Marker LGALS9 for diagnosing nasopharyngeal squamous carcinoma and application thereof

Abstract

The invention discloses a marker LGALS9 for diagnosing nasopharyngeal squamous carcinoma and application thereof, and belongs to the technical field of molecular diagnosis. The invention finds that the LGALS9 molecule has significant high expression in nasopharyngeal squamous cell carcinoma tissues, prompts that the LGALS9 can be used as a unique marker for diagnosing the nasopharyngeal squamous cell carcinoma, further constructs a diagnosis model by taking the LGALS9 as the marker, preferably detects the expression level of the marker LGALS9 in the cancer tissues by adopting an immunohistochemical method, and unexpectedly finds that the LGALS9 molecule can be used for diagnosing the nasopharyngeal squamous cell carcinoma tissues. The diagnostic performance of the diagnostic model constructed according to an immunohistochemical detection result is superior to that of a model constructed by adopting a protein spectrum detection method. The marker LGALS9 and the diagnostic model provided by the invention provide technical support for tissue traceability of nasopharyngeal squamous carcinoma, and have great clinical transformation potential.

Description

Technical Field

[0001] This invention belongs to the field of molecular diagnostic technology, specifically relating to LGALS9, a biomarker for diagnosing nasopharyngeal squamous cell carcinoma, and its applications. Background Technology

[0002] Cancers of Unknown Primary (CUP) are a type of pathologically diagnosed metastatic malignant tumor, but the primary site cannot be determined using conventional methods such as laboratory tests, imaging techniques, and histological examination. Statistics show that CUP accounts for approximately 2%-10% of all cancer cases, ranking among the most common malignant tumors in terms of incidence and mortality. The incidence of CUP increases significantly with age. Studies report the highest proportion of CUP in the respiratory and digestive organs, with the liver being the most frequently recorded single metastatic site. A significant proportion of cases still lack a specific primary site record in cancer registry data. The traditional treatment for CUP is broad-spectrum chemotherapy, with a relatively poor overall prognosis. Therefore, accurately determining the tissue origin of CUP is of significant clinical importance, providing guidance for the diagnosis and treatment of CUP patients.

[0003] Squamous cell carcinoma is a common histological subtype of CUP (cleft liposome). Clinically, the primary sites of squamous cell carcinoma mainly originate from the nasopharynx, head and neck, esophagus, lung, and cervix. Currently, for the tissue-based diagnosis of primary and metastatic nasopharyngeal carcinoma, clinicopathology often uses in situ hybridization (ISH) to detect small RNAs encoded by Epstein-Barr virus (EBV) infection (EBER) to indicate tissue origin. This method identifies EBV infection characteristics to determine whether the tumor tissue originated from the nasopharynx. However, with further research, researchers have found that EBV infection-related phenotypes can also be detected in squamous cell carcinomas from various non-nasopharyngeal organs. For example, lymphoepithelioma-like carcinoma, a special subtype of squamous cell carcinoma closely related to EBV infection, is widely distributed in multiple organ systems such as the head and neck, esophagus, and lung, leading to a significant decrease in the specificity of EBER detection, which cannot meet the clinical needs for precise tracing of the origin of nasopharyngeal carcinoma.

[0004] Research data indicates that EBER, as the only differential diagnostic marker for nasopharyngeal carcinoma, has limitations in both sensitivity and specificity. For example, EBER shows 100% negative expression in esophageal squamous cell carcinoma, but only 66% positive in metastatic nasopharyngeal carcinoma; false positive cases of EBER exist in cervical, lung, and head and neck squamous cell carcinomas, respectively, at 2.5%, 5.1%, and 2.7%; in a multicenter validation cohort, the overall detection rate of EBER for nasopharyngeal carcinoma metastases is less than 70%, making it difficult to address the complex heterogeneity of tumors. These results suggest an urgent need to develop a diagnostic marker and method with better specificity and sensitivity to improve the accuracy of nasopharyngeal carcinoma tissue tracing.

