A liver cancer diagnosis product, device and application

Abstract

The application belongs to the field of medical diagnosis, and particularly relates to a liver cancer diagnosis product, device and application. The diagnosis product comprises reagents for detecting protein markers and metabolic markers, wherein the protein markers comprise alpha-fetoprotein, and the metabolic markers comprise nicotinamide and a combination of at least one selected from the group consisting of 2-hydroxy-3-methylbutyric acid, acetoacetic acid, tyrosine, phenylalanine, carbamoyl phosphate, glyoxylic acid, uridine, linoleic acid and 1-methyluric acid. The application establishes and verifies a high-performance liver cancer metabolomics diagnosis model, and reveals a nicotinamide-driven pro-tumor metabolic axis, thereby providing a transformation tool for early detection of liver cancer, and having important clinical significance and application prospect.

Description

Technical Field

[0001] This invention belongs to the field of medical diagnostics, specifically relating to a liver cancer diagnostic product, device, and application. Background Technology

[0002] Liver cancer is one of the most common malignant tumors. Although the widespread use of hepatitis B vaccines has gradually changed the pattern of liver cancer incidence, the 5-year survival rate for liver cancer patients is still only 15.1%. Early diagnosis is key to improving the long-term survival chances of liver cancer patients, but current clinical diagnostic techniques have significant limitations.

[0003] The pathological classification of liver cancer mainly includes hepatocellular carcinoma (HCC), intrahepatic cholangiocarcinoma (ICC), and mixed HCC and ICC. Among these, ICC, as the second most common primary liver tumor, accounts for 10% to 20% of newly diagnosed liver cancers. Its global incidence and related mortality have increased significantly in recent decades, with an extremely low 5-year survival rate. Patients with ICC have a lower long-term survival rate than those with HCC, which is closely related to the high propensity for regional and distant metastasis and the lack of effective systemic treatment options. However, most current liver cancer diagnostic research focuses on HCC, generally neglecting the existence of ICC, thus limiting the overall diagnostic efficacy of clinical liver cancer.

[0004] Currently, the diagnosis or risk assessment of liver cancer still has the following shortcomings: (1) Limitations of traditional biomarkers and screening technologies: Alpha-fetoprotein (AFP), as a widely used biomarker for liver cancer, suffers from low sensitivity and high false negative rate. When AFP is used alone (with a cutoff value of 400 U / L), the sensitivity is only 20%-60%. Even when used in combination with abdominal ultrasound, the sensitivity only increases to 60%-65%, while the specificity drops to 70%-95%, which is insufficient to meet the needs of early and accurate diagnosis.

[0005] (2) The scope of existing metabolomics research is limited: the vast majority of studies only focus on HCC and do not include ICC, which accounts for 10%-20% of liver cancer, in the research scope. As a result, the diagnostic model cannot cover a wider range of liver cancer patients and its applicability is limited.

[0006] (3) Deficiencies in the sample and validation of existing metabolomics studies: Most current studies have small sample sizes and are influenced by local diets and lifestyles, resulting in strong regionality in metabolite detection results. Studies are mostly limited to specific regions and do not cover representative areas nationwide, making it difficult to promote and apply them. At the same time, there is a lack of completely independent external validation. Although some studies have conducted validation, the independence of the validation cohort is insufficient, which limits the reliability of the results and their clinical translational value.

[0007] (4) Efficiency and cost issues of existing detection technologies: The following drawbacks exist when using LC-MS for metabolite detection: ① The detection efficiency is low, and a single sample analysis takes 30 minutes to several hours. It also requires time to process column equilibration and gradient elution, making it unsuitable for rapid screening of large-scale samples.

[0008] ② The sample requirement is large, requiring tens to hundreds of microliters of serum, which is not suitable for small sample volumes.

[0009] ③ The process is complex, and the sample needs to undergo pretreatment steps such as desalting, chromatographic separation or derivatization, which can easily introduce human error. In addition, the consumption of reagents (such as organic solvents and chromatographic columns) is large, resulting in high long-term costs.

[0010] ④ Chromatographic separation may result in the loss of some metabolites (such as volatile substances), making it difficult to achieve global analysis of complex metabolic networks such as bile acids and sphingolipids in liver cancer.

[0011] (5) Limited clinical translation potential: Existing technologies have shortcomings in terms of integration with clinical processes and point-of-care testing. For example, the compatibility of LC-MS testing processes with routine clinical testing needs to be improved, making it difficult to achieve applications such as real-time analysis and rapid intraoperative testing.

[0012] In evaluating the accuracy of liver cancer screening technologies, sensitivity, specificity, and receiver operating characteristic (ROC) curves are core indicators. Ideal biomarkers should possess characteristics such as sensitivity, specificity, ease of use, low cost, high reproducibility, rapid detection, correlation with tumor stage, and easy sample acquisition (e.g., blood, urine). However, currently used biomarkers still fall short of these requirements. Summary of the Invention

[0013] To address the technical problems existing in the prior art, this invention provides a diagnostic metabolic biomarker covering multiple types of liver cancer and its applications. This invention utilizes metabolomics technology to design a detection scheme with superior diagnostic efficacy, improving the sensitivity and specificity of liver cancer diagnosis and reducing the false negative rate. Simultaneously, it constructs a diagnostic model that can simultaneously cover HCC and ICC, expanding the applicable patient population. The large sample size of this invention covers representative regions nationwide, and a completely independent external validation cohort is established to enhance the reliability of the results and its promotional value. Furthermore, nanoparticle-enhanced laser desorption / ionization mass spectrometry (NPELDI-MS) detection technology is used to optimize the detection process, enhancing its clinical translation potential.

[0014] In a first aspect, the present invention provides a liver cancer diagnostic or assessment product comprising reagents for detecting protein biomarkers and metabolic biomarkers, wherein the protein biomarkers include alpha-fetoprotein, and the metabolic biomarkers include nicotinamide and a combination of at least one selected from the following: 2-hydroxy-3-methylbutyric acid, acetoacetic acid, tyrosine, phenylalanine, carbamoyl phosphate, glyoxylic acid, uridine, linoleic acid, alanine, and 1-methyluric acid.

[0015] In a second aspect, the present invention provides an apparatus for liver cancer diagnosis or risk assessment, comprising: The data acquisition unit is configured to acquire detection data on the amount of biomarkers in a subject sample, the biomarkers including protein biomarkers and metabolic biomarkers, wherein the protein biomarkers include alpha-fetoprotein, and the metabolic biomarkers include nicotinamide and a combination of at least one of the following: 2-hydroxy-3-methylbutyric acid, acetoacetic acid, tyrosine, phenylalanine, carbamoyl phosphate, glyoxylic acid, uridine, linoleic acid, alanine, and 1-methyluric acid; The data processing unit is loaded with a trained machine learning model for liver cancer diagnosis and is configured to use the detection data to perform liver cancer diagnosis or risk assessment.

[0016] Thirdly, the present invention provides a computer-readable storage medium storing a computer program, which is implemented when executed by a processor: The method involves obtaining detection data on the amount of biomarkers in subject samples, the biomarkers including protein biomarkers and metabolic biomarkers, wherein the protein biomarkers include alpha-fetoprotein, and the metabolic biomarkers include nicotinamide and a combination of at least one of the following: 2-hydroxy-3-methylbutyric acid, acetoacetic acid, tyrosine, phenylalanine, carbamoyl phosphate, glyoxylic acid, uridine, linoleic acid, alanine, and 1-methyluric acid. The detection data is input into a trained machine learning model for liver cancer diagnosis to achieve liver cancer diagnosis or risk assessment.

