A method for constructing a differential model of microvascular invasion in liver cancer and a method for testing the predicted value of microvascular invasion in liver cancer

Abstract

The present invention provides a method for constructing a differential model of hepatocellular carcinoma microvascular invasion and a method for testing the predicted value of hepatocellular carcinoma microvascular invasion, which relates to the technical field of metabolomics. The present invention provides positive serum samples and negative serum samples of hepatocellular carcinoma microvascular invasion; measure the 1 HNMR spectra of the positive serum samples and negative serum samples, and use the ratio of the full-spectrum data in the 1 HNMR spectra to the peak height or integral of the bile acid characteristic peaks as the independent variable, and the positive or negative of hepatocellular carcinoma microvascular invasion as the observation variable to construct an orthogonal partial least squares discriminant analysis (OPLS-DA) model. The present invention can construct the model by using nuclear magnetic resonance hydrogen spectrum technology in combination with conventional data processing software and data modeling software. The construction method is simple, and it can quickly and accurately predict the positive or negative of hepatocellular carcinoma microvascular invasion with one analysis, and has broad application prospects.

Description

Technical Field

[0001] The present invention relates to the technical field of metabolomics, and particularly relates to a method for constructing a differential diagnosis model for microvascular invasion of liver cancer and a method for testing the prediction value of microvascular invasion of liver cancer. Background Art

[0002] Hepatocellular carcinoma (HCC) is one of the most common malignant tumors globally. Despite the continuous optimization of surgical resection, local ablation, and comprehensive treatment regimens, and the successive introduction of molecular targeted therapy drugs, the mortality rate of liver cancer remains high. On the one hand, due to the lack of obvious early symptoms of liver cancer, most patients are already in the advanced stage at the time of diagnosis, and they have almost lost the possibility of radical surgery or transplantation; on the other hand, the postoperative recurrence rate and metastasis rate of patients are extremely high.

[0003] Microvascular invasion (MVI) is considered one of the most risk factors for liver cancer recurrence, indicating the presence of highly invasive biological behavior of the tumor. MVI refers to the presence of cancer cell nests in the lumen of vessels lined by endothelial cells under the microscope, mainly in the portal vein branches (including intra-capsular vessels). Pathological grading method: M0: no MVI is found; M1 (low-risk group): ≤ 5 MVIs, and occurring in the liver tissue adjacent to the cancer; M2 (high-risk group): > 5 MVIs, or MVI occurring in the liver tissue far from the cancer.

[0004] Currently, the detection of MVI still mainly relies on postoperative pathological examination. If the presence of MVI can be predicted before surgery and active intervention can be given at an early stage, it will surely improve the prognosis of patients. Therefore, preoperative prediction of MVI is of great significance for improving the long-term survival rate of HCC patients and reducing the postoperative recurrence rate. Currently, there is no reported method for quickly differentiating MVI. Summary of the Invention

[0005] In view of this, the purpose of the present invention is to provide a method for constructing a differential diagnosis model for microvascular invasion of liver cancer and a method for testing the prediction value of microvascular invasion of liver cancer. The model constructed by the present invention can achieve rapid differentiation of microvascular invasion of liver cancer.

[0006] To achieve the above-mentioned invention purpose, the present invention provides the following technical solutions:

[0007] The present invention provides a method for constructing a differential diagnosis model for microvascular invasion of liver cancer, including the following steps:

[0008] Provide positive serum samples and negative serum samples of microvascular invasion of liver cancer;

[0009] Measure the 1 HNMR spectra of the positive serum samples and negative serum samples, with the 1The ratio of the full-spectrum data in the HNMR spectrum to the peak height or integral of the characteristic peaks of bile acids is used as the independent variable, and the positive or negative microvascular invasion of liver cancer is used as the observation variable. Orthogonal partial least squares discriminant analysis is used for supervised clustering analysis to obtain the OPLS-DA model.

[0010] Preferably, for the determination of positive and negative serum samples 1 The method for obtaining the HNMR spectrum of serum samples includes the following steps:

[0011] (1) Mix the positive and negative serum samples with methanol and phosphate buffer respectively. Vortex, sonicate and centrifuge the resulting mixture in sequence. Take the supernatant and concentrate it to dryness to obtain serum metabolites.

