A biomarker for predicting the risk of lymph node metastasis in PTC and its application
Through the gene panel and LASSO regression algorithm composed of 17 biomarkers, a PTC lymph node metastasis prediction model was constructed, which solved the problem that the risk of neck lymph node metastasis in the existing technology cannot be effectively predicted, and achieved high sensitivity and specificity sexual risk assessment reduces unnecessary treatment burden.
Patent Information
- Application Number
- CN202410432708.8
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2024-04-11
- Publication Date
- 2024-11-22
- Estimated Expiration
- 2044-04-11
AI Technical Summary
The existing technology cannot effectively predict the risk of neck lymph node metastasis in patients with papillary thyroid carcinoma (PTC), low sensitivity and specificity of imaging examinations, and low evidence quality in preventive central lymph node dissection surgery, resulting in excessive or underestimation of lymph nodes. Transfer risk.
A gene panel composed of 17 biomarkers was used to construct a PTC lymph node metastasis prediction model based on the LASSO regression algorithm, and the risk score was calculated through gene expression matrix analysis, which was used to evaluate the risk of lymph node metastasis before surgery.
It improves the sensitivity and specificity of PTC lymph node metastasis risk prediction, provides accurate evaluation results, provides reliable evidence for clinical treatment strategies, and reduces the risk of unnecessary excessive intervention and no metastasis.
Smart Images

Figure CN118064594B_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the field of biomedical technology, and in particular to a biomarker for predicting the risk of PTC lymph node metastasis and an application thereof. Background Art
[0002] Thyroid cancer is the most common malignant tumor in the endocrine system, accounting for about 1.3% of all malignant tumors in the body. Among them, papillary thyroid cancer (PTC) accounts for 80% to 90% of thyroid malignant tumors. PTC originates from papillary thyroid epithelial cells. Common symptoms include thyroid nodules, neck nodules, hoarseness, and cervical lymphadenopathy. Although PTC is less malignant and generally has a good prognosis after treatment, about 20% of PTC patients still experience recurrence and metastasis. If accompanied by extranodal invasion of lymph nodes, the 10-year recurrence rate is as high as 40% to 70%, which seriously affects the quality of life of patients and even endangers their lives, and imposes a financial and mental burden on the patients' families.
[0003] PTC is prone to cervical lymph node metastasis in the early stage when the tumor is small. Preoperative diagnosis of whether a patient has cervical lymph node metastasis is currently based on imaging examinations, but imaging examinations cannot evaluate lymph nodes smaller than 0.5 cm. There is also a lack of biomarkers with high sensitivity and specificity.
[0004] Prophylactic Central Neck Dissection (pCND) is a surgical procedure that aims to preventively clear the central lymph nodes in the thyroid region in patients with early thyroid cancer who have not found lymph node metastasis to reduce the risk of cervical lymph node metastasis. It can find 30% to 70% of occult lymph node metastases (not found by physical examination and imaging). However, the current evidence quality level for pCND is not high, which may lead to underestimation of patients who do not have clinical lymph node metastasis but actually have lymph node metastasis; moreover, domestic guidelines also believe that pCND can be performed as long as the nerves and parathyroid glands are well protected, but in fact, it may lead to excessive intervention and overestimation of the clinical risk of lymph node metastasis, that is, clinical assessment of lymph node metastasis, but there is no lymph node metastasis in reality. Therefore, based on the evaluation results of pCND, diagnosis and treatment according to lymph node metastasis will also impose a burden on patients in terms of physical, psychological and economic aspects.
