Methods for constructing cancer assessment models, as well as cancer assessment systems, media, and electronic devices.

Abstract

This invention discloses a method for constructing a cancer assessment model, as well as a cancer assessment system, media, and electronic device. The assessment model developed in this invention significantly improves the diagnostic efficacy of hepatitis B virus-related liver cancer, especially demonstrating excellent predictive performance in early-stage liver cancer and patients with negative serum biomarkers. Compared with single biomarkers and existing scoring models, the cancer assessment model constructed in this invention has significant advantages in comprehensive indicators such as AUC, sensitivity, and specificity. Although GP73, introduced in the model, has a negative coefficient, its synergistic effect with other biomarkers effectively improves diagnostic performance. The assessment model constructed in this invention provides a reliable tool for non-invasive prediction and early screening of liver cancer, and has significant clinical application value.

Description

Technical Field

[0001] This invention relates to the field of disease assessment, and more specifically to cancer assessment models and their construction methods, as well as cancer assessment systems, computer-readable storage media, and electronic devices based on these models. Background Art

[0002] Hepatocellular carcinoma (HCC) is one of the most common malignant tumors worldwide. Although ultrasound, magnetic resonance imaging (MRI), and liver biopsy have greatly improved the accuracy of HCC diagnosis, these techniques have certain limitations, such as high cost, invasiveness, and the inability to detect small tumors <2 cm in a timely manner, which greatly limits their application. Currently, various molecular markers, such as alpha-fetoprotein (AFP), AFP-L3 isoform, and abnormal prothrombin (DCP), are widely used in the clinical diagnosis of HCC. However, the detection accuracy of individual molecular markers is still insufficient. Multiple studies have shown that the combined use of multiple markers can significantly improve the sensitivity and specificity of liver cancer diagnosis. Currently, the GALAD (composed of sex, age, AFP, AFP-L3%, and DCP) multi-marker diagnostic model for HCC, first developed by German researchers, has demonstrated good diagnostic efficacy in multi-center studies, across different countries and populations, making the GALAD model highly universal. Based on this, domestic scholars have developed the C-GALAD scoring system, which is more suitable for Chinese patients, but its specificity and sensitivity in diagnosing liver cancer are limited.

[0003] The information in the background section is merely intended to illustrate the general background of the invention and should not be construed as an admission or implication in any way that such information constitutes prior art known to those skilled in the art. Summary of the Invention

[0004] To address at least some of the technical problems existing in the prior art, this invention provides a method for constructing a cancer assessment model based on the combination of multiple indicators. This method not only analyzes the correlation between each indicator and the disease, but also analyzes the correlation between the indicators themselves, excluding closely correlated indicators, and incorporating new indicators through specific multifactor analysis, thereby significantly improving the sensitivity and accuracy of the assessment model. Specifically, this invention includes the following:

[0005] A first aspect of the present invention provides a method for constructing a cancer assessment model, comprising the following steps:

[0006] (1) Group the samples according to the symptoms;

[0007] (2) Obtain multiple indicators that differ between different groups, and analyze the correlation between each indicator and the disease, as well as the correlation between each indicator, to obtain the initial screening indicators.

[0008] (3) Based on the following formula, a multivariate analysis of the initial screening indicators was performed to obtain the cancer assessment model: logit(P)=ln[P / (1-P)]=β0+β1x1+β2x2+......+βmxm;

[0009] Where P represents the probability of disease A, 1-P represents the probability of disease B, β0 represents the intercept term, and β1, β2, and βm each represent the coefficients of the corresponding indicators.

[0010] In some implementations, according to the method for constructing a cancer assessment model based on the present invention, the conditions include chronic hepatitis B, HBV-related cirrhosis, and newly diagnosed liver cancer patients.

[0011] In some implementations, the method for constructing a cancer assessment model according to the present invention includes indicators comprising patient basic characteristics and serological indicators.

[0012] In some implementations, according to the method for constructing a cancer assessment model according to the present invention, indicators that show significant differences between disease groups and have no or low correlation with each other are selected as initial screening indicators.

[0013] In a second aspect, the present invention provides a cancer assessment model, wherein the assessment model is constructed by the method described in the first aspect.

[0014] In some embodiments, according to the cancer assessment model of the present invention, the assessment model formula is:

[0015] logit(P) = -10.837 + 0.102 × age + 2.754 × gender (male = 1, female = 0) - 2.019 × log GP73 + 0.922 × log(AFP) + 2.468 × log(DCP).

[0016] A third aspect of the present invention provides a cancer assessment system comprising:

[0017] The data acquisition unit is configured to acquire the subject's indicator data;

[0018] A data processing unit includes the cancer assessment model described in the second aspect and is configured to process the indicator data acquired in the data acquisition unit using the cancer assessment model to obtain prediction results.

[0019] In some embodiments, according to the cancer assessment system of the present invention, the indicator data includes the subject's age, sex, and levels of AFP, DCP, and GP73.

[0020] A fourth aspect of the present invention provides a computer-readable storage medium storing a computer program that includes the cancer assessment model described in the second aspect, and which, when executed by a processor, implements the following method to assess cancer:

[0021] The subject's indicator data are acquired from the data acquisition unit;

[0022] The cancer assessment model is used to process the indicator data acquired in the data acquisition unit to obtain prediction results.