[0017] Fourthly, the present invention provides the use of a detection reagent in the preparation of products for early diagnosis or risk assessment of liver cancer, the detection reagent comprising reagents for detecting protein biomarkers and metabolic biomarkers, wherein the protein biomarkers include alpha-fetoprotein, and the metabolic biomarkers include nicotinamide and a combination of at least one selected from the following: 2-hydroxy-3-methylbutyric acid, acetoacetic acid, tyrosine, phenylalanine, carbamoyl phosphate, glyoxylic acid, uridine, linoleic acid, alanine, and 1-methyluric acid.

[0018] This invention offers the following advantages: In some embodiments, the invention achieves superior diagnostic accuracy through the combined use of nine metabolic markers (2-hydroxy-3-methylbutyric acid, tyrosine, carbamoyl phosphate, nicotinamide, glyoxylic acid, uridine, linoleic acid, alanine, and 1-methyluric acid) with alpha-fetoprotein, with area under the curve (AUC) values ​​of 0.92 and 0.93 in the discovery and validation cohorts, respectively. This invention demonstrates effectiveness in early detection and consistent performance across both HCC and ICC subtypes. Mendelian randomization analysis supports a potential causal relationship between circulating nicotinamide (NAM) (a key metabolite in this feature combination) and liver cancer risk. Functional studies indicate that NAM, through NAD-dependent... + The mechanism by which / SIRT1 promotes HIF1α expression, thereby activating glycolysis and the MAPK signaling pathway, and subsequently promoting LC proliferation and metastasis. (Regarding NAD...) + Pharmacological inhibition of the / SIRT1 / HIF1α axis can reverse these pro-cancer effects. In summary, this invention establishes and validates a high-performance hepatocellular carcinoma metabolomics diagnostic model, reveals the NAM-driven pro-tumor metabolic axis, provides a translational tool for the early detection of hepatocellular carcinoma, and offers in-depth insights into its metabolic mechanisms. Attached Figure Description

[0019] Figure 1 This diagram illustrates the differential metabolite profile, model development, and correlation analysis in hepatocellular carcinoma (HCC). (A) A volcano plot shows 340 metabolic features in HCC patients compared to non-HCC controls, with significantly upregulated signals (red), significantly downregulated signals (blue), and insignificant signals (grey). (B) Pathway enrichment analysis of 31 differentially expressed metabolites. (C) A Naive Bayes model for diagnosing HCC constructed using 31 differentially expressed metabolites. (D) Ranking of the 31 differentially expressed metabolites based on LASSO coefficients. (E) A violin plot of 11 differentially expressed metabolites with LASSO coefficients > 0.20 between HCC and non-HCC samples. (F) A Naive Bayes model constructed using 5-fold cross-validation of the 11 differentially expressed metabolites. (G) Spearman correlation matrix between demographic, clinical, and differentially expressed metabolite indicators. Only correlations with |r| > 0.3 and FDR < 0.05 are shown. * p <0.05,** p <0.01, *** p <0.001.

[0020] Figure 2The construction and validation of hepatocellular carcinoma (HCC) diagnostic models are shown. (A) Receiver operating characteristic (ROC) curves used to construct and validate the diagnostic performance of HCC: alpha-fetoprotein (AFP, green), a Naive Bayes model based on 9 metabolites (9M, orange), and a combined Naive Bayes model integrating 9 metabolites and AFP (9M+AFP, blue). (B) ROC curves for the construction and validation of the 9M-based Naive Bayes model for HCC diagnosis in an AFP-negative population. (C) ROC curves for the construction and validation of the 9M+AFP Naive Bayes model for HCC diagnosis in a hepatitis B surface antigen (HBsAg)-positive population. (D) ROC curves for the construction and validation of the 9M+AFP Naive Bayes model in diagnosing early-stage HCC (stages I and II). (E) ROC curves for the construction and validation of the 9M+AFP Naive Bayes model in diagnosing hepatocellular carcinoma (HCC, red) and intrahepatic cholangiocarcinoma (ICC, blue).

[0021] Figure 3 The study demonstrates that NAM promotes the proliferation, migration, and invasion of hepatocellular carcinoma cells. (A) Cell viability of HepG2 cells treated with a specified concentration of NAM for 72 hours. (B) Box plots show the differences in NAM levels among patients with stage I to IV hepatocellular carcinoma after adjusting for age, sex, HBsAg status, and location. Statistical analysis using analysis of covariance shows the overall adjusted values. p Values ​​(left). Box plots show the differences in NAM levels between hepatocellular carcinoma patients with and without distant metastases after adjusting for age, sex, HBsAg status, and location. Statistical analysis was performed using the Wilcoxon rank-sum test with covariate correction (right). (C) Representative images and quantifications of migration and invasion assays in HepG2 cells treated with specified concentrations of NAM. (D) Volcano plots showing genes significantly upregulated (red) and downregulated (blue) (fold change ≥ 2) in HepG2 cells treated with 200 μM NAM. p <0.05). (E) KEGG enrichment analysis of differentially expressed genes from (D), highlighting significant enrichment of HIF-1α, MAPK, and glycolysis pathways. (F) mRNA expression levels of SLC2A1 (encoding Glut1) and VEGFA (encoding VEGF) in HepG2 cells treated with increasing concentrations of NAM. (G) Western blot analysis of protein expression in HepG2 cells after treatment with different concentrations of NAM. (H) Intracellular lactate (LA) production in HepG2 cells after treatment with different concentrations of NAM. * p <0.05,** p <0.01, *** p <0.0001; ns, not significant.

[0022] Figure 4 The results show that NAM upregulates HIF1α expression via the NAD+ / SIRT1-dependent pathway. (A) Cell viability of HepG2 cells treated with 200 μM NAM, with or without co-treatment with 50 μM Daporinad (a NAMPT inhibitor). (B) Western blot analysis of protein expression in HepG2 cells after treatment with different concentrations of Daporinad or EX527 (a SIRT1 inhibitor). (C) Western blot analysis of protein expression in HepG2 cells treated with 200 μM NAM, with or without treatment with 50 μM Daporinad and 50 μM EX527. (D) Intracellular lactate (LA) production in HepG2 cells after treatment with 200 μM NAM, with or without co-treatment with 50 μM Daporinad and 50 μM EX527. (E) Representative images and quantifications of migration and invasion assays using HepG2 cells treated with 200 μM NAM, with or without co-treatment with 50 μM Daporinad and 50 μM EX527. (F) Representative images and quantifications of migration and invasion assays using HepG2 cells treated with 200 μM NAM, with or without co-treatment with 20 μM PX478 (an HIF1α inhibitor).

[0023] Figure 5 The study demonstrates the association between NAM and hepatitis B virus-associated hepatocellular carcinoma (HCC). (A) Four groups were included: a healthy control group (HC), patients with chronic hepatitis B (CHB), patients with non-HBV-associated HCC, and patients with HBV-associated HCC. The overall difference in serum NAM levels among the four groups was statistically significant (Kruskal-Wallis rank-sum test). p <0.001). Pairwise comparisons were performed using the Wilcoxon rank-sum test with Bonferroni correction. (B) Receiver operating characteristic (ROC) curves assessing the diagnostic performance of serum NAM in distinguishing HBV-hepatocellular carcinoma patients from CHB. Detailed Implementation

[0024] Various exemplary embodiments of the present invention will now be described in detail. This detailed description should not be considered as a limitation of the present invention, but rather as a more detailed description of certain aspects, features, and embodiments of the present invention.

