Diagnostic, prognostic markers for non-alcoholic fatty liver disease and uses thereof

Abstract

The application discloses a diagnostic and prognostic marker of non-alcoholic fatty liver disease and application thereof, and the diagnostic and prognostic marker comprises a protein biomarker, a metabolite biomarker and a lipid biomarker in serum, a protein biomarker, a metabolite biomarker and a lipid biomarker in urine, and a metabolite biomarker and a lipid biomarker in feces. The biomarker disclosed by the application can be used for non-invasive diagnosis of non-alcoholic fatty liver disease, and is especially used for non-invasive diagnosis of whether the non-alcoholic fatty liver disease has liver fibrosis and whether the non-alcoholic fatty liver disease has significant liver fibrosis (F2). In addition, the biomarker disclosed by the application can be used as a prognostic biomarker of non-alcoholic fatty liver disease.

Description

Technical Field

[0001] This invention relates to diagnostic and prognostic biomarkers for non-alcoholic fatty liver disease, and particularly to new non-invasive diagnostic methods for non-alcoholic fatty liver disease. Background Technology

[0002] Nonalcoholic fatty liver disease (NAFLD) has become a significant global public health issue and is also one of the most common liver diseases in my country. The prevalence of NAFLD in my country increased from 18.2% in 2007 to 29.8% in 2017. The NAFLD spectrum includes simple fatty liver (NAFL), nonalcoholic steatohepatitis (NASH), nonalcoholic fatty liver cirrhosis, and liver cancer. Approximately 25% of NAFL patients develop NASH as the disease progresses, eventually progressing to cirrhosis and liver cancer, seriously threatening human physical and mental health and quality of life. In a systematic review and meta-analysis of 1495 NAFLD patients, the risk of all-cause mortality, particularly death from liver disease, increased significantly with the occurrence and severity of liver fibrosis. A multicenter longitudinal study found that compared to F0, fibrosis grade ≥F1 increased the risk of all-cause mortality and liver transplantation in patients with alcoholic fatty liver disease. Liver fibrosis is an important prognostic indicator for non-alcoholic fatty liver disease (NAFLD); therefore, effective screening for fibrosis in NAFLD patients can help facilitate timely intervention and treatment to improve patient outcomes.

[0003] Currently, liver biopsy remains the gold standard for diagnosing non-alcoholic fatty liver disease (NAFLD) because it accurately assesses the degree of hepatic steatosis, hepatocellular damage, inflammation, necrosis, and fibrosis, thus diagnosing the extent of liver disease, assessing prognosis, and guiding follow-up time. However, its clinical application is limited by drawbacks such as trauma and complications, sampling errors, and differences in pathologist observations, resulting in low patient acceptance and high testing costs. Various predictive models combining clinical data and different combinations of serum fibrosis markers can roughly determine the presence of significant fibrosis (≥F2) and advanced liver fibrosis (F3, F4). However, existing serological-based liver fibrosis prediction models do not meet the "diagnostic accuracy reporting standards." Liver elasticity values ​​(LSM) measured by vibration-controlled transient elastography (VCTE) based on FibroScan help differentiate between no / mild liver fibrosis (F0, F1) and advanced liver fibrosis (F3, F4), but there is still no universally accepted threshold for diagnosing cirrhosis. Currently, fibrosis prediction models based on serology and imaging have poor diagnostic efficacy and cannot meet clinical diagnostic needs.

[0004] Proteomics, metabolomics, lipidomics, and fecal metagenomics have been applied to diagnose non-alcoholic fatty liver disease (NAFLD) and liver fibrosis, and to predict the occurrence of cirrhosis and liver cancer. These methods have identified many biomarkers with good diagnostic efficacy for NAFLD and liver fibrosis. Therefore, by integrating multi-omics technologies (proteomics, metabolomics, and lipidomics) and machine learning algorithms, diagnostic and predictive models for NAFLD, NASH, and liver fibrosis can be established, thus replacing liver biopsies for diagnosis and follow-up. Summary of the Invention

[0005] To address the problems existing in current diagnostic techniques, this invention provides diagnostic and prognostic biomarkers for non-alcoholic fatty liver disease (NAFLD). This invention utilizes blood, urine, and stool samples from 130 patients with NAFLD confirmed by liver biopsy and 40 healthy volunteers. Employing multi-omics analysis (proteomics, metabolomics, lipidomics, etc.) and machine learning models, a non-invasive diagnostic model was established for NAFLD, the presence of liver fibrosis, and the presence of significant liver fibrosis (≥F2). NAFLD patients were divided into training and testing cohorts. This invention provides serum, urine, and stool biomarkers for diagnosing NAFLD, including the presence or absence of liver fibrosis and significant liver fibrosis (≥F2).

[0006] Compared with the control group, serum samples from patients with non-alcoholic fatty liver disease (NAFLD) showed differential expression of 103 proteins, with 35 proteins upregulated and 68 proteins downregulated. Serum samples from NAFLD patients also showed differential expression of 168 metabolites, with 131 metabolites upregulated and 37 downregulated. Furthermore, serum samples from NAFLD patients showed differential expression of 163 lipids, with 157 lipids upregulated and 6 lipids downregulated. Based on this data, a assay containing six serum proteins—THBS1, F13A1, DSG2, VSIG4, PPBP, and FGB—can accurately diagnose NAFLD. The area under the ROC curve (AUC) for the training cohort was 0.999, and for the testing cohort, it was 0.997. One group containing seven serum metabolites—dehydrocholic acid, docosahexaenoic acid, 5,6-dehydroarachidonic acid, L-glutamic acid, Phe-Phe, 4-ethyloctanoic acid, and Asp-Phe—demonstrated accurate diagnostic ability for non-alcoholic fatty liver disease (NAFLD), with an AUC of 1.0 in both the training and testing cohorts. Another group containing five serum lipids—carnitine C14-OH, carnitine C16:1-OH, TG (16:0_18:1_18:1), TG (16:0_18:1_22:6), and FFA (18:1)—demonstrated good diagnostic ability for NAFLD, with an AUC of 0.998 in the training cohort and 1.0 in the testing cohort. Using serum protein biomarkers, metabolite biomarkers, and lipid biomarkers for the combined diagnosis of non-alcoholic fatty liver disease (NAFLD), we found that a group containing 2 proteins (F13A1, VSIG4), 3 metabolites (L-glutamic acid, Phe-Phe, Asp-Phe), and 3 lipids [carnitine C16:1-OH, TG (16:0_18:1_22:6), FFA (18:1)] can accurately diagnose NAFLD. The area under the ROC curve (AUC) in both the training and testing cohorts was 1.0.

[0007] Compared with patients with non-alcoholic fatty liver disease (NAFLD) without fibrosis, patients with fibrotic NAFLD showed differential expression of 231 proteins in their serum (153 upregulated and 78 downregulated); differential expression of 58 metabolites (30 upregulated and 28 downregulated); and differential expression of 54 lipids (3 upregulated and 51 downregulated). Based on this data, we found that a set of five serum proteins [GDI1, ADH1C, SERPINB10, EMILIN1, isocitrate dehydrogenase (NADP(+))1(IDH1)] has good accuracy in diagnosing the presence or absence of liver fibrosis in NAFLD. In the training cohort, the area under the ROC curve (AUC) was 0.882, with a sensitivity of 71.0% and a specificity of 65.2%. In the test cohort, the AUC was 0.826, with a sensitivity of 88.9% and a specificity of 88.9%. A combination of six serum metabolites (indole-3-acetic acid, DL-O-tyrosine, (+)-prosopinine, Phe Thr Thr, succinic acid, and acetylhomoserine) can diagnose the presence of liver fibrosis in patients with non-alcoholic fatty liver disease. In the training cohort, the area under the ROC curve (AUC) was 0.880, with a sensitivity of 69.6% and a specificity of 92.6%. In the test cohort, the AUC was 0.836, with a sensitivity of 69.6% and a specificity of 77.8%. A combination of 10 lipid biomarkers [carnitine C5:1-2OH, CE (20:2), Cer (d18:2 / 23:0), DG (20:0-18:0), SM (d18:1 / 21:1), SM (d18:2 / 26:1), TG (18:2-18:3-18:3), FFA (14:0), PC (14:0-18:1), and PC (20:2-20:4)] can diagnose liver fibrosis in patients with non-alcoholic fatty liver disease. In the training cohort, the area under the ROC curve (AUC) was 0.853, with a sensitivity of 78.3% and a specificity of 81.5%. In the test cohort, the AUC was 0.802, with a sensitivity of 78.3% and a specificity of 77.8%. A combination of five serum proteins (GDI1, ADH1C, SERPINB10, EMILIN1, IDH1), two metabolites (indole-3-aceticacid, DL-O-tyrosine), and four lipids [Cer (d18:2 / 23:0), SM (d18:1 / 21:1), TG (18:2_18:3_18:3), PC (20:2_20:4)] demonstrated high accuracy in diagnosing the presence of liver fibrosis in patients with non-alcoholic fatty liver disease.The area under the ROC curve (AUC) in the training cohort was 0.943, with a sensitivity of 82.6% and a specificity of 88.9%. The area under the ROC curve (AUC) in the test cohort was 0.899, with a sensitivity of 69.6% and a specificity of 88.9%.

[0008] Compared with patients with non-alcoholic fatty liver disease (NAFLD) without significant liver fibrosis, patients with NAFLD with significant liver fibrosis (≥F2) showed differential expression of 17 serum proteins, with 3 proteins upregulated and 14 proteins downregulated; differential expression of 26 serum metabolites, with 5 metabolites upregulated and 21 metabolites downregulated; and differential expression of 163 serum lipids, with 5 lipids upregulated and 158 lipids downregulated. Based on this data, a panel of 10 serum proteins (KITLG, YWHAE, HSPA8, GSTO1, ARHGDIA, RAC1, LCN 2, GNPTG, PROCR, CAT) can accurately diagnose significant liver fibrosis (≥F2). In the training cohort, the area under the ROC curve (AUC) was 0.979, with a sensitivity of 93.5% and a specificity of 88.3%. In the test cohort, the AUC was 0.827, with a sensitivity of 83.3% and a specificity of 96.2%. Contains 15 metabolites (laserpitin, 2-arachidonoyl-1-stearoyl-sn-glycero-3-phosphoethanolamine, 2-hydroxyisocaproic acid, Ile Ile Glu Glu Val, 5-pentyl-1,4-dioxan-2-one, teleocidin B-1, 8-(3-Octyl-2-oxiranyl)octanoic a cid, leupeptin, 7-methylguanosine, 20,26-dihydroxyecdysone, DL-O-tyrosine, rivastigmine, phosphocholin e, p-coumarylalcohol 4-O-glucoside and Ala Leu The Tyr combination was able to diagnose significant fibrosis (≥F2). In the training cohort, the area under the ROC curve (AUC) was 0.984, with a sensitivity of 94.8% and a specificity of 93.5%. In the test cohort, the AUC was 0.622, with a sensitivity of 33.3% and a specificity of 76.9%.A combination of 12 lipids [Cer(d18:1 / 26:0), DG(18:2_22:6), LPC(0:0 / 18:1), TG(18:0_18:0_18:0), TG(18:2_18:2_22:4), TG(18:2_18:3_22:6), carnitine C20:3, taurochenodeoxycholic acid, PC(18:2_20:5), CerP(d18:1 / 16:0), CerP(d18:1 / 16:1) and LPA(16:0)] can diagnose significant fibrosis (≥F2). In the training cohort, the area under the ROC curve (AUC) was 0.962, with a sensitivity of 97.4% and a specificity of 89.6%. In the test cohort, the AUC was 0.878, with a sensitivity of 66.7% and a specificity of 96.2%. A biomarker combination comprising 5 proteins (GSTO1, LCN2, GNPTG, PROCR, CAT), 3 metabolites (2-arachidon oyl-1-stearoyl-sn-glycero-3-phosphoethanolamine, 5-pentyl-1,4-dioxan-2-one, rivastigmine), and 4 lipids [LPC (0:0 / 18:1), carnitine C20:3, PC (18:2-20:5), LPA (16:0)] was able to diagnose significant fibrosis (≥F2) in patients with non-alcoholic fatty liver disease. In the training cohort, the area under the ROC curve (AUC) was 0.988, with a sensitivity of 97.4% and a specificity of 93.5%. In the test cohort, the AUC was 0.846, with a sensitivity of 83.3% and a specificity of 76.9%.

[0009] Compared with the control group, patients with non-alcoholic fatty liver disease (NAFLD) showed differential expression of 315 proteins in their urine, with 206 upregulated and 109 downregulated; differential expression of 1659 metabolites in their urine, with 919 upregulated and 740 downregulated; and differential expression of 83 lipids in their urine, with 80 upregulated and 3 downregulated. Based on this data, a cohort containing five urinary proteins (COL4A2, AQP2, FAM3B, APLP2, and PCDHGC5) can accurately diagnose NAFLD. The area under the ROC curve (AUC) for the training cohort was 0.955, and for the testing cohort, it was 0.852. A group containing four urinary metabolites—alpha-Methylene-gamma-butyrolactone, (2S)-3-Hydroxypropane-1,2-diyl didecanoate, Genipin, and Ophiobolin F—was able to accurately diagnose patients with non-alcoholic fatty liver disease. The area under the ROC curve (AUC) in the training cohort was 1.000, and the area under the ROC curve (AUC) in the test cohort was also 1.000. A set of seven urinary lipids, including TG (14:0_16:0_18:0), TG (16:0_16:0_18:0), TG (18:1_18:1_18:1), BMP (22:6_22:6), FFA (24:6), PG (18:1_18:1), and FFA (30:1), can accurately diagnose patients with non-alcoholic fatty liver disease. The area under the ROC curve (AUC) in the training cohort was 0.959, and the AUC in the test cohort was 0.788.

[0010] Compared with patients with non-alcoholic fatty liver disease (NAFLD) without fibrosis, patients with fibrotic NAFLD showed differences in the expression of 30 proteins in their urine (22 upregulated and 8 downregulated); 190 metabolites showed differences (109 upregulated and 81 downregulated); and 14 lipids showed differences (11 upregulated and 3 downregulated). Based on these data, a set of two urinary proteins, FLRT1 and NOTCH3, can be used to diagnose whether NAFLD patients have liver fibrosis. The area under the ROC curve (AUC) in the training cohort was 0.740, and in the testing cohort it was 0.655. A group containing two urinary metabolites (4R)-4-[(3R,5R,8R,9S,10S,13R,14S,17R)-10,13-dimethyl-3-sulfooxy-2,3,4,5,6,7,8,9,11,12,14,15,16,17-tetradecahydro-1H-cyclopenta[a]phenanthren-17-yl]pentanoic Acid, [(2S)-2-pentadecanoyloxy-3-tetradecanoyloxypropyl](7Z,10Z,13Z,16Z,19Z)-docosa-7,10,13,16,19-pentaenoate can diagnose liver fibrosis. The area under the ROC curve (AUC) in the training cohort was 0.772, and the AUC in the test cohort was 0.736. A combination of four lipid markers—MG (20:0), BMP (18:1-22:6), PGF2α, and FFA (24:4)—can diagnose whether patients with non-alcoholic fatty liver disease have liver fibrosis. The AUC in the training cohort was 0.904, and the AUC in the test cohort was 0.643. A combination of one protein FLRT1 and two lipids, MG (20:0) and FFA (24:4), can diagnose whether patients with non-alcoholic fatty liver disease have liver fibrosis. The area under the ROC curve in the training cohort was AUC = 0.904, and the area under the ROC curve in the test cohort was AUC = 0.708.

[0011] Compared with non-alcoholic fatty liver disease (NAFLD) patients without significant liver fibrosis, NAFLD patients with significant fibrosis (≥F2) showed differential expression of 25 proteins in urine (4 upregulated and 21 downregulated); differential expression of 82 metabolites in urine (24 upregulated and 58 downregulated); and differential expression of 5 lipids in urine (all upregulated). Based on this data, a group containing 7 urinary proteins—AMY2A, PECAM1, ENG, ARSF, FLRT1, GMPR2, and MAN1B1—could distinguish significant liver fibrosis (≥F2). The area under the ROC curve (AUC) in the training cohort was 0.902, and the AUC in the test cohort was 0.485. A group containing 24 urinary metabolites: 5-Chloro-1,3-dihydro-1-(4-piperidinyl)-2H-benzimidazol-2-one, 3-Methoxybenzoic acid, 17-phenyl trinor PGF2 cycloprop ylmethyl amide, Hexaethylene glycol, Leu-Gly-Leu, 3-O-Coumaroylquinic acid, Gossypin; Gossypetin 8-O-glucoside, Palmitoyl Serotonin, Phellamurin, Isorhamnetin 3-glucoside, Lys Trp Lys, Meclofenamate sodium, Lunarine, Asp GlnAsn Asp, PA (16:0 / 22:6(4Z,7Z,10Z,13Z,16Z,19Z)), Canaliculatol, 2-Ethyl-4-methylthiazole, Morellinol, 4-(dimethylamino)azobenzene n-oxide, 1-Aminopropan-2-ol, Kalkitoxin thioamid e alcohol, 2,3,5-Triphenyltetrazolium chloride, morpholin-4-ium-4-methoxyphenyl-(morpholino)-phosphinodit hioate, and Triamifos can diagnose significant fibrosis (≥F2). The area under the ROC curve (AUC) in the training cohort was 1.000, and the area under the ROC curve (AUC) in the test cohort was 0.605.Three lipid biomarkers, DG (14:0–16:1), Glycoursodeoxycholic acid, and Taurocholic acid, can diagnose whether patients with non-alcoholic fatty liver disease (NAFLD) have significant liver fibrosis (≥F2). The area under the ROC curve (AUC) in the training cohort was 0.803, and the AUC in the test cohort was 0.265. A combination of one protein (ENG) and four metabolites (3-Methoxybenzoic acid, Leu-Gly-Leu, Isorhamnetin 3-glucoside, and 1-Aminopropan-2-ol) can diagnose whether patients with NAFLD have significant liver fibrosis (≥F2). The AUC in the training cohort was 0.964, and the AUC in the test cohort was 0.577.

[0012] Compared with the control group, 1293 metabolites in the feces of patients with non-alcoholic fatty liver disease (NAFLD) showed differences, with 828 metabolites upregulated and 465 metabolites downregulated. 169 lipids in the feces of patients with NAFLD also showed differences, with 91 lipids upregulated and 78 lipids downregulated. Five fecal biomarkers (6-(3,4-dihydroxy-6-methyl-5-oxooxan-2-yl)-5,7-dihydroxy-2-(4-hydroxy-3-methoxyphenyl)-4H-chromen-4-one, Tris(2-chloroethyl)phosphate; LC-ESI-ITFT; MS2; CE, 8-Epi-prostaglandin F2 alpha, Benzanthrone, and PYRROLIDINE) demonstrated good diagnostic ability for NAFLD, with an AUC of 0.997 in the training cohort and 0.945 in the testing cohort. A combination of seven lipids—LPE (0:0 / 18:1), PE (P-16:0_18:1), PE (P-18:0_18:1), PG (18:1_20:2), PA (18:1_22:6), LNAPE (18:1 / N-17:0), and CE (20:5)—can accurately diagnose non-alcoholic fatty liver disease. The area under the ROC curve (AUC) in the training cohort was 0.978, and the AUC in the test cohort was 0.852. A group containing five metabolites—6-(3,4-dihydroxy-6-methyl-5-oxooxan-2-yl)-5,7-dihydroxy-2-(4-hydroxy-3-methoxyphenyl)-4H-chromen-4-one, Tris(2-chloroethyl)phosphate, LC-ESI-ITFT, MS2, CE, 8-Epi-prostaglandin F2alpha, Benzanthrone, and PYRROLIDINE—and three lipids—PE (P-16:0-18:1), PA (18:1-22:6), and CE (20:5)—can accurately diagnose non-alcoholic fatty liver disease. The area under the ROC curve (AUC) in the training cohort was 1.000, and the AUC in the test cohort was 0.945.

[0013] Compared with non-alcoholic fatty liver disease (NAFLD) patients without liver fibrosis, NAFLD patients with liver fibrosis showed differences in the expression of 156 metabolites in feces, with 54 upregulated and 102 downregulated; and differences in the expression of 114 lipids in feces, with 110 upregulated and 4 downregulated. Based on this data, a group containing 19 fecal metabolites includes 6-[4-(1H-Imidazol-1-yl)phenoxy]-N,N-dimethyl-1-hexanamin e,dihydrochloride, Phe LeuAsn, Capric Acid (C10:0), (4E)-1-(4-hydroxy-3-methoxyphenyl)dec-4-en-3-one, HisGlu His, Rimexolone, 13-OxoODE; 13-Keto-9Z,11E-octadecadienoic acid, 1,2-Dihydroxyheptadec-16-yn-4-yl acetate, Tyr Pro Ile, LPC (18:1 / 0:0), Fenproporex, 2-Phenylethanol, Ethyl butyrate, Codeine, Leu-Ala-Val, 3,4-Dimethylstyrene, Soyasaponin Bb, His Asn Phe Lys, NG, and NG-Dimethyl-L-arginine can diagnose whether patients with non-alcoholic fatty liver disease have liver fibrosis. The area under the ROC curve (AUC) in the training cohort was 1.000, and the area under the ROC curve (AUC) in the test cohort was 0.300. A set of 14 fecal lipid biomarkers, including TG (16:0_18:0_20:0), TG (16:0_18:1_20:2), DG (18:3_18:3), Cer (t18:0 / 24:0), MGDG (16:0_18:2), DGDG (16:0_20:1), DGDG (16:0_18:2), Taurocholic acid, LPI (18:0), CerP (d18:1 / 14:1), LPI (17:0), FFA (26:0), FFA (32:0), and FFA (32:1), can diagnose whether patients with non-alcoholic fatty liver disease have liver fibrosis. The area under the ROC curve (AUC) in the training cohort was 1.000, and the AUC in the test cohort was 0.483.A group containing 21 biomarkers: Phe Leu Asn, Capric Acid (C10:0), (4E)-1-(4-hydroxy-3-methoxyphenyl)dec-4-en-3-one, His GluHis, 1,2-Dihydroxyheptadec-16-yn-4-yl acetate, Tyr Pro Ile, LPC (18:1 / 0:0), Fenproporex, Codeine, Leu-Ala-Val, 3,4-Dimethylstyrene, Soyasaponin Bb, His AsnPhe Lys, NG,NG-Dimethyl-L-arginine, Cer (t18:0 / 24:0), MGDG (16:0-18:2), DGDG (16:0-20:1), DGDG (16:0-18:2), Taurocholicacid, LPI (18:0), F FA(32:0) can diagnose whether non-alcoholic fatty liver disease has fibrosis. The area under the ROC curve (AUC) in the training cohort was 1.000, and the area under the ROC curve (AUC) in the test cohort was 0.586.

[0014] Compared with patients with non-alcoholic fatty liver disease (NAFLD) without significant liver fibrosis, patients with NAFLD with significant fibrosis (≥F2) showed differences in the expression of 64 metabolites in feces, with 22 upregulated and 42 downregulated; and 39 lipids showed differences in fecal expression. These biomarkers can be used to diagnose NAFLD, determine whether NAFLD has fibrosis, and identify significant fibrosis. Based on this data, a group containing seven fecal metabolites—Artemisinin, LysThr Leu, 1-Dehydro-12-gingerdione, Ciloszolinol, Penicillic acid, MetHis, Sarcoaldesterol A; (3beta,5alpha,6beta,11alpha)-Gorgostane-3,5,6,11-tetrol—can diagnose whether patients with non-alcoholic fatty liver disease have significant liver fibrosis (≥F2). The area under the ROC curve (AUC) in the training cohort was 0.981, and the AUC in the test cohort was 0.430. A set of four fecal lipid markers, DG (16:0-20:4), LPE (17:0), PE (P-14:0-18:1), and norcholic acid, can diagnose whether patients with non-alcoholic fatty liver disease have significant liver fibrosis (≥F2). The area under the ROC curve in the training cohort was AUC = 0.808, and the area under the ROC curve in the test cohort was AUC = 0.523. A group containing 7 metabolites (Artemisinin, Lys Thr Leu, 1-De hydro-12-gingerdione, Cilostabol, Penicillic acid, Met His, Sarcoaldesterol A; (3beta,5alpha,6beta,11alpha)-Gorgostane-3,5,6,11-tetrol) and 4 lipids (DG (16:0-20:4), LPE (17:0), PE (P-14:0-18:1), and norcholic acid) can diagnose whether non-alcoholic fatty liver disease has significant liver fibrosis (≥F2). The area under the ROC curve (AUC) in the training cohort was 0.500, and the area under the ROC curve (AUC) in the test cohort was also 0.500. Attached Figure Description

[0015] Figure 1 ROC curves of serum biomarkers for diagnosing non-alcoholic fatty liver disease (NAFLD) in the training and testing cohorts;

[0016] Figure 2ROC curves of serum markers for diagnosing non-alcoholic fatty liver disease (NAFLD) with or without fibrosis and significant liver fibrosis (≥F2) in the training and testing cohorts.

[0017] Figure 3 ROC curves of urinary biomarkers for diagnosing non-alcoholic fatty liver disease (NAFLD) in the training and testing cohorts.

[0018] Figure 4 ROC curves of urinary markers for diagnosing non-alcoholic fatty liver disease (NAFLD) in training and testing cohorts, indicating the presence or absence of fibrosis and significant liver fibrosis (≥F2).

[0019] Figure 5 ROC curves for fecal biomarkers used to diagnose non-alcoholic fatty liver disease (NAFLD) in the training and testing cohorts.

[0020] Figure 6 ROC curves of fecal biomarkers for diagnosing non-alcoholic fatty liver disease (NAFLD) in training and testing cohorts, indicating the presence or absence of fibrosis and significant liver fibrosis (≥F2). Detailed Implementation

[0021] This invention utilizes blood, urine, and stool samples from 130 patients with non-alcoholic fatty liver disease (NAFLD) confirmed by liver biopsy and 40 healthy volunteers. Employing multi-omics analysis (proteomics, metabolomics, lipidomics, etc.) and machine learning models, a non-invasive diagnostic model for NAFLD, the presence of liver fibrosis, and the presence of significant liver fibrosis (≥F2) was established. NAFLD patients were divided into training and testing cohorts. This invention provides serum, urine, and stool biomarkers for diagnosing NAFLD, including the presence of liver fibrosis and the presence of significant liver fibrosis (≥F2).

[0022] (1) Serum protein biomarkers for the diagnosis of non-alcoholic fatty liver disease

[0023] This invention utilizes proteomics to discover that, compared to the control group, serum samples from patients with non-alcoholic fatty liver disease (NAFLD) showed differential expression of 103 proteins (Welch t-test, p-value < 0.01, fold change > 2.0 or < 1 / 2), with 35 proteins downregulated and 68 proteins downregulated (Table 1). Based on this data, a set of six serum proteins—THBS1, F13A1, DSG2, VSIG4, PPBP, and FGB—can accurately diagnose NAFLD patients. The area under the ROC curve (AUC) in the training cohort was 0.999, and the AUC in the test cohort was 0.997. Figure 1 A).

[0024] Table 1

[0025]

[0026]

[0027]

[0028] (2) Serum metabolite biomarkers for the diagnosis of non-alcoholic fatty liver disease

[0029] This invention utilizes metabolomics to identify 168 different metabolites in serum samples from patients with non-alcoholic fatty liver disease (NAFLD) compared to the control group (Welch t-test, p-value < 0.01, fold change > 2.0 or < 1 / 2), with 131 metabolites upregulated and 37 downregulated (Table 2). Based on this data, a group containing seven serum metabolites—dehydrocholic acid, docosahexaenoic acid, 5,6-dehydroarachidonic acid, L-glutamic acid, Phe-Phe, 4-ethyloctanoic acid, and Asp-Phe—demonstrated accurate diagnostic ability for NAFLD, with an AUC of 1.0 in both the training and testing cohorts. Figure 1 B).

[0030] Table 2

[0031]

[0032]

[0033]

[0034]

[0035] (3) Serum lipid biomarkers for the diagnosis of non-alcoholic fatty liver disease

[0036] Lipidomics analysis revealed 163 different lipid profiles in serum samples from patients with non-alcoholic fatty liver disease (NAFLD) compared to the control group (Welch t-test, p-value < 0.01, fold change > 2.0 or < 1 / 2), with 157 lipids upregulated and 6 lipids downregulated (Table 3). Based on this data, a cohort containing five serum lipids—carnitine C14-OH, carnitine C16:1-OH, TG (16:0_18:1_18:1), TG (16:0_18:1_22:6), and FFA (18:1)—demonstrated good diagnostic ability for NAFLD, with an AUC of 0.998 in the training cohort and 1.0 in the testing cohort. Figure 1 C).

[0037] Table 3

[0038]

[0039]

[0040]

[0041] (4) Combined diagnosis of non-alcoholic fatty liver disease using serum protein biomarkers, metabolite biomarkers, and lipid biomarkers.

[0042] Using serum protein biomarkers, metabolite biomarkers, and lipid biomarkers for the combined diagnosis of non-alcoholic fatty liver disease (NAFLD), we found a group containing two proteins (F13A1, VSIG4), three metabolites (L-glutamic acid, Phe-Phe, Asp-Phe), and three lipids [carnitine C16:1-OH, TG (16:0-18:1-22:6), FFA (18:1)] that can accurately diagnose NAFLD. The area under the ROC curve (AUC) was 1.0 in both the training and testing cohorts. Figure 1 D).

[0043] (5) Serum protein biomarkers used to diagnose non-alcoholic fatty liver disease with or without liver fibrosis and significant liver fibrosis (≥F2).

[0044] Proteomics analysis revealed 231 differentially expressed serum proteins in patients with non-alcoholic fatty liver disease (NAFLD) and those without fibrosis (Welch t-test, p-value < 0.05), with 153 proteins upregulated and 78 proteins downregulated (Table 4). We found a group of five serum proteins [GDI1, ADH1C, SERPINB10, EMILIN1, isocitrate dehydrogenase (NADP(+))1(IDH1)] with good accuracy in diagnosing the presence or absence of liver fibrosis in NAFLD. In the training cohort, the area under the ROC curve (AUC) was 0.882, with a sensitivity of 71.0% and a specificity of 65.2%. In the test cohort, the AUC was 0.826, with a sensitivity of 88.9% and a specificity of 88.9%. Figure 2 A).

[0045] Table 4

[0046]

[0047]

[0048]

[0049]

[0050]

[0051] We found that compared with patients with non-alcoholic fatty liver disease (NAFLD) without significant liver fibrosis, patients with NAFLD with significant fibrosis (≥F2) showed different expression levels of 17 proteins in their serum (Welch t-test, pvalue < 0.05), of which 3 proteins were upregulated and 14 proteins were downregulated (Table 5). We also found that a group of 10 serum proteins (KITLG, YWHAE, HSPA8, GSTO1, ARHGDIA, RAC1, LCN2, GNPTG, PROCR, CAT) could accurately diagnose significant liver fibrosis (≥F2). In the training cohort, the area under the ROC curve (AUC) was 0.979, with a sensitivity of 93.5% and a specificity of 88.3%. In the test cohort, the AUC was 0.827, with a sensitivity of 83.3% and a specificity of 96.2%. Figure 2 B).

[0052] Table 5

[0053]

[0054] (6) Serum metabolite biomarkers used to diagnose non-alcoholic fatty liver disease with or without liver fibrosis and significant liver fibrosis.

[0055] Metabolomics analysis revealed 58 different serum metabolites compared to non-alcoholic fatty liver disease (NAFLD) patients with liver fibrosis (Welch t-test, p value < 0.05), with 30 metabolites upregulated and 28 downregulated (Table 6). We found that a combination of six serum metabolites (indole-3-aceticacid, DL-O-tyrosine, (+)-prosopinine, Phe Thr Thr, succinic acid, and acetylhomoserine) could diagnose the presence of liver fibrosis in NAFLD patients. In the training cohort, the area under the ROC curve (AUC) was 0.880, with a sensitivity of 69.6% and a specificity of 92.6%. In the test cohort, the AUC was 0.836, with a sensitivity of 69.6% and a specificity of 77.8%. Figure 2 C).

[0056] Table 6

[0057]

[0058]

[0059] We found that compared with patients with non-alcoholic fatty liver disease without significant liver fibrosis, patients with significant fibrosis (≥F2) non-alcoholic fatty liver disease had different serum expression levels of 26 metabolites (Welch t test, p value <0.05), of which 5 metabolites were upregulated and 21 metabolites were downregulated (Table 7). A combination of 15 metabolites (laserpitin, 2-arachidonoyl-1-stearoyl-sn-glycero-3-phosphoethanolamine, 2-hydroxyisocaproic acid, Ile Glu Val, 5-pentyl-1,4-dioxan-2-one, teleocin B-1, 8-(3-Octyl-2-oxiranyl)octanoic acid, leupeptin, 7-methylguanosine, 20,26-dihydroxyecdysone, DL-O-tyrosine, rivastigmine, phosphocholine, p-coumaryl alcohol 4-O-glucoside, and Ala Leu Tyr) was able to diagnose significant fibrosis (≥F2). In the training cohort, the area under the ROC curve (AUC) was 0.984, with a sensitivity of 94.8% and a specificity of 93.5%. In the test cohort, the AUC was 0.622, with a sensitivity of 33.3% and a specificity of 76.9%. Figure 2 D).

[0060] Table 7

[0061]

[0062] (7) Serum lipid biomarkers used to diagnose non-alcoholic fatty liver disease with or without liver fibrosis and significant liver fibrosis (≥F2).

[0063] Lipomics analysis revealed that 54 metabolites were different in the serum of patients with non-fibrotic non-alcoholic fatty liver disease compared with those without fibrotic non-alcoholic fatty liver disease (Welch t test, p value < 0.05), of which 3 lipids were upregulated and 51 lipids were downregulated (Table 8). We found that a combination of 10 serum lipid biomarkers [carnitine C5:1-2OH, CE (20:2), Cer (d18:2 / 23:0), DG (20:0-18:0), SM (d18:1 / 21:1), SM (d18:2 / 26:1), TG (18:2-18:3-18:3), FFA (14:0), PC (14:0-18:1), and PC (20:2-20:4)] can diagnose the presence of liver fibrosis in patients with non-alcoholic fatty liver disease. In the training cohort, the area under the ROC curve (AUC) was 0.853, with a sensitivity of 78.3% and a specificity of 81.5%. In the test cohort, the AUC was 0.802, with a sensitivity of 78.3% and a specificity of 77.8%. Figure 2 E).

[0064] Table 8

[0065]

[0066]

[0067] We found that compared with patients with non-alcoholic fatty liver disease without significant liver fibrosis, patients with significant fibrosis (≥F2) non-alcoholic fatty liver disease showed differences in the expression levels of 163 lipids in their serum (Welch t test, p value <0.05), of which 5 lipids were upregulated and 158 metabolites were downregulated (Table 9). A combination of 12 lipids [Cer(d18:1 / 26:0), DG(18:2_22:6), LPC(0:0 / 18:1), TG(18:0_18:0_18:0), TG(18:2_18:2_22:4), TG(18:2_18:3_22:6), carnitine C20:3, taurochenodeoxycholic acid, PC(18:2_20:5), CerP(d18:1 / 16:0), CerP(d18:1 / 16:1) and LPA(16:0)] was able to diagnose significant fibrosis (≥F2). In the training cohort, the area under the ROC curve (AUC) was 0.962, with a sensitivity of 97.4% and a specificity of 89.6%. In the test cohort, the AUC was 0.878, with a sensitivity of 66.7% and a specificity of 96.2%. Figure 2 F).

[0068] Table 9

[0069]

[0070]

[0071]

[0072]

[0073] (8) Combined diagnosis of non-alcoholic fatty liver disease using serum protein, metabolite, and lipid biomarkers, including the presence or absence of liver fibrosis and significant liver fibrosis (≥F2).

[0074] We conducted a multi-omics analysis combining proteomics, metabolomics, and lipidomics to explore a set of biomarkers with higher accuracy in diagnosing liver fibrosis in patients with non-alcoholic fatty liver disease (NAFLD). We found a combination of 5 serum proteins (GDI1, ADH1C, SERPINB10, EMILIN1, IDH1), 2 metabolites (indole-3-acetic acid, DL-O-tyrosine), and 4 lipids [Cer(d18:2 / 23:0), SM(d18:1 / 21:1), TG(18:2_18:3_18:3), PC(20:2_20:4)] that demonstrated high accuracy in diagnosing the presence of liver fibrosis in NAFLD patients. In the training cohort, the area under the ROC curve (AUC) was 0.943, with a sensitivity of 82.6% and a specificity of 88.9%. In the test cohort, the AUC was 0.899, with a sensitivity of 69.6% and a specificity of 88.9%. Figure 2 G).

[0075] We discovered a biomarker combination containing 5 proteins (GSTO1, LCN2, GNPTG, PROCR, CAT), 3 metabolites (2-arachidonoyl-1-stearoyl-sn-glycero-3-phosphoethanolamine, 5-pentyl-1,4-dioxan-2-one, rivastigmine), and 4 lipids [LPC (0:0 / 18:1), carnitine C20:3, PC (18:2-20:5), LPA (16:0)] that can diagnose the presence or absence of significant fibrosis (≥F2) in patients with non-alcoholic fatty liver disease. In the training cohort, the area under the ROC curve (AUC) was 0.988, with a sensitivity of 97.4% and a specificity of 93.5%. In the test cohort, the AUC was 0.846, with a sensitivity of 83.3% and a specificity of 76.9%. Figure 2 H).

[0076] (9) Urinary protein biomarkers for the diagnosis of non-alcoholic fatty liver disease

[0077] This invention utilizes proteomics to discover that, compared to the control group, 315 proteins showed differential expression in urine samples from patients with non-alcoholic fatty liver disease (NAFLD) (Welch t-test, p-value < 0.05, fold change > 2.0 or < 1 / 2), with 206 proteins upregulated and 109 downregulated (Table 10). Based on this data, a set of five urinary proteins—COL4A2, AQP2, FAM3B, APLP2, and PCDHGC5—can accurately diagnose NAFLD. The area under the ROC curve (AUC) in the training cohort was 0.955, and in the test cohort, it was 0.852. Figure 3 A).

[0078] Table 10

[0079]

[0080]

[0081]

[0082]

[0083]

[0084]

[0085] (10) Urinary metabolite biomarkers for the diagnosis of non-alcoholic fatty liver disease

[0086] This invention utilizes metabolomics to discover that, compared to the control group, NAFLD patients showed differential expression of 1659 metabolites in their urine (Welch t test, p value < 0.01, fold change > 3.0 or < 1 / 3), with 919 upregulated and 740 downregulated (Table 11). One group containing four urinary metabolites—alpha-Methylene-gamma-butyrolactone, (2S)-3-Hydroxypropane-1,2-diyl didecanoate, Genipin, and Ophiobolin F—was used to accurately diagnose non-alcoholic fatty liver disease. The area under the ROC curve (AUC) in the training cohort was 1.000, and the area under the ROC curve (AUC) in the test cohort was also 1.000. Figure 3 B).

[0087] Table 11

[0088]

[0089]

[0090]

[0091]

[0092]

[0093]

[0094]

[0095]

[0096]

[0097]

[0098]

[0099]

[0100]

[0101]

[0102]

[0103]

[0104]

[0105]

[0106]

[0107]

[0108]

[0109]

[0110]

[0111]

[0112]

[0113]

[0114]

[0115] (11) Urinary lipid biomarkers for the diagnosis of non-alcoholic fatty liver disease

[0116] This invention utilizes lipidomics to discover that, compared with the control group, patients with non-alcoholic fatty liver disease (NAFLD) showed differences in 83 lipids in their urine (Welch t-test, p value < 0.05), with 80 lipids upregulated and 3 downregulated (Table 12). Based on this data, a set of seven urinary lipids—TG (14:0-16:0-18:0), TG (16:0-16:0-18:0), TG (18:1-18:1-18:1), BMP (22:6-22:6), FFA (24:6), PG (18:1-18:1), and FFA (30:1)—can accurately diagnose NAFLD. The area under the ROC curve (AUC) in the training cohort was 0.959, and the AUC in the test cohort was 0.788. Figure 3 C).