[0023] A fifth aspect of the present invention provides an electronic device comprising: a processor and a memory, wherein:

[0024] The memory is used to store the executable instructions of the processor;

[0025] The processor is configured to perform the following method by executing the executable instructions:

[0026] The subject's indicator data is obtained from the data acquisition unit, and the cancer assessment model described in the second aspect is used to process the indicator data obtained from the data acquisition unit to obtain the prediction result.

[0027] This invention provides a method, system, medium, and electronic device for constructing a cancer assessment model. The developed assessment model significantly improves the diagnostic efficacy of hepatitis B virus-related liver cancer, especially demonstrating excellent predictive performance in early-stage liver cancer and patients with negative serum biomarkers. Compared with single biomarkers (AFP, DCP, GP73) and existing scoring models (GALAD, C-GALAD), the cancer assessment model constructed in this invention has significant advantages in comprehensive indicators such as AUC, sensitivity, and specificity. Although GP73 has a negative coefficient in the model, its synergistic effect with other biomarkers effectively improves diagnostic performance. Studies have shown that the model developed in this invention provides a reliable tool for non-invasive prediction and early screening of liver cancer, and has important clinical application value. Attached Figure Description

[0028] Figure 1 Correlation heatmap among DCP, AFP, AFP-L3%, and GP73.

[0029] Figure 2 A comparison of the diagnostic value of DCP, AFP, GP73, G-GADA, C-GALAD, and GALAD scores in liver cancer (A) and early-stage liver cancer (A).

[0030] Figure 3A comparison of the diagnostic value of DCP, AFP, GP73, G-GADA, C-GALAD, and GALAD scores in differentiating between liver cirrhosis and liver cancer. Among them, (A) liver cancer, (B) early-stage liver cancer.

[0031] Figure 4 A comparison of the diagnostic value of DCP, AFP, GP73, G-GADA, C-GALAD, and GALAD scores in differentiating chronic hepatitis B from liver cancer. Among them, (A) liver cancer, (B) early-stage liver cancer.

[0032] Figure 5 The role of DCP, AFP, GP73, G-GADA, C-GALAD, and GALAD scores in the diagnosis of liver cancer in AFP- and DCP-negative patients. Detailed Implementation

[0033] Various exemplary embodiments of the present invention will now be described in detail. This detailed description should not be considered as limiting the present invention, but rather as a more detailed description of certain aspects, features, and embodiments of the present invention.

[0034] It should be understood that the terminology used in this invention is merely for describing particular embodiments and is not intended to limit the invention. Furthermore, with respect to numerical ranges in this invention, it should be understood that the upper and lower limits of the range and each intermediate value between them are specifically disclosed. Any stated value or intermediate value within a stated range, as well as each smaller range between any other stated value or intermediate value within said range, are also included in this invention. The upper and lower limits of these smaller ranges may be independently included or excluded from the range.

[0035] Unless otherwise stated, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains. While only preferred methods and materials have been described herein, any methods and materials similar or equivalent to those described herein may be used in the implementation or testing of this invention. All references to this specification are incorporated by way of citation to disclose and describe methods and / or materials associated with those references. In the event of any conflict with any incorporated reference, the content of this specification shall prevail.

[0036] Methods for constructing cancer assessment models

[0037] One aspect of the present invention provides a method for constructing a cancer assessment model, comprising the following steps:

[0038] (1) Group the samples according to the symptoms;

[0039] (2) Obtain multiple indicators that differ between different groups, and analyze the correlation between each indicator and the disease, as well as the correlation between each indicator, to obtain the initial screening indicators.

[0040] (3) Based on the following formula, a multifactor analysis of the initial screening indicators is performed to obtain the cancer assessment model: logit(P)=ln[P / (1-P)]=β0+β1x1+β2x2+......+βmxm.

[0041] In step (1) of this invention, samples are grouped according to patients with chronic hepatitis B, HBV-related cirrhosis, and newly diagnosed liver cancer. Patients already suffering from chronic hepatitis B, HBV-related cirrhosis, or newly diagnosed liver cancer are accurately diagnosed using standards known in the art. Liver cancer patients can be further divided into early and late stages. The sample type is not particularly limited and can be a fluid sample (blood sample) or a tissue sample from the patient.

[0042] In step (2) of this invention, multiple indicators showing differences between different groups are obtained, including but not limited to basic patient characteristics (such as patient gender, age, etc.) and serological indicators (such as DCP, AFP, GP73, AFP-L3%). Then, the correlation between each indicator and the disease, as well as the correlation between the indicators themselves, are analyzed to obtain initial screening indicators. In one specific embodiment, the correlation between each indicator and the disease refers to a significant correlation between gender, age, DCP, AFP, GP73, AFP-L3%, and the disease. In-depth research revealed that AFP-L3% is correlated with AFP, DCP, GP73, and gender; therefore, this indicator is discarded. Meanwhile, GP73 shows no significant correlation with DCP and a weak correlation with AFP and AFP-L3%, making it a primary screening indicator. This significantly improves the problem of low diagnostic value (low AUC value) caused by using GP73 as an indicator or reference factor. In a preferred embodiment, the initial screening indicators include gender, age, DCP, AFP, and GP73.