[0025] 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.

[0026] 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.

[0027] Diagnostic products In a first aspect, the present invention provides a liver cancer diagnostic or risk assessment product comprising reagents for detecting protein biomarkers and metabolic biomarkers, wherein the protein biomarkers include alpha-fetoprotein, and the metabolic biomarkers include nicotinamide and a combination of at least one selected from the following (e.g., 2, 3, 4, 5, 6, 7, 8, 9, or 10): 2-hydroxy-3-methylbutyric acid, acetoacetic acid, tyrosine, phenylalanine, carbamoyl phosphate, glyoxylic acid, uridine, linoleic acid, alanine, and 1-methyluric acid.

[0028] Unless otherwise stated, the term "detection" as used herein refers to methods that include determining the presence and / or quantification of a marker in a sample. It should be noted that the term "quantity" as used herein should be interpreted broadly: it can refer to quantitative or semi-quantitative results, and it can refer to absolute or relative content.

[0029] In this invention, among the aforementioned differential metabolic markers, compared to the amount of each metabolite in the control (healthy population), nicotinamide was significantly upregulated in liver cancer patients, 2-hydroxy-3-methylbutyric acid was significantly downregulated in liver cancer patients, acetoacetic acid was significantly downregulated in liver cancer patients, tyrosine was significantly upregulated in liver cancer patients, phenylalanine was significantly downregulated in liver cancer patients, carbamoyl phosphate was significantly downregulated in liver cancer patients, glyoxylic acid was significantly upregulated in liver cancer patients, uridine was significantly downregulated in liver cancer patients, linoleic acid was significantly downregulated in liver cancer patients, alanine was significantly downregulated in liver cancer patients, and 1-methyluric acid was significantly downregulated in liver cancer patients.

[0030] In some embodiments, the reagent comprises a reagent for nanoparticle-enhanced laser desorption / ionization mass spectrometry (LAMS), which may use reagents known in the art for nanoparticle-enhanced LMS. In a preferred embodiment, the nanoparticles comprise iron(III) oxide (Fe3O4) nanoparticles.

[0031] In some implementations, the reagents used to detect protein biomarkers include a detection antibody or fragment thereof for alpha-fetoprotein. The sequence of the antibody is not particularly limited, and commercially available detection antibodies can be used.

[0032] In some embodiments, the metabolic markers include 2-hydroxy-3-methylbutyric acid, tyrosine, carbamoyl phosphate, nicotinamide, glyoxylic acid, uridine, linoleic acid, and alanine.

[0033] In some embodiments, the metabolic markers include acetoacetic acid, phenylalanine, carbamoyl phosphate, nicotinamide, glyoxylic acid, uridine, linoleic acid, alanine, and 1-methyluric acid. In some preferred embodiments, the metabolic markers include 2-hydroxy-3-methylbutyric acid, tyrosine, carbamoyl phosphate, nicotinamide, glyoxylic acid, uridine, linoleic acid, alanine, and 1-methyluric acid.

[0034] In some implementations, the test sample includes a fluid sample from the subject, examples of which include, but are not limited to, one or more of the following: whole blood, serum, plasma, blood components, follicular fluid, saliva, urine, milk (including colostrum and breast milk), ascites, amniotic fluid, tears, body fluids, or other fluids produced by the body. In a preferred implementation, the test sample described herein is serum.

[0035] In some implementations, the product includes, but is not limited to, chips, test strips, or reagent kits.

[0036] In a second aspect, the present invention provides an apparatus for liver cancer diagnosis or risk assessment, comprising: The data acquisition unit is configured to acquire detection data on the amount of biomarkers in a subject sample, the biomarkers including protein biomarkers and metabolic biomarkers, wherein the protein biomarkers include alpha-fetoprotein, and the metabolic biomarkers include nicotinamide and a combination of at least one of the following: 2-hydroxy-3-methylbutyric acid, acetoacetic acid, tyrosine, phenylalanine, carbamoyl phosphate, glyoxylic acid, uridine, linoleic acid, alanine, and 1-methyluric acid; The data processing unit is loaded with a trained machine learning model for liver cancer diagnosis and is configured to use the detection data to diagnose liver cancer.

[0037] In some implementations, the liver cancer diagnostic machine learning model is obtained by including the following steps: (1) Obtain detection data of the metabolic markers and the protein markers; (2) The LASSO regression scoring method was used to rank differential metabolites and screen out significantly differential metabolites with a score ≥0.20; (3) A machine learning model for liver cancer diagnosis was constructed using the Naive Bayes machine learning model.

[0038] Thirdly, the present invention provides a computer-readable storage medium storing a computer program, which is implemented when executed by a processor: The method involves obtaining detection data on the amount of biomarkers in subject samples, the biomarkers including protein biomarkers and metabolic biomarkers, wherein the protein biomarkers include alpha-fetoprotein, and the metabolic biomarkers include nicotinamide and a combination of at least one of the following: 2-hydroxy-3-methylbutyric acid, acetoacetic acid, tyrosine, phenylalanine, carbamoyl phosphate, glyoxylic acid, uridine, linoleic acid, alanine, and 1-methyluric acid. The detection data is input into a trained machine learning model for liver cancer diagnosis to achieve liver cancer diagnosis.

[0039] Fourthly, the present invention provides the use of a detection reagent in the preparation of products for early diagnosis or risk assessment of liver cancer, the detection reagent comprising reagents for detecting protein biomarkers and metabolic biomarkers, wherein the protein biomarkers include alpha-fetoprotein, and the metabolic biomarkers include nicotinamide and a combination of at least one selected from the following: 2-hydroxy-3-methylbutyric acid, acetoacetic acid, tyrosine, phenylalanine, carbamoyl phosphate, glyoxylic acid, uridine, linoleic acid, alanine, and 1-methyluric acid.

[0040] In any of the above embodiments of the present invention, the diagnostic product or device can be used for the early diagnosis of liver cancer, i.e., to determine whether a subject is a liver cancer patient or whether the subject has a high or low risk of developing liver cancer. In some embodiments, the subject includes AFP-negative individuals.

[0041] In any of the above embodiments of the present invention, the diagnostic product or device can be used for the early diagnosis of hepatocellular carcinoma (HCC), i.e., to determine whether a subject is a patient with hepatocellular carcinoma or whether the subject has a high or low risk of developing hepatocellular carcinoma.

[0042] In the above embodiments of the present invention, the diagnostic product or device can be used for the early diagnosis of intrahepatic cholangiocarcinoma (ICC), that is, to determine whether a subject is a patient with intrahepatic cholangiocarcinoma or whether the subject has a high or low risk of having intrahepatic cholangiocarcinoma.

