A diagnostic biomarker for liver cancer based on exosomal miRNA and its application

Abstract

This invention provides a diagnostic or prognostic biomarker for liver cancer, a liver cancer diagnostic kit based thereon, and their applications. The liver cancer diagnostic or prognostic biomarker of this invention involves biologically significant miRNA pairs in serum or plasma exosomes. A liver cancer diagnostic model constructed using the expression ratio of these miRNA pairs as a model feature exhibits high diagnostic sensitivity and specificity for liver cancer and shows significant potential in early liver cancer screening, thus possessing important practical significance for early diagnosis or prognostic prediction of liver cancer.

Description

[0001] This application is a divisional application of Chinese Patent Application No. 202311541101.5, filed on November 17, 2023, entitled "A diagnostic biomarker for liver cancer based on exosomal miRNA and its application therein". Technical Field

[0002] This invention belongs to the field of medical testing technology, specifically relating to a liver cancer diagnostic biomarker based on exosomal miRNA and its application. Background Technology

[0003] Liver cancer is one of the most common malignant tumors in clinical practice.

[0004] Because liver cancer is characterized by its insidious onset, early-stage liver cancer is difficult to detect, while late-stage liver cancer progresses rapidly, has limited treatment options, is difficult to cure, and has a poor prognosis, resulting in a low survival rate for liver cancer in my country. Studies have shown that patients diagnosed with early-stage liver cancer can achieve a 5-year survival rate of 70-75% through radical treatments such as liver transplantation, surgical tumor resection, or local ablation therapy. Therefore, in addition to basic prevention measures such as vaccination, early screening for liver cancer is crucial for reducing its incidence and mortality.

[0005] Currently, the most commonly used and traditional diagnostic techniques for liver cancer include imaging and serum alpha-fetoprotein (AFP). Imaging includes ultrasound and CT scans; however, imaging is typically used to diagnose intrahepatic space-occupying lesions and lacks sensitivity and specificity for detecting early-stage liver cancer or nodular cirrhosis without substantial lesions. Furthermore, these diagnostic methods are subjective, depending on operator experience and equipment sensitivity. Serum AFP is a commonly used and important indicator for diagnosing liver cancer and monitoring treatment efficacy. However, the sensitivity of AFP alone is only 40-65%, making it less than ideal as an early diagnostic indicator. The GALAD model, utilizing protein biomarkers such as AFP and DCP, pioneered multi-indicator combined modeling in China, improving the sensitivity and specificity for early-stage liver cancer. Therefore, molecular diagnostic methods are a powerful supplement to the auxiliary diagnosis of liver cancer.

[0006] Exosomes are small vesicles secreted by cells and entering body fluids or the extracellular environment. With a diameter of only 30-100 nanometers, they play a crucial role in intercellular communication. Protected by a lipid bilayer, exosomes can stably exist in various biological fluids and cell culture media. miRNAs are present within the contents of exosomes and can be released from cancer cells into body fluids through exosome encapsulation, without being degraded by ribosomal enzymes. Previous studies have shown that over half of the protein molecules in the serum of liver cancer patients exhibit fucosylation. The relationship between these proteins with specific glycosylation changes and the development and progression of liver disease may play a significant role in the diagnosis, treatment, and prognosis of liver diseases. Therefore, analyzing the relationship between miRNAs in fucosylated exosomes and the occurrence and development of liver cancer will contribute to the development of more targeted diagnostic biomarkers for liver cancer. Summary of the Invention

[0007] Purpose of the invention

[0008] In view of the problems existing in the above-mentioned prior art methods, the present invention aims to provide a highly sensitive and specific diagnostic or prognostic biomarker for liver cancer, a kit based thereon, or its application, including: its application in constructing models for the diagnosis, efficacy or prognostic assessment of liver cancer, or its application in the preparation of drugs or kits for the diagnosis, efficacy or prognostic assessment of liver cancer.

[0009] Solution

[0010] To achieve the above objectives, the present invention provides the following technical solution:

[0011] In a first aspect, the present invention provides a diagnostic or prognostic biomarker for liver cancer, said diagnostic or prognostic biomarker comprising a miRNA pair consisting of any two of the following miRNAs, or any combination of such miRNA pairs:

[0012] hsa-let-7a, hsa-miR-21, hsa-miR-125a, hsa-miR-150, hsa-miR-200a, hsa-miR-483, hsa-miR-199a, hsa-miR-200a, hsa-miR-429, hsa-miR-126, hsa-miR-381, hsa-miR-185, hsa-miR-215, hsa-miR-374a.