[0043] In step (3) of the present invention, gender, age, DCP, AFP, and GP73 are used as variables, based on logit(P)=ln[P / (1-P)]=β0+β1x1+β2x2+......+βmxm, where P represents the probability of disease A, 1-P represents the probability of disease B, β0 represents the intercept term, and β1, β2, and βm each represent the coefficient of the corresponding index.

[0044] In this invention, the final formula for the cancer assessment model is: logit(P) = -10.837 + 0.102 × age + 2.754 × gender (male = 1, female = 0) - 2.019 × log GP73 + 0.922 × log(AFP) + 2.468 × log(DCP).

[0045] Cancer assessment models and systems containing them

[0046] In one aspect of the present invention, a cancer assessment model constructed by the above method is provided, the formula of which is: logit(P)=-10.837+0.102×age+2.754×gender (male=1, female=0)-2.019×log GP73+0.922×log(AFP)+2.468×log(DCP).

[0047] In another aspect, the present invention provides a cancer assessment system (also referred to as a "cancer assessment device") comprising:

[0048] The data acquisition unit is configured to acquire the subject's indicator data;

[0049] The data processing unit includes the aforementioned cancer assessment model and is configured to process the indicator data acquired in the data acquisition unit using the cancer assessment model to obtain prediction results.

[0050] In this invention, the data acquisition unit can be any suitable device or instrument capable of acquiring the subject's age, sex, and the levels of AFP, DCP, and GP73. For example, it could be an electronic information storage unit containing basic patient information, medical records, follow-up data, etc., or a reagent kit or instrument capable of measuring or quantifying AFP, DCP, and GP73. An exemplary data acquisition unit includes a multi-mode input port. This port can receive information data from various sources, including the subject's age, sex, and the levels of AFP, DCP, and GP73, both online and offline. Another exemplary data acquisition unit includes a port for connection to a serological marker assay instrument (e.g., a chemiluminescence immunoassay analyzer).

[0051] In this invention, the data processing unit may further include an evaluation unit. The form of the evaluation unit is not limited, as long as it includes the evaluation model of this invention. For example, the evaluation unit is a processor. Preferably, the evaluation unit is communicatively connected to the data acquisition unit, thereby enabling it to retrieve data from the data acquisition unit and input it into the evaluation model for calculation.

[0052] The present invention further provides a computer storage medium (readable storage medium) or cloud for use by, or in conjunction with, an instruction execution system, apparatus, or device (such as a computer-based system, a processor-included system, or other system that can fetch and execute instructions from, an instruction execution system, apparatus, or device). For the purposes of this invention, a "computer-readable medium" can be any means that can contain, store, communicate, propagate, or transmit programs for use by, or in conjunction with, an instruction execution system, apparatus, or device. More specific examples of computer-readable media include: an electrical connection (electronic device) having one or more wires, a portable computer disk drive (magnetic device), random access memory (RAM), read-only memory (ROM), erasable and editable read-only memory (EPROM or flash memory), fiber optic devices, and portable optical disc read-only memory (CDROM). Alternatively, the computer-readable medium may be paper or other suitable media on which the program can be printed, since the program can be obtained electronically, for example, by optically scanning the paper or other medium, followed by editing, interpreting, or otherwise processing as necessary, and then stored in a computer memory.

[0053] It should be understood that various parts of the present invention can be implemented in hardware, software, firmware, or a combination thereof. In the above embodiments, multiple steps or methods can be implemented in software or firmware stored in memory and executed by a suitable instruction execution system. For example, if implemented in hardware, as in another embodiment, it can be implemented using any one or a combination of the following techniques known in the art: discrete logic circuits having logic gates for implementing logical functions on data signals, application-specific integrated circuits (ASICs) having suitable combinational logic gates, programmable gate arrays (PGAs), field-programmable gate arrays (FPGAs), etc.

[0054] Those skilled in the art will understand that all or part of the steps of the method implementing the above embodiments can be implemented by a program instructing related hardware. The program can be stored in a computer-readable storage medium, and when executed, the program includes one or a combination of the steps of the method embodiments.

[0055] Furthermore, the functional units in the various embodiments of this invention can be integrated into a single processing module, or each unit can exist physically separately, or two or more units can be integrated into a single module. The integrated module can be implemented in hardware or as a software functional module. If the integrated module is implemented as a software functional module and sold or used as an independent product, it can also be stored in a computer-readable storage medium. The aforementioned storage medium can be a read-only memory, a hard disk, or an optical disk, etc.

[0056] Example

[0057] I. Materials and Methods

[0058] 1. Research Subjects

[0059] This study included 332 non-hepatocellular carcinoma (HCC) cases treated at Mengchao Hepatobiliary Hospital of Fujian Medical University from June 2015 to June 2020, as well as 70 patients initially diagnosed with HCC without prior treatment. The 332 non-HCC patients included 209 patients with chronic hepatitis B and 123 patients with HBV-related cirrhosis. All 70 HCC patients met the criteria of the "Guidelines for the Diagnosis and Treatment of Primary Liver Cancer" published by the Chinese Society of Clinical Oncology, and were confirmed by histopathology or imaging. Early-stage HCC was defined as a single tumor diameter not exceeding 5 cm, or the total diameter of multiple tumors not exceeding 3 cm. Late-stage HCC was defined as exceeding the Milan criteria, such as a single tumor larger than 5 cm, or the total diameter of multiple tumors exceeding 3 cm. The 70 HCC patients included 34 cases of early-stage HCC and 36 cases of intermediate-to-late-stage HCC; 7 cases with distant metastasis and 63 cases without distant metastasis; 23 cases with microvascular invasion and 47 cases without microvascular invasion.