[0012] (2) Mix the serum metabolites with deuterated reagent. Vortex, sonicate and centrifuge the resulting mixture in sequence to obtain the supernatant. Take the supernatant for nuclear magnetic resonance detection to obtain 1 the HNMR spectrum; tetramethylsilane is added as an internal standard in the deuterated reagent.

[0013] Preferably, the volume ratios of the positive and negative serum samples to methanol are 1:4 - 20 respectively.

[0014] Preferably, the pH value of the phosphate buffer is 2 - 8 and the concentration is 0.05 - 3 mol / L; the dosage ratios of the positive and negative serum samples to the phosphate buffer are 1.0 mL:50 - 200 μL respectively.

[0015] Preferably, in step (1), the vortex oscillation time is 3 - 15 min; the sonication time is 5 - 60 min; the centrifugation speed is 8000 - 13000 rpm and the time is 5 - 20 min; the concentration method is rotary evaporation, the temperature of the rotary evaporation is 40 - 55 °C and the time is 3 - 6 h.

[0016] Preferably, the conditions for nuclear magnetic resonance detection in step (2) include: a 5 mm multinuclear broadband observation probe, an observation frequency of 600.13 MHz, a spectral width SW of 19.8 ppm, a spectral center point O1P of 4 - 10 ppm, and the number of scans NS of 64 - 256 times.

[0017] Preferably, the chemical shift of the characteristic peaks of bile acids is 0.70 - 0.74 ppm.

[0018] Preferably, the number of independent variables is 10K - 60K.

[0019] Preferably, the orthogonal variable component of the OPLS-DA model is 3-10, R2Y is 0.6-0.9, and Q2 is 0.2-0.6.

[0020] The present invention also provides a test method for predicting the value of microvascular invasion in liver cancer, comprising the following steps:

[0021] Determine the 1 HNMR spectrum of the serum sample to be tested, and use the ratio of the full-spectrum data in the HNMR spectrum of the serum sample to be tested 1 to the peak height or integral of the bile acid characteristic peak as the input factor, and input it into the OPLS-DA model obtained by the construction method described in the above technical solution to calculate the predicted value.

[0022] The present invention provides a method for constructing a differential model of microvascular invasion in liver cancer, comprising the following steps: providing positive serum samples and negative serum samples of microvascular invasion in liver cancer, and determining the 1 HNMR spectra of the positive serum samples and negative serum samples, using the ratio of the full-spectrum data in the 1 HNMR spectra to the peak height or integral of the bile acid characteristic peak as the independent variable, and using positive or negative microvascular invasion in liver cancer as the observation variable, and performing supervised clustering analysis by orthogonal partial least squares discriminant analysis to obtain the OPLS-DA model. The model (i.e., the OPLS-DA model) obtained by using the construction method provided by the present invention can quickly distinguish positive or negative microvascular invasion in liver cancer, realize the prediction of microvascular invasion in liver cancer before surgery, and is of great significance for improving the long-term survival rate of patients with hepatocellular carcinoma and reducing the postoperative recurrence rate. Especially when using 1 the ratio of the full-spectrum data of HNMR to the peak height or integral of the bile acid characteristic peak as the independent variable, that is, using the method of calibrating the full-spectrum of bile acid metabolites 1 HNMR, not only can the influence of certain "ghost peaks" on the authenticity of the data be eliminated, but also the relative quantitative data of serum metabolites can be obtained by one detection, without further identification and absolute quantification of specific metabolites, and has the characteristics of "integrity" (detecting metabolites containing H without bias) and "ambiguity" (without absolute quantification of specific metabolites). While ensuring accuracy, it greatly reduces the detection cost. The present invention can construct the model by using nuclear magnetic resonance hydrogen spectrum technology in combination with conventional data processing software and data modeling software. The construction method is simple, and it can quickly and accurately predict positive or negative microvascular invasion in liver cancer by one analysis, and has broad application prospects. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] Figure 1 For the original 1 HNMR spectrum of the serum sample numbered 1 in the M0 sample group of Example 1 and the calibrated1 HNMR spectrum Figure 1 Among them, (a) is the original 1 HNMR spectrum, and (b) is the HNMR spectrum after calibration of the characteristic peaks of bile acids; 1 HNMR spectrum;