[0005] The inventors previously proposed a set of biomarkers for predicting the origin of squamous cell carcinoma in patent applications CN 116798520 A and CN 117969855 A. These biomarkers consist of 39 proteins. The inventors used proteomic profiling to detect the expression levels of these 39 proteins in cancer tissue and constructed a support vector machine model. In the training set, the model achieved an AUC of 0.994 for the predicted ROC curve of nasopharyngeal squamous cell carcinoma, and in the external validation set, the model achieved an AUC of 0.941 for the predicted ROC curve of nasopharyngeal squamous cell carcinoma.

[0006] Those skilled in the art will recognize that proteomic profiling requires large-scale detection equipment and specialized personnel, resulting in significant time consumption and high costs. The inventor's prior art, involving proteomic profiling of 39 proteins, is labor-intensive and expensive, making its clinical application difficult. Building upon previous research, the inventors believe it is necessary to streamline the number of biomarkers and optimize the detection method. They propose a simple and low-cost approach to biomarker detection, reducing workload and costs while maintaining high predictive accuracy, thereby increasing the clinical translation potential of the diagnostic technology. Summary of the Invention

[0007] One objective of this invention is to provide a highly specific biomarker for the diagnosis of nasopharyngeal squamous cell carcinoma. Proteomics studies indicate that galectin 9 (LGALS9) is significantly highly expressed in nasopharyngeal squamous cell carcinoma tissues, suggesting that LGALS9 can be used as the sole biomarker for the diagnosis of nasopharyngeal squamous cell carcinoma. Another objective of this invention is to provide a model for the diagnosis of nasopharyngeal squamous cell carcinoma based on the biomarker LGALS9 and a method for constructing such a model. This invention preferably employs immunohistochemistry to detect the expression level of the biomarker LGALS9 in cancerous tissues. This detection method is not only simple and low-cost, but the inventors unexpectedly discovered that the diagnostic model constructed based on the immunohistochemical detection results outperforms the proteomic detection method previously used by the inventors. The biomarker LGALS9 and the diagnostic model provided by this invention provide technical support for the tissue tracing of nasopharyngeal squamous cell carcinoma and are of great significance for the clinical translation of this technology.

[0008] The objective of this invention is achieved through the following technical solution: In a first aspect, the present invention provides the use of a substance for detecting the LGALS9 biomarker in the preparation of a product for diagnosing nasopharyngeal squamous cell carcinoma, characterized in that the substance for detecting the LGALS9 biomarker is any reagent required for detecting LGALS9 protein levels or gene levels.

[0009] In some embodiments of the present invention, the reagents for detecting LGALS9 protein levels are those required for detecting LGALS9 protein levels using proteometry, immunohistochemistry (IHC), enzyme-linked immunosorbent assay (ELISA), Western blotting, or immunofluorescence.

[0010] In some embodiments of the present invention, the reagents for detecting the LGALS9 gene level are the reagents required for detecting the LGALS9 gene level using RT-PCR, RT-qPCR, gene chip detection, DNA blotting, or in situ hybridization.

[0011] Preferably, the substance used to detect the marker LGALS9 is selected from any one of the following: (1) Reagents for detecting LGALS9 protein levels using proteometry; (2) Reagents for detecting LGALS9 protein levels using immunohistochemistry.

[0012] In the most preferred embodiment of the present invention, the substance used to detect the LGALS9 biomarker is a reagent for detecting LGALS9 protein levels using immunohistochemistry. The reagent includes an anti-LGALS9 antibody; in a specific embodiment of the present invention, it is a rabbit anti-human LGALS9 monoclonal antibody.

[0013] Furthermore, the reagents for detecting LGALS9 protein levels also include other conventional reagents known to those skilled in the art for immunohistochemical detection. Specifically, these other conventional reagents are selected from xylene analogues, ethanol, 3% H2O2 solution, 1% BSA blocking solution, DAB chromogenic reagent, hematoxylin, and horseradish peroxidase-labeled goat anti-mouse IgG.

[0014] The products used for diagnosing squamous cell carcinoma of nasopharyngeal origin include, but are not limited to, models, chips, detection platforms, reagent kits, test strips, or membrane strips.

[0015] In a specific embodiment of the present invention, the diagnosis of nasopharyngeal squamous cell carcinoma includes the following steps: (1) The level of LGALS9 protein in the sample was detected by immunohistochemistry; (2) When LGALS9 is expressed positively in the sample to be tested, it suggests that the sample may be nasopharyngeal squamous cell carcinoma.