[0060] 2. Inclusion and Exclusion Criteria

[0061] 2.1 Inclusion criteria:

[0062] (1) Study subjects must meet the following diagnostic criteria: Primary liver cancer is diagnosed according to the "Guidelines for the Diagnosis and Treatment of Primary Liver Cancer (2011, 2017 Editions)" and confirmed by liver tissue biopsy or cytological examination. Cirrhosis and chronic hepatitis are diagnosed according to the "Guidelines for the Prevention and Treatment of Chronic Hepatitis B (2015, 2019 Editions)". According to the Milan criteria, liver cancer can be further divided into early and late stages. In addition, the following conditions must be met: a. Age 18 years and above, regardless of gender; b. HBsAg and / or HBV DNA positive for more than 6 months;

[0063] (2) Liver cancer patients must be diagnosed for the first time;

[0064] (3) The study subjects must not have severe liver or kidney dysfunction;

[0065] (4) Patients must participate in the study voluntarily and be able to fully understand and sign the informed consent form.

[0066] 2.2 Exclusion criteria:

[0067] (1) Diagnosed with other viral infections (such as HIV) before enrollment;

[0068] (2) Has a history of major organ transplantation, received warfarin treatment, or has undergone surgery, ablation, radiotherapy or chemotherapy for liver cancer, liver metastasis or intrahepatic bile duct cancer;

[0069] (3) Has a history of pancreatitis, or currently has clinical or laboratory evidence to support the diagnosis of pancreatitis, or shows symptoms and laboratory abnormalities related to pancreatitis;

[0070] (4) No frozen serum samples are available, or the serum volume is insufficient to complete the biomarker detection;

[0071] (5) Unable to understand or refuse to sign the informed consent form, or unable to comply with the research protocol requirements;

[0072] (6) Other circumstances deemed unsuitable for participation in the study by the researchers. The research team strictly adhered to the above diagnostic, inclusion, and exclusion criteria during the study. This study has been approved by the Ethics Committee of Mengchao Hepatobiliary Hospital, Fujian Medical University (Approval No.: 2017-014-01). All patients were informed about the study and voluntarily signed the study consent form.

[0073] 3. Methods

[0074] 3.1 Specimen Collection and Testing

[0075] In this embodiment, peripheral blood samples were collected from patients meeting the study criteria. The collected blood samples were centrifuged at 1000 r / min for 10 minutes to separate the serum. The serum was aliquoted and immediately frozen at -80°C until testing. Serum AFP, DCP, and GP73 levels were analyzed using a chemiluminescent immunoassay (MAGLUMI). ® AFP (CLIA), MAGLUMI ® DCP (CLIA) and MAGLUMI ® GP73 (CLIA) was measured, and the testing procedure was strictly performed according to the specific storage conditions and usage guidelines in the kit instructions. The cutoff values ​​were age >50, DCP ≤40 mAU / mL, and AFP <20 ng / mL. Since there is no unified standard for the reference range of GP73 in clinical practice, this example mainly aims to compare and analyze the relationship between its levels and the diagnosis of liver cancer.

[0076] 3.2 Observation Indicators

[0077] GALAD model value calculation: logit(P) = -10.08 + 0.09 × [age] + 1.67 × [gender (male = 1, female = 0)] + 2.34 × log10 [AFP] + 0.04 × AFP-L3 (%) + 1.33 × log10 [DCP];

[0078] C-GALAD model value calculation: logit(P)=-11.501+0.733×[gender(male=1, female=0)]+0.099×[age]+0.073×[AFP-L3(%)]+0.840×log[AFP(μg / L)]+2.346×log[DCP(mAU / mL)].

[0079] 4. Statistical Analysis

[0080] Based on the distribution characteristics of each variable, appropriate descriptive statistical methods were selected, and the normality of the data was assessed using the Kolmogorov-Smirnov (KS) test. For normally distributed quantitative data, the mean and standard deviation (X±SD) were used for description; for non-normally distributed variables, the median and interquartile ranges [M (P25, P75)] were used. Categorical variables were expressed as percentages, and differences between groups were compared using the χ² test. Statistical analysis and graphing were performed using SPSS 26.0, GraphPad Prism 9.0, and R language version 3.4.4. In this example, MedCalc software was used to plot the receiver operating characteristic (ROC) curve, calculate the area under the ROC curve, determine the optimal cutoff value using the Youden index, and calculate sensitivity and specificity. Spearson correlation analysis was performed. All statistical tests were two-tailed. P < 0.05 was considered statistically significant.

[0081] II. Results

[0082] 1. Comparison of baseline characteristics of subjects

[0083] In this embodiment, there were significant differences in age, sex, and various biomarkers (DCP, AFP, GP73, AFP-L3%) among different groups (CHB, LC, HCC) (P<0.05). The HCC group was predominantly male, with a mean age of 54 years. Serum DCP, AFP, and AFP-L3% levels in HCC patients were significantly higher than in the CHB and LC groups. GP73 levels also showed significant differences between the CHB group and the LC and HCC groups (P<0.05), but no statistically significant difference between the LC and HCC groups. These results provide potential discriminant variables for the diagnostic model (Table 1).

