A kit for RT-qPCR-assisted diagnosis of gastric cancer based on eight microRNA fluorescent probes

Abstract

The present invention provides a kit for assisting the diagnosis of gastric cancer based on 8 kinds of microRNA fluorescent probes RT-qPCR, and belongs to the technical field of disease diagnostic reagents. The present invention provides a kit for assisting the diagnosis of gastric cancer based on 8 kinds of microRNA fluorescent probes RT-qPCR, including a real-time fluorescence quantitative PCR detection primer probe set and an internal reference primer; the 8 kinds of microRNAs include miR-223-3p, miR-186-5p, miR-18a-5p, miR-106b-5p, miR-20a-5p, miR-451a, miR-7976 and miR-151a-3p. The kit not only has the characteristics of high detection repeatability, but also has good detection specificity, which is relatively consistent with the clinical gold standard diagnosis results, and can be used for clinical auxiliary diagnosis of gastric cancer.

Description

Technical Field

[0001] The present invention belongs to the technical field of disease diagnostic reagents, and specifically relates to a kit for assisting in the diagnosis of gastric cancer based on 8 microRNA fluorescent probes RT-qPCR. Background Art

[0002] Gastric cancer has the characteristics of high incidence and low detection rate. At present, the diagnosis method of gastric cancer is to combine the patient's clinical manifestations, through gastroscopy and gastric mucosal biopsy, which is the gold standard for gastric cancer diagnosis.

[0003] Studies have shown that microRNA plays an important regulatory role in the occurrence and development of tumors. Malignant tumors are the result of the interaction between genetic and environmental factors, and environmental factors play a greater role. MicroRNA is a large class of regulatory factors between the changing environment and the unchanging genetic material, and is the main bridge connecting environmental factors and genetic factors. Therefore, microRNA may be an ideal biomarker for malignant tumors. In addition, the expression level of some microRNAs in the blood can change significantly during the lesion period before cancer formation; microRNA is very stable in the blood, and plasma microRNA is relatively stable under harsh conditions such as freezing and thawing, high temperature storage (up to 37°C), acidic conditions and RNAse digestion. Compared with protein markers and mRNA expression profiles, abnormal expression of microRNA appears earlier and can more accurately distinguish tumor types, which is more beneficial for early diagnosis and more suitable as a tumor diagnostic marker.

[0004] At present, the research on miRNA as a biomarker for gastric cancer is still in the laboratory research stage at home and abroad, and has not been successfully applied to the clinical diagnosis of gastric cancer. There are two main methods for detecting miRNAs: quantitative PCR technology and gene chip technology. Quantitative PCR technology is currently the most widely used and economical method for detecting miRNAs. However, due to the disadvantages of short miRNA sequences and low levels in the blood, it is very difficult to accurately detect miRNAs. In so many studies, the research results of circulating miRNAs have extremely low reproducibility. The test results of different laboratories are not comparable, and may even produce opposite results. For example, a review of 11 studies showed that of the 31 miRNAs associated with heart failure identified in one study, only 5 miRNAs could be repeated in another study, but no miRNA could be repeated in more than two studies (Romaineetal.2015). Therefore, miRNA has not been successfully used in the clinical diagnosis of cancer. It fully shows that the existing miRNA-qPCR detection technology has serious defects, and it must be subversively innovated to achieve its successful application in clinical practice. Summary of the invention

[0005] In view of this, the purpose of the present invention is to provide a kit for RT-qPCR-assisted diagnosis of gastric cancer based on 8 microRNA fluorescent probes, which not only has the characteristics of high repeatability, but also has good detection specificity and sensitivity, and has significant clinical application value.

[0006] The present invention provides a kit for RT-qPCR-assisted diagnosis of gastric cancer based on 8 kinds of microRNA fluorescent probes, including a real-time fluorescent quantitative PCR detection primer probe set and an internal reference primer;

[0007] The real-time fluorescence quantitative PCR detection primer probe set includes a universal probe, a universal reverse primer, and a forward primer for detecting at least one of the following eight microRNAs: miR-223-3p, miR-186-5p, miR-18a-5p, miR-106b-5p, miR-20a-5p, miR-451a, miR-7976, and miR-151a-3p;

[0008] The nucleotide sequence of the universal probe is shown in SEQ ID NO: 1;

[0009] The nucleotide sequence of the universal reverse primer is shown in SEQ ID NO: 2;

[0010] The nucleotide sequence of the forward primer for detecting miR-223-3p is shown in SEQ ID NO: 3;

[0011] The nucleotide sequence of the forward primer for detecting miR-186-5p is shown in SEQ ID NO: 4;

[0012] The nucleotide sequence of the forward primer for detecting miR-18a-5p is shown in SEQ ID NO: 5;

[0013] The nucleotide sequence of the forward primer for detecting miR-106b-5p is shown in SEQ ID NO: 6;

[0014] The nucleotide sequence of the forward primer for detecting miR-20a-5p is shown in SEQ ID NO: 7;

[0015] The nucleotide sequence of the forward primer for detecting miR-451a is shown in SEQ ID NO: 8;

[0016] The nucleotide sequence of the forward primer for detecting miR-7976 is shown in SEQ ID NO:9;

[0017] The nucleotide sequence of the forward primer for detecting miR-151a-3p is shown in SEQ ID NO: 10;

[0018] The nucleotide sequences of the internal reference primers are shown in SEQ ID NO: 15 to SEQ ID NO: 44.

[0019] Preferably, the 5' end of the universal probe is modified with a fluorescent dye group, and the 3' end of the universal probe is modified with a quencher group.

[0020] Preferably, it comprises 2×Boostmix enzyme and real-time fluorescence quantitative PCR reaction solution.

[0021] Preferably, polyadenylation and cDNA amplification reagents and cDNA enrichment reagents are also included.

[0022] Preferably, the polyadenylation and cDNA amplification reagents include universal reverse transcription primers and reverse transcription amplification premix.