[0016] In this invention, the sample to be tested is an ex vivo tumor tissue taken from the patient to be tested, which can be a fresh sample or a formalin-fixed paraffin-embedded sample.

[0017] Secondly, the present invention provides a method for constructing a model for diagnosing nasopharyngeal squamous cell carcinoma, characterized in that a support vector machine model is constructed using the immunohistochemical detection results of LGALS9 protein as variables.

[0018] In a specific embodiment of the present invention, the immunohistochemical detection result of the LGALS9 protein is a qualitative result.

[0019] In one embodiment of the present invention, the method for constructing the model includes the following specific steps: (1) Divide the samples into training set and validation set. Divide the training set into k disjoint folds. Select any fold as the test set to judge the performance of the support vector machine model. Use the remaining k-1 fold samples as the training set to train the model. Train the model using the svm function. (2) Input the immunohistochemical qualitative detection results of LGALS9 protein in the training set samples, construct the support vector machine model, and evaluate the model performance using the receiver operating characteristic curve (ROC). (3) The model was evaluated using the immunohistochemical detection results of LGALS9 protein from the validation set samples.

[0020] Thirdly, the present invention provides a model, characterized in that the model is constructed by the method described in the second aspect of the present invention.

[0021] Furthermore, the model is used to diagnose squamous cell carcinoma of nasopharyngeal origin, and includes an information acquisition module and a diagnostic module.

[0022] The information acquisition module is used to perform the step of acquiring information on the expression level of LGALS9 protein in the sample to be tested. The diagnostic module is used to perform the step of determining whether the sample to be tested is a squamous cell carcinoma of nasopharyngeal origin based on the LGALS9 protein expression level information.

[0023] Fourthly, the present invention provides an application of the model described in the third aspect of the present invention in the preparation of a product for diagnosing nasopharyngeal squamous cell carcinoma, characterized in that the application includes the following steps: S1) Immunohistochemistry was used to detect the LGALS9 protein level in the sample to be tested, and the qualitative results of LGALS9 expression in the sample to be tested were obtained. S2) Perform data analysis and model calculation on the immunohistochemical detection results obtained in step S1); S3) Determine whether the sample to be tested is a squamous cell carcinoma of nasopharyngeal origin based on the probability score calculated by the model.

[0024] Based on the immunohistochemical detection results of LGALS9 protein, the diagnostic accuracy of LGALS9 protein level detection using immunohistochemistry for nasopharyngeal squamous cell carcinoma was 77.3% in the internal training set and 90% in the external validation set. Furthermore, a support vector machine model constructed using the immunohistochemical results of LGALS9 protein from the internal training set was used to evaluate the model's ROC curve, as shown in the figure. Figure 3 As shown, AUC = 0.812. A support vector machine model constructed using the immunohistochemical results of LGALS9 protein from the external validation set samples was used to evaluate the model's ROC curve, as shown in the figure. Figure 4 As shown, AUC=0.882.

[0025] As can be seen, compared with the traditional EBER method for diagnosing nasopharyngeal squamous cell carcinoma, the diagnostic method provided by this invention has higher specificity and sensitivity. Furthermore, compared with the inventor's prior diagnostic method that used proteomic profiling to detect 39 protein biomarkers, this invention uses only immunohistochemistry to detect one protein biomarker, achieving a diagnostic accuracy of 81.4-90% for nasopharyngeal squamous cell carcinoma, reducing detection costs and improving detection efficiency. Attached Figure Description

[0026] Figure 1 Expression levels of LGALS9 protein in different squamous cell carcinoma cohorts.

[0027] Figure 2 ROC curves of LGALS9 protein detected by mass spectrometry in the training set, internal validation set, and external validation set cohorts.

[0028] Figure 3 ROC curve of LGALS9 protein detected by immunohistochemistry in the Fudan cohort.

[0029] Figure 4 ROC curve of LGALS9 protein detected by immunohistochemistry in the Macau cohort. Detailed Implementation

[0030] 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, and 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.