[0024] Figure 2 It is the HNMR spectrum after calibration of the characteristic peaks of bile acids in the serum samples numbered 50 in the M0 sample group and numbered 100 in the MVI sample group in Example 1, 1 HNMR spectrum, Figure 1 Among them, A is the HNMR spectrum after calibration of the characteristic peaks of bile acids in the serum sample numbered 50 in the M0 sample group, 1 and B is the HNMR spectrum after calibration of the characteristic peaks of bile acids in the serum sample numbered 100 in the MVI sample group; 1 HNMR spectrum;

[0025] Figure 3 It is the OPLS-DA score plot of the calibrated characteristic peaks of bile acids in M0 (numbers 1-55, 105-109) and MVI (numbers 56-104, 120-124) in Example 1; 1 HNMR full-spectrum OPLS-DA score plot;

[0026] Figure 4 It is the result plot of 200 permutation tests of the OPLS-DA model constructed in Example 1;

[0027] Figure 5 It is the prediction plot of unknown samples by the OPLS-DA model constructed based on bile acid calibration in Example 2 1 HNMR full spectrum;

[0028] Figure 6 It is the prediction plot of M0 / MVI classification by the OPLS-DA model constructed based on bile acid calibration in Example 2 1 HNMR full spectrum;

[0029] Figure 7 It is the prediction plot of unknown samples by the OPLS-DA model constructed based on 1 the normalized values of HNMR full-spectrum data in Comparative Example 1;

[0030] Figure 8 It is the prediction plot of M0 / MVI classification by the OPLS-DA model constructed based on 1 the normalized values of HNMR full-spectrum data in Comparative Example 1;

[0031] Figure 9 It is the prediction plot of M0 / MVI classification by the OPLS-DA model constructed based on choline calibration in Comparative Example 2 1 HNMR full spectrum. Detailed implementation mode

[0032] The present invention provides a method for constructing a differential diagnosis model of hepatocellular carcinoma microvascular invasion, comprising the following steps:

[0033] Providing positive serum samples and negative serum samples of hepatocellular carcinoma microvascular invasion;

[0034] Measuring the 1 HNMR spectra of the positive serum samples and negative serum samples, using the ratio of the full-spectrum data in the 1 HNMR spectra to the peak height or integral of the bile acid characteristic peaks as the independent variable, and the positive or negative of hepatocellular carcinoma microvascular invasion as the observation variable, and performing supervised clustering analysis using orthogonal partial least squares discriminant analysis to obtain an OPLS-DA model.

[0035] The present invention provides positive serum samples and negative serum samples of hepatocellular carcinoma microvascular invasion; measuring the 1 HNMR spectra of the positive serum samples and negative serum samples. Before measuring the 1 HNMR spectra, the present invention preferably thaws the positive serum samples and negative serum samples at room temperature respectively. In the present invention, the method for measuring the 1 HNMR spectra of each positive serum sample and negative serum sample preferably comprises the following steps:

[0036] (1) Mixing the positive serum samples and negative serum samples with methanol and phosphate buffer solution respectively, subjecting the obtained mixture to vortex oscillation, ultrasound and centrifugation in sequence, taking the supernatant and concentrating it to dryness to obtain serum metabolites;

[0037] (2) Mixing the serum metabolites with deuterated reagent, subjecting the obtained mixture to vortex oscillation, ultrasound and centrifugation in sequence to obtain a supernatant; taking the supernatant for nuclear magnetic resonance detection to obtain 1 HNMR spectra; tetramethylsilane is added as an internal standard in the deuterated reagent.

[0038] In the present invention, the volume ratio of the positive serum sample and the negative serum sample to methanol is preferably 1:4 to 20. In the present invention, the pH value of the phosphate buffer is preferably 2 to 8, and the concentration is preferably 0.05 to 3 mol / L; the dosage ratios of the positive serum sample and the negative serum sample to the phosphate buffer are preferably 1.0 mL:50 to 200 μL respectively. In the present invention, the time of vortex oscillation is preferably 3 to 15 min; the time of ultrasonic treatment is preferably 5 to 60 min; the centrifugation speed is preferably 8000 to 13000 rpm, and the time is preferably 5 to 20 min. In the present invention, the concentration method is preferably rotary evaporation, the temperature of the rotary evaporation is preferably 40 to 55 °C, and the time is preferably 3 to 6 h. The method for treating samples in the present invention is simple.