[0043] The above technical solution will be further explained in detail below: Identification and Validation of Differential Metabolites Serum metabolite measurements were logarithmically transformed and normalized. In internal analysis, based on the normality test results, a two-tailed t-test or a two-tailed Wilcoxon signed-rank test was first used to calculate... p Values ​​were used for preliminary screening of differentially expressed metabolites. To avoid multicollinearity, the Benjamini-Hochberg (BH) method was used for further analysis. p Values ​​were adjusted using multiple comparisons to calculate the false discovery rate (FDR) (P. adjusted). Orthogonal partial least squares discriminant analysis (OPLS-DA) was further used to demonstrate differences between the case and control groups. Subsequently, Lasso regression was performed to calculate importance scores for all metabolic features, sorted in descending order. First, duplicate features and features corresponding to unknown metabolites were excluded, further filtered by removing features with Lasso regression scores below 0.20. Then, the number of selected metabolic features was optimized to improve the model's area under the receiver operating characteristic (AUC). To ensure robustness and mitigate overfitting, a Naive Bayes classifier was trained using this combination and evaluated using 5-fold cross-validation on the discovery set. The diagnostic performance of the model was evaluated based on sensitivity, specificity, and AUC on both the discovery set and the independent external validation set.

[0044] Construction and validation of diagnostic models (1) The Naive Bayes algorithm was implemented using the data mining toolkit Orange (version 3.36.1). For feature evaluation based on Lasso regression scores (LRScore), the “Rank” widget in Orange was used, which was connected to the Lasso regression model to obtain the LRScore for each metabolic feature. To ensure model robustness, hierarchical 5-fold cross-validation was used to train and optimize the Naive Bayes model on the discovery dataset.

[0045] (2) To further validate the model, it was subsequently tested on independent, external validation datasets. All other data analyses were performed using R software (v4.4.2). The diagnostic task was defined as binary classification, and the Naive Bayes model output a continuous probability score (ranging from 0 to 1) representing the likelihood of each patient having cancer. Propensity score was used to correct for potential confounding factors, including age, sex, and location. Furthermore, the diagnostic performance of the metabolite combination was specifically evaluated in subgroups of individuals who were HBsAg positive and AFP negative.

[0046] Association between biomarker combinations and clinical characteristics Pathway enrichment analysis was performed using the pathway analysis function in MetaboAnalyst 6.0. Nine clinical endpoints were included, such as AFP, albumin, alanine aminotransferase (ALT), aspartate aminotransferase (AST), cholinesterase (CHE), globulin, prothrombin time (PT), prothrombin activity (PTA), and total bilirubin (TBIL). Spearman correlation analysis was performed on the clinical endpoints and metabolites, with an FDR < 0.05 indicating a significant correlation.

[0047] Cell Culture and Reagents Human hepatocellular carcinoma cells (HepG2) were purchased from the American Type Culture Collection (ATCC) and cultured in minimum essential medium (MEM) supplemented with 10% fetal bovine serum (FBS) at 37°C under a humid atmosphere of 5% CO2. NAM (#S1899), Daporinad (#S2799), EX-527 (#S1541), and PX-478 dihydrochloride (#S7612) were all purchased from Selleck.

[0048] Cell viability assay HepG2 cells at 80% confluence were collected by digestion with 0.25% trypsin, centrifuged, and resuspended to a density of 1–3 × 10⁶ cells / year. 4 Cells / mL. The cell suspension was seeded into 96-well plates (100 μL per well) and allowed to adhere for 12–16 hours. Subsequently, cells were treated with a series of NAM concentrations. After 72 hours of incubation, 10 μL of CCK-8 solution was added to each well to determine cell viability. After further incubation for 2–4 hours, absorbance was measured at 450 nm using a microplate reader (Multiskan FC, Thermo Fisher Scientific, USA).

[0049] Cell migration and invasion assay Cell migration was assessed using Transwell chambers with an 8 μm pore size. Serum-starved HepG2 cells were resuspended in serum-free medium and seeded into the upper chamber. The lower chamber was filled with complete medium supplemented with 10% FBS as a chemical inducer. After incubation at 37°C for 48 hours, unmigrated cells on the upper surface of the membrane were carefully removed with a cotton swab. Cells that migrated to the lower surface were fixed with 4% paraformaldehyde for 15 minutes, stained with 0.1% crystal violet for 20 minutes, rinsed with PBS, and air-dried. The number of migrating cells was quantified by counting in three randomly selected microscopic fields from each chamber. For invasion assays, the same procedure was followed, except that the Transwell chambers were pre-coated with Matrigel to simulate the extracellular matrix barrier.

[0050] NAD+ and NADH detection and determination Using a commercially available NAD+ / NADH assay kit (Beyotime), HepG2 cells were measured after NAM treatment (1 × 10⁻⁶ cells per sample) according to the manufacturer's protocol. 6 NAD+ and NADH levels within individual cells. Briefly, cells were lysed using 200 µL of ice-cold lysis buffer. For total NAD+ / NADH measurement, 20 µL of lysis buffer was directly transferred to a 96-well plate. For NADH-specific measurement, one portion of the lysis buffer was heated at 60°C for 30 min to decompose NAD+ before spotting. Then, 90 µL of alcohol dehydrogenase working solution was added to each well, and the plate was incubated at 37°C for 10 min, followed by 10 µL of chromogenic solution and incubation at 37°C for another 30 min. Absorbance was measured at 450 nm using a microplate reader. NAD+ concentration was calculated by subtracting the NADH value from the total NAD+ / NADH value. A standard curve was run simultaneously for quantification.

[0051] Immunoprecipitation (IP) and immunoblotting (IB) HepG2 cells were collected and lysed on ice for 30 min in RIPA buffer (50 mM Tris-HCl pH 7.4, 1% NP40, 0.5% Na-deoxycholate, 0.1% SDS, 150 mM NaCl, 2 mM EDTA, and 50 mM NaF) supplemented with a protease inhibitor mixture (Sigma) and benzyl sulfonyl fluoride (Sigma). Cell lysates were clarified by centrifugation at 12,000 rpm for 20 min at 4°C, and protein concentration was determined by BCA assay (Beyotime). In the subsequent IB step, an equal volume of protein (20–40 μg) was used for each sample. For IP, the primary antibody was incubated with lysis buffer overnight at 4°C. Protein G agarose was then added, and incubation continued for 2 h at 4°C. The agarose beads were then washed three times with lysis buffer to elute the immunoprecipitate, which was then analyzed by IB. The antibodies used were: anti-HIF1α (#20690-1-AP, Proteintech; 1:2000 dilution), anti-Glut1 (#21829-1-AP, Proteintech; 1:1000 dilution), anti-VEGF (#66828-1-Ig, Proteintech; 1:1000 dilution), anti-phospho-ERK1 / 2 (#4370, CST; 1:1000 dilution), and anti-ER... K1 / 2 (#4695, CST; 1:1000 dilution), anti-Acetyl-p53K382 (#ab75754, abcam; 1:500 dilution), anti-pan-acetyl (#66289-1-Ig, Proteintech; 1:500 dilution), anti-VHL (#24756-1-AP, 1:1000 dilution), and anti-β-actin (#66009-1-Ig, 1:10000 dilution).

[0052] Lactic acid detection and determination The concentration of L-lactic acid (LA) in HepG2 cell culture supernatant was quantified using a commercially available L-lactic acid colorimetric assay kit (Elabscience, E-BC-K044-M). In short, the cell supernatant was collected and used directly without dilution. For the assay, 5 µL of supernatant was added to each well of a microplate, followed by 100 µL of freshly prepared enzyme working solution (buffer and enzyme stock solution mixed at a 10:1 ratio) and 20 µL of chromogenic reagent. The reaction mixture was incubated at 37°C for 10 minutes. The reaction was then terminated by adding 180 µL of stop solution. After mixing for 5 seconds, the absorbance was measured at 530 nm using a microplate reader. The LA concentration (mmol / L) in the sample was calculated based on a standard curve generated using L-lactic acid standards of known concentrations.