[0031] Example 1: Constructing a model for diagnosing nasopharyngeal squamous cell carcinoma based on proteomic detection data. I. Research Methods 1. Enrolled patients This study protocol followed the guidelines set forth in the Declaration of Helsinki and obtained written informed consent from all participants.

[0032] Fudan Cohort: From June 2012 to November 2021, a total of 325 patients with squamous cell carcinoma were enrolled at Fudan University Cancer Hospital for this study. These included squamous cell carcinoma originating from the nasopharynx (n=69), head and neck (n=56), lung (n=63), esophagus (n=63), and cervical (n=74). This study was approved by the Medical Ethics Committee of Fudan University Cancer Hospital (Ethics Number: 2012228-Exp10).

[0033] External validation cohort: From January 2016 to December 2021, a total of 63 patients with squamous cell carcinoma were enrolled for this study at Shanghai East Hospital, Ruijin Hospital affiliated to Shanghai Jiao Tong University, Renji Hospital affiliated to Shanghai Jiao Tong University, Zhongshan Hospital affiliated to Fudan University, Shanghai Tongren Hospital, and the First Affiliated Hospital of Zhejiang University School of Medicine. Among them, squamous cell carcinoma of nasopharyngeal origin (n=13), squamous cell carcinoma of head and neck origin (n=10), squamous cell carcinoma of lung origin (n=12), squamous cell carcinoma of esophagus origin (n=14), and squamous cell carcinoma of cervix origin (n=14).

[0034] 2. Preparation of protein samples 2.1) Sample collection: Squamous cell carcinoma tissue samples were collected and stored in clean EP tubes; 2.2) Lysis of samples: Add 500 μL of lysis buffer to the EP tube containing the tissue sample and homogenize the sample. The lysis buffer formula is: 1% (w / v) DOC, 10 mM TCEP, 40 mM CAA, 100 mM Tris, pH 8.5. 2.3) Ultrasonic disruption: Place the sample tube on an ice-water mixture and sonicate under the following conditions: 5 min, 3 s on, 3 s off, and 30% power. 2.4) Protein separation: Centrifuge the sample at 4℃, 16,000g, for 10min, and retain the supernatant; 2.5) Protein quantification: The protein concentration in the supernatant from step 2.4 was determined using Nanodrop, and 50 μg of protein was transferred to a new EP tube; 2.6) Reductive alkylation: Take 100 μg of protein solution and add 10 mM dithiothreitol to a final concentration. The reaction is carried out at 56 °C for 30 min. Then, add 10 mM iodoacetamide to a final concentration at room temperature and carry out alkylation reaction in the dark for 30 min. 2.7) Protein washing: Add the sample after the reaction in step 2.6 to a 10KD ultrafiltration tube and centrifuge at 4℃ and 14,000g for 2 min to ensure that the protein is completely bound to the membrane. Wash the protein sample on the membrane with 50mM ammonium bicarbonate, 300μL / time, and centrifuge at 4℃ and 14,000g for 20 min. Repeat twice. 2.8) Proteolytic digestion: Dilute the sample to 1 μg / μL with 100 mM TEAB, add intracellular protease Lys-C at a mass ratio of 1:100 (enzyme:protein), and digest at 37℃ for 2 h. Then add trypsin at a mass ratio of 1:50 (enzyme:protein) and digest overnight at 37℃. 2.9) Peptide collection: Add the sample after the reaction in step 2.8 to a 10KD ultrafiltration tube, centrifuge at 14,000g for 20min, retain the peptides in the collection tube, add 200μL of mass spectrometry water, centrifuge and wash once, retain the peptides collected the second time, combine the peptides collected in the two collections and vacuum dry them, and use the obtained product for mass spectrometry detection.

[0035] 3. Mass spectrometry detection of protein samples Instrument: QExactive HF mass spectrometer; Column type: 50cm C18 chromatographic separation column (2μm, 75μm, x500mm, Thermo Fisher Scientific, USA); Flow rate: 200 nL / min; Mobile phase: Solution A is H2O:FA = 99.8:0.2, Solution B is CAN:FA = 99.8:0.2; Total separation time: 65 min.