[0023] Preferably, the nucleotide sequence of the universal reverse transcription primer is shown in SEQ ID NO: 11;

[0024] The reverse transcription amplification premix comprises at least one of the following: 5×RT buffer, ATP, inhibitor, polyA enzyme, reverse transcriptase and stabilizer.

[0025] Preferably, the cDNA enrichment reagent includes hybridization primers, exonuclease I, chloroform, cDNA enrichment primer I, enhancer, 2×Boostmix enzyme, UDG enzyme, and cDNA enrichment primer II.

[0026] Preferably, the nucleotide sequence of the cDNA enrichment primer 1 is shown in SEQ ID NO: 12;

[0027] The nucleotide sequence of the cDNA enrichment primer II is shown in SEQ ID NO: 13;

[0028] The nucleotide sequences of the hybridization primers are shown in SEQ ID NO:14 and SEQ ID NO:53 to SEQ ID NO:149.

[0029] The present invention provides a diagnostic system for assisting in the diagnosis of gastric cancer, comprising:

[0030] The first module is used to receive the original data of fluorescent quantitative PCR of the sample to be tested based on the kit described in the above technical solution and the information of the patient from whom the sample to be tested comes;

[0031] The second module is used to calculate the gastric cancer probability and non-gastric cancer probability of the sample to be tested based on the fluorescent quantitative PCR raw data transmitted by the first module;

[0032] The third module: determines the diagnosis result of the sample to be tested based on the calculation result of the second module; and

[0033] The fourth module is used to output the patient information of the source of the sample to be tested in the first module and the diagnosis result of the sample to be tested in the third module to the terminal.

[0034] Preferably, the calculation method in the second module includes a support vector machine machine learning algorithm.

[0035] The present invention provides a kit for RT-qPCR-assisted diagnosis of gastric cancer based on 8 microRNA fluorescent probes, including a real-time fluorescent quantitative PCR detection primer probe set and an internal reference primer. The present invention first studies and analyzes the microRNA expression of cancer tissues and plasma samples of non-gastric cancer patients and gastric cancer patients, and screens out microRNAs with significant differences (miR-223-3p, miR-186-5p, miR-18a-5p, miR-106b-5p, miR-20a-5p, miR-451a, miR-7976 and miR-151a-3p). Based on these 8 different microRNAs, the forward primers of qPCR are designed respectively, and the universal primers and universal probes are combined to perform Taqman fluorescence quantitative PCR amplification. At the same time, the present invention also provides internal reference primers, and uses multiple miRNAs sequences that are stably expressed in cancer tissues and plasma samples of non-gastric cancer patients and gastric cancer patients to design internal reference primers to calibrate the expression of other miRNAs, thereby reducing the variation in miRNA detection. It can be seen that the kit provided by the present invention is obtained by integrating the design of polyadenylation RT-qPCR and Taqman technology, and has high repeatability. More than 2,000 gastric cancer and non-gastric cancer clinical samples were detected with the kit. After comparing the test results with the clinical gold standard diagnosis results, the specificity was 90.8% and the sensitivity was 90.7%. By comparing the test results with the three cancer markers (CEA, CA125, CA72-4) for auxiliary diagnosis of cancer, it was found that the results were highly consistent, indicating that this kit has significant clinical application value. The RT-qPCR gastric cancer detection kit using 8 microRNA fluorescent probes can be used for clinical auxiliary diagnosis of gastric cancer. It has the characteristics of high sensitivity, high specificity, and high accuracy, and can significantly improve the detection rate of gastric cancer patients. At the same time, the kit is an in vitro diagnostic product, and only 2 ml of peripheral blood is needed for detection. The detection method is quick and simple, and can improve the detection efficiency, avoiding invasive damage and discomfort to patients caused by invasive examinations such as gastroscopy (biopsy). This kit can be used in conjunction with the existing equipment and instruments of the hospital's PCR laboratory, without the need to purchase new equipment, thereby reducing the hospital's use costs; it can cover groups with infectious diseases and those who are not suitable for gastroscopy.

[0036] Furthermore, the present invention also specifically defines that the fluorescent quantitative PCR detection reagent also includes a 2×Boost mix enzyme. Since PCR is an exponential amplification, the DNA template can be amplified by tens of billions of times. The existence of non-specific amplification is fatal to quantitative PCR. Unfortunately, the hot-start polymerase commonly used on the market is not a complete hot start, which can lead to serious non-specific amplification. The vast majority of RT-qPCR results are derived from technical noise. In order to overcome the shortcoming that the hot-start polymerase commonly used on the market is not a complete hot start, the present invention provides a new 2×USQmiR (Universal and Specific Quantification of miRNA) Mix enzyme. The reaction system of USQmiR greatly improves the specificity of PCR amplification and solves the problem of non-specific amplification in detection.

[0037] The present invention provides a diagnostic system for assisting in the diagnosis of gastric cancer, comprising a first module: for receiving the original data of fluorescence quantitative PCR of the sample to be tested based on the kit described in the above technical solution and the information of the patient from whom the sample to be tested comes; a second module: for calculating the probability of gastric cancer and the probability of non-gastric cancer of the sample to be tested based on the original data of fluorescence quantitative PCR transmitted by the first module; a third module: for determining the diagnostic result of the sample to be tested based on the calculation result of the second module; and a fourth module: for outputting the patient information from which the sample to be tested comes in the first module and the diagnostic result of the sample to be tested in the third module to a terminal. The present invention uses a computer artificial intelligence model for result analysis. The present invention first screens the miRNAs with the most significant differential expression in the plasma of patients with gastric cancer and non-gastric cancer, and then uses different numbers (1 to 27) of miRNAs to perform machine learning classification of gastric cancer and non-gastric cancer (gastric discomfort such as gastritis, gastric ulcer, gastric erosion, etc.) through 10-fold cross validation, and screens out machine learning classification based on 8 specific miRNAs with high accuracy, and screens the support vector machine (SVM) machine learning algorithm as the optimal machine learning classifier through accuracy and Kappa value. Based on the ROC curve of the verification results of the best model for diagnosing gastric cancer, the AUC area of ​​the SVM algorithm is above 92%, and the sensitivity and specificity are both above 90%. It shows that the machine learning model constructed by the present invention can effectively identify gastric cancer and non-gastric cancer (gastric discomfort such as gastritis, gastric ulcer, gastric erosion and other cancers including lung cancer, breast cancer and intestinal cancer), and the biological repeatability (using different samples) of the model is very good. BRIEF DESCRIPTION OF THE DRAWINGS