[0005] Thyroid fine-needle aspiration biopsy (FNAB) is a common minimally invasive diagnostic method for thyroid diseases. It punctures suspicious thyroid nodules under ultrasound guidance, takes part of the tissue for pathological biopsy, and then judges the benign / malignant nature of the thyroid nodules based on the pathological results. A large amount of evidence-based medicine has proven that FNAB is currently the most sensitive and specific method for evaluating thyroid nodules. However, although FNAB can be used to identify the nature of thyroid nodules before surgery, if the nodules are diagnosed as PTC, it is still not possible to determine whether the tumor has metastasized to the cervical lymph nodes through pathological results. That is, FNAB is still limited in evaluating whether PTC has metastasized to the cervical lymph nodes, and there is still room for expansion in predicting disease prognosis. Summary of the invention
[0006] The present invention is intended to provide a biomarker for predicting the risk of PTC lymph node metastasis and its application, using a gene panel composed of 17 biomarkers for predicting the risk of PTC lymph node metastasis, and constructing a prediction model for PTC lymph node metastasis based on the LASSO regression algorithm, while also performing risk scoring and calculation on PTC lymph node metastasis, which can accurately evaluate whether cervical lymph node metastasis occurs in the early stage of PTC, so as to solve the problems in the background technology.
[0007] In order to achieve the above object, the present invention provides the following technical solutions:
[0008] A biomarker for predicting the risk of lymph node metastasis of PTC, including 17 biomarkers, and the 17 biomarkers are: TACSTD2, AGR2, CYP1B1, ATF3, KRT17, SPOCK3, DSC3, S100A9, PLAUR, F2R, LAMP5, SERPINE1, ASPN, HLA-DQB1, C1QC, MARCO, and LPL.
[0009] Furthermore, the 17 biomarkers constitute a gene panel for risk prediction of PTC lymph node metastasis.
[0010] The above-mentioned biomarker for predicting the risk of PTC lymph node metastasis is used to construct a prediction model for PTC lymph node metastasis.
[0011] Furthermore, the prediction model is constructed based on the LASSO regression algorithm.
[0012] Furthermore, the method for constructing the prediction model comprises the following steps:
[0013] S1. The expression matrix of 17 biomarkers of multiple PTC patients was randomly divided into training set and test set according to proportion;
[0014] S2. Use the predict function in the glment package of R language to model the training set and the test set respectively, and select the parameter lambda.min during modeling.
[0015] Furthermore, the model of the constructed prediction model is:
[0016] Risk score for the presence of lymph node metastasis = -13.69 - 0.12x TACSTD2 - 0.1x AGR2 + 0.24x CYP1B1 + 0.13x ATF3 + 0.05x KRT17 + 0.03x SPOCK3 + 0.01x DSC3 + 0.16x S100A9 + 0.61x PLAUR + 0.14x F2R + 0.01x LAMP5 + 0.06x SERPINE1 + 0.13x ASPN - 0.07x HLA-DQB1 - 0.18x C1QC + 0.17x MARCO + 0.16x LPL;
[0017] Among them, the numerical value of the risk score is positively correlated with the probability of evaluating whether lymph node metastasis exists.
[0018] The above-mentioned biomarker for predicting the risk of PTC lymph node metastasis is used for the risk score of PTC lymph node metastasis.
[0019] Furthermore, the risk assessment method comprises the following steps:
[0020] A1. Collect 50-150 mg of tissue puncture specimens from patients diagnosed with PTC by FNAB and store them in a cryopreserved state at -90 to -70°C.