[0039] In the present invention, the deuterated reagent is preferably deuterated methanol or heavy water; there is no particular requirement for the dosage of the deuterated reagent in the present invention, as long as it can dissolve the serum metabolites. In the present invention, the conditions of vortex oscillation, ultrasonic treatment and centrifugation are preferably the same as those in the above technical solution, and will not be elaborated here. In the present invention, the conditions of nuclear magnetic resonance detection preferably include: a 5 mm multinuclear broadband observation probe, an observation frequency of 600.13 MHz, a spectral width SW of 19.8 ppm, a spectral center point O1P of 4 to 10 ppm, and a number of scans NS of 64 to 256 times. After 1 obtaining the 1 HNMR spectrum, the present invention also preferably preprocesses the

[0040] HNMR spectrum; the preprocessing includes baseline calibration, removing interference peaks such as water peaks and solvent residual peaks, and there is no particular requirement for the preprocessing method in the present invention, and a method well-known to those skilled in the art can be used. In the examples of the present invention, a self-written APP small program based on matlab is used.

[0040] After obtaining the 1 HNMR spectra of the positive serum sample and the negative serum sample, the present invention uses the ratio of the full-spectrum data (i.e., the intensity of each data point in the spectrum) in the 1 HNMR spectrum to the peak height or integral of the bile acid characteristic peak as the independent variable, and the positive or negative of liver cancer microvascular invasion as the observation variable, and performs supervised clustering analysis using orthogonal partial least squares discriminant analysis to obtain an OPLS-DA model. In the present invention, the chemical shift of the bile acid characteristic peak is 0.70 to 0.74 ppm; the present invention uses 1 the ratio of the HNMR full-spectrum data to the peak height or integral of the bile acid characteristic peak as the independent variable, that is, uses the bile acid metabolite 1The method for calibrating the full 1H NMR spectrum can eliminate the influence of certain "ghost peaks" on the authenticity of data compared with the conventional full-spectrum or integral normalization method. In the present invention, the number of independent variables is preferably 10K - 60K. In the present invention, the OPLS-DA model can be constructed using the SIMCA software well-known to those skilled in the art, that is, the ratio of the full-spectrum data in the 1 1H NMR spectrum to the peak height or integral of the characteristic peaks of bile acids is input into the SIMCA software. In the present invention, the number of orthogonal variable components of the OPLS-DA model is preferably 3 - 10, R2Y is preferably 0.6 - 0.9, and Q2 is preferably 0.2 - 0.6.

[0041] The present invention provides a method for testing the predicted value of microvascular invasion in liver cancer, comprising the following steps: measuring the 1 1H NMR spectrum of the serum sample to be tested, using the ratio of the full-spectrum data in the 1 1H NMR spectrum of the serum sample to be tested to the peak height or integral of the characteristic peaks of bile acids as the input factor, inputting it into the OPLS-DA model obtained by the construction method described in the above technical solution, and calculating the predicted value.

[0042] In the present invention, the method for measuring the 1 1H NMR spectrum of the serum sample to be tested is the same as the method for measuring the 1 1H NMR spectrum of the positive and negative serum samples, which will not be elaborated here. In the present invention, the average value of the classification scores of the negative serum samples of microvascular invasion in liver cancer in the OPLS-DA model is about 1.0. Therefore, if the predicted value of the serum sample to be tested is greater than or equal to 0.5, it is close to the classification score value of the negative serum sample of microvascular invasion in liver cancer, and it is determined that the serum sample to be tested is negative for microvascular invasion in liver cancer; if the predicted value is less than 0.5, it is determined that the serum sample to be tested is positive for microvascular invasion in liver cancer.

[0043] The present invention uses nuclear magnetic resonance hydrogen spectroscopy technology combined with conventional data processing software and data modeling software to construct a model, which can quickly and accurately predict positive or negative microvascular invasion in liver cancer with a single analysis, and has broad application prospects.

[0044] The following will describe in detail the construction method of the microvascular invasion discrimination model in liver cancer and the testing method of the predicted value of microvascular invasion in liver cancer provided by the present invention in conjunction with the embodiments, but they should not be construed as limiting the protection scope of the present invention.

[0045] In each embodiment, MVI represents positive microvascular invasion in liver cancer, and M0 represents negative microvascular invasion in liver cancer.