[0038] Figure 1 Flow chart of the qPCR detection technology for miRNA developed for the present invention;

[0039] Figure 2 This is a graph showing the expression level detection results of 98 miRNAs using the miRNA qPCR detection method developed by the present invention;

[0040] Figure 3 A resolution result diagram of the miRNA qPCR detection method developed by the present invention;

[0041] Figure 4 Screening computer artificial intelligence model technology flow chart for the present invention;

[0042] Figure 5 The results of machine learning classification of gastric cancer and non-gastric cancer samples using 10-fold cross validation based on different numbers of miRNAs;

[0043] Figure 6 The accuracy and Kappa value results of classifying gastric cancer and non-gastric cancer samples using five machine learning classifiers;

[0044] Figure 7 ROC curve diagram of the validation results of the final model for diagnosing gastric cancer;

[0045] Figure 8 This is a statistical result diagram of the performance parameters of the final model for the verification results of gastric cancer diagnosis;

[0046] Fig. 9 This is the ROC curve diagram of the final model validation results for machine learning dataset 1;

[0047] Fig.10 This is the ROC curve diagram of the final model validation results for machine learning dataset 2;

[0048] Fig.11 The ROC curve diagram of the final model validation results for machine learning dataset 3;

[0049] Fig.12 Schematic diagram of AImiR software operation flow and ROC curve;

[0050] Fig.13 Diagnostic report for AImiR software;

[0051] Fig.14 This is a comparison of the relative expression of miRNA in normal plasma (Normal) and hemolytic plasma (Hemolysis);

[0052] Fig.15 This is a comparison chart of the results of testing the same batch of clinical samples using the developed qPCR detection method for miRNA and the gold standard detection method. DETAILED DESCRIPTION

[0053] The present invention provides a kit for RT-qPCR-assisted diagnosis of gastric cancer based on 8 kinds of microRNA fluorescent probes, comprising a real-time fluorescent quantitative PCR detection primer probe group and an internal reference primer; the real-time fluorescent quantitative PCR detection primer probe group comprises a universal probe, a universal reverse primer and a forward primer for detecting at least one of the following 8 kinds of microRNAs: miR-223-3p, miR-186-5p, miR-18a-5p, miR-106b-5p, miR-20a-5p, miR-451a, miR-7976 and miR-151a-3p;

[0054] The nucleotide sequence of the universal probe is as shown in SEQ ID NO: 1 (ACC + AT + CA + AT + CG + TG + TG), wherein the bases with "+" are locked nucleic acids (LNA); the nucleotide sequence of the universal reverse primer is shown in SEQ ID NO: 2 (ATAGATTGAGTGGAGCGTAGGGTTTCGG);

[0055] The nucleotide sequence of the forward primer for detecting miR-223-3p is shown in SEQ ID NO: 3 (CTGTGTCAGTTTGTCAAATACCCCA);

[0056] The nucleotide sequence of the forward primer for detecting miR-186-5p is shown in SEQ ID NO: 4 (CTAACACAAAGAATTCTCCTTTTGGGCT);

[0057] The nucleotide sequence of the forward primer for detecting miR-18a-5p is shown in SEQ ID NO: 5 (ACAGAAGTTAAGGTGCATCTAGTGCAGATAG);

[0058] The nucleotide sequence of the forward primer for detecting miR-106b-5p is shown in SEQ ID NO: 6 (TCTCTCTAAAGTGCTGACAGTGCAGAT);

[0059] The nucleotide sequence of the forward primer for detecting miR-20a-5p is shown in SEQ ID NO: 7 (GAACGTAAAGTGCTTATAGTGCAGGTAG);

[0060] The nucleotide sequence of the forward primer for detecting miR-451a is shown in SEQ ID NO: 8 (ACTCCAAACCGTTACCATTACTGAGTT);

[0061] The nucleotide sequence of the forward primer for detecting miR-7976 is shown in SEQ ID NO: 9 (TGAGTGCCCTGAGACTTTTGCTC);

[0062] The nucleotide sequence of the forward primer for detecting miR-151a-3p is shown in SEQ ID NO: 10 (AGGACTAGACTGAAGCTCCTTGAGG);

[0063] The nucleotide sequences of the internal reference primers are shown in SEQ ID NO: 15 to SEQ ID NO: 44.

[0064] In the present invention, the 5' end of the universal probe is preferably modified with a fluorescent dye group, and the 3' end of the universal probe is modified with a quencher group. The present invention does not specifically limit the types of the fluorescent dye group and the quencher group, and fluorescent groups and quenchers well known in the art can be used, for example, common fluorescent groups include FAM, HEX and TET, and common quenchers include TAMRA and BHQ1.

[0065] In the present invention, the sources of the primer probe set and the internal reference primer are preferably commissioned to be synthesized by a gene synthesis company. In the embodiment of the present invention, the primer probe set and the internal reference primer are commissioned to be synthesized, modified and purified by Sangon Biotech (Shanghai) Co., Ltd.

[0066] In the present invention, the kit preferably includes 2×Boost mix enzyme and real-time fluorescence quantitative PCR reaction solution. The 2×Boost mix enzyme can preferably refer to the specific quantitative PCR reaction mixture in Example 1 recorded in the patent No. ZL 201910219827.4, the patent name of which is a specific quantitative PCR reaction mixture, miRNA quantitative detection kit and detection method.