[0084] Table 1. Baseline characteristics analysis of the study population

[0085]

[0086] * indicates a difference compared to the CHB group, and # indicates a difference compared to the LC group.

[0087] 2. Correlation analysis between serological markers and clinical characteristics of liver cancer

[0088] Based on the cutoff values ​​of serological indicators, liver cancer patients were divided into two groups: a DCP low expression group (n=14) and a DCP high expression group (n=56) based on a DCP cutoff value of 40 mAU / mL; and an AFP low expression group (n=30) and an AFP high expression group (n=40) based on an AFP cutoff value of 20 ng / mL.

[0089] Further analysis showed that serum DCP levels in liver cancer patients were significantly correlated with tumor size and Milan stage (P<0.05); serum AFP levels were significantly correlated with tumor size, microvascular invasion, and Milan stage (P<0.05); serum AFP-L3% levels were significantly correlated with patient age, sex, tumor size, distant metastasis, microvascular invasion, and Milan stage (P<0.05); and serum GP73 levels were significantly correlated with tumor size, tumor number, microvascular invasion, and Milan stage (P<0.05). Specific results are shown in Table 2.

[0090] Table 2. Correlation between serum DCP, AFP, AFP-L3%, and GP73 levels and clinical and tumor characteristics in patients with liver cancer.

[0091]

[0092] 3. Correlation analysis between clinical and serological indicators of patients

[0093] Correlation analysis was performed on patient age and serum DCP, AFP, AFP-L3%, and GP73 levels. The results showed that AFP was weakly correlated with patient gender, DCP, and GP73 (P<0.05), and strongly correlated with AFP-L3% (P<0.05). DCP showed no significant correlation with GP73, but was weakly correlated with patient gender and AFP-L3% (P<0.05). AFP-L3% also showed a weak correlation with GP73 and patient gender (P<0.05). In conclusion, AFP-L3% was correlated with the other three serological indicators and patient gender. Figure 1 ).

[0094] 4. Establishment of the G-GADA model

[0095] Patient age, sex, DCP, AFP, AFP-L3%, and GP73 were used as independent variables, and whether or not the patient had liver cancer (non-liver cancer = 0, liver cancer = 1) was used as the dependent variable. The following formula was used: logit(P) = ln[P / (1-P)] = β0 + β1x1 + β2x2 + β3x3 + β4x4 + β5x5. Where P represents the probability of liver cancer, 1-P represents the probability of not having liver cancer, β0 represents the intercept term, and β1, β2, β3, β4, and β5 each represent the coefficients of their respective independent variables. After a series of screenings, the variables ultimately included in the model were age, sex, DCP, AFP, and GP73, and the following equation was established: logit(P) = −10.837 + 0.102 × age + 2.754 × sex (male = 1, female = 0) - 2.019 × log GP73 + 0.922 × log(AFP) + 2.468 × log(DCP) (Table 3). The significant difference between this model and the GALAD model is that GP73 replaces AFP-L3%. GP73, as an easily detectable biomarker, provides greater convenience and potential advantages for the widespread clinical application of the model. To reflect this improvement while retaining its association with the GALAD model, this model is named the "G-GADA model," an optimized version that incorporates the innovative biomarker GP73 into the GALAD model. This naming not only highlights the innovation in variable selection but also emphasizes the model's advantages in clinical application.

[0096] Table 3. Multivariate analysis of patient age, gender, DCP, AFP, AFP-L3%, and GP73.

[0097]

[0098] 5. Comparison of the diagnostic value of the G-GADA model with other indicators

[0099] 5.1 The efficacy of DCP, AFP, GP73, G-GADA, C-GALAD and GALAD score in diagnosing liver cancer and early-stage liver cancer

[0100] The diagnostic value of DCP, AFP, GP73, G-GADA, C-GALAD, and GALAD score for liver cancer was evaluated using ROC curves. The results showed that the AUCs of single serum markers DCP, AFP, and GP73 were 0.888, 0.708, and 0.532, respectively. In contrast, the AUC of the G-GADA model reached 0.941, which was significantly better than the single markers and the C-GALAD and GALAD scoring models (AUC: 0.915, 0.890) (Z G-GADA - C-GALAD = 2.558, Z G-GADA - GALAD = 3.593). p =0.0105、p= 0.0003). In the early diagnosis of liver cancer, the G-GADA model also demonstrated excellent diagnostic ability. The AUC of the G-GADA model reached 0.908, which was significantly better than the single index and the C-GALAD and GALAD (AUC: 0.848, 0.814) scoring models (Z G-GADA - C-GALAD = 4.580, Z G-GADA - GALAD = 4.388); p< 0.0001 p= (0.0001). Overall, the G-GADA model performed best in terms of comprehensive performance indicators such as AUC, sensitivity, and specificity in both liver cancer and early-stage liver cancer diagnosis, demonstrating higher diagnostic efficiency and showcasing its clinical application potential in liver cancer screening (see Table 4). Figure 2 .

[0101] Table 4. Comparison of the diagnostic value of DCP, AFP, GP73, G-GADA, C-GALAD, and GALAD scores in liver cancer and early-stage liver cancer.