[0053] Data Analysis The inclusion of metabolic biomarkers is expected to improve predictive performance, achieving an AUC of 0.80. At a significance level of α = 0.05 and a tolerance for error (d) of 0.10, at least 62 hepatocellular carcinoma cases and 62 controls are required to ensure statistical power. Therefore, the sample sizes of the training set and independent external validation set in this invention meet the necessary methodological requirements. All analyses were performed using R v4.4.2, two-sided. P <0.05 is considered significant.

[0054] Example 1 1. Research Design and Participant Recruitment The study was conducted in two phases: a discovery phase and an external validation phase. Liver cancer was diagnosed through histopathological examination, and exclusion criteria included any history of cancer, radiation therapy, or chemotherapy. Control subjects were defined as individuals whose non-liver cancer diagnosis was confirmed by abdominal ultrasound during their medical visit.

[0055] Inclusion and exclusion criteria for the study population: The case group consisted of liver cancer patients aged 35-70 years who had not been diagnosed with other cancers within the past 5 years, and who had not undergone surgery, radiotherapy, chemotherapy, or drug treatment at the time of blood collection, and who volunteered to participate in this study. The control group consisted of patients with cirrhosis, hepatitis B, and healthy individuals who had not been diagnosed with cancer within the past 5 years, and were matched for sample collection location and time, gender, age difference within 5 years, and batch effect. Patients with the following conditions were excluded: comorbidities, including hematological disorders, dysfunction of vital organs such as the lungs and liver; and hemolysis.

[0056] The final cohort consisted of 2149 eligible participants. The findings cohort comprised 1924 subjects from 11 clinical centers across different regions, including 1013 hepatocellular carcinoma (HCC) patients and 911 non-HCC controls. The HCC patients included 833 HCC patients, 107 ICC patients, and 5 patients with a mixture of HCC and ICC. For external validation, a completely independent dataset was compiled from two centers, including 100 treatment-naïve HCC patients and 125 non-HCC controls.

[0057] 2. Serum sample collection (1) Draw blood from a peripheral venous vein into a 5 mL anticoagulant-free vacuum blood collection tube. Ensure a sufficient volume (5 mL) is collected. If using a serum tube containing a coagulant, ensure that the blood is in full contact with the coagulant on the tube wall to accelerate coagulation. All patients must have their blood drawn on an empty stomach to avoid interference from lipids, sugars, and other components in their diet on serum metabolites and biochemical indicators.

[0058] (2) After collection, gently invert the blood collection tube 3-5 times (only for serum tubes containing anticoagulants, the purpose is to make the blood and the anticoagulant on the tube wall mix thoroughly; if it is a regular serum tube without anticoagulants, the inversion step can be omitted). Avoid vigorous shaking throughout the process to prevent blood cell rupture and hemolysis, which will affect the subsequent test results.

[0059] (3) If coagulation and centrifugation cannot be performed immediately, the blood sample should be temporarily stored at room temperature (20-25℃) (do not refrigerate at 4℃, as low temperature will inhibit the activity of coagulation factors and prolong the coagulation time). It is essential to ensure that the coagulation operation begins within 2 hours after blood collection to avoid prolonged storage of blood and degradation of components.

[0060] (4) Place the blood collection tube vertically at room temperature and allow it to stand. Adjust the standing time according to the tube type: 30-60 minutes for ordinary serum tubes without anticoagulants, and 15-30 minutes for serum tubes containing coagulants, until the blood is completely coagulated (judgment criteria: a complete gel-like clot forms on the surface of the blood, there is no flowing liquid blood at the bottom of the tube, and there is no obvious gap between the clot and the tube wall). After coagulation, place the blood collection tube in a centrifuge and centrifuge at 3000-3500 rpm (approximately 1000-1500×g) at 4℃ for 10-15 minutes. After centrifugation, carefully extract the upper light yellow transparent liquid (serum) and transfer it to a 2 mL cell cryopreservation tube. During the operation, avoid touching the middle layer of white fibrin clots or the lower layer of dark red blood cells with the tip of the pipette. If impurities are accidentally aspirated, centrifugation must be repeated.

[0061] (5) Clearly mark the sample information (such as patient name, hospital ID, blood collection date, time, and sample type "serum") on the EP tube with a marker to ensure the uniqueness of the marking and that it strictly corresponds to the patient's hospital ID to avoid sample confusion.

[0062] (6) After labeling, if no further processing is to be carried out, the cryovials should be placed on ice (for short-term storage, not exceeding 1 hour) or immediately placed in a -20°C freezer for temporary storage (not exceeding 24 hours). It is strictly forbidden to leave them at room temperature for a long time (more than 30 minutes) to prevent changes in serum enzyme activity, metabolite degradation or microbial contamination.

[0063] (7) Place the labeled cryovials into a 10×10 cryovial box in sequence. Attach labels to the side and top of the cryovial box, indicating the sample type "serum," the date of receipt, and the sample group (e.g., "normal human serum," "hepatitis patient serum," "liver cancer patient serum"). Serum samples from normal individuals, hepatitis patients, cirrhosis patients, and liver cancer patients should be placed separately in different cryovial boxes to avoid cross-contamination and facilitate subsequent classification and retrieval.

[0064] (8) After the cryopreservation box is full or a single sample is temporarily stored for no more than 24 hours, transfer the cryopreservation box to a -80°C freezer for long-term storage, awaiting subsequent transport or testing. Before the cryopreservation box is full and placed in the -80°C freezer, the cryopreservation box and the EP tube inside must be kept on ice at all times to ensure that the serum sample does not leave the low temperature environment and to avoid repeated freeze-thaw cycles (repeated freeze-thaw cycles will cause serum protein denaturation and metabolite decomposition, affecting the accuracy of the test).

[0065] 3. Serum AFP quantification Serum AFP concentrations were analyzed in batches after sample collection to minimize batch-to-batch variability. Serum AFP concentrations were measured using a standardized chemiluminescent immunoassay. Analysis was performed on a Gi2000 automated immunoassay system using a specific kit (Beijing Haomai Biotechnology Co., Ltd., catalog number: AFP100L). All procedures were strictly performed according to the manufacturer's protocols. A cutoff value of 20.0 ng / mL was used to define a positive result.

[0066] 4. Detection and screening of metabolic biomarkers (1) Collect blood samples from liver cancer patients and non-liver cancer controls (cirrhosis, chronic hepatitis B, and healthy individuals), separate serum, and perform analysis using NPELDI-MS technology. The specific implementation method is as follows: First, Fe3O4 nanoparticles for enhancing mass spectrometry signals were synthesized via a hydrothermal method. The steps were as follows: 0.6 g of trisodium citrate dihydrate and 2.4 g of ferric chloride hexahydrate were dissolved in 80 mL of ethylene glycol and sonicated for 100 minutes to obtain a homogeneous orange solution. Then, 3.84 g of anhydrous sodium acetate was added to the mixture, followed by sonication for another 45 minutes. The dissolved solution was transferred to a 200 mL stainless steel hydrothermal reactor lined with polytetrafluoroethylene (PTFE) and hydrothermally reacted at 200°C in a drying oven for 10 hours. Afterward, the system was allowed to cool naturally to room temperature for approximately 10 hours. Finally, the precipitate was collected, washed twice each with deionized water and anhydrous ethanol, and dried at 60°C for 24 hours to obtain Fe3O4 nanoparticles.