[0036] The dried peptide fragments were thoroughly dissolved in loading buffer (H2O:CH3OH:FA = 94.8:5:0.2), centrifuged at 12000 r / min for 10 min, and then analyzed by mass spectrometry. For the remaining specific steps, please refer to the "III. Mass Spectrometry Detection of Gastric Cancer Protein Samples" section of the specific implementation method in CN108445097A.

[0037] 4. Mass spectrometry data analysis of protein samples 4.1) The spectra obtained by mass spectrometry identification were searched using Proteome Discover (version 1.4, Thermo Scientific) with Mascot 2.3 to obtain the original files, and the spectra were matched with the NCBI human Ref-sequence protein database (2013 version).

[0038] The search parameters are set as follows: 2 miscleavage sites, dynamic modifications including methionine oxidation, N-terminal acetylation, and reductive alkylation on cysteine, peptide length containing at least 7 amino acids, peptide fraction at least 10, and peptide order set to high. The primary ion bias is set to 20 ppm, and the secondary ion bias to 50 mmu. An integrated reverse library is used for evaluation, and an FDR (false detection rate) of less than 1% is considered acceptable.

[0039] For the remaining specific steps, please refer to the "IV. Mass Spectrometry Data Analysis of Gastric Cancer Protein Samples" section of the specific implementation method in CN108445097A.

[0040] 4.2) Peptide quality control Select the target protein peptide that meets any of the following conditions for subsequent analysis: Condition 1: US ≥ 1 and S ≥ 2; Condition 2: S≥3; Condition 3: Ion score ≥ 40.

[0041] In US, U stands for Unique peptide, meaning that this peptide segment is not shared with other proteins and is the only peptide segment of this protein; S stands for Strict peptide, which is the Mascot Ion score—the ion score is greater than 20, that is, the strictness of identification in secondary spectroscopy; US stands for Unique strict peptide, which is a peptide segment that simultaneously satisfies the conditions of being the only peptide segment of the protein and having a Mascot ion score greater than 20.

[0042] 4.3) Data Standardization The iBAQ value of the high-confidence protein is preferably calculated using the label-free quantitative iBAQ method based on peak area. Then, the iFOT value is obtained by calculating the ratio of the quantitative values ​​of each identified protein to all identified proteins. Preferably, the iBAQ value of a protein is the sum of the peak areas of all corresponding peptides of that protein / the theoretical number of peptides. According to this method, the expression profiles of 39 proteins (see CN117969855 A) were obtained for subsequent analysis.

[0043] Furthermore, the researchers discovered that in nasopharyngeal squamous cell carcinoma samples, LGALS9, ATP2A3, TMEM205, ISYNA1, PDK1, MYO6, PRSS1, DTYMK, and SERPINB9 were specifically upregulated; while CDKN2A, LGALS7, LGALS3, PKP3, CCDC6, PKP1, TOMM22, and HSPB6 were specifically downregulated.

[0044] II. Research Results 1. A predictive model was constructed using 39 biomarker proteins. Based on the 39 characteristic proteins obtained above, technicians constructed a support vector machine model to predict whether the tissue sample was nasopharyngeal squamous cell carcinoma. Data published in CN 117969855 A shows that, based on protein proteomic analysis data, the AUC of the predicted ROC curve for nasopharyngeal squamous cell carcinoma was 0.994 in the validation set of the Fudan cohort; and 0.941 in the external validation set.

[0045] The inventors discovered that although the model constructed using 39 proteins as biomarkers has high predictive accuracy, detecting these 39 characteristic proteins using proteomics requires large-scale detection equipment, and the large volume of detection and analysis is time-consuming, resulting in high costs and limiting its clinical application. Therefore, the inventors first considered reducing the number of biomarker proteins to decrease the workload and cost of detection.

[0046] 2. Constructing a predictive model using LGALS9 biomarker proteins. Building upon this foundation, technicians selected specific biomarkers from previously detected characteristic proteins of nasopharyngeal squamous cell carcinoma to construct models for predicting the disease. Mass spectrometry data for LGALS9 protein in cervical squamous cell carcinoma, esophageal squamous cell carcinoma, lung squamous cell carcinoma, nasopharyngeal squamous cell carcinoma, and head and neck squamous cell carcinoma are shown below. Figure 1 As shown. From Figure 1 The data shows that LGALS9 protein is significantly highly expressed in nasopharyngeal squamous cell carcinoma among the five types of squamous cell carcinoma mentioned above. Therefore, researchers selected LGALS9 as the sole biomarker to construct a support vector machine model.