[0046] Example 1

[0047] Construct the OPLS-DA model:

[0048] Collection and pretreatment of serum samples with known positive or negative microvascular invasion in hepatocellular carcinoma: The MVI sample group was from patients with hepatocellular carcinoma with MVI aged 30 - 80 years old, and the M0 sample group was hepatocellular carcinoma patients with matched gender and age and negative MVI. There were 60 cases in the M0 sample group (numbered 1 - 55, 105 - 109), and 54 cases in the MVI sample group (numbered 56 - 104, 120 - 124). Identification criteria for serum samples MVI and M0: Referring to the diagnostic criteria of MVI in the 2015 hepatocellular carcinoma pathology guidelines, the pretreatment of pathological sections all referred to the protocol of Shanghai Eastern Hepatobiliary Hospital. The pathological sections of all hospitals were electronically stored, and whether they contained MVI was determined by pathological experts according to the same criteria. Thaw 114 serum samples at room temperature. Add 200 μL of each serum sample to 1.5 mL of methanol and 20 μL of 3 mol / L phosphate buffer (pH = 7.4). Vortex for 3 min, sonicate for 10 min, and centrifuge at 13000 rpm for 10 min in sequence. Then take 1.2 mL of the supernatant and concentrate and evaporate it to dryness at 50 °C for 6 h in a vacuum rotary evaporator to obtain serum metabolites; then add 600 μL of deuterated methanol (containing 0.03 wt% TMS as an internal standard) to the serum metabolites for reconstitution, vortex for 3 min, sonicate for 3 min, and centrifuge at 13000 rpm for 10 min. Then take 550 μL of the supernatant and transfer it to a 5 mm NMR tube for measurement.

[0049] Determination of serum metabolites: The measuring instrument used was a Bruker AV - 600 MHz nuclear magnetic resonance spectrometer, with a 5 mm multinuclear broadband observation (BBO) probe, and the observation frequency was 600.13 MHz. The detailed experimental parameters were as follows: SW was 19.8 ppm, O1P was 6.175 ppm, and NS was 64. The nuclear magnetic resonance proton spectra of each serum sample in the M0 sample group and the MVI sample group were measured.

[0050] Process of NMR data processing: Use a self - developed APP small program based on matlab to perform baseline calibration on the measured proton spectrum, remove interference peaks such as water peaks and solvent residue peaks. Integrate the characteristic peaks of bile acids, and then 1 Ratio all data points in the HNMR spectrum to the integration of bile acids (bile acid calibration), specifically as follows: x j,new = x j,old / Int bile, where x j,new represents the new variable after bile acid calibration, x j,old, is the original variable, and Int bile represents the integration value of the bile acid characteristic peak at 0.70 - 0.74 ppm, obtaining 1 the relative quantitative value of the HNMR full spectrum, as shown in Figure 1 and Figure 2 shown.Figure 1 Among them, (a) is the original 1 HNMR spectrum of the serum sample numbered 1 in the M0 sample group, and (b) is the 1 HNMR spectrum of the serum sample numbered 1 in the M0 sample group after calibration of the characteristic peaks of bile acids; Figure 2 Among them, A represents the 1 HNMR spectrum of the serum sample numbered 50 in the M0 sample group after calibration of the characteristic peaks of bile acids, and B represents the 1 HNMR spectrum of the serum sample numbered 100 in the MVI sample group after calibration of the characteristic peaks of bile acids.

[0051] M0 / MVI discrimination analysis of HNMR full spectrum based on calibration of characteristic peaks of bile acids: An OPLS-DA model was established using the above calibration data to perform discrimination analysis on the serum. The orthogonal variable component of the model was 5, R2Y was 0.892, and Q2 was 0.228. As 1 shown, Figure 3 in the figure, legend 1 represents the M0 group, and legend 2 represents the MVI group. Figure 3 Among them, the MVI group and the M0 group are clearly separated. The MVI group is all aggregated on the right side of the OPLS-DA score plot, and the M0 group is all aggregated on the left side of the OPLS-DA score plot, indicating that there are obvious differences in the Figure 3 M0 / MVI of the HNMR full spectrum based on calibration of characteristic peaks of bile acids. 1 HNMR full spectrum M0 / MVI has obvious differences.