[0067] In the present invention, the kit preferably further comprises a polyadenylation and cDNA amplification reagent and a cDNA enrichment reagent. The polyadenylation and cDNA amplification reagent preferably comprises a universal reverse transcription primer and a reverse transcription amplification premix. The nucleotide sequence of the universal reverse transcription primer is preferably as shown in SEQ ID NO: 11 (GCCTCTGACTCCAGGATCTGTAGACCTCCATCCGAGACA CACGATTGATGGTTTTTTTTTTTTTTTTTTVN). The reverse transcription amplification premix preferably comprises at least one of the following: 5×RT buffer, ATP, polyA enzyme, reverse transcriptase and stabilizer. The cDNA enrichment reagent preferably comprises a hybridization primer, exonuclease I, chloroform, cDNA enrichment primer I, enhancer, 2×Boost mix enzyme, UDG enzyme, cDNA enrichment primer II, and streptavidin-modified magnetic beads. The nucleotide sequence of the cDNA enrichment primer I is preferably as shown in SEQ ID NO: 12. The nucleotide sequence of the cDNA enrichment primer II is preferably as shown in SEQ ID NO: 13. The nucleotide sequence of the hybridization primer is preferably as shown in SEQ ID NO:14 and SEQ ID NO:53 to SEQ ID NO:149.

[0068] In the present invention, the kit is preferably a positive control, a negative control, a calibrator 1, a calibrator 2, a calibrator 3, a calibrator 4, a calibrator 5 and a calibrator 6. The positive control is preferably a pre-amplified cDNA product. The negative control is plasma from healthy people.

[0069] In the present invention, a method for assisting in the diagnosis of gastric cancer based on the kit preferably comprises four steps: polyadenylation and cDNA reaction, cDNA enrichment, real-time fluorescence quantitative PCR and result analysis, and the specific operations are as follows:

[0070] 1. Polyadenylation and cDNA Reaction

[0071] MicroRNA is extracted from human plasma and adenylated using poly(A) polymerase. The polyadenylated microRNA is reverse transcribed into cDNA using a universal reverse transcription primer with poly(T) at the 3' end.

[0072] In the present invention, the extraction of microRNA from human plasma is preferably performed using a nucleic acid extraction kit produced by the applicant's company.

[0073] 2. cDNA Enrichment

[0074] Using cDNA as a template, PCR amplification was first performed using hybrid primers, and the amplified products were subjected to a first pre-amplification using hybrid primers / cDNA enrichment primers I and a second pre-amplification using hybrid primers / cDNA enrichment primers II to enrich the cDNA template.

[0075] 3. Real-time fluorescence quantitative PCR

[0076] With the enriched cDNA as template, qPCR amplification was performed using forward primers, universal reverse primers and universal probes for detecting any one of the eight miRNAs. During the amplification process, the probe was complementary to the cDNA chain containing the microRNAs (miR-223-3p, miR-186-5p, miR-18a-5p, miR-106b-5p, miR-20a-5p, miR-451a, miR-7976, miR-151a-3p) template, and the fluorescent dye group at the 5' end of the probe was cleaved by the Taq enzyme and released a fluorescent signal. The intensity of the detected fluorescent signal can reflect the content of miR-223-3p, miR-186-5p, miR-18a-5p, miR-106b-5p, miR-20a-5p, miR-451a, miR-7976, and miR-151a-3p in the sample.

[0077] 4. Results Analysis

[0078] The fluorescence quantitative PCR instrument detection data is imported into the miRNA quantitative PCR data machine learning processing software (machine learning model), and the diagnosis results are output by computer calculation.

[0079] In the present invention, the software is preferably AImiR software. For a demonstration of AImiR software, see the link https: / / jiawangr.shinyapps.io / AImiRFinalmodel / .

[0080] In the present invention, the algorithm of the machine learning model is preferably a support vector machine machine learning algorithm. In an embodiment of the present invention, the most significantly differentially expressed miRNAs in the plasma of gastric cancer and non-gastric cancer patients are first screened out, and then, using different numbers (1-27) of differential miRNAs, 10-fold cross validation (10-FoldCross Validation) gastric cancer and non-gastric cancer machine learning classification is performed, and the machine learning classification accuracy of 8 specific miRNAs is the highest. The forward primers are designed using the 8 specific miRNAs screened, and the corresponding miRNA expression levels in the plasma samples of a single gastric cancer patient and non-gastric cancer patient are detected by RT-qPCR method. The amplification results are respectively input into 5 machine learning classifiers to classify gastric cancer and non-gastric cancer. According to the accuracy and Kappa value, the svm machine learning algorithm is the best. The ROC curve of the machine learning model constructed based on the svm machine learning algorithm for diagnosing gastric cancer, the AUC area (equivalent to recognition ability) of the SVM algorithm is above 92%, and the sensitivity and specificity are above 90%, and the biological repeatability of the model is also very good.

[0081] In the present invention, the method for determining the diagnosis result is preferably as follows: the result output by the machine learning model for each sample is the probability of gastric cancer and the probability of non-gastric cancer (gastric discomfort such as gastritis, gastric ulcer, gastric erosion, etc.) (the sum of the probabilities is 100%). If the probability of gastric cancer is greater than 50%, the sample is judged to be positive, i.e., gastric cancer, otherwise the sample is judged to be negative. If the probabilities of the two are equal, the result is judged to be uncertain (IND, indeterminate). That is, the diagnosis results include three types: positive, negative and uncertain.