[0102]

[0103] 5.2 The efficacy of DCP, AFP, GP73, G-GADA, C-GALAD, and GALAD scores in differentiating between liver cancer and liver cirrhosis

[0104] Further subgroup analysis was conducted to evaluate the performance of the G-GADA model and other indicators in the differential diagnosis of cirrhosis and hepatocellular carcinoma (HCC), as well as early-stage HCC. In the diagnosis of cirrhosis and HCC, the AUC values ​​of the single serum biomarkers DCP, AFP, and GP73 were 0.891, 0.725, and 0.520, respectively. The G-GADA model, however, had an AUC of 0.916, a sensitivity of 88.57%, and a Youden index of 0.7475, significantly outperforming other single biomarkers and the C-GALAD (AUC: 0.885) and GALAD (AUC: 0.866) scoring models (ZG-GADA - C-GALAD = 2.113, ZG-GADA - GALAD = 2.633). p< 0.0346 p=0.0085). In the differentiation between cirrhosis and early-stage liver cancer, the AUC values ​​of single serum markers AFP, DCP, and GP73 were 0.601, 0.815, and 0.654, respectively. The AUC value of the G-GADA model was 0.870, with specificity, positive predictive value, and negative predictive value of 86.18%, 62.29%, and 94.43%, respectively, and a Youden index of 0.6853, which were significantly better than single indicators and the C-GALAD (AUC: 0.797) and GALAD (AUC: 0.773) scoring models (Z G-GADA - C-GALAD = 3.576, Z G-GADA - GALAD = 3.300). p< 0.0003 p= (0.0010). Overall, the G-GADA model demonstrated the best comprehensive diagnostic performance. This result further validates the high efficacy of the G-GADA model in the differential diagnosis of liver cancer (see Table 5). Figure 3 .

[0105] Table 5. Comparison of the diagnostic value of DCP, AFP, GP73, G-GADA, C-GALAD, and GALAD scores in differentiating between cirrhosis and liver cancer.

[0106]

[0107] 5.3 The efficacy of DCP, AFP, GP73, G-GADA, C-GALAD, and GALAD scores in differentiating between liver cancer and chronic hepatitis B

[0108] In the diagnosis of chronic hepatitis B and liver cancer, the G-GADA model demonstrated excellent diagnostic performance, with an AUC of 0.955, sensitivity and specificity of 88.57% and 95.22%, respectively, and a Youden index of 0.8379. In contrast, among single serum biomarkers, DCP had an AUC of 0.887 and outstanding specificity (97.61%), but its overall diagnostic ability was still inferior to the G-GADA model. The AUCs of the C-GALAD and GALAD models were 0.933 and 0.905, respectively, and their specificity, positive predictive value, and Youden index were all lower than those of G-GADA, with statistically significant differences (ZG-GADA - C-GALAD = 2.324, ZG-GADA - GALAD = 3.582). p= 0.0201 p=0.0003). In the diagnosis of chronic hepatitis B and early-stage liver cancer, the G-GADA model also performed best, with an AUC value of 0.931. In comparison, the AUCs of the C-GALAD and GALAD models were 0.878 and 0.838, respectively, and were lower than G-GADA in terms of specificity, positive predictive value, negative predictive value, and comprehensive indicators (Z G-GADA - C-GALAD = 3.947, Z G-GADA - GALAD = 4.266). p= 0.0001 p< (0.0001). Overall, the G-GADA model showed the best performance in diagnosing liver cancer in individuals with chronic hepatitis B, demonstrating its important clinical value in early screening and accurate diagnosis of liver cancer. See Table 6. Figure 4 .

[0109] Table 6. Comparison of the diagnostic value of DCP, AFP, GP73, G-GADA, C-GALAD, and GALAD scores in differentiating chronic hepatitis B from liver cancer.

[0110]

[0111] 5.4 The role of DCP, AFP, GP73, G-GADA, C-GALAD, and GALAD scores in the diagnosis of hepatocellular carcinoma in AFP- and DCP-negative patients.

[0112] In the diagnosis of hepatocellular carcinoma in AFP-negative (<20 ng / mL) patients, although the G-GADA model showed no statistically significant difference in AUC compared to the C-GALAD and GALAD models (G-GADA: 0.900, C-GALAD: 0.879, GALAD: 0.870) (p=0.0568, p=0.0789), G-GADA outperformed other models in specificity (93.01%), positive predictive value (55.94%), negative predictive value (97.67%), and Youden index (0.7301). In contrast, among single serum biomarkers, DCP performed best with an AUC of 0.849 and high specificity (86.76%), but its overall performance was significantly lower than that of G-GADA. In DCP-negative (≤40 mAU / mL) patients, the G-GADA model demonstrated a good balance between sensitivity (78.57%) and specificity (66.89%). Compared to the higher specificity of C-GALAD (87.71%) and GALAD (92.86%), the G-GADA model showed superior overall diagnostic performance, effectively compensating for the imbalance between sensitivity and specificity between C-GALAD and GALAD. Overall, in AFP- or DCP-negative patients, the G-GADA model demonstrated stronger comprehensive diagnostic capabilities compared to the C-GALAD and GALAD models. This further highlights the clinical value of G-GADA, especially for patients with negative serum biomarkers (see Table 7). Figure 5 .

[0113] Table 7. Comparison of the diagnostic value of DCP, AFP, GP73, G-GADA, C-GALAD, and GALAD scores in the differential diagnosis of chronic hepatitis B and liver cancer.