[0021] A2. Take out the specimen preserved in step A1, add 500-1000ul Trizol to homogenize the sample, place it at room temperature for 5-10 minutes, then add 150-250ul chloroform, shake vigorously for 15-30 seconds, and place it at room temperature for 3-5 minutes;
[0022] A3, the solution obtained in step A2 was first centrifuged at 3-5°C, 10000-14000rpm for 15-20min, then the aqueous phase was transferred to a new centrifuge tube, and 400-500ul of isopropanol was added to mix, and the mixture was placed at room temperature for 10-15min; then the mixture was centrifuged at 3-5°C, 10000-14000rpm for 10-15min, the supernatant was removed, the RNA gel precipitate at the bottom of the tube was retained, and the RNA precipitate was washed with 1ml of 75% ethanol, and the precipitate was resuspended; finally, the mixture was centrifuged at 3-5°C, 8000-10000rpm for 5-10min, the supernatant was discarded, and the RNA precipitate was dried at room temperature, and then 10-20ul of nuclease-free water was added to dissolve it;
[0023] A4. Utilization The purity of the RNA sample obtained in step A3 is detected by spectrophotometer, and the integrity and concentration of the RNA sample are detected by Agilent 2100 RNA Nano 6000 Assay Kit;
[0024] A5. For RNA samples that have passed the test in step A4, select magnetic beads with polythymine and T repeat oligonucleotides to enrich and purify them, then add lysis buffer Fragmentation Buffer to the purified messenger RNA, use six-base random primers to reverse transcribe into double-stranded DNA, and purify the product;
[0025] A6. The double-stranded DNA purified in step A5 is subjected to end repair, polyadenylic acid tail addition, adapter ligation, and polymerase chain reaction amplification in sequence. After the amplified product passes the quality inspection, it is subjected to denaturation and single-stranded circularization, and the circularized product is subjected to quality inspection;
[0026] A7, preparing DNA nanoballs from the cyclized products that have passed the quality inspection in step A6, and sequencing the DNA nanoballs by loading them into a DNBSEQ-T7 platform sequencer;
[0027] A8, the raw data in fastq format obtained by sequencing in step A7 is cleaned by quality control to obtain cleaned data, and then the Hisat2 alignment index file is obtained, and Hisat2 alignment and samtools format conversion are performed, and the featureCounts tool is used to count to obtain the sequencing fragment count matrix; and the sequencing fragment count matrix is converted into a cpm matrix for standardization, and the ID identifier used by the Ensembl database is converted into a gene name to obtain a gene expression matrix for subsequent analysis;
[0028] A9. Extract the expression matrix of the 17 genes included in the risk score, and calculate the risk score of the sample based on the standardized expression level of each gene.
[0029] Furthermore, the calculated risk score is compared with the risk score threshold obtained from the PTC lymph node metastasis prediction model to classify patients as having a high risk or a low risk of lymph node metastasis.
[0030] Furthermore, based on the comparison results between the risk score and the risk score threshold obtained by the PTC lymph node metastasis prediction model, the method for dividing lymph node metastasis into high risk or low risk is: a risk score greater than or equal to 0.54 is high risk, and a risk score less than 0.54 is low risk.
[0031] The beneficial effects of the technical solution are:
[0032] 1. A prediction model for PTC lymph node metastasis is constructed based on the 17 biomarkers provided by the present invention, which has high sensitivity and specificity; it can significantly evaluate the presence or absence of lymph node metastasis before surgery, and the evaluation accuracy is high; the construction of the prediction model is highly operable, with a clear process and convenient calculation;
[0033] 2. The risk scoring method provided by the present invention can accurately assess the risk of lymph node metastasis before surgery by performing gene expression analysis on puncture biopsy specimens of PTC patients, providing a reliable basis for the selection of clinical treatment strategies. BRIEF DESCRIPTION OF THE DRAWINGS
[0034] Figure 1 In the embodiment of the present invention, fresh tumor tissue specimens are collected from patients with papillary thyroid carcinoma after surgical resection;
[0035] In the figure, ① is the information of 36 patients without lymph node metastasis among 87 samples of common eukaryotic transcriptome sequencing;
[0036] ② 51 of the 87 samples sequenced for common eukaryotic transcriptomes had information on patients with lymph node metastasis;
[0037] Figure 2 The expression levels of 17 biomarkers in PTC patients with and without lymph node metastasis in the embodiments of the present invention are shown in FIG.
[0038] In the figure, ① is the expression level of TACSTD2; ② is the expression level of AGR2; ③ is the expression level of CYP1B1; ④ is the expression level of ATF3; ⑤ is the expression level of KRT17; ⑥ is the expression level of SPOCK3; ⑦ is the expression level of DSC3; ⑧ is the expression level of S100A9; ⑨ is the expression level of PLAUR; ⑩ is the expression level of F2R; is the expression level of LAMP5; is the expression level of SERPINE1; is the expression level of ASPN; is the expression level of HLA-DQB1; is the expression level of C1QC; is the expression level of MARCO; The expression level of LPL
[0039] Figure 3 This is a comparison chart of risk scores in PTC patients with and without lymph node metastasis in an embodiment of the present invention;
[0040] Figure 4 This is a ROC curve diagram of predicting lymph node metastasis of PTC patients based on 17 genes in different sets in an embodiment of the present invention;
[0041] In the figure, ① is the ROC curve in the training set; ② is the ROC curve in the test set. DETAILED DESCRIPTION
[0042] The present invention is further described in detail below in conjunction with the accompanying drawings and embodiments:
[0043] A biomarker for predicting the risk of PTC lymph node metastasis, including 17 biomarkers, wherein the 17 biomarkers are: TACSTD2, AGR2, CYP1B1, ATF3, KRT17, SPOCK3, DSC3, S100A9, PLAUR, F2R, LAMP5, SERPINE1, ASPN, HLA-DQB1, C1QC, MARCO, and LPL; among them, the 17 biomarkers constitute a gene panel for predicting the risk of PTC lymph node metastasis.