[0013] In a preferred embodiment, the diagnostic or prognostic biomarker for liver cancer comprises a pair of miRNAs selected from the following:

[0014] hsa-miR-200a / hsa-miR-150;

[0015] hsa-miR-483 / hsa-miR-199a;

[0016] hsa-miR-200a / hsa-miR-199a;

[0017] hsa-miR-150 / hsa-miR-429;

[0018] hsa-miR-126 / hsa-miR-200a;

[0019] hsa-miR-199a / hsa-miR-429;

[0020] hsa-miR-381 / hsa-miR-200a;

[0021] hsa-miR-185 / hsa-miR-429;

[0022] hsa-miR-215 / hsa-miR-199a;

[0023] hsa-miR-125a / hsa-miR-215;

[0024] hsa-let-7a / hsa-miR-21;

[0025] hsa-miR-200a / hsa-miR-374a;

[0026] hsa-miR-125a / hsa-miR-21;

[0027] hsa-miR-21 / hsa-miR-150; and,

[0028] Any one, several, or all of the aforementioned items in combination.

[0029] In a preferred embodiment, the diagnostic or prognostic biomarker for liver cancer comprises a pair of miRNAs selected from the following:

[0030] hsa-let-7a / hsa-miR-21;

[0031] hsa-miR-125a / hsa-miR-21;

[0032] hsa-miR-21 / hsa-miR-150;

[0033] hsa-miR-200a / hsa-miR-150; and,

[0034] Any one, two, three, or all of the aforementioned items.

[0035] In some specific implementations, the diagnostic or prognostic biomarkers for liver cancer include combinations of hsa-let-7a / hsa-miR-21, hsa-miR-125a / hsa-miR-21, hsa-miR-21 / hsa-miR-150, and hsa-miR-200a / hsa-miR-150.

[0036] In some other specific embodiments, the diagnostic or prognostic biomarkers for liver cancer include combinations of hsa-let-7a / hsa-miR-21, hsa-miR-200a / hsa-miR-150, and hsa-miR-125a / hsa-miR-21.

[0037] In some other specific embodiments, the diagnostic or prognostic biomarkers for liver cancer include a combination of hsa-miR-125a / hsa-miR-21 and hsa-miR-21 / hsa-miR-150.

[0038] Preferably, the miRNA in the miRNA pair is a miRNA from serum or plasma, and more preferably a miRNA from serum or plasma fucosylated exosomes.

[0039] More preferably, the diagnostic or prognostic markers for liver cancer further include any one or more of AFP, AFP-L3, and DCP; preferably, AFP, AFP-L3, or DCP are corresponding proteins in serum, plasma, or whole blood.

[0040] In a second aspect, the present invention provides the application of the diagnostic or prognostic biomarkers for liver cancer as described in the first aspect above in the construction of models for the diagnosis, efficacy or prognostic assessment of liver cancer.

[0041] In a specific implementation scheme, the expression ratio of the miRNA pairs is used as a model feature, and machine learning methods are used for modeling.

[0042] The machine learning method mentioned can be a commonly used machine learning method in the field of modeling, such as logistic regression, support vector machine, tree model, neural network, etc.

[0043] Thirdly, the present invention provides the use of the diagnostic or prognostic biomarkers for liver cancer as described in the first aspect above, or their detection reagents, in the preparation of medicaments or kits for the diagnosis, efficacy or prognostic assessment of liver cancer.

[0044] Fourthly, the present invention provides a drug or kit for the diagnosis, efficacy or prognosis assessment of liver cancer, the drug or kit comprising the diagnostic or prognostic biomarkers for liver cancer as described in the first aspect above or their detection reagents.

[0045] In patients with liver cancer, the expression ratio of the above miRNA pairs in serum fucosylated exosomes was significantly different from that in healthy individuals.

[0046] Beneficial effects

[0047] The diagnostic or prognostic biomarkers for liver cancer provided by this invention involve biologically significant miRNA pairs in serum or plasma exosomes. The expression ratio of these miRNA pairs is used as a characteristic of the liver cancer diagnostic model, avoiding the instability of using the expression level of a single miRNA as a characteristic, thereby enabling more accurate diagnosis or prognostic assessment. The liver cancer diagnostic model constructed using the expression ratio of the miRNA pairs as a model characteristic has high diagnostic sensitivity and specificity (above 85% and 90%, respectively), and also has significant potential for early liver cancer screening. Attached Figure Description

[0048] One or more embodiments are illustrated by way of example with reference to the accompanying drawings, and these illustrative examples are not intended to limit the embodiments. Here, the specific term "illustrative" means "serving as an example, embodiment, or illustration." Any embodiment illustrated as "illustrative" is not necessarily to be construed as superior to or better than other embodiments.