[0052] The constructed OPLS-DA model was subjected to 200 permutation tests to further test the validity of the model (completed by SMICA software), and the results are as Figure 4 shown. Figure 4 Among them, R2 represents the interpretability of the model for the classification variable Y, and Q2 represents the predictability of the model. Figure 4 It shows that Q2 is negative (-0.241), indicating that the original model has good robustness and there is no overfitting phenomenon, and it can be used for the prediction of unknown samples.

[0053] Example 2

[0054] The established OPLS-DA model in Example 1 was used to predict unknown (samples not used for OPLS-DA modeling) MVI / M0 serum samples, and the method is as follows:

[0055] Collection of unknown MVI / M0 serum samples: Another 20 serum samples of MVI / M0 with matched gender and age were collected, including 10 in the M0 group and 10 in the MVI group, and they were numbered respectively.

[0056] The unknown MVI / M0 serum samples were measured according to the method of Example 1 to obtain the samples 1 H-NMR spectra, and the data were processed. The established OPLS-DA model in Example 1 was used to predict the unknown samples, and the results are as Figure 5 shown. In Figure 5 "Workset" represents the samples for constructing the OPLS-DA model in Example 1, and "Predictionset" represents the prediction samples in Example 2. It can be seen that the prediction samples are basically divided into two categories. One category (numbered 110-119) clusters with the M0 samples in Example 1, and the other category (numbered 125-134) clusters with the MVI samples in Example 1. It is preliminarily predicted that the samples numbered 110-119 are M0 serum samples, and the samples numbered 125-134 are MVI serum samples. From Figure 5 it is difficult to clearly see whether the classification is correct. Therefore, a graph is made with the prediction score values of all samples, and the results are as Figure 6 shown. From Figure 6 it can be seen that for the prediction of 10 M0 samples, 7 have scores greater than 0.5, 3 are misclassified, and the correct rate is 70%; while for 10 MVI prediction samples, 8 have classification scores less than 0.5, and the correct rate is 80%. The average prediction accuracy of 20 unknown samples is 75%. Therefore, this method has the advantages of simple sample processing, relatively high prediction accuracy, and can achieve rapid identification of MVI / M0 in one analysis, and has certain clinical application value.

[0057] Comparative Example 1

[0058] 1 HNMR spectrum data preprocessing method: For all data points in the " 1 HNMR spectra" in Example 1 and Example 2, the ratio of the integral of bile acids is calculated (bile acid calibration), specifically as follows: x j,new = x j,old / Int bile , where x j,new represents the new variable after bile acid calibration, x j,old, is the original variable, and Int bile represents the integral value of the bile acid characteristic peak at 0.70-0.74 ppm, obtaining 1 the relative quantitative value of the HNMR full spectrum" is changed to " 1 For all data points in the " j,new HNMR spectra", the ratio of the data point intensity sum is calculated (full spectrum normalization), specifically as follows: x j,old = x j / sum j,new , where x j,old, represents the new variable after full spectrum normalization calibration, x j,old, is the original variable, and sumj The sum of the intensities representing each data point is obtained to 1 "HNMR full-spectrum normalization value", and the rest is the same as in Example 1 and Example 2.

[0059] Using the above 1 HNMR full-spectrum normalization value to establish an OPLS-DA model, the orthogonal variable component of the model is 3, R2Y is 0.847, Q2 is 0.173, and 20 unknown samples (the same as in Example 2) are predicted. The results are as Figure 7 shown. From Figure 7 it can be seen that the clustering of samples in the MVI group and the M0 group is not as obvious as that of the bile acid calibration ( Figure 3 ), and more importantly, plotting the prediction score values of all samples in the comparative example, as Figure 8 shown, the accuracy of predicting 10 M0 samples is only 40% (only 4 samples have scores greater than 0.5), with a relatively high false positive rate. The accuracy of predicting MVI is consistent with the bile acid calibration spectrum (8 samples have classification scores less than 0.5), which is 80%. The average prediction accuracy of 20 unknown samples is 60%, lower than the bile acid calibration method.