[0044] The above-mentioned biomarker for predicting the risk of PTC lymph node metastasis is used to construct a prediction model for PTC lymph node metastasis; wherein the prediction model is constructed based on the LASSO regression algorithm, and the specific construction method includes the following steps:
[0045] S1. The expression matrix of 17 biomarkers of multiple PTC patients was randomly divided into training set and test set according to proportion;
[0046] S2. Use the predict function in the glment package of R language to model the training set and the test set respectively, and select the parameter lambda.min during modeling.
[0047] The model of the constructed prediction model is:
[0048] Risk score for the presence of lymph node metastasis = -13.69 - 0.12x TACSTD2 - 0.1x AGR2 + 0.24x CYP1B1 + 0.13x ATF3 + 0.05x KRT17 + 0.03x SPOCK3 + 0.01x DSC3 + 0.16x S100A9 + 0.61x PLAUR + 0.14x F2R + 0.01x LAMP5 + 0.06x SERPINE1 + 0.13x ASPN - 0.07x HLA-DQB1 - 0.18x C1QC + 0.17x MARCO + 0.16x LPL;
[0049] Among them, the numerical value of the risk score is positively correlated with the probability of evaluating whether lymph node metastasis exists.
[0050] The above-mentioned biomarker for predicting the risk of PTC lymph node metastasis is used for risk scoring of PTC lymph node metastasis; wherein the risk scoring method comprises the following steps:
[0051] A1. Collect 50-150 mg of tissue puncture specimens from patients diagnosed with PTC by FNAB and store them in a cryopreserved state at -90 to -70°C.
[0052] A2. Take out the specimen preserved in step A1, add 500-1000ul Trizol to homogenize the sample, place it at room temperature for 5-10 minutes, then add 150-250ul chloroform, shake vigorously for 15-30 seconds, and place it at room temperature for 3-5 minutes;
[0053] A3, the solution obtained in step A2 was first centrifuged at 3-5°C, 10000-14000rpm for 15-20min, then the aqueous phase was transferred to a new centrifuge tube, and 400-500ul of isopropanol was added to mix, and the mixture was placed at room temperature for 10-15min; then the mixture was centrifuged at 3-5°C, 10000-14000rpm for 10-15min, the supernatant was removed, the RNA gel precipitate at the bottom of the tube was retained, and the RNA precipitate was washed with 1ml of 75% ethanol, and the precipitate was resuspended; finally, the mixture was centrifuged at 3-5°C, 8000-10000rpm for 5-10min, the supernatant was discarded, and the RNA precipitate was dried at room temperature, and then 10-20ul of nuclease-free water was added to dissolve it;
[0054] A4. Utilization The purity of the RNA sample obtained in step A3 is detected by spectrophotometer, and the integrity and concentration of the RNA sample are detected by Agilent 2100 RNA Nano 6000 Assay Kit;
[0055] A5. For RNA samples that have passed the test in step A4, select magnetic beads with polythymine and T repeat oligonucleotides to enrich and purify them, then add lysis buffer Fragmentation Buffer to the purified messenger RNA, use six-base random primers to reverse transcribe into double-stranded DNA, and purify the product;
[0056] A6. The double-stranded DNA purified in step A5 is subjected to end repair, polyadenylic acid tail addition, adapter ligation, and polymerase chain reaction amplification in sequence. After the amplified product passes the quality inspection, it is subjected to denaturation and single-stranded circularization, and the circularized product is subjected to quality inspection;
[0057] A7, preparing DNA nanoballs from the cyclized products that have passed the quality inspection in step A6, and sequencing the DNA nanoballs by loading them into a DNBSEQ-T7 platform sequencer;
[0058] A8, the raw data in fastq format obtained by sequencing in step A7 is cleaned by quality control to obtain cleaned data, and then the Hisat2 alignment index file is obtained, and Hisat2 alignment and samtools format conversion are performed, and the featureCounts tool is used to count to obtain the sequencing fragment count matrix; and the sequencing fragment count matrix is converted into a cpm matrix for standardization, and the ID identifier used by the Ensembl database is converted into a gene name to obtain a gene expression matrix for subsequent analysis;
[0059] A9. Extract the expression matrix of the 17 genes included in the risk score, and calculate the risk score of the sample based on the standardized expression level of each gene.