[0049] Figure 1 This is a heatmap of the expression ratios of the 14 paired miRNAs screened in Example 1 on NGS samples.

[0050] Figure 2 The model effect of the combination of expression ratios of the 14 paired miRNAs screened in Example 1 on NGS samples: (a) shows the ROC curve, and (b) shows the confusion matrix.

[0051] Figure 3 The box plot of the expression ratios of the four paired miRNAs screened in Example 1 on NGS data is as described in Example 2; where T and N represent the hepatocellular carcinoma group and the non-hepatocellular carcinoma control group, respectively.

[0052] Figure 4 Box plot of the CT value difference of the three paired miRNAs screened in Example 1 for qPCR, as described in Example 3; where T and N represent the liver cancer group and the non-liver cancer control group, respectively.

[0053] Figure 5 The diagram shows the diagnostic performance of a liver cancer diagnostic model constructed based on a preferred combination of paired miRNAs; (a) shows the ROC curve, (b) shows the confusion matrix, and (c) shows the Venn diagram of positive samples.

[0054] Figure 6 The figure shows the effect of different stages of liver disease on a liver cancer diagnostic model constructed based on the optimal combination of paired miRNAs; where (a) shows the ROC curve and (b) shows the confusion matrix.

[0055] Figure 7 The diagram shows the diagnostic efficacy of a liver cancer diagnostic model constructed based on a preferred combination of paired miRNAs and a triple detection index for liver cancer. Figure (a) shows the ROC curve, Figure (b) shows the confusion matrix, and Figure (c) shows the Venn diagram of positive samples.

[0056] Figure 8 The figure shows the effect of different stages of liver disease on the liver cancer diagnostic model constructed based on the optimal combination of paired miRNAs and the triple detection index of liver cancer; wherein, (a) shows the ROC curve and (b) shows the confusion matrix.

[0057] Figure 9 Box plot of the CT value difference of qPCR for the two paired miRNAs screened in Example 1, as described in Example 6; where T and N represent the liver cancer group and the non-liver cancer control group, respectively.

[0058] Figure 10 The diagram shows the diagnostic performance of the liver cancer diagnostic model constructed based on the preferred combination of paired miRNAs; where (a) shows the ROC curve and (b) shows the confusion matrix.

[0059] Figure 11 The diagram shows the diagnostic efficacy of a liver cancer diagnostic model constructed based on a preferred combination of paired miRNAs and a triple detection index for liver cancer; Figure (a) shows the ROC curve, and Figure (b) shows the confusion matrix.

[0060] Figure 12 The diagram shows the diagnostic efficacy of a liver cancer model constructed based on the preferred combination of paired miRNAs and AFP; where (a) shows the ROC curve and (b) shows the confusion matrix.

[0061] Figure 13 Venn diagrams showing the predictions of 160 liver cancer samples using the single AFP indicator (i.e., “AFP”), a diagnostic model based on a preferred combination of paired miRNAs (i.e., “miR”), and a diagnostic model based on a preferred combination of paired miRNAs combined with AFP (i.e., “miR+AFP”). Detailed Implementation

[0062] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention. Unless otherwise expressly stated, throughout the specification and claims, the term "comprising" or its variations such as "including" or "comprising of," etc., will be understood to include the stated elements or components, and does not exclude other elements or other components.

[0063] Furthermore, to better illustrate the present invention, numerous specific details are provided in the following detailed embodiments. Those skilled in the art should understand that the present invention can be practiced without certain specific details. In some embodiments, materials, elements, methods, and means well known to those skilled in the art are not described in detail in order to highlight the spirit of the invention.

[0064] Unless otherwise specified, the reagents, kits, raw materials, and equipment used in the following examples can all be purchased commercially. Unless otherwise specified, the experimental or detection methods involved in this invention are conventional experimental or detection methods in the art, or can be performed with reference to the corresponding kits or product instructions.

[0065] Example 1: Screening for characteristic miRNAs of liver cancer based on serum exosome data.

[0066] 1. Collection and preparation of clinical data and samples

[0067] A retrospective analysis was conducted, and serum samples were collected from patients with liver cancer, patients with benign liver disease, and healthy individuals. Sample information is shown in Table 1.