[0082] The present invention provides a diagnostic system for assisting in the diagnosis of gastric cancer, comprising:

[0083] The first module is used to receive the original data of fluorescent quantitative PCR of the sample to be tested based on the kit described in the above technical solution and the information of the patient from whom the sample to be tested comes;

[0084] The second module is used to calculate the gastric cancer probability and non-gastric cancer probability of the sample to be tested based on the fluorescent quantitative PCR raw data transmitted by the first module;

[0085] The third module: determines the diagnosis result of the sample to be tested based on the calculation result of the second module; and

[0086] The fourth module is used to output the patient information of the source of the sample to be tested in the first module and the diagnosis result of the sample to be tested in the third module to the terminal.

[0087] In the present invention, the calculation method in the second module preferably includes a support vector machine machine learning algorithm.

[0088] The following is a detailed description of a kit for RT-qPCR-assisted diagnosis of gastric cancer based on eight microRNA fluorescent probes provided by the present invention in conjunction with the examples, but they should not be construed as limiting the scope of protection of the present invention.

[0089] Example 1

[0090] The miRNA qPCR detection method developed by Jiansheng Biotechnology, the technical flow chart is shown in Figure 1 .

[0091] 1. Materials and Methods

[0092] 1.1 Reagents:

[0093] 5×RT buffer, polyA enzyme, ATP, reverse transcriptase, exonuclease I, exonuclease II, 2×Boostmix, ROX (carboxy-X-rhodamine), positive quality control (plasma cDNA pre-amplification product of gastric cancer patients), negative quality control;

[0094] 1.2 Sequence

[0095] Universal probe: ACC+AT+CA+AT+CG+TG+TG (SEQ ID NO: 1);

[0096] Universal reverse primer: ATAGATTGAGTGGAGCGTAGGGTTTCGG (SEQ ID NO: 2);

[0097] Forward primer: A forward primer designed according to the miRNA sequence to be detected, see SEQ ID NO: 3 to SEQ ID NO: 10, see Table 1 for details.

[0098] Table 18 miRNAs sequences and designed forward primer sequences

[0099]

[0100]

[0101] The universal reverse transcription primer: GCCTCTGACTCCAGGATCTGTAGACCTCCAT CCGAGACACACGATTGATGGTTTTTTTTTTTTTTTTTTVN (SEQ ID NO: 11).

[0102] cDNA Enrichment Primer I:

[0103] TCTACAGATCCTGGCCTCTGACTCCAGGATCTGTAGACCTCCATCCGAGACACACGAT (SEQ ID NO: 12);

[0104] cDNA enrichment primer II: GCCTCTGACTCCAGGATCTGTAGAC (SEQ ID NO: 13);

[0105] Internal reference primers: Sequences are shown in Table 2.

[0106] Table 2 Internal reference primer sequences

[0107]

[0108]

[0109] Hybridization primers: See Table 3:

[0110] Table 3 Hybridization primer sequences

[0111]

[0112]

[0113]

[0114] 2. Instruments and equipment:

[0115] Ordinary PCR instrument, fluorescent quantitative PCR instrument, high-speed refrigerated centrifuge, vortex instrument, handheld centrifuge, pipette (specifications: 2.5μl, 10μl, 100μl, 200μl).

[0116] 3. Experimental Methods

[0117] 3.1 microRNA extraction

[0118] MicroRNA was extracted from plasma according to the instructions of nucleic acid extraction reagent provided by Jiansheng Medical Technology Co., Ltd.

[0119] 3.2 Polyadenylation reaction and cDNA synthesis (polyA reaction and reverse transcription reaction)

[0120] (1) PolyA reaction steps

[0121] In each labeled PCR tube, add 4 μl 5×RT buffer (ThermoFisher), 2 μl ATP, 0.5 μl RNA inhibitor, 1 μl polyA enzyme (New England Biolab), and 12.5 μl RNA sample in turn, mix by inverting 5-10 times, add the corresponding numbered microRNA after centrifugation, and the total volume is 20 μl. Incubate on the PCR instrument for 30 minutes.

[0122] (2) Reverse transcription steps

[0123] Add reverse transcription primers and corresponding numbered polyA templates to the PCR tube. Set the PCR program to 65°C, 5 min, run the program, in the above reverse transcription PCR tube, 5×RT buffer 2μl, 1.6M trehalose 4.5μl, Actinomycin D (1mg / ml)*1.2μl, μl, T4gp32 / RecA / ATP mixed solution*1.5μl, RNA Inhibitor (40000U / ml) 0.3μl, MaximaHRT enzyme (50U / μl) 1.5μl, 10mM dNTPs*1.5μl, 10μM RT primer 1.5μl and PolyA template 16μl, totaling 30μl.

[0124] Set the PCR program as follows: 42°C for 15 min, 50°C for 30 min, 55°C for 30 min, 60°C for 30 min, 65°C for 30 min, 85°C for 5 min; run on the machine.

[0125] Add an equal volume of chloroform to the PCR tube on the PCR instrument and immediately vortex for 1 minute. Centrifuge the sample tube using a high-speed refrigerated centrifuge and pipette the supernatant into a labeled new PCR tube.

[0126] 3.3. Pre-amplification PCR operation steps

[0127] (1) Primer hybridization reaction

[0128] In a PCR tube, add 1000 mM concentration of hybridization primers and the above reverse transcription product stock solution.

[0129] Insert the PCR tube containing the reaction solution into the float and place it in boiling water.

[0130] Add 2.5 μl of exonuclease I to each tube and place it in a PCR instrument for digestion. After digestion, add an equal volume of chloroform. Centrifuge using a high-speed refrigerated centrifuge and gently pipette the supernatant into a new, labeled PCR tube.

[0131] (2) Chloroform extraction pre-amplification PCR reaction-1

[0132] In a PCR tube, add 10 μl 2×Boost mix, 1 μl hybridization primer, 1 μl cDNA enrichment primer I, 1 μl enhancer (Tth RecA, 0.2 ug / μl) and 7 μl hybridized cDNA template.