[0047] The LGALS9 mass spectrometry data from the Fudan cohort were divided into a training set and an internal validation set in a 7:3 ratio. Support vector machines and cross-validation were used to train the model on the training set, the internal validation set for model evaluation, and the external validation set for model validation to obtain the ROC curve. For details, please refer to CN 117969855 A. The results are as follows: Figure 2 As shown in the figure, the model built based on proteomic data using LGALS9 as a marker has AUCs of 0.884, 0.87, and 0.712 on the training set, internal validation set, and external validation set, respectively. This model's predictive performance for nasopharyngeal squamous cell carcinoma is lower than that of the combination of 39 feature proteins, and the model's performance evaluation stability is poor, failing to meet clinical needs.

[0048] Based on this, the technicians unexpectedly discovered that using immunohistochemistry to qualitatively detect LGALS9 protein resulted in a higher accuracy rate in identifying nasopharyngeal squamous cell carcinoma. Furthermore, a support vector machine model could be constructed based on the qualitative detection results, which significantly outperformed the prediction model constructed based on proteomic detection data in predicting nasopharyngeal squamous cell carcinoma. Moreover, as those skilled in the art know, the immunohistochemical detection method is simpler and easier to perform in clinical practice.

[0049] Example 2: A model for predicting nasopharyngeal squamous cell carcinoma was constructed based on immunohistochemical detection results. I. Research Methods 1. Patient enrollment Fudan Cohort: From June 2012 to December 2021, a total of 359 patients with squamous cell carcinoma were enrolled at Fudan University Cancer Hospital for this study. These included squamous cell carcinoma of the nasopharynx (n=44), cervical (n=82), esophageal (n=80), lung (n=82), and head and neck (n=71). This study was approved by the Medical Ethics Committee of Fudan University Cancer Hospital (ethics number: 2309282-22), and the study protocol followed the guidelines set forth in the Declaration of Helsinki. Written informed consent was obtained from all participants.

[0050] External validation cohort: From July 2002 to August 2024, a total of 150 patients with squamous cell carcinoma were enrolled at the Conde de São Januário Hospital in Macau (ethics number: 0053 / MEC / N / 2024) for this study. Among them, there were nasopharyngeal squamous cell carcinoma (n=30), head and neck squamous cell carcinoma (n=30), lung squamous cell carcinoma (n=30), esophageal squamous cell carcinoma (n=30), and cervical squamous cell carcinoma (n=30).

[0051] 2. Immunohistochemical detection Immunohistochemical staining of tumor tissue samples was performed using standard methods to verify the expression of LGALS9 protein in the tissue. The immunohistochemical detection method included the following steps: 1) Pre-treat tissue sections according to conventional methods, including dewaxing, dehydration to hydration, antigen retrieval, and serum blocking; 2) Add rabbit anti-human LGALS9 monoclonal antibody (dilution ratio 1:200) to the tissue section and incubate overnight at 4°C; 3) Remove from the water, allow to warm to room temperature for 15 minutes, and wash with 0.01M PBS 4 times, 5 minutes each time; 4) Add secondary antibody to the tissue section and incubate at 37°C for 30 min; 5) Wash three times with 0.01 mol / L PBS, 5 min each time; 6) Develop with DAB for 2-10 minutes, then observe under a microscope; 7) Stop the color development with double distilled water, then counterstain with hematoxylin for 10 seconds; 8) After differentiation, tap water turns blue, soak in distilled water, dehydrate and become transparent, then cover with a coverslip; 9) Observe the positive / negative staining under a microscope, and randomly select 5 fields of view and take pictures.

[0052] II. Research Results In the Fudan cohort, LGALS9 protein expression was as follows: 77.3% of nasopharyngeal squamous cell carcinomas (34 / 44) showed positive expression of LGALS9; 12.2% (10 / 82) of cervical squamous cell carcinomas; 13.8% (11 / 80) of esophageal squamous cell carcinomas; 15.9% (13 / 82) of lung squamous cell carcinomas; and 16.9% (12 / 71) of head and neck squamous cell carcinomas.