[0068] As shown in Table 1, the dataset is divided into six cohorts: the NGS cohort is used for the discovery of characteristic miRNA pairs in liver cancer; and the five PCR cohorts are used for modeling and validation of two preferred combinations.

[0069] Liver cancer patients meeting the following inclusion criteria were primarily selected: early-stage cases treatable by radical surgery; pathological diagnosis confirming the tumor's Edmoson staging; complete basic information for all cases; exclusion of other causes of chronic liver disease such as alcoholic fatty liver disease and autoimmune liver disease; exclusion of pregnancy, germ cell tumors, malignant tumors of other organs, severe infectious diseases, and diseases of other vital organs. Benign liver disease samples included hepatitis, cirrhosis, and intrahepatic hemangioma, excluding post-liver cancer surgery, liver transplantation, and malignant tumors of other organs.

[0070] Table 1. Statistics of Selected Cases:

[0071]

[0072]

[0073] 2. Isolation and purification of exosomes

[0074] Blood was collected in 5 mL vacuum blood collection tubes (coagulant + separating gel) and centrifuged at 1800×g for 10 minutes at room temperature within 2 hours after sampling. The resulting serum was centrifuged at 3000×g for 10 minutes at 4°C to remove any cell debris and apoptotic bodies. The precipitate was then discarded, and the supernatant serum was extracted and stored at -80°C, avoiding repeated freeze-thaw cycles. Exosomes were isolated using the GlyExo-Capture kit, in which the magnetic beads used were the company's patented "A lectin-magnetic carrier coupling complex for separating glycosylated exosomes from clinical samples".

[0075] 3. Extraction and NGS sequencing of exosomal miRNAs

[0076] use Following the instructions in the Mini Kit, total RNA was extracted from extracellular vesicles and evaluated using the high-sensitivity RNA kit of the Qsep100 automated nucleic acid analysis system. Then, the exosomal small RNAs were compiled into cDNA libraries using the Illumina NEBNext small RNA library preparation kit. Library fragments of the desired size were selected and purified using an E-Gel Power Snap electrophoresis system and an E-Gel SizeSelect II gel. After checking the quality and concentration of the cDNA libraries, 75nt single-end sequencing was performed on an Illumina NextSeq 550 sequencing system, with each library generating more than 10M reads.

[0077] 4. Construct a miRNA interaction network and analyze NGS sequencing data. Through single-factor screening and genetic algorithm screening, identify characteristic miRNA pairs for liver cancer.

[0078] (1) Construction of miRNA interaction network

[0079] 1) Obtain the target sites of miRNAs from the miRTarBase database;

[0080] 2) Based on the miRNA target information obtained in step 1), transcription factors that can serve as miRNA targets are screened from the human transcription factor databases hTFtarget and AnimalTFDB.

[0081] 3) Based on the transcription factors screened in step 2), obtain the miRNAs that they further regulate through bioinformatics methods or public databases, thereby constructing the miRNA-TF-miRNA interaction relationship and obtaining the miRNA interaction network.

[0082] (2) Analyze NGS sequencing data and prepare miRNA quantification data.

[0083] By analyzing the above NGS sequencing data, we obtained miRNA quantification data for each sample in a comprehensive sample set that includes disease samples and control samples:

[0084] First, raw sequencing files in FASTQ format are required for quality control. Based on the quality control results, cutadapt software is used to remove adapters and low-quality reads. For the quality control-qualified data, the exceRpt small RNA analysis workflow is used for annotation and quantification to obtain the expression matrix. Based on the expression levels in the expression matrix, miRNAs with excessively low counts are filtered out (specific analysis is required for each case) to obtain corrected miRNA quantification data.

[0085] (3) Construction of miRNA expression level ratio characteristics

[0086] Based on the miRNA interaction network constructed above and the prepared miRNA quantitative data, the expression ratio of miRNA pairs with interaction relationships in each sample was calculated.

[0087] To avoid a denominator of 0, a "1" is added to the denominator when calculating the miRNA-to-expression ratio. The calculation formula is as follows:

[0088] miRNA_a / miRNA_b = counts a / (count b +1)

[0089] (4) Screening of characteristic miRNA pairs

[0090] Characteristic miRNA pairs for liver cancer were obtained through single-factor screening and genetic algorithm screening. The basic process is as follows:

[0091] i) Single-factor screening: Compare the expression ratio of each miRNA pair between the disease biological sample group and the normal biological sample group. Use scipy.stats.ttest_ind in Python to calculate the p value, and use statsmodels.stats.multitest.fdrcorrection to correct the p value. For the corrected p value p-adjusted, screen based on a threshold of 0.05.