[0060] Finally, the calculated risk score was compared with the risk score threshold obtained from the PTC lymph node metastasis prediction model and classified as high risk or low risk of lymph node metastasis: a risk score greater than or equal to 0.54 was classified as high risk, and a risk score less than 0.54 was classified as low risk.
[0061] The specific implementation process is as follows:
[0062] The present invention detects and obtains the expression information of each gene in PTC tumor tissue based on the DNBSEQ-T7 platform, and the specific steps are as follows:
[0063] 1. Collection of experimental materials:
[0064] Fresh tumor tissue specimens (50-150 mg, the size of a green bean) were collected from 87 patients with papillary thyroid carcinoma (36 patients without lymph node metastasis and 51 patients with lymph node metastasis) after surgical resection. A total of 87 samples were placed in cryopreservation tubes and stored at -80°C.
[0065] 2. Main experimental methods:
[0066] The common eukaryotic transcriptome sequencing technology is a method of obtaining the transcript sequence and expression information of specific cells or tissues comprehensively and quickly based on the existing reference genome of eukaryotic organisms, and then comprehensively studying gene functions at the messenger RNA level to reveal specific biological processes. Figure 1 As shown, a total of 87 samples from the PTC patient group without lymph node metastasis and the PTC patient group with lymph node metastasis were selected for reference common eukaryotic transcriptome sequencing.
[0067] 3. Specific experimental steps:
[0068] (1) RNA extraction
[0069] The samples were taken out from -80°C, 500-1000ul Trizol was added to each sample for homogenization, and the samples were placed at room temperature for 5-10 minutes. 200ul chloroform was added, and the samples were shaken vigorously for 15-30s and placed at room temperature for 3-5 minutes. Then, the samples were centrifuged at 4°C, 12000rpm for 15 minutes, the aqueous phase was transferred to a new centrifuge tube, and isopropanol with twice the volume of chloroform (400ul) was added to mix, and the samples were placed at room temperature for 10 minutes. The samples were centrifuged at 4°C, 12000rpm for 10 minutes, the supernatant was removed, the RNA gel precipitate at the bottom of the tube was retained, and the RNA precipitate was washed with 1ml 75% ethanol and resuspended. Finally, the samples were centrifuged at 4°C, 8000rpm for 5 minutes, and the supernatant was discarded. After the RNA precipitate was dried at room temperature, 10-20ul nuclease-free water was added to dissolve it.
[0070] use The purity of RNA samples (OD260 / 280 ≥ 1.8, OD260 / 230 ≥ 1.8) was detected by a spectrophotometer (IMPLEN, CA, USA); the integrity and concentration of RNA samples were detected using Agilent 2100 RNA Nano 6000 Assay Kit (Agilent Technologies, CA, USA).
[0071] (2) Library construction and sequencing
[0072] After the RNA sample is qualified, since the 3' end of the eukaryotic messenger RNA has a polyadenylic acid tail structure, magnetic beads with polythymine and T repeat oligonucleotides are selected for enrichment and purification; Fragmentation Buffer is added to the purified messenger RNA to fragment it into short fragments, and then the fragmented messenger RNA is used as a template and reverse transcribed into double-stranded DNA with six-base random primers, and the product is purified.
