0064] Step 6: Spot the sample according to the sample priority strategy;

[0065] Step 7: Use enhanced laser desorption ionization mass spectrometry NPELDI-MS for mass spectrometry detection to obtain the metabolic fingerprint of the sample.

[0066] Example 1: Rapid detection of FFPE (clinical routine samples) tissue metabolic fingerprints based on NPELDI-MS

[0067] Tissue dewaxing optimization: Comparing three dewaxing methods, non-dewaxing, xylene dewaxing, and 70°C heating dewaxing, the 70°C heating method was selected, and the total ion current TIC was significantly increased, p<0.05. (Non-dewaxing 2.25×10 7 , xylene dewaxing 1.87×10 7 , 70°C heating dewaxing 3.00×10 7 ),

[0068] Metabolite extraction: Using methanol / water (80 / 20, v / v) extractant, the TIC reached 3.07×10 7 , which was significantly better than other solvents (see Table 1 for details, p<0.05).

[0069] Table 1 TIC of mass spectrometry detection of FFPE with different extraction solvents

[0070]

[0071] Sample loading: The "sample-first" spotting method (3.9×10 7 TIC, a 70% improvement over the matrix-first method) is adopted. Among them, for the chip microarray automation, a 384-sample / chip design is used to support high-throughput detection (3 seconds / sample). During spotting, specifically, inorganic nanoparticles are prepared into a nanoparticle matrix solution of 1 mg / mL with deionized water; spotting is carried out on the mass spectrometry target plate, 1 μL of each extracted tissue sample is spotted, and dried at room temperature; then, 1 μL of the nanoparticle matrix solution is spotted on the dried tissue sample and dried at room temperature.

[0072] Among them, the inorganic nanoparticles are Fe3O4 nanoparticles. The Fe3O4 nanoparticles (pore size 3.5 nm) are designed to selectively enrich small molecule metabolites (<1000 Da). Through laser desorption direct ionization, without chromatographic separation, the Fe3O4 nanoparticles are synthesized by the solvothermal method. The nanoparticles selectively capture small molecule metabolites (<1000 Da) through pore size, and laser desorption direct ionization is used to avoid chromatographic separation. High-temperature dewaxing reduces the interference of paraffin, and the methanol / water extractant can efficiently dissolve polar metabolites.

[0073] The single-sample detection time is 40 minutes, and the sample consumption is 0.016 mm 3 (1 / 150 of the traditional method). It is highly consistent with the metabolic profile of frozen tissues (Pearson correlation) (cosine similarity 0.88 - 0.94, as Figure 2 shown), avoiding the need for cryopreservation.

[0074] Example 2: Construction of a diagnostic and typing model based on the metabolic fingerprint of gastric cancer FFPE tissues

[0075] Metabolites were extracted and detected from a total of 284 tissues, including paired cancer tissues (GCT) and adjacent tissues (ANT) from 142 gastric cancer patients. The obtained metabolic fingerprints were preprocessed, including data resampling, spectral line smoothing, baseline correction, spectral peak alignment, and missing value filling, to obtain m / z signals, as Figure 3 (cancer tissue GCT, adjacent tissue ANT) shown.

[0076] The tissue sample data collected was divided into a training set and a test set. The machine learning algorithm was optimized for parameters and model training on the training set using 20-fold repeated 5-fold cross-validation to obtain the performance of the algorithm on the training set. The trained model was used to make predictions on the test set to obtain the performance of the algorithm on the test set, as Figure 4 shown.

[0077] Based on the tissue metabolic fingerprint, cancer tissues and adjacent tissues were distinguished, corresponding to Figure 4Results: For the training set, the AUC was 0.999 (95% CI: 0.981 - 1.000), and for the test set, the AUC was 0.979 (95% CI: 0.921 - 0.998).

[0078] Subtyping: A subtyping model was constructed based on 29 HER2-positive tissues confirmed by immunohistochemistry (IHC 3+ or IHC 2+ combined with positive FISH) in gastric cancer tissue samples (GCT). The positive cases were randomly divided into a training set and a test set at a ratio of 7:3, and an algorithm model was constructed based on tissue metabolic fingerprints (TMFs) to distinguish HER2-positive and negative subtypes. The results showed (as Figure 5 shown) that the AUC of this model in the discovery set reached 0.963 (95% CI: 0.917 - 1.000), and the AUC of the validation set was 0.945 (95% CI: 0.867 - 1.000), indicating that the tissue metabolic fingerprint model can accurately identify HER2-positive gastric cancer, providing an auxiliary decision-making basis at the metabolic level for the selection of clinical treatment regimens (such as trastuzumab targeted therapy).