[0092] ii) For the selected miRNA pairs, calculate the logarithm of the change in the expression ratio of the disease biological sample group relative to the expression ratio of the normal biological sample group, i.e., log2FoldChange, and select an appropriate threshold for log2FoldChange according to the actual situation to further screen suitable targets.

[0093] The genetic algorithm screening is performed more than 100 times.

[0094] After the above screening procedure, 14 paired miRNAs were obtained, and their details are shown in Table 2 below:

[0095] hsa-miR-200a / hsa-miR-150 hsa-miR-199a / hsa-miR-429 hsa-let-7a / hsa-miR-21 hsa-miR-483 / hsa-miR-199a hsa-miR-381 / hsa-miR-200a hsa-miR-200a / hsa-miR-374a hsa-miR-200a / hsa-miR-199a hsa-miR-185 / hsa-miR-429 hsa-miR-125a / hsa-miR-21 hsa-miR-150 / hsa-miR-429 hsa-miR-215 / hsa-miR-199a hsa-miR-21 / hsa-miR-150 hsa-miR-126 / hsa-miR-200a hsa-miR-125a / hsa-miR-215 .

[0096] In addition, a heatmap of the expression ratios of these 14 paired miRNAs was plotted on the NGS cohort, see [link to heatmap]. Figure 1 ; Figure 1 The heatmap shows that the expression ratios of these 14 paired miRNAs performed well in the clustering of cancer and non-cancer samples in the NGS cohort. This suggests that these 14 miRNA pairs, and any combination thereof, have the potential to serve as biomarkers for liver cancer.

[0097] To verify this, the model performance of the combination of these 14 miRNA pairs is then validated using an example. Specifically, the training and test sets are randomly divided in a 7:3 ratio on the NGS queue. A random forest model is used for modeling on the training set, and validation is performed on the test set. The ROC curve and confusion matrix of the model built based on the above 14 miRNA pair combinations are shown below. Figure 2 ; Figure 2 The results show that the AUCs for the training and validation sets are 0.98 and 0.93, respectively, and the sensitivity and specificity are 90.12% and 96.04%, and 88.24% and 90.91%, respectively, indicating that the established model has good classification performance.

[0098] 5. Further screening for characteristic miRNAs of liver cancer.

[0099] Recursive feature elimination was performed on the expression ratios of the 14 paired miRNAs selected above, ultimately yielding the 8 paired miRNA combinations with the highest cross-validation AUC. Using qPCR to confirm the cohort samples, the 8 paired miRNAs were validated, and 4 paired miRNAs with consistent trends and significant differences between cancer and control groups were selected, as shown below:

[0100] hsa-let-7a / hsa-miR-21;

[0101] hsa-miR-125a / hsa-miR-21;

[0102] hsa-miR-200a / hsa-miR-150;

[0103] hsa-miR-21 / hsa-miR-150.

[0104] Example 2: Box plot visualization of four paired miRNAs using NGS quantitative data.

[0105] Box plots were tested and generated on the NGS cohort for the four paired miRNAs selected in Example 1. The results are as follows: Figure 3 As shown.

[0106] Depend on Figure 3 The box plots show that the expression ratios of these four paired miRNAs differed significantly between the liver cancer group and the control group (p<0.5), indicating that these four paired miRNAs have the potential to become biomarkers for liver cancer.

[0107] Example 3: Verification of the hepatocellular carcinoma diagnostic performance of the preferred combination of paired miRNAs by RT-qPCR.

[0108] Among the four paired miRNAs screened in Example 1, three paired miRNAs, hsa-let-7a / hsa-miR-21, hsa-miR-125a / hsa-miR-21, and hsa-miR-200a / hsa-miR-150, were selected as the preferred combination. They were validated in-center and multi-center validation cohorts using RT-qPCR, and the sample information is shown in Table 1.

[0109] The specific verification method is as follows:

[0110] First, the first-strand cDNA was synthesized. The reaction system and reaction conditions are as follows:

[0111] Reaction system and volume:

[0112]

[0113]

[0114] Reaction conditions and time:

[0115] Reaction conditions Time (minutes) 42℃ 60 85℃ 5 4℃ ∞ .