[0073] The purified double-stranded DNA is then subjected to end repair, polyadenylic acid tail addition, adapter ligation, and polymerase chain reaction amplification. After the amplified product passes the quality inspection, it is denatured and single-stranded cyclized, and the cyclized product is subjected to quality inspection.
[0074] The qualified circularization products are used to make DNA nanoballs and loaded onto the DNBSEQ-T7 platform sequencer for sequencing.
[0075] (3) Identification and screening of biomarkers
[0076] After the raw data in fastq format obtained by the test is cleaned by quality control, the cleaned data can be obtained; then the Hisat2 alignment index file is obtained, Hisat2 alignment and samtools format conversion are performed, and the featureCounts tool is used to count the sequencing fragment count matrix. The sequencing fragment count matrix is converted into a cpm matrix for standardization, and the ID identifier used by the Ensembl database is converted into a gene name to obtain a gene expression matrix for subsequent analysis.
[0077] Based on the above gene expression matrix, the LASSO algorithm was used to screen the gene set obtained from the "single-cell sequencing of papillary thyroid carcinoma" study (including 144 genes in epithelial, endothelial, fibroblastic, myeloid, T cell immune and other cellular components), and the threshold was set to P value < 0.05. Therefore, a total of 17 genes were included, including TACSTD2, AGR2, CYP1B1, ATF3, KRT17, SPOCK3, DSC3, S100A9, PLAUR, F2R, LAMP5, SERPINE1, ASPN, HLA-DQB1, C1QC, MARCO, and LPL, to form a panel.
[0078] 2. Biomarker Detection Efficacy
[0079] 1. Comparison of gene expression levels of TACSTD2, AGR2, CYP1B1, ATF3, KRT17, SPOCK3, DSC3, S100A9, PLAUR, F2R, LAMP5, SERPINE1, ASPN, HLA-DQB1, C1QC, MARCO and LPL in PTC patients without lymph node metastasis and with lymph node metastasis:
[0080] (1) Main methods: The R language was used for statistical analysis to compare the two groups of data. The levels of TACSTD2, AGR2, CYP1B1, ATF3, KRT17, SPOCK3, DSC3, S100A9, PLAUR, F2R, LAMP5, SERPINE1, ASPN, HLA-DQB1, C1QC, MARCO and LPL in the tumor tissues of the PTC patient group with lymph node metastasis were compared with those of the PTC patient group without lymph node metastasis. The Wilcoxon signed-rank test was used to test whether the differences between them were statistically significant.
[0081] 2) Results: Figure 2 As shown, based on the detection of common eukaryotic transcriptome sequencing technology, the levels of KRT17, DSC3, PLAUR, SERPINE1, ASPN and LPL in PTC patients with lymph node metastasis were significantly higher than those in patients without lymph node metastasis, and the level of HLA-DQB1 was significantly lower than that in patients without lymph node metastasis.
[0082] 2. PTC lymph node metastasis prediction model based on machine learning and calculation of risk score.
[0083] 1) Specific methods: The expression matrix of 17 genes in 87 PTC patients was divided into a training set (n=68, 26 cases without lymph node metastasis and 42 cases with lymph node metastasis) and a test set (n=19, 10 cases without lymph node metastasis and 9 cases with lymph node metastasis) in a ratio of 8:2. The least absolute shrinkage and selection operation regression algorithm was used, and the lambda.min parameter was selected to construct the model.
[0084] (2) Results: Figure 3 As shown in the figure, the prediction model included all 17 genes, and obtained the constant term and the coefficient of each gene. Using the coefficients given by the prediction model, a risk score for predicting lymph node metastasis in PTC patients was constructed, that is, risk score = -13.69-0.12x TACSTD2-0.1x AGR2+0.24x CYP1B1+0.13x ATF3+0.05x KRT17+0.03xSPOCK3+0.01x DSC3+0.16x S100A9+0.61x PLAUR+0.14x F2R+0.01x LAMP5+0.06xSERPINE1+0.13x ASPN-0.07x HLA-DQB1-0.18x C1QC+0.17x MARCO+0.16x LPL; the risk score of the PTC patient group with lymph node metastasis was significantly higher than that of the PTC patient group without lymph node metastasis.