[0067] For NPELDI-MS analysis, 1 μL of serum sample diluted 10-fold was spotted onto the target plate and air-dried at room temperature. Subsequently, 1 μL of nanoparticle solution (dissolved in deionized water at a concentration of 1 mg / mL) was applied to the pre-spotted sample and air-dried again at room temperature. NPELDI-MS analysis was performed in positive ion reflectance mode with the following optimized parameters: laser intensity 35% of maximum power, accelerating voltage 20 kV, and pulse repetition frequency 200 Hz. All acquired mass spectra were exported in raw data format for subsequent processing.

[0068] Mass spectrometry data were analyzed using PyCharm Community (version 2022.3). Spectral preprocessing included baseline correction and Savitzky-Golay smoothing to reduce noise interference. Only peaks with a signal-to-noise ratio exceeding 3.0 and appearing in more than two-thirds of the replicate samples were retained as metabolic characteristic peaks.

[0069] The structures of metabolites were validated by Fourier transform ion cyclotron resonance mass spectrometry, followed by biological identification by querying human metabolomics databases. The Hiplot online platform was used for t-distributed random neighborhood embedding, principal component analysis, and uniform manifold approximation and projection to visualize the differences in metabolomics profiles between and within the hepatocellular carcinoma (HCC) and non-HCC groups.

[0070] (2) The serum metabolite measurements were logarithmically transformed and standardized. In the internal data analysis, based on the normality test results, a two-tailed t-test or a two-tailed Wilcoxon rank-sum test was used to calculate the p-value, and differentially differentiated metabolites were initially screened to obtain a group of statistically significant differentially differentiated metabolites. To avoid false positives from multiple comparisons, the Benjamini-Hochberg method was used. p The values ​​are corrected, and the error detection rate is calculated.

[0071] Preliminary screening identified 37 potentially significant features (FDR < 0.05) from 340 detected metabolic features. Figure 1 A) corresponds to 31 annotated metabolites—15 of which are upregulated and 16 downregulated in patients with liver cancer. Pathway enrichment analysis showed that these metabolites are involved in pathways such as phenylalanine / tyrosine metabolism, tryptophan metabolism, arginine biosynthesis, linoleic acid metabolism, and taurine / linotenic acid metabolism. Figure 1 B). Using a Naive Bayes model with all 31 metabolites, the AUC reached 0.83 when distinguishing between individuals with liver cancer and those without. Figure 1 C).

[0072] The differentially expressed metabolites were further refined using the LASSO regression scoring method, and metabolites with scores ≥ 0.20 were selected. The combination of 11 metabolites (11-M) consisted of nicotinamide, 2-hydroxy-3-methylbutyric acid, acetoacetic acid, tyrosine, phenylalanine, carbamoyl phosphate, glyoxylic acid, uridine, linoleic acid, alanine, and 1-methyluric acid. Figure 1 (D and E). To prevent overfitting, 5-fold cross-validation was used to evaluate the model in the discovery cohort. After adjusting for age, sex, and location, the reduced 11-M AUC reached 0.85 ( Figure 1 F) confirms its robust diagnostic performance.

[0073] Analysis of 11 metabolites and clinical parameters revealed significant associations between key metabolites and liver function indicators. Phenylalanine and tyrosine levels were positively correlated with PT and AST, and negatively correlated with PTA, CHE, and albumin. This pattern suggests impaired hepatic synthetic function and hepatocellular damage. Figure 1 G). These aromatic amino acids showed a significant positive correlation with AFP levels, further suggesting a potential link to tumor invasiveness. Serum albumin (a reliable surrogate indicator of functional liver quality and synthetic reserves) showed a significant negative correlation with several other combined metabolites, including acetoacetic acid, glyoxylic acid, NAM, and uridine. Figure 1 G). This model emphasizes that the dysregulatory metabolic features captured by the selected combinations are closely related to a background of progressive liver failure (a hallmark of advanced hepatocellular carcinoma).

[0074] 5. Construction and validation of liver cancer diagnostic models (1) The selected metabolite combinations were optimized. Given the high correlation between phenylalanine and tyrosine (r = 0.96) and between 2-hydroxy-3-methylbutyric acid and acetoacetic acid (r = 0.91), a simplified combination of 9 metabolite combinations (9-M) was obtained by removing phenylalanine and acetoacetic acid from each pair.

[0075] (2) Based on the nine metabolic biomarkers (2-hydroxy-3-methylbutyric acid, tyrosine, carbamoyl phosphate, nicotinamide, glyoxylic acid, uridine, linoleic acid, alanine, and 1-methyluric acid) screened in step (1), a liver cancer diagnostic model was constructed using a Naive Bayes machine learning model. The combined model maintained diagnostic performance comparable to the original 11-M, with AUC values ​​of 0.85 and 0.84 in the discovery cohort and the independent external validation cohort, respectively. Figure 2 A). Crucially, the 9-M combination demonstrated strong discriminatory power in clinically challenging subgroups of AFP-negative individuals in both cohorts. Figure 2 B) addresses a key limitation of current screening methods.

[0076] (3) Based on the nine metabolic biomarkers screened in step (1), and combined with the traditional biomarker AFP, a liver cancer diagnostic model (9-M+AFP joint model) was constructed using a Naive Bayes machine learning model. The joint model significantly improved diagnostic accuracy. It achieved excellent AUC values ​​of 0.92 and 0.93 in the discovery set and external validation, respectively. Figure 2A) Compared to using AFP alone, this model represents a significant improvement. Notably, the combined model dramatically increased sensitivity from 0.39 in the detection cohort to 0.83 and from 0.42 in the validation cohort to 0.83, while maintaining relatively high specificity. This model also effectively distinguishes between liver cancer and chronic hepatitis B (CHB) in high-risk populations. Figure 2 C) demonstrates excellent clinical applicability, addressing a common diagnostic challenge in practice. Furthermore, it maintains high performance in detecting early-stage (Stage I and Stage II) disease. Figure 2 D), the sensitivity of the detection cohort was 0.84, and that of the validation cohort was 0.78. Finally, the model showed consistently high accuracy in both HCC and ICC, the two major histological subtypes. Figure 2 E) supports its use as a comprehensive "liquid biopsy" tool for primary liver cancer screening.

[0077] 6. Nicotinamide (NAM) promotes the proliferation, migration, and invasion of liver cancer cells. Although NAM acts as a protective agent in skin cancer development, its serum levels were significantly elevated in a liver cancer cohort, suggesting a potential pathogenic role in this context. In vitro functional assays confirmed that NAM promotes liver cancer cell proliferation in a concentration-dependent manner. Figure 3 A). Mendelian randomization analysis further supports this pro-tumor association, suggesting a potential causal relationship between genetically predicted circulating NAM levels and increased risk of liver cancer.