[0116] Then, miRNA qPCR was performed on an ABI 7500 real-time quantitative PCR system. The reaction system and conditions are as follows:

[0117] Reaction system and volume:

[0118]

[0119] Reaction conditions and time:

[0120]

[0121] The primer sequences used in the above RT-qPCR program are as follows:

[0122]

[0123]

[0124] The results are as follows:

[0125] Box plots showing the CT value differences for paired miRNA qPCRs across different qPCR cohorts are shown below. Figure 4 ;Depend on Figure 4 It can be seen that the trends of these three paired miRNAs are consistent with those of the NGS data, both within and outside the center, and are relatively stable across different centers. This indicates that the three selected paired miRNAs are also applicable to the RT-qPCR cohort, showing significant differences between the hepatocellular carcinoma group and the non-hepatocellular carcinoma control group, without any bias in the detection method or significant inter-individual differences.

[0126] Example 4: Constructing a liver cancer diagnostic model based on a preferred combination of paired miRNAs and verifying its diagnostic performance.

[0127] To evaluate the diagnostic efficacy of the preferred combination of the three paired miRNAs described above for liver cancer, this embodiment utilizes logistic regression for modeling on a training cohort and validates the results on both an in-center and multi-center validation set. The specific method is as follows:

[0128] The sample information used is shown in Table 1: the qPCR confirmation cohort was used as the training set, and the in-center validation cohort and multi-center validation cohort were used as validation sets. A logistic regression model was established for the three paired miRNAs in the training set, and the cutoff was determined according to the principle of maximizing the Youden exponent in the training set. The overall performance of the model was evaluated using the AUC, sensitivity, and specificity of the ROC curve. Venn diagrams were used to examine the crossover between model-predicted positivity and AFP positivity in samples with actual diagnostic results of liver cancer. Based on the above methods, statistical analyses were performed on samples from the entire cohort and separately extracted samples from early-stage liver cancer and benign liver diseases.

[0129] The diagnostic performance of the model is as follows Figure 5 As shown; Figure 5The results show that the AUCs for the training set, the in-center validation set, and the multi-center validation set are 0.996, 0.951, and 0.944, respectively. Figure 5 (a)), with specific cutoffs, the sensitivity and specificity were 94.34%, 87.50%, 86.76% and 96.20%, 93.48%, 93.14%, respectively. Figure 5 (b) This indicates that the model performance of the three paired miRNAs is at a high level and the performance is basically stable across different centers.

[0130] Simultaneously, this embodiment also detected AFP in positive samples and compared the predictive performance with that of the aforementioned model. The results showed that the overall sensitivity of this model in predicting liver cancer reached 88.54%, while the sensitivity for AFP was only 61.26%. Figure 5 (c)).

[0131] In addition, by Figure 5 (c) It can also be concluded that the liver cancer diagnostic model based on the combination of these three paired miRNAs has strong complementarity with AFP, and can detect 87.76% of samples that were not detected by AFP. This indicates that the model based on the above three paired miRNAs has a strong predictive effect on liver cancer and can serve as a diagnostic biomarker for liver cancer.

[0132] Since the purpose of early liver cancer screening is to detect early-stage liver cancer lesions in patients with benign liver disease, this embodiment extracted samples of early-stage liver cancer and benign liver disease separately, and examined the prediction effect on the model. The results are as follows: Figure 6 As shown; Figure 6 The results showed that the AUC of 0.936 for early-stage liver cancer samples and benign liver disease samples was 84.29% and 92.20%, respectively, both of which were at a high level. This indicates that the optimal combination of the three paired miRNAs we screened has great potential in the screening of early-stage liver cancer.

[0133] Example 5: Constructing a liver cancer diagnostic model based on a preferred combination of paired miRNAs and a triple detection index for liver cancer (AFP, AFP-L3, and DCP).

[0134] In this embodiment, joint modeling is performed using the three paired miRNAs and the triple detection indicators for liver cancer (AFP, AFP-L3, and DCP). The sample information used for the training and validation sets is consistent with that in Example 4. The specific modeling method is as follows:

[0135] The sample information used is shown in Table 1: the qPCR confirmation cohort was used as the training set, and the in-center validation cohort and multi-center validation cohort were used as validation sets. A logistic regression model was established for the three paired miRNAs in the training set, and the cutoff was determined according to the principle of maximizing the Youden exponent in the training set. The overall performance of the model was evaluated using the AUC, sensitivity, and specificity of the ROC curve. Venn diagrams were used to examine the crossover between model-predicted positivity and AFP positivity in samples with actual diagnostic results of liver cancer. Based on the above methods, statistical analyses were performed on samples from the entire cohort and separately extracted samples from early-stage liver cancer and benign liver diseases.