[0085] 3. The performance of the 17-gene panel risk score as a biomarker for predicting cervical lymph node metastasis in tumor tissues of PTC patients detected by common eukaryotic transcriptome sequencing technology was tested.
[0086] (1) Specific methods: The receiver operating characteristic curve (hereinafter referred to as ROC curve) was used to determine the classifier performance index of the 17-gene panel as a biomarker for cervical lymph node metastasis in PTC patients. The ROC curves of the training set and the test set were drawn using graphpad software. The control group was set as PTC patients without lymph node metastasis, and the disease group was set as the risk score of PTC patients with lymph node metastasis. A 95% confidence interval was set, and the results were output as a percentage, i.e., the area under the curve (the area under the ROC curve and the coordinate axis, hereinafter referred to as AUC). At the same time, the risk of lymph node metastasis was divided into high and low according to the threshold.
[0087] (2) Results: Figure 4 As shown in the figure, calculated by ROC curve, the AUC of the risk score is 0.8947 in the training set and 07111 in the test set. The threshold is 0.54. A risk score ≥ 0.54 indicates high risk, and a risk score < 0.54 indicates low risk. This panel has the potential to become a gene panel for predicting cervical lymph node metastasis of papillary thyroid carcinoma based on common eukaryotic transcriptome sequencing technology and machine learning technology.
[0088] The above is only an embodiment of the present invention, and the common knowledge such as the known specific technical solutions or characteristics in the solution is not described in detail here. It should be pointed out that for those skilled in the art, several modifications and improvements can be made without departing from the technical solution of the present invention, which should also be regarded as the protection scope of the present invention, and these will not affect the effect of the implementation of the present invention and the practicality of the patent. The scope of protection required by this application shall be based on the content of its claims, and the specific implementation methods and other records in the specification can be used to interpret the content of the claims.
Claims
1. Application of a biomarker for predicting the risk of PTC lymph node metastasis in constructing a prediction model for PTC lymph node metastasis, characterized in that: The gene panel was composed of the expression levels of 17 biomarkers TACSTD2, AGR2, CYP1B1, ATF3, KRT17, SPOCK3, DSC3, S100A9, PLAUR, F2R, LAMP5, SERPINE1, ASPN, HLA-DQB1, C1QC, MARCO, and LPL, and the prediction model constructed was: Risk score for presence of lymph node metastasis = -13.69 -0.12 x TACSTD2 -0.1 x AGR2 + 0.24 xCYP1B1 + 0.13 x ATF3 + 0.05 x KRT17 + 0.03 x SPOCK3 + 0.01 x DSC3 + 0.16 xS100A9 + 0.61 x PLAUR + 0.14 x F2R + 0.01 x LAMP5 + 0.06 x SERPINE1 + 0.13 xASPN -0.07 x HLA-DQB1 -0.18 x C1QC + 0.17 x MARCO + 0.16 x LPL; Among them, the numerical value of the risk score is positively correlated with the probability of evaluating whether there is lymph node metastasis; According to the comparison results of the risk score and the risk score threshold obtained by the PTC lymph node metastasis prediction model, the method of dividing the high risk or low risk of lymph node metastasis is as follows: a risk score greater than or equal to 0.54 is high risk, and a risk score less than 0.54 is low risk.
2. The use of a biomarker for predicting the risk of PTC lymph node metastasis according to claim 1 in constructing a prediction model for PTC lymph node metastasis, characterized in that: The prediction model is constructed based on the LASSO regression algorithm.
3. The use of a biomarker for predicting the risk of PTC lymph node metastasis in constructing a prediction model for PTC lymph node metastasis according to claim 1, characterized in that: The method for constructing the prediction model comprises the following steps: S1. The expression matrix of 17 biomarkers of multiple PTC patients was randomly divided into training set and test set according to proportion; S2. Use the predict function in the glment package of R language to model the training set and the test set respectively, and select the parameter lambda.min during modeling.