[0114]

[0115] III. Discussion

[0116] Currently, the diagnosis of hepatocellular carcinoma (HCC) mainly relies on imaging examinations such as ultrasound, supplemented by serum tumor markers (AFP, AFP-L3, and DCP). However, imaging has insufficient sensitivity for early HCC and is affected by operator subjectivity. Computed tomography (CT) or magnetic resonance imaging (MRI) examinations have problems such as radiation risks, high costs, and false positive signals due to nonspecificity. Several scholars have proposed establishing a combined diagnostic mathematical model using multiple biochemical indicators to improve the diagnostic accuracy of HCC. Best et al. constructed the GALAD scoring model based on AFP, AFP-L3, DCP levels, age, and sex, demonstrating its diagnostic efficacy is higher than that of a single indicator. Research by Feng Wenxing et al. showed that the AUC of the GALAD model for diagnosing HCC is superior to ultrasound examination. Domestically, Liu et al. developed the C-GALAD model based on Chinese patient data, which has superior diagnostic performance compared to traditional models. The GALAD model includes three biomarkers (AFP, AFP-L3, and DCP) and two demographic characteristics (age and sex). AFP is widely upregulated in hepatocellular regeneration, primary liver cancer cells, and embryonal carcinoma tissues, and can serve as a biomarker for diagnosing HCC. Some scholars believe its diagnostic performance is good, but 30% of diagnosed HCC patients are still AFP-negative. AFP-L3 is a subtype of AFP, secreted solely by HCC cells, and is the main form of AFP in the serum of HCC patients. Its level is correlated with tumor differentiation and can effectively reflect the growth rate and size of HCC tumors. This invention also confirms its correlation with tumor size, the presence of distant metastases, microvascular invasion, and tumor stage. Under normal circumstances, DCP is usually undetectable in the serum of healthy individuals because it is mainly produced by malignant tumor hepatocytes. Therefore, DCP is considered a highly specific HCC biomarker and is widely recommended for liver cancer screening, especially showing potential advantages in AFP-negative patients. However, the clinical application of DCP in HCC remains controversial. In recent years, GP73 has been widely used as a novel biomarker for HCC detection. In normal livers, GP73 is mainly expressed in bile duct epithelial cells, while its expression in hepatocytes is relatively low. Studies have shown that the AUC value of GP73 is as high as 0.89 in chronic hepatitis B, chronic hepatitis C, non-alcoholic fatty liver disease, autoimmune hepatitis, and the resulting cirrhosis. Therefore, serum GP73 has good diagnostic value for hepatitis and cirrhosis caused by different etiologies. Other studies have shown that GP73 has a sensitivity of 74.6%, a specificity of 97.4%, and a positive predictive value of 82.7% in diagnosing hepatocellular carcinoma, with serum expression higher than in control groups and adjacent tissues. Therefore, GP73 is not only a good serum marker for liver fibrosis and cirrhosis, but is also widely used to differentiate between healthy individuals, benign liver lesions, and liver cancer.

[0117] This study found that the expression level of GP73 in patients with chronic hepatitis B, cirrhosis, and hepatocellular carcinoma (HCC) exhibits a non-linear trend. There was no significant difference in GP73 expression between LC and HCC patients, while it was lowest in LC patients. This trend reflects the pathophysiological changes during disease progression. Cirrhosis is a significant risk factor for HCC, and HCC development is a cumulative process with increasing age, typically progressing from chronic hepatitis B to cirrhosis and then to HCC. In this study, although a significant age difference was observed between the CHB and HCC groups, the age difference between the LC and HCC groups was not significant. This result may suggest that GP73 reflects the degree of liver fibrosis or chronic inflammation more than a direct marker of HCC development. However, in the multivariate analysis model (C-GADA), the coefficient for GP73 was negative (OR < 1), suggesting that elevated GP73 levels may be negatively correlated with the probability of HCC diagnosis. This phenomenon may stem from the interaction effects among multiple variables, particularly the high expression of GP73 in cirrhosis patients, which may weaken its independent diagnostic value in distinguishing HCC from non-HCC patients. This result highlights the importance of considering the interactions between variables in multivariate models and suggests that the clinical application of GP73 requires comprehensive evaluation in conjunction with the patient's specific pathological stage and other biomarkers to more accurately reflect its diagnostic value for liver cancer. Furthermore, the partial functional overlap between GP73 and biomarkers such as AFP in liver cancer may reduce its independent discriminative power. The results of this study indicate that AFP, DCP, and GP73, three HCC-specific biomarkers and two risk factors, as independent risk factors for liver cancer diagnosis, produce a synergistic effect when combined. This further demonstrates that although GP73 performs poorly as a single variable, it has a synergistic effect in the combined model, optimizing overall diagnostic performance.

[0118] This invention comprehensively evaluated the application value of the G-GADA model in the diagnosis of liver cancer and early-stage liver cancer using ROC curve analysis. The results showed that the G-GADA model has significant advantages in diagnostic performance. Specifically, in liver cancer diagnosis, the AUC of G-GADA reached 0.941, significantly higher than that of single serum biomarkers DCP, AFP, and GP73, as well as the scoring models GALAD and C-GALAD (all P < 0.05). This indicates that G-GADA, by combining multiple indicators, not only improves diagnostic accuracy but also has higher clinical value in practical applications. Notably, in the diagnosis of early-stage liver cancer, the AUC of the G-GADA model reached 0.908, with a specificity as high as 91.87%, superior to other diagnostic models in both sensitivity and specificity. This result is particularly crucial because timely diagnosis of early-stage liver cancer is essential for improving patient prognosis, while the performance of single serum biomarkers such as DCP, AFP, and GP73 in the diagnosis of early-stage liver cancer is relatively limited. The G-GADA model, through multi-parameter integration, effectively compensates for the shortcomings of single biomarkers, further improving the accuracy of early-stage liver cancer screening. Furthermore, subgroup analyses further validated the stability and advantages of G-GADA. In the subgroups of chronic hepatitis B and hepatocellular carcinoma, and cirrhosis and hepatocellular carcinoma, the AUC of G-GADA was significantly superior to that of single biomarker and GALAD and C-GALAD scoring models. Moreover, the G-GADA model also demonstrated good performance in diagnosing AFP and DCP-negative patients. In patients with AFP levels below 20 ng / mL, the AUC of G-GADA reached 0.900, with a specificity of 93.01%. This result indicates that even with low AFP levels, G-GADA can still provide high diagnostic accuracy (Youden index 0.7301), suggesting its potential application in AFP-negative patients. Simultaneously, in patients with DCP ≤40, the AUC of G-GADA was also higher than other models, further validating its applicability in patients with low DCP levels. Overall, the G-GADA model demonstrated stable and superior diagnostic performance in subgroups with different disease states and biomarker levels, further demonstrating its clinical practicality and broad applicability.

[0119] The results of this invention show that the G-GADA model, by integrating multiple biomarkers such as AFP, DCP, and GP73, effectively distinguishes between patients with liver cancer and those without liver cancer. Furthermore, by optimizing the diagnostic contribution of GP73 through phased analysis, it provides more accurate and reliable support for early screening of liver cancer.

[0120] IV. Conclusion

[0121] The G-GADA model developed in this invention significantly improves the diagnostic efficacy of HBV-related hepatocellular carcinoma by introducing GP73, a novel serum biomarker, demonstrating excellent predictive performance, especially in early-stage hepatocellular carcinoma and patients with negative serum biomarkers. Compared with single biomarkers (AFP, DCP, GP73) and existing scoring models (GALAD, C-GALAD), G-GADA has certain advantages in comprehensive indicators such as AUC, sensitivity, and specificity. Although GP73 has a negative coefficient in the model, its synergistic effect with other biomarkers effectively improves diagnostic performance. The G-GADA model of this invention provides a reliable tool for non-invasive prediction and early screening of hepatocellular carcinoma and has significant clinical application value.

[0122] Although the invention has been described with reference to exemplary embodiments, it should be understood that the invention is not limited to the disclosed exemplary embodiments. Various adjustments or changes may be made to the exemplary embodiments described in this specification without departing from the scope or spirit of the invention. The scope of the claims should be interpreted in the broadest possible sense to cover all modifications and equivalent structures and functions.

Claims

1. A liver cancer assessment system for AFP or DCP-negative patient populations, characterized in that, include: The data acquisition unit is configured to acquire the subject's indicator data, which includes the subject's age, gender, and levels of AFP, DCP, and GP73. The data processing unit includes a liver cancer assessment model and is configured to process the indicator data acquired in the data acquisition unit using the liver cancer assessment model to obtain a prediction result; wherein, the liver cancer assessment model is: logit(P)=-10.837+0.102×age+2.754×gender-2.019×log(GP73)+0.922×log(AFP)+2.468×log(DCP), where in gender, male=1, female=0, and in logit(P), P represents the probability of liver cancer; AFP negative means serum AFP < 20 ng / mL, and DCP negative means serum DCP ≤ 40 mAU / mL.

2. A computer-readable storage medium, characterized in that, It stores a computer program containing a liver cancer assessment model for AFP or DCP negative patient populations: logit(P) = -10.837 + 0.102 × age + 2.754 × gender - 2.019 × log(GP73) + 0.922 × log(AFP) + 2.468 × log(DCP), where gender is male = 1 and female = 0, and in logit(P), P represents the probability of liver cancer; Furthermore, when executed by the processor, the following methods are implemented to assess liver cancer: The subject's indicator data are obtained from the data acquisition unit, including the subject's age, gender, and levels of AFP, DCP, and GP73. The liver cancer assessment model is used to process the indicator data acquired in the data acquisition unit to obtain prediction results; AFP negative means serum AFP < 20 ng / mL, and DCP negative means serum DCP ≤ 40 mAU / mL.

3. An electronic device, characterized in that, include: Processor and memory, wherein: The memory is used to store the executable instructions of the processor; The processor is configured to perform the following method by executing the executable instructions: The subject's indicator data is obtained from the data acquisition unit. The indicator data includes the subject's age, gender, and the levels of AFP, DCP, and GP73. The indicator data obtained from the data acquisition unit is processed using a liver cancer assessment model for AFP or DCP-negative patients to obtain prediction results. The model is: logit(P) = -10.837 + 0.102 × age + 2.754 × gender - 2.019 × log(GP73) + 0.922 × log(AFP) + 2.468 × log(DCP), where male = 1 and female = 0 in gender, and P in logit(P) represents the probability of liver cancer; AFP negative means serum AFP < 20 ng / mL, and DCP negative means serum DCP ≤ 40 mAU / mL.

Images