[0060] Comparative Example 2

[0061] 1 HNMR spectrum data preprocessing method: For all data points in the HNMR spectra in Example 1 and Example 2, the ratio of the integral of choline metabolites is calculated (choline calibration), specifically as follows: x 1 = x j,new / Int j,old where x choline represents the new variable after choline calibration, x j,new is the original variable, and Int j,old, represents the integral value of the choline characteristic peak at 3.21 - 3.24 ppm, obtaining choline the relative quantitative value of choline calibration for the HNMR full spectrum, and the rest is the same as in Example 1 and Example 2. 1 Using the above choline calibration

[0062] HNMR full spectrum to establish an OPLS-DA model, the orthogonal variable component of the model is 5, R2Y is 0.933, Q2 is 0.195, and the above 20 unknown samples (the same as in Example 2) are predicted. The results are as 1 shown. The accuracy of predicting 10 M0 samples is 50% (5 samples have scores greater than 0.5); for 10 MVI predicted samples, 7 samples have classification scores less than 0.5, and the correct rate is 70%. The average prediction accuracy of 20 unknown samples is 60%, lower than the bile acid calibration method. Figure 9 shown. The accuracy of predicting 10 M0 samples is 50% (5 samples have scores greater than 0.5); for 10 MVI predicted samples, 7 samples have classification scores less than 0.5, and the correct rate is 70%. The average prediction accuracy of 20 unknown samples is 60%, lower than the bile acid calibration method.

[0063] As can be seen from the above embodiments, the present invention uses nuclear magnetic resonance hydrogen spectrum technology in combination with conventional data processing software and data modeling software to construct a model, and can quickly and accurately predict positive or negative microvascular invasion of liver cancer through one analysis, having broad application prospects.

[0064] The above are only the preferred embodiments of the present invention. It should be noted that for those of ordinary skill in the art, without departing from the principle of the present invention, several improvements and refinements can still be made, and these improvements and refinements should also be regarded as the protection scope of the present invention.

Claims

1. A method for constructing a differential diagnosis model of microvascular invasion in liver cancer, characterized in that Comprising the following steps: Providing positive serum samples and negative serum samples of hepatocellular carcinoma microvascular invasion; Measure the 1 HNMR spectra of the positive and negative serum samples, and use the ratio of the full-spectrum data in the 1 HNMR spectrum to the peak height or integral of the bile acid characteristic peaks as the independent variable, and the positive or negative of hepatocellular carcinoma microvascular invasion as the observation variable. Perform supervised clustering analysis using orthogonal partial least squares discriminant analysis to obtain the OPLS-DA model.

2. The method according to claim 1, characterized in that, The method for measuring the 1 HNMR spectra of positive and negative serum samples comprises the following steps: (1) Mixing the positive serum samples and negative serum samples with methanol and phosphate buffer respectively, vortexing, ultrasonically treating and centrifuging the obtained mixture in sequence, taking the supernatant and concentrating it to dryness to obtain serum metabolites; (2) Mix the serum metabolite with a deuterated reagent, vortex, ultrasonicate, and centrifuge the resulting mixture in sequence to obtain a supernatant; take the supernatant for nuclear magnetic resonance detection to obtain 1 an HNMR spectrum; add tetramethylsilane as an internal standard to the deuterated reagent.

3. The method according to claim 2, wherein The volume ratios of the positive serum samples and negative serum samples to methanol are 1:4 to 20 respectively.

4. The method according to claim 2, characterized in that, The pH value of the phosphate buffer is 2 to 8, and the concentration is 0.05 to 3 mol / L; the dosage ratios of the positive serum samples and negative serum samples to the phosphate buffer are 1.0 mL:50 to 200 μL respectively.

5. The method according to claim 2, wherein In the step (1), the time of vortexing is 3 to 15 min; the time of ultrasonically treating is 5 to 60 min; the speed of centrifuging is 8000 to 13000 rpm, and the time is 5 to 20 min; the concentration is rotary evaporation, the temperature of the rotary evaporation is 40 to 55 °C, and the time is 3 to 6 h.

6. The method according to claim 2, wherein The conditions for nuclear magnetic resonance detection in the step (2) include: a 5 mm multinuclear broadband observation probe, an observation frequency of 600.13 MHz, a spectral width SW of 19.8 ppm, a spectral center point O1P of 4 to 10 ppm, and a number of scans NS of 64 to 256 times.

7. The method according to claim 1, characterized in that The chemical shift of the characteristic peak of the bile acid is 0.70 to 0.74 ppm.

8. The method according to claim 1, wherein The number of independent variables is 10K to 60K.

9. The method according to claim 1, characterized in that The orthogonal variable components of the OPLS-DA model are 3 to 10, R2Y is 0.6 to 0.9, and Q2 is 0.2 to 0.6.

Images