[0136] The diagnostic performance of the model is as follows Figure 7 As shown; by Figure 7 From the ROC curves in (a), we can see that the AUCs for the training set, the in-center validation set, and the multi-center validation set are 1, 0.983, and 0.969, respectively; Figure 7 (b) The confusion matrix shows that the sensitivity and specificity at a specific cutoff are 100%, 92.19%, 90.44% and 97.47%, 100%, 95.67%, respectively; and, from Figure 7 The Venn diagram of the positive samples in (c) shows that this model can detect 95.48% of AFP-positive and 88.78% of AFP-negative cancer samples. This indicates that the model based on the above three paired miRNA pairs and the triple detection index for liver cancer has a strong predictive effect on liver cancer.

[0137] Furthermore, this embodiment also tested the performance of early-stage liver cancer samples compared to benign liver disease samples on the model, and the results are as follows: Figure 8 As shown; Figure 8 The results showed that the AUC of 0.958 for early-stage liver cancer samples and benign liver disease samples was 91.43% and 95.41%, respectively. This indicates that the three paired miRNAs we screened, combined with the triple detection index for liver cancer, have significant potential for screening early-stage liver cancer.

[0138] Example 6: Verification of the hepatocellular carcinoma diagnostic performance of the preferred combination of paired miRNAs 2 by RT-qPCR.

[0139] Of the four paired miRNAs screened in Example 1, two paired miRNAs (hsa-miR-125a / hsa-miR-21 and hsa-miR-21 / hsa-miR-150) were selected as preferred combination two. They were validated using RT-qPCR in both in-center and multi-center validation cohorts; sample information is shown in Table 1. The amplification primers and reaction system are described in Example 3. The box plot of the CT value difference between the two paired miRNAs in the qPCR is shown in [Table 1]. Figure 9 ; Figure 9The results of qPCR validation of the preferred combination of the two paired miRNAs showed a significant difference between the hepatocellular carcinoma group and the non-hepatocellular carcinoma control group.

[0140] Example 7: Constructing a liver cancer diagnostic model based on a preferred combination of paired miRNAs and validating its diagnostic performance.

[0141] To evaluate the diagnostic efficacy of the preferred combination two, consisting of the two paired miRNAs described above, for liver cancer, this embodiment utilizes logistic regression for modeling on the training cohort and validates it on the validation set. The specific method is as follows:

[0142] The sample information used is shown in Table 1: the preferred combination qPCR training cohort was used as the training set, and the preferred combination qPCR validation cohort was used as the validation set. A logistic regression model was established for the two paired miRNAs in the training set, and the cutoff was determined according to the principle of maximizing the Youden exponent in the training set. The overall performance of the model was evaluated using the AUC of the ROC curve, as well as its sensitivity and specificity.

[0143] The diagnostic performance of the model is as follows Figure 10 As shown; Figure 10 The results show that the AUCs for the training and validation sets are 0.916 and 0.922, respectively. Figure 10 (a)), with specific cutoffs, the sensitivity and specificity were 84.38% and 84.38%, and 88.05% and 91.59%, respectively. Figure 10 (b) This indicates that the diagnostic efficacy of the model based on two paired miRNAs is at a high level and the efficacy is basically stable across different batches of data.

[0144] Example 8: A liver cancer diagnostic model was constructed based on a preferred combination of paired miRNAs and a triple detection index for liver cancer (AFP, AFP-L3, and DCP), and its diagnostic performance was verified.

[0145] In this embodiment, joint modeling is performed using the two paired miRNAs and the triple markers for liver cancer detection (AFP, AFP-L3, and DCP). The sample information used for the training and validation sets is consistent with that in Example 7. The specific modeling method is as follows:

[0146] The sample information used is shown in Table 1: the preferred combination qPCR training cohort was used as the training set, and the preferred combination qPCR validation cohort was used as the validation set. Logistic regression models were established for the two paired miRNAs and triplet detection in the training set, and the cutoff was determined according to the principle of maximizing the Youden exponent of the training set. The overall performance of the model was evaluated by the AUC of the ROC curve, as well as its sensitivity and specificity.

[0147] The diagnostic performance of the model is as follows Figure 11 As shown; Figure 11The results show that the AUCs for the training and validation sets are 0.968 and 0.959, respectively. Figure 11 (a)), with a specific cutoff, the sensitivity and specificity were 91.67% and 85.94%, and 94.34% and 94.39%, respectively. Figure 11 (b) This indicates that the model based on these two paired miRNAs and the triple detection index for liver cancer has a strong predictive effect on liver cancer.

[0148] Example 9: Constructing a liver cancer diagnostic model based on a preferred combination of paired miRNAs and the liver cancer marker (AFP), and validating its diagnostic performance.