[0133] Set the PCR instrument program as follows: ①37℃10min, ②95℃10min, ③(95℃10s, 65℃1min)3cycles, ④(95℃10s, 62℃1min)3cycles, ⑤(95℃10s, 58℃2min)2cycles, ⑥(95℃10S, 60℃2min)2cycles, ⑦(95℃10s, 65℃1min), 12cycles, ⑧72℃10min, ⑨99℃, 20min; run the machine.

[0134] After PCR amplification, add an equal volume of chloroform to the PCR tube. Centrifuge using a high-speed refrigerated centrifuge and gently pipette the supernatant into a new labeled PCR tube. Use a pipette to pipette the liquid into another new labeled PCR tube and immediately proceed with downstream experiments.

[0135] (3) Chloroform extraction pre-amplification PCR reaction-2

[0136] In a PCR tube, add 5 μl of 2×Boost mix enzyme, 0.5 μl of hybridization primer, 0.5 μl of cDNA enrichment primer II, 0.5 μl of enhancer and 3.5 μl of the above template.

[0137] The PCR instrument was set to the following program: ① 37°C for 10 min, ② 95°C for 10 min, ③ (95°C for 10 s, 65°C for 1 min), 12 cycles, ④ 72°C for 10 min, 72°C for 5 min, ⑧ 99°C for 20 min, and the instrument was put into operation.

[0138] After PCR amplification is completed, add an equal volume of chloroform to the PCR tube. Centrifuge using a high-speed refrigerated centrifuge and gently pipette the supernatant into a labeled new PCR tube. Add 2.5 μl of exonuclease II to the PCR tube and incubate.

[0139] After PCR amplification, add an equal volume of chloroform to the PCR tube, centrifuge using a high-speed refrigerated centrifuge, and gently pipette the supernatant into a labeled new PCR tube.

[0140] 4. Real-time fluorescence quantitative PCR

[0141] Prepare miRNAs reaction solution according to the number of samples to be tested, and each 10μl reaction system contains 5μl of 2×Boost mix enzyme. According to the arrangement of the test samples, add the prepared reaction solution, diluted template to be tested, negative and positive quality control products and standards to each well in the 384-well plate.

[0142] After the sample is added, seal the plate with a special membrane for PCR fluorescence quantitative detection, centrifuge, and test on the machine.

[0143] 5. Result analysis:

[0144] The fluorescence quantitative PCR instrument detection data is imported into the miRNA quantitative PCR data machine learning processing software (machine learning model), and the computer completes the calculation and outputs the detection results.

[0145] Example 2

[0146] The miRNA qPCR detection method developed by Jiansheng Bio is highly reproducible

[0147] The detection method described in Example 1 was used to detect the expression levels of 98 miRNAs in the plasma exosomes of gastric cancer patients and healthy controls.

[0148] Results Figure 2 The results showed that most miRNAs did not change between gastric cancer patients and healthy controls, and only a few miRNAs had changes in expression levels. The expression of miRNAs in the two samples was highly correlated (R 2 =0.9987), indicating that quantitative PCR technology has high repeatability and accuracy.

[0149] Example 3

[0150] The miRNA qPCR detection method developed by Jiansheng Bio has high resolution

[0151] In Example 2, miRNA qPCR detection method was used to detect miRNA in cancer samples and healthy control samples to obtain amplification curves. The amplification curves were analyzed to find expression differences after internal reference correction. The internal reference correction method can be found in the prior art (Vandesompele, J. et al. Accurate normalization of real-time quantitative RT-PCR data by geometric averaging of multiple internal control genes. Genome biology 3, RESEARCH0034 (2002)).

[0152] Depend on Figure 3It can be seen that the expression level of 98 miRNAs in the plasma of cancer patients is only 1.06 times that in the plasma of healthy people. Such a small difference can be detected (p<0.01), indicating that Jiansheng miRNA qPCR detection technology has a high resolution.

[0153] Example 4

[0154] Screening Methods for Computer Artificial Intelligence Models

[0155] 1. Differential expression in plasma of patients with gastric cancer and non-gastric cancer

[0156] Gastric cancer plasma samples and non-gastric cancer patient (gastric discomfort such as gastritis, gastric ulcer, gastric erosion) plasma samples were mixed, and the expression levels of 1152 miRNAs were detected using the detection method of Example 1, and the miRNA species with differential expression (p < 0.05) in the two samples were selected. A single sample was selected to screen 27 differentially expressed miRNAs; the differentially expressed miRNAs were sorted according to importance using the caret package of R language. The process is shown in Figure 4 .

[0157] Results Figure 5 . Using different numbers (1 to 27) of miRNAs, machine learning classification (Recursive Feature Elimination (RFE)) of gastric cancer and non-gastric cancer (gastric discomfort such as gastritis, gastric ulcer, gastric erosion, etc.) was performed through 10-fold cross validation. The accuracy of machine learning classification using 8 miRNAs is better than or equal to that of more than 8 miRNAs.

[0158] 2. Selection of machine learning algorithms

[0159] Forward primers were designed for the 8 screened miRNAs, and the primer sequences are shown in Table 1.

[0160] The RT-qPCR method described in Example 1 was used to detect the corresponding miRNA expression levels in plasma samples of individual gastric cancer patients and non-gastric cancer patients. The expression data were then input into five machine learning classifiers (SVM, KNN, RF, CART, IDA) to classify gastric cancer and non-gastric cancer. The results are shown in Figure 6 ,Judging from the accuracy and Kappa value, the results show that the SVM machine learning algorithm is the best.

[0161] Example 5

[0162] The SVM machine learning algorithm selected in Example 4 was used to construct the final machine learning model for gastric cancer diagnosis. The plasma samples of gastric cancer patients and non-gastric cancer patients were tested for qPCR results based on 8 miRNAs, using the same method as in Example 4. Among them, the plasma samples of gastric cancer patients were derived from the Institute of Oncology, Chinese Academy of Medical Sciences and the Institute of Oncology, Peking University; the plasma samples of non-gastric cancer patients were derived from Beijing Hospital and the Second People's Hospital of Dongying City, Shandong Province. The final machine learning model for gastric cancer diagnosis was used to calculate and analyze the test results, and the ROC curve was constructed.