[0078] Serum NAM levels gradually increase as the tumor stage progresses. Figure 3 B). Most notably, serum NAM concentrations were significantly higher in patients with distant metastases than in patients without metastases. Figure 3 B) indicates that NAM is involved in the metastatic process. Transwell assays functionally validated this, showing that NAM treatment significantly enhanced the migration and invasion of liver cancer cells. Figure 3 C).

[0079] RNA sequencing was performed on NAM-treated liver cancer cells. The results showed that 1952 genes were upregulated and 1042 genes were downregulated. Figure 3 D). Pathway enrichment analysis of these 2994 differentially expressed genes showed significant activation of the HIF1α signaling pathway, MAPK pathway, and glycolysis pathway, all of which are recognized drivers of cancer progression. Figure 3 E). Subsequent molecular validation confirmed that NAM upregulates HIF1α protein expression, subsequently activating its downstream MAPK and glycolysis pathways (E). Figure 3 FH provides a mechanistic basis for its pro-tumor effect in liver cancer.

[0080] 7. NAM upregulates HIF1α expression via the NAD+ / SIRT1-dependent pathway. NAM is a key precursor for intracellular NAD+ biosynthesis, and its pro-proliferative effect has been observed to be completely eliminated by the NAMPT inhibitor Daporinad. Figure 4 A). Besides providing fuel for proliferation, elevated NAD+ levels are also a necessary cofactor for activating the deacetylase sirtuin. Pharmacological inhibition of NAD+ synthesis or SIRT1 activity (using EX-527) effectively suppresses HIF1α protein levels and the expression of its downstream targets. Figure 4 B). In these experiments, the activity of SIRT1 was verified by the acetylation level of its classical substrate p53 at Lys382 site and the overall histone acetylation level. Figure 3 G and Figure 4 B). NAM-induced HIF1α upregulation and downstream pathway activation can be reversed by Daporinad and EX-527 ( Figure 4 C and D). Furthermore, it was demonstrated that the NAM-driven pro-migration and pro-invasion phenotypes can be reversed by inhibiting the NAD+ / SIRT1 / HIF1α axis (C and D). Figure 4 E and F). Immunoprecipitation experiments confirmed that NAM treatment reduced the acetylation level of HIF1α and simultaneously weakened its interaction with VHL.

[0081] 8. NAM is associated with hepatitis B virus (HBV)-related liver cancer. Compared with HC, serum NAM levels were significantly elevated in the CHB group. Notably, NAM levels were also significantly higher in the HBV-hepatocellular carcinoma group than in the non-HBV-hepatocellular carcinoma group. Figure 5 A). Importantly, NAM concentrations were significantly higher in HBV-hepatocellular carcinoma patients than in CHB patients, while no significant difference was observed between non-HBV-hepatocellular carcinoma patients and HC patients. Figure 5 A). This pattern suggests that NAM accumulation is specifically associated with HBV-related malignant progression, rather than being a universal feature across all hepatocellular carcinoma subtypes. Furthermore, serum NAM levels alone can effectively differentiate between HBV-related hepatocellular carcinoma and CHB, with an AUC of 0.83 (…). Figure 5 B). In summary, these data indicate that elevated NAM levels are closely associated with HBV infection and further increase during the progression from chronic hepatitis to liver cancer.

[0082] Example 2 In this embodiment, after removing 1-methyluric acid, the last item on the LASSO regression score, the remaining eight metabolites (2-hydroxy-3-methylbutyric acid, tyrosine, carbamoyl phosphate, nicotinamide, glyoxylic acid, uridine, linoleic acid, and alanine) achieved an AUC of 0.85 on the internal validation set and 0.82 on the external validation set when used alone. When combined with AFP, the diagnostic model achieved an AUC of 0.92 on the internal validation set and 0.93 on the external validation set. In AFP-negative individuals, the diagnostic model using only the metabolites achieved an AUC of 0.87 on the internal validation set and 0.82 on the external validation set. In early-stage liver cancer, the diagnostic model combined with AFP achieved an AUC of 0.92 on the internal validation set and 0.93 on the external validation set. In hepatitis B surface antigen-positive individuals, the diagnostic model combined with AFP achieved an AUC of 0.92 on the internal validation set and 0.88 on the external validation set.

[0083] This embodiment further validates the classification of liver cancer. In HCC diagnosis, the internal set AUC is 0.94 and the external set AUC is 0.92; in ICC diagnosis, the internal set AUC is 0.85 and the external set AUC is 0.94.

[0084] Example 3 Based on the differential metabolite correlation analysis in Example 1, phenylalanine and tyrosine (r=0.96) and 2-hydroxy-3-methylbutyric acid and acetoacetic acid (r=0.91) showed strong correlations. Therefore, in this example, tyrosine was replaced with phenylalanine and 2-hydroxy-3-methylbutyric acid was replaced with acetoacetic acid, thereby constructing a machine learning model with nine new metabolites (acetoacetic acid, phenylalanine, carbamoyl phosphate, nicotinamide, glyoxylic acid, uridine, linoleic acid, alanine, and 1-methyluric acid).

[0085] The results showed that the AUC of the diagnostic model constructed using metabolites alone reached 0.84 in the internal validation set and 0.83 in the external validation set. With the addition of AFP, the AUC of the diagnostic model reached 0.92 in the internal validation set and 0.94 in the external validation set. In the AFP-negative population, the AUC of the diagnostic model constructed using metabolites alone reached 0.86 in the internal validation set and 0.83 in the external validation set. In early-stage liver cancer, the AUC of the diagnostic model combined with AFP reached 0.92 in the internal validation set and 0.93 in the external validation set. In the hepatitis B surface antigen-positive population, the AUC of the diagnostic model combined with AFP reached 0.92 in the internal validation set and 0.88 in the external validation set.

[0086] This embodiment further validates the classification of liver cancer. In HCC diagnosis, the internal set AUC is 0.94 and the external set AUC is 0.93; in ICC diagnosis, the internal set AUC is 0.85 and the external set AUC is 0.95.

[0087] Example 4 This embodiment illustrates a device for liver cancer diagnosis or risk assessment, which includes at least: The data acquisition unit is configured to acquire detection data on the amount of biomarkers in a subject sample, the biomarkers including protein biomarkers and metabolic biomarkers, wherein the protein biomarkers include alpha-fetoprotein, and the metabolic biomarkers include nicotinamide and a combination of at least one of the following: 2-hydroxy-3-methylbutyric acid, acetoacetic acid, tyrosine, phenylalanine, carbamoyl phosphate, glyoxylic acid, uridine, linoleic acid, alanine, and 1-methyluric acid; The data processing unit is loaded with a trained machine learning model for liver cancer diagnosis and is configured to use the detection data to diagnose liver cancer.

[0088] In this embodiment, the machine learning model for liver cancer diagnosis is obtained by the following steps: (1) Obtain detection data of differential metabolic markers and protein markers; (2) Differential metabolites were ranked using the LASSO regression scoring method, and significantly different metabolites with scores ≥ 0.20 were screened out; (3) A machine learning model for liver cancer diagnosis was constructed using the Naive Bayes machine learning model.

[0089] In some implementations, the data processing unit further includes a preset threshold for comparing the detected value with a corresponding threshold. The specific value of the preset threshold is not particularly limited; it can be the amount of each of the aforementioned biomarkers in a healthy population. Normal biomarker levels in a population can be obtained by testing a sufficient number of individuals, or can be found in existing clinical literature.