[0079] Example 3: Construction of a prognostic prediction model based on gastric cancer tissue metabolic fingerprints

[0080] The present invention developed a metabolite prognostic prediction (TMP) scoring system for gastric cancer prognosis prediction. The included patients were randomly assigned to the discovery group or the validation group at a ratio of 3:1 for further analysis. In the discovery cohort, univariate Cox regression analysis was first performed to identify metabolite characteristics associated with overall survival. Subsequently, through multivariate Cox regression analysis, metabolites independently associated with overall survival were further selected and used to construct TMP. There were significant metabolic differences between the high TMP group and the low TMP group, and the specific method was as follows:

[0081] By integrating the data of 142 gastric cancer patients with complete survival information, the samples were randomly divided (7:3) into a discovery set (n = 99) and a validation set (n = 43), and a gastric cancer prognosis prediction model (TMP scoring system) based on tissue metabolic fingerprints (TMFs) was constructed. First, in the discovery set, univariate Cox regression was used to screen metabolic features significantly associated with the overall survival of patients, and metabolic features with p < 0.2 were screened. Subsequently, through multivariate Cox regression analysis, with a threshold of p < 0.05, 5 key metabolic features, namely metabolites, were determined, and the TMP score was established. The 5 metabolites screened were malondialdehyde, an oxidative stress marker; benzoic acid, a microbial co-metabolite; indoleacetic acid, a tryptophan derivative; hexadecanamide ethanol; and stearic acid. After dividing the patients into high and low TMP groups according to the median score, Kaplan-Meier survival analysis showed that the median survival time of the low TMP group was significantly better than that of the high TMP group (log-rank test, p < 0.05), verifying the prognostic discrimination efficacy of the model (as Figure 6 shown).

[0082] The 5 annotated metabolites in the model are involved in oxidative regulation, microbiota interaction, and lipid metabolism pathways, suggesting their potential biological associations with gastric cancer progression. This model provides a new tool at the metabolic level for individualized prognostic assessment of gastric cancer.

[0083] The preferred specific embodiments of the present invention have been described in detail above. It should be understood that those of ordinary skill in the art can make many modifications and variations based on the concept of the present invention without creative labor. Therefore, all technical solutions that can be obtained by those skilled in the art in the technical field of the present invention through logical analysis, reasoning, or limited experiments based on the concept of the present invention on the basis of the prior art should fall within the protection scope determined by the claims.

Claims

1. A method for detecting tissue metabolic fingerprints based on solid-phase mass spectrometry, characterized in that It includes the following steps: Step 1: Place the tissue sample in a methanol solution to release metabolites; Step 2: Heat the sample to dewax; Step 3: Ice-bath the sample; Step 4: Centrifuge the sample to remove wax; Step 5: Collect the supernatant as a sample; Step 6: Perform sample loading using the sample first spotting method; Step 7: Conduct mass spectrometry detection to obtain the metabolic fingerprint of the tissue sample.

2. The method according to claim 1, characterized in that, 80% methanol is used in Step 1.

3. The method according to claim 1, characterized in that In Step 2, specifically, the sample is heated at 70 °C.

4. The method according to claim 1, wherein In Step 6, specifically, inorganic nanoparticles are made into a nanoparticle matrix solution with deionized water. First, each extracted sample is spotted on a mass spectrometry target plate and dried at room temperature; then, the nanoparticle matrix solution is spotted on the dried sample and dried at room temperature.

5. Use of the method according to any one of claims 1-4 in constructing a gastric cancer diagnosis and typing model, characterized in that, First, preprocess the metabolic fingerprint of the tissue obtained from gastric cancer patients; then divide the collected metabolic fingerprint data into a training set and a test set, optimize the parameters and train the model of the machine learning algorithm on the training set using cross-validation to obtain the performance of the algorithm on the training set, and predict the trained model on the test set to obtain the performance of the algorithm on the test set, thus obtaining a gastric cancer diagnosis and typing model.

6. The application according to claim 5, characterized in that The tissue samples used for collecting metabolic fingerprints are paired cancer tissues of gastric cancer patients, namely GCT, and adjacent cancer tissues, namely ANT.

7. The application according to claim 5, characterized in that The preprocessing includes data resampling, spectral line smoothing, baseline correction, spectral peak alignment, and missing value filling to obtain the m / z signal.

8. The application according to claim 5, characterized in that, Specifically, 5-fold cross-validation with 20 repetitions is used to optimize the parameters and train the model of the machine learning algorithm on the training set.

9. Use of the method according to any one of claims 1-4 in constructing a prognostic prediction model for gastric cancer metabolites, characterized in that, Randomly assign gastric cancer patient samples with complete survival information to a discovery group and a validation group. Perform univariate Cox regression analysis on the metabolic fingerprint data set of the discovery group to screen metabolic features significantly associated with the overall survival of patients, and then through multivariate Cox regression analysis, select metabolites independently associated with the overall survival, and use the selected metabolites to construct a gastric cancer prognosis prediction TMP scoring model.

10. The metabolite combination in the gastric cancer metabolite prognosis prediction model according to claim 9, characterized in that, The metabolite combination consists of malondialdehyde, an oxidative stress marker, benzoic acid, a microbial co-metabolite, indoleacetic acid, a tryptophan derivative, hexadecanamide ethanol, and stearic acid.

Images