[0053] Using the Fudan University queue as the training set, the training set S is divided into k disjoint subsets / folds. Any fold is selected in turn as the test set to judge the performance of the support vector machine model. The remaining k-1 fold samples are used as the training set to train the model. At the same time, each model iterates through the hyperparameter nu values ​​in turn. That is, there are k models under each hyperparameter. The model is trained using a linear kernel function.

[0054] The specific operation process is as follows: Load the R language e1071 package, and train the model using the svm function for each pair of training sets (k-1 sets of data) and test sets (1 set of data). Set the parameter scale to TRUE, the parameter probability to T, the parameter nu to a user-defined number in the range of 0-1, the parameter type to "nu-classification", and the parameter kernel to "linear". Based on the average test accuracy of k models under a certain hyperparameter as described above, the model performance is taken as k, preferably k = 10. The hyperparameter nu is then successively set to 0.1, 0.2, 0.3, 0.4, 0.5, 0.6, 0.7, 0.8, and the hyperparameter nu that results in the best model performance is selected as the optimal hyperparameter. In this example, it is 0.4.

[0055] The immunohistochemical qualitative results of LGALS9 protein from the Fudan cohort training set were input, and a support vector machine model was constructed using the obtained hyperparameters, nu-classification, and linear kernel. Receiver operating characteristic (ROC) curves were plotted using MedCalc software, and the area under the curve (AUC) was calculated. The results are as follows: Figure 3 As shown, AUC = 0.812 (95% CI 0.747-0.878).

[0056] In the Macau cohort, LGALS9 protein expression was as follows: LGALS9 was positively expressed in 90% (27 / 30) of nasopharyngeal squamous cell carcinomas, but negatively expressed in 86.7% of cervical squamous cell carcinomas (26 / 30), 86.7% of esophageal squamous cell carcinomas (26 / 30), 73.3% of lung squamous cell carcinomas (22 / 30), and 76.6% of head and neck squamous cell carcinomas (23 / 30).

[0057] Using the Macau cohort as an external validation set, the efficacy of the above training set model in predicting nasopharyngeal squamous cell carcinoma was verified. Receiver operating characteristic (ROC) curves were plotted, and the results are as follows: Figure 4 As shown, the area under the receiver operating characteristic curve (AUC) is 0.882 (95% CI 0.815-0.949).

[0058] Immunohistochemical results showed that in the Fudan and Macau cohorts, the diagnostic accuracy of LGALS9 for nasopharyngeal squamous cell carcinoma detected by immunohistochemistry was 77.3% and 90%, respectively, which was significantly higher than the diagnostic accuracy of LGALS9 for nasopharyngeal squamous cell carcinoma detected by proteomic analysis.

[0059] Screening of targets for nasopharyngeal squamous cell carcinoma using immunohistochemistry: In previous studies, this invention discovered that, in addition to LGALS9 protein, SERPINB9 is also specifically highly expressed in nasopharyngeal squamous cell carcinoma. Therefore, the present invention's technicians used SERPINB9 as a specific biomarker and employed immunohistochemical methods for detection. The detection results are as follows: In the Fudan cohort, the positive expression rate of SERPINB9 was 88.9% (32 / 36) in nasopharyngeal squamous cell carcinoma, 81.8% (36 / 44) in cervical squamous cell carcinoma, 75.0% (33 / 44) in esophageal squamous cell carcinoma, 77.3% (34 / 44) in lung squamous cell carcinoma, and 88.2% (30 / 34) in head and neck squamous cell carcinoma. Due to sample detachment from some tissue microarrays, the above data only represent the results of valid sample detection and analysis.

[0060] Theoretically, SERPINB9 should show high positive expression in nasopharyngeal squamous cell carcinoma, but the above test results show no significant difference in the positive expression rate of SERPINB9 among all sources of squamous cell carcinoma. This phenomenon indicates that proteomics detection methods and immunohistochemical detection methods are not universally applicable for specific biomarker proteins. In other words, SERPINB9 biomarkers obtained through proteomics may not be effective when detected using immunohistochemical methods. The inventors believe that the main reason is that proteomics detects peptide sequences, and identification is possible as long as the sequence is unique. However, many factors affect the results of immunohistochemical detection, such as antibody quality, post-translational modifications (PTMs), and protein isoforms.