[0149] In this embodiment, the two paired miRNAs and the liver cancer marker (AFP) are used for joint modeling. The sample information of the training and validation sets used is consistent with that in Example 7. The specific modeling method is as follows:

[0150] The sample information used is shown in Table 1: the preferred combination qPCR training cohort was used as the training set, and the preferred combination qPCR validation cohort was used as the validation set. A logistic regression model was established for the two paired miRNAs and AFP in the training set, and the cutoff was determined according to the principle of maximizing the Youden exponent in the training set. The overall performance of the model was evaluated by the AUC of the ROC curve, as well as its sensitivity and specificity.

[0151] The diagnostic performance of the model is as follows Figure 12 As shown; Figure 12 The results show that the AUCs for the training and validation sets are 0.966 and 0.942, respectively. Figure 12 a) At a specific cutoff, the sensitivity and specificity were 90.63% and 85.94%, and 91.20% and 92.52%, respectively. Figure 12 b). This indicates that the model built based on these two paired miRNAs and the liver cancer marker AFP also has a good predictive effect on liver cancer.

[0152] Furthermore, in this embodiment, the diagnostic performance of the model constructed based on the two paired miRNAs, the model based on AFP only, and the model constructed based on the two paired miRNAs combined with AFP were compared. The results are as follows: Figure 13 As shown; Figure 13 The results showed that in 160 liver cancer samples, the sensitivity of AFP was only 60.62%, while the model based on two paired miRNAs combined with AFP could detect 97.94% of AFP-positive samples and 74.60% of AFP-negative samples. This indicates that the optimal combination of paired miRNAs has a good complementary effect on AFP, a commonly used clinical liver cancer marker, and the model based on paired miRNAs combined with AFP has a higher detection rate for liver cancer.

[0153] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A diagnostic biomarker for liver cancer, characterized in that, The liver cancer diagnostic markers consist of the following miRNAs and AFP, AFP-L3, and DCP: hsa-let-7a, hsa-miR-21, hsa-miR-125a, hsa-miR-150, hsa-miR-200a; In the diagnosis of liver cancer, the expression ratios of hsa-let-7a to hsa-miR-21, hsa-miR-125a to hsa-miR-21, and hsa-miR-200a to hsa-miR-150 were used as analytical features.

2. The liver cancer diagnostic marker according to claim 1, characterized in that, The miRNA is a miRNA found in serum or plasma.

3. The liver cancer diagnostic marker according to claim 2, characterized in that, The miRNA is a miRNA from fucosylated exosomes in serum or plasma.

4. The liver cancer diagnostic marker according to any one of claims 1-3, characterized in that, The AFP is the corresponding protein in serum, plasma, or whole blood.

5. The use of the liver cancer diagnostic marker or its detection reagent according to any one of claims 1-4 in the preparation of a kit for liver cancer diagnosis.

6. A reagent kit for the diagnosis of liver cancer, characterized in that, The kit includes a detection reagent for the liver cancer diagnostic biomarker according to any one of claims 1-4, wherein the detection reagent includes primers and probes for detecting the expression level of each miRNA in the liver cancer diagnostic biomarker.

7. The reagent kit according to claim 6, characterized in that, The kit includes primers and probes for detecting the expression levels of each miRNA in the liver cancer diagnostic biomarkers, wherein, The forward primer used to detect the expression level of hsa-let-7a is shown in SEQ ID NO:1, the reverse primer is shown in SEQ ID NO:7, and the probe is shown in SEQ ID NO:6; The forward primer used to detect the expression level of hsa-miR-21 is shown in SEQ ID NO:2, the reverse primer is shown in SEQ ID NO:7, and the probe is shown in SEQ ID NO:6; The forward primer used to detect the expression level of hsa-miR-125a is shown in SEQ ID NO:3, the reverse primer is shown in SEQ ID NO:7, and the probe is shown in SEQ ID NO:6; The forward primer used to detect the expression level of hsa-miR-150 is shown in SEQ ID NO:4, the reverse primer is shown in SEQ ID NO:7, and the probe is shown in SEQ ID NO:6; The forward primer used to detect the expression level of hsa-miR-200a is shown in SEQ ID NO:5, the reverse primer is shown in SEQ ID NO:7, and the probe is shown in SEQ ID NO:

6.

8. The kit according to claim 6 or 7, characterized in that, In patients with liver cancer, the expression ratio of the miRNA pairs in serum fucosylated exosomes was significantly different from that in healthy individuals.

Images