[0163] The ROC curve of the final model for the diagnosis of gastric cancer is shown in Figure 7 . There is no overlap in the samples of Validation Experiment-1, Validation Experiment-2, Validation Experiment-3 and Validation Experiment-4. The AUC area (equivalent to the recognition ability) obtained by the final model based on the SVM algorithm is above 92%. The sensitivity and specificity are both above 90% (Validation Experiment-3 is close). This shows that these four models can effectively identify gastric cancer and non-gastric cancer (gastritis, gastric ulcer, gastric erosion and other stomach discomfort and other cancers including lung cancer, breast cancer and colorectal cancer), and the biological repeatability of the model (using different samples) is very good. In sharp contrast to our high repeatability, the current detection results of circulating miRNAs have extremely low repeatability.

[0164] Example 6

[0165] The performance parameter analysis results of the final detection model for diagnosing gastric cancer in Example 5 are shown in Tables 4 and Figure 8 .

[0166] Table 4

[0167]

[0168] Figure 8 It is the performance parameter of the verification result of the final model for diagnosing gastric cancer. The kappa value represents the consistency between the model classification result and the gold standard diagnosis result. The closer to 1, the better the consistency. Usually, 0.75 or above means excellent consistency results, and 0.8-1 is almost completely consistent. The average kappa value of the final model verification is 0.8116, indicating that the model classification result is almost consistent with the gold standard diagnosis result. Other parameters also prove this point. Performance Because it is a verification experiment of the final model, there is only one error rate, which belongs to the final model.

[0169] The AUC area of ​​the SVM algorithm is 0.996, the accuracy, Kappa, sensitivity and specificity are all above 97% (Kappa is close to) and the P value is <4.324e-16. According to these important parameters for judging the machine learning ability, the algorithm performance meets the algorithm design requirements.

[0170] Example 7

[0171] Final model evaluation on machine learning dataset 1

[0172] Dataset 1 includes a total of 288 clinical samples, including 144 cases of gastric cancer and 144 cases of non-gastric cancer (gastric discomfort such as gastritis, gastric ulcer, gastric erosion, etc.).

[0173] The ROC curve results are shown in Fig. 9 The AUC of the diagnostic model based on the SVM algorithm is 0.996, and the accuracy, Kappa, sensitivity and specificity are all above 97% (Kappa is close to) and the P value is <4.324e-16. According to these important parameters for judging the machine learning ability, the algorithm performance meets the algorithm design requirements.

[0174] Example 8

[0175] Final model evaluation on machine learning dataset 2

[0176] Dataset 2 includes a total of 428 clinical samples, including 213 cases of gastric cancer and 215 cases of non-gastric cancer (gastric discomfort such as gastritis, gastric ulcer, gastric erosion, etc.).

[0177] The ROC curve results are shown in Fig.10 The AUC area of ​​the diagnostic model of the SVM algorithm is 0.970, and the accuracy, Kappa, sensitivity and specificity are all above 90% (Kappa is close to) and the P value is <2.59e-15. According to these important parameters for judging the machine learning ability, the algorithm performance meets the algorithm design requirements.

[0178] Example 9

[0179] Final model evaluation on machine learning dataset 3

[0180] Dataset 3 includes a total of 1119 clinical samples, including 480 cases of gastric cancer and 639 cases of non-gastric cancer (gastric discomfort such as gastritis, gastric ulcer, gastric erosion, etc.).

[0181] The ROC curve results are shown in Fig.11 The AUC of the gastric SVM algorithm was 0.922, and the accuracy, Kappa, sensitivity, and specificity were close to 90%, Kappa was close to 80%, and P value was <2e-16. Based on these important parameters for judging the machine learning ability, the algorithm performance met the algorithm design requirements.

[0182] Example 10

[0183] A kit for RT-qPCR-assisted diagnosis of gastric cancer based on 8 microRNA fluorescent probes. Specific reagents are shown in Table 5.

[0184] Table 5 Reagents and specifications

[0185]

[0186]

[0187] Embodiment 11

[0188] The computer program for constructing the final machine learning model for gastric cancer diagnosis is made into AImiR software. For a demonstration of AImiR software, see the link https: / / jiawangr.shinyapps.io / AImiRFinalmodel / . For the operation method of AImiR software, see Fig.12 .

[0189] The diagnostic report output by AImiR software is shown in Fig.13 The report consists of two parts. The upper part is the diagnosis result of the clinical sample and the probability of it being gastric cancer or non-gastric cancer. If the probability is greater than the threshold, it is gastric cancer, otherwise it is non-gastric cancer. The lower part is the ROC curve of the machine learning model used, showing the sensitivity, specificity and optimal threshold. It can provide clinicians with the performance of the machine learning model used to judge the reliability of the diagnosis results of clinical samples. Fig.13 As shown in the figure, the sensitivity, specificity and AUC of the machine learning model used are all above 94%, and the clinical diagnosis results are relatively reliable. In addition, the upper part can display the diagnosis results of different samples by turning pages, but the ROC curve is always displayed. This makes it convenient for users to check the results.

[0190] Example 12

[0191] Comparison of relative expression of miRNA in normal plasma and hemolytic plasma

[0192] The normal plasma and hemolytic plasma were derived from the same person with the same number. The expression levels of 8 miRNAs in the normal plasma and hemolytic plasma were detected using the kit of Example 10, and the detection results were calculated and analyzed using the AImiR software developed in Example 11.

[0193] Results Fig.14, where A is the calibrated data (corrected by internal reference); B is the uncalibrated data (not calibrated by internal reference). It can be seen that the absolute amount of most miRNAs increased significantly after hemolysis (B), but most miRNAs did not change much after calibration (A). This result shows that the absolute amount of most miRNAs increased greatly after hemolysis (average = 35 times), but most miRNAs did not change much after calibration (average = 0.8 times). Hemolysis has a great impact on the lack of proper calibration, but has little impact on the proper calibration.

[0194] Example 13

[0195] The miRNA qPCR detection method developed by Jiansheng Bio was used to detect clinical samples

[0196] The clinical samples included 728 gastric cancer (gastric cancer plasma samples were obtained from the Cancer Institute of the Chinese Academy of Medical Sciences and the Cancer Institute of Peking University; non-gastric cancer plasma samples were obtained from Beijing Hospital and the Second People's Hospital of Dongying City, Shandong Province) and non-gastric cancer specimens. The detection method was the same as in Example 12. At the same time, the same clinical samples were tested using the traditional gold standard method (microscopic examination and pathological biopsy). The positive coincidence rate, negative coincidence rate, sensitivity and specificity were calculated respectively.

[0197] Among them, the definition of positive coincidence rate is the ratio of true positives to the number of positive samples predicted by the model; negative coincidence rate is the ratio of true negatives to the number of negative samples predicted by the model. Sensitivity is the ratio of true positives to the number of positive samples in the gold standard; specificity is the ratio of true negatives to the number of negative samples in the gold standard.

[0198] The results of the above examples show that there are 65 inconsistent test results between the miRNA qPCR detection method and the gold standard detection method, and the overall detection compliance rate is 97.39%. Among them, 307 positive samples were detected, 40 of which were negative, and the detection positive compliance rate was 92.60%; 421 negative samples were detected, 25 of which were positive, and the detection negative compliance rate was 90.83% (see Fig.15 ). The P value compared with the random result was 7.E-10, which is far below 0.05, indicating that the recognition result of machine learning is statistically significant. The consistency of the test results of six groups (patient samples collected at 6 different times or locations) with the gold standard diagnosis results.

[0199] The above is only a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principle of the present invention. These improvements and modifications should also be regarded as the scope of protection of the present invention.

Claims

1. A kit for RT-qPCR-assisted diagnosis of gastric cancer based on 8 microRNA fluorescent probes, characterized in that: It includes a real-time fluorescence quantitative PCR detection primer probe set and an internal reference primer; The real-time fluorescence quantitative PCR detection primer probe set includes a universal probe, a universal reverse primer, and a forward primer for detecting at least one of the following eight microRNAs: miR-223-3p, miR-186-5p, miR-18a-5p, miR-106b-5p, miR-20a-5p, miR-451a, miR-7976, and miR-151a-3p; The nucleotide sequence of the universal probe is shown in SEQ ID NO: 1; The nucleotide sequence of the universal reverse primer is shown in SEQ ID NO: 2; The nucleotide sequence of the forward primer for detecting miR-223-3p is shown in SEQ ID NO: 3; The nucleotide sequence of the forward primer for detecting miR-186-5p is shown in SEQ ID NO: 4; The nucleotide sequence of the forward primer for detecting miR-18a-5p is shown in SEQ ID NO: 5; The nucleotide sequence of the forward primer for detecting miR-106b-5p is shown in SEQ ID NO: 6; The nucleotide sequence of the forward primer for detecting miR-20a-5p is shown in SEQ ID NO: 7; The nucleotide sequence of the forward primer for detecting miR-451a is shown in SEQ ID NO: 8; The nucleotide sequence of the forward primer for detecting miR-7976 is shown in SEQ ID NO:9; The nucleotide sequence of the forward primer for detecting miR-151a-3p is shown in SEQ ID NO: 10; The nucleotide sequences of the internal reference primers are shown in SEQ ID NO: 15 to SEQ ID NO:

44.

2. The kit according to claim 1, characterized in that The 5' end of the universal probe is modified with a fluorescent dye group, and the 3' end of the universal probe is modified with a quencher group.

3. The kit according to claim 1, characterized in that Includes 2×Boost mix enzyme and real-time fluorescence quantitative PCR reaction solution.

4. The kit according to any one of claims 1 to 3, characterized in that Also included are reagents for polyadenylation and cDNA amplification and reagents for cDNA enrichment.

5. The kit according to claim 4, characterized in that: The polyadenylation and cDNA amplification reagents include universal reverse transcription primers and reverse transcription amplification premix.

6. The kit according to claim 5, characterized in that The nucleotide sequence of the universal reverse transcription primer is shown in SEQ ID NO: 11; The reverse transcription amplification premix comprises at least one of the following: 5×RT buffer, ATP, inhibitor, polyA enzyme, reverse transcriptase and stabilizer.

7. The kit according to claim 4, characterized in that: The cDNA enrichment reagents include hybridization primers, exonuclease I, chloroform, cDNA enrichment primer I, enhancer, 2×Boost mix enzyme, UDG enzyme, and cDNA enrichment primer II.

8. The kit according to claim 7, characterized in that: The nucleotide sequence of the cDNA enrichment primer 1 is shown in SEQ ID NO: 12; The nucleotide sequence of the cDNA enrichment primer II is shown in SEQ ID NO: 13; The nucleotide sequences of the hybridization primers are shown in SEQ ID NO:14 and SEQ ID NO:53 to SEQ ID NO:

149.

9. A diagnostic system for assisting in the diagnosis of gastric cancer, characterized in that: The diagnostic system comprises: The first module is used to receive the original data of fluorescent quantitative PCR detected by the kit according to any one of claims 1 to 5 and the information of the patient from whom the sample was derived; The second module is used to calculate the gastric cancer probability and non-gastric cancer probability of the sample to be tested based on the fluorescent quantitative PCR raw data transmitted by the first module; The third module: determines the diagnosis result of the sample to be tested based on the calculation result of the second module; and The fourth module is used to output the patient information of the source of the sample to be tested in the first module and the diagnosis result of the sample to be tested in the third module to the terminal.

10. The diagnostic system according to claim 6, characterized in that: The calculation method in the second module includes a support vector machine machine learning algorithm.

Images