[0061] In summary, the inventive contribution of this invention lies in screening a single biomarker protein from multiple biomarker proteins that can be detected using immunohistochemistry. This not only simplifies experimental procedures and reduces detection costs, but also achieves higher predictive and diagnostic efficacy for nasopharyngeal squamous cell carcinoma.

[0062] The above specific embodiments are merely illustrative of the invention and do not represent a limitation thereof. Those skilled in the art will recognize that other variations of the specific structure of this invention are possible.

Claims

1. The application of a substance that detects the biomarker LGALS9 in the preparation of products for diagnosing nasopharyngeal squamous cell carcinoma, characterized in that, The substance used to detect the LGALS9 biomarker is any reagent required to detect LGALS9 protein or gene levels; the products used to diagnose nasopharyngeal squamous cell carcinoma include models, chips, detection platforms, kits, test strips, or membrane strips.

2. The application according to claim 1, characterized in that, The reagents used to detect LGALS9 protein levels are those required for detecting LGALS9 protein levels using proteometry, immunohistochemistry, enzyme-linked immunosorbent assay, Western blotting, or immunofluorescence. The reagents used to detect the LGALS9 gene level are those required for detecting the LGALS9 gene level using RT-PCR, RT-qPCR, gene chip detection, DNA blotting, or in situ hybridization.

3. The application according to claim 1, characterized in that, The substance used to detect the marker LGALS9 is selected from any one of the following: (1) Reagents for detecting LGALS9 protein levels using proteometry; (2) Reagents for detecting LGALS9 protein levels using immunohistochemistry.

4. The application according to claim 3, characterized in that, The substance used to detect the LGALS9 biomarker is a reagent for detecting LGALS9 protein levels using immunohistochemistry, and the reagent includes an anti-LGALS9 antibody.

5. The application according to claim 1, characterized in that, The diagnosis of nasopharyngeal squamous cell carcinoma includes the following steps: (1) The level of LGALS9 protein in the sample was detected by immunohistochemistry; (2) When LGALS9 is expressed positively in the sample to be tested, it indicates that the sample is a squamous cell carcinoma of nasopharyngeal origin.

6. A method for constructing a model for diagnosing nasopharyngeal squamous cell carcinoma, characterized in that, A support vector machine model was constructed using the immunohistochemical qualitative detection results of LGALS9 protein as variables.

7. The method according to claim 6, characterized in that, The method for constructing the model includes the following specific steps: (1) Divide the samples into training set and validation set. Divide the training set into k disjoint folds. Select any fold as the test set to judge the performance of the support vector machine model. Use the remaining k-1 fold samples as the training set to train the model. Train the model using the svm function. (2) Input the immunohistochemical qualitative detection results of LGALS9 protein in the training set samples, construct the support vector machine model, and evaluate the model performance using the receiver operating characteristic curve; (3) The model was evaluated using the immunohistochemical detection results of LGALS9 protein from the validation set samples.

8. A model, characterized in that, The model is constructed by the method described in any one of claims 6 or 7.

9. The model according to claim 8, characterized in that, The model is used to diagnose squamous cell carcinoma of nasopharyngeal origin, and includes an information acquisition module and a diagnostic module; The information acquisition module is used to perform the step of acquiring information on the expression level of LGALS9 protein in the sample to be tested. The diagnostic module is used to perform the step of determining whether the sample to be tested is a squamous cell carcinoma of nasopharyngeal origin based on the LGALS9 protein expression level information.

10. The use of the model according to claim 8 or 9 in the preparation of products for diagnosing nasopharyngeal squamous cell carcinoma, characterized in that, The application includes the following steps: S1) Immunohistochemistry was used to detect the LGALS9 protein level in the sample to obtain the qualitative results of LGALS9 expression in the sample. S2) Perform data analysis and model calculation on the immunohistochemical detection results obtained in step S1); S3) Determine whether the sample to be tested is a squamous cell carcinoma of nasopharyngeal origin based on the probability score calculated by the model.

Images