[0090] Those skilled in the art will understand that the various exemplary embodiments described in this invention can be implemented by software or by combining software with necessary hardware. Therefore, specific embodiments of the invention can be embodied in the form of a software product, which can be stored on a non-volatile storage medium or a non-transitory computer-readable storage medium (such as a CD-ROM, USB flash drive, portable hard drive, etc.) or on a network, including several instructions to cause a computing device (such as a personal computer, server, mobile terminal, or network device, etc.) to execute the method according to the invention.

[0091] In exemplary embodiments, the program product of the present invention can employ any combination of one or more readable media. A readable medium can be a readable signal medium or a readable storage medium. A readable storage medium can be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination thereof. More specific examples of readable storage media include, but are not limited to: electrical connections having one or more wires, portable disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fibers, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination thereof.

[0092] In this embodiment, the computer-readable storage medium stores a computer program that is implemented when executed by a processor: The method involves obtaining detection data on the amount of biomarkers in subject samples, the biomarkers including protein biomarkers and metabolic biomarkers, wherein the protein biomarkers include alpha-fetoprotein, and the metabolic biomarkers include nicotinamide and a combination of at least one of the following: 2-hydroxy-3-methylbutyric acid, acetoacetic acid, tyrosine, phenylalanine, carbamoyl phosphate, glyoxylic acid, uridine, linoleic acid, alanine, and 1-methyluric acid. The detection data is input into a trained machine learning model for liver cancer diagnosis to achieve liver cancer diagnosis.

[0093] Accordingly, based on the same inventive concept, the present invention also provides an electronic device.

[0094] In an exemplary embodiment, the electronic device is manifested as a general-purpose computing device. Components of the electronic device may include, but are not limited to: at least one processor, at least one memory, and a bus connecting different system components (including the memory and the processor). The memory stores program code that can be executed by the processing unit, causing the processing unit to perform at least some steps of the method described in this invention, wherein the processor includes at least the data processing unit described in this invention. The memory may include a readable medium in the form of volatile storage units, such as random access memory (RAM) and / or cache memory units, and may further include read-only memory units (ROM).

[0095] The memory of the present invention may also include a program / utility having a set (at least one) of program modules, including but not limited to: an operating system, one or more application programs, other program modules, and program data, each or some combination of these examples may include an implementation of a network environment.

[0096] A bus can represent one or more of several types of bus structures, including a memory bus or memory controller, a peripheral bus, a graphics acceleration port, a processing unit, or a local bus that uses any of the various bus structures.

[0097] Electronic devices can also communicate with one or more external devices (such as keyboards, pointing devices, Bluetooth devices, etc.), and with one or more devices that enable users to interact with the electronic device, and / or with any device that enables the electronic device to communicate with one or more other computing devices (such as routers, modems, etc.).

[0098] This communication can be achieved through input / output (I / O) interfaces. Furthermore, the electronic device can communicate with one or more networks (e.g., local area networks (LANs), wide area networks (WANs), and / or public networks, such as the Internet) via a network adapter. The network adapter communicates with other modules of the electronic device via a bus. It should be understood that, although not shown herein, other hardware and / or software modules can be used in conjunction with the electronic device, including but not limited to: microcode, device drivers, redundant processing units, external disk drive arrays, RAID systems, tape drives, and data backup storage systems.

[0099] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. These modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A liver cancer diagnostic product, characterized in that, The invention includes reagents for detecting protein biomarkers and metabolic biomarkers, wherein the protein biomarkers include alpha-fetoprotein, and the metabolic biomarkers include nicotinamide and a combination of at least one of the following: 2-hydroxy-3-methylbutyric acid, acetoacetic acid, tyrosine, phenylalanine, carbamoyl phosphate, glyoxylic acid, uridine, linoleic acid, alanine, and 1-methyluric acid.

2. The liver cancer diagnostic product according to claim 1, characterized in that, The reagents include those for nanoparticle-enhanced laser desorption / ionization mass spectrometry detection.

3. The liver cancer diagnostic product according to claim 1, characterized in that, The reagents include antibodies or fragments thereof used for the detection of alpha-fetoprotein.

4. The liver cancer diagnostic product according to claim 1, characterized in that, The metabolic markers include 2-hydroxy-3-methylbutyric acid, tyrosine, carbamoyl phosphate, nicotinamide, glyoxylic acid, uridine, linoleic acid, and alanine. Preferably, the metabolic markers include acetoacetic acid, phenylalanine, carbamoyl phosphate, nicotinamide, glyoxylic acid, uridine, linoleic acid, alanine, and 1-methyluric acid; Preferably, the metabolic markers include 2-hydroxy-3-methylbutyric acid, tyrosine, carbamoyl phosphate, nicotinamide, glyoxylic acid, uridine, linoleic acid, alanine, and 1-methyluric acid.

5. The liver cancer diagnostic product according to claim 1, characterized in that, The samples to be tested include fluid samples from the subjects.

6. The liver cancer diagnostic product according to claim 1, characterized in that, The products include chips, test strips, or reagent kits.

7. A device for liver cancer diagnosis or risk assessment, characterized in that, include: A data acquisition unit is configured to acquire detection data on the amount of biomarkers in a subject sample, the biomarkers including protein biomarkers and metabolic biomarkers, wherein the protein biomarkers include alpha-fetoprotein, and the metabolic biomarkers include nicotinamide and a combination of at least one of the following: 2-hydroxy-3-methylbutyric acid, acetoacetic acid, tyrosine, phenylalanine, carbamoyl phosphate, glyoxylic acid, uridine, linoleic acid, alanine, and 1-methyluric acid; The data processing unit is loaded with a trained machine learning model for liver cancer diagnosis and is configured to use the detection data to perform liver cancer diagnosis or risk assessment.

8. The device for liver cancer diagnosis or risk assessment according to claim 7, characterized in that, The machine learning model for liver cancer diagnosis is obtained through the following steps: (1) Obtain detection data of the metabolic markers and the protein markers; (2) Differential metabolites were ranked using the LASSO regression scoring method, and significantly different metabolites with scores ≥ 0.20 were screened out; (3) A machine learning model for liver cancer diagnosis was constructed using the Naive Bayes machine learning model.

9. A computer-readable storage medium, characterized in that, It stores a computer program, which is implemented when executed by a processor: The method involves obtaining detection data on the amount of biomarkers in subject samples, the biomarkers including protein biomarkers and metabolic biomarkers, wherein the protein biomarkers include alpha-fetoprotein, and the metabolic biomarkers include nicotinamide and a combination of at least one of the following: 2-hydroxy-3-methylbutyric acid, acetoacetic acid, tyrosine, phenylalanine, carbamoyl phosphate, glyoxylic acid, uridine, linoleic acid, alanine, and 1-methyluric acid. The detection data is input into a trained machine learning model for liver cancer diagnosis to achieve liver cancer diagnosis or risk assessment.

10. The use of the test reagent in the preparation of products for early diagnosis or risk assessment of liver cancer, characterized in that, The detection reagent includes reagents for detecting protein biomarkers and metabolic biomarkers, wherein the protein biomarkers include alpha-fetoprotein, and the metabolic biomarkers include nicotinamide and a combination of at least one of the following: 2-hydroxy-3-methylbutyric acid, acetoacetic acid, tyrosine, phenylalanine, carbamoyl phosphate, glyoxylic acid, uridine, linoleic acid, alanine, and 1-methyluric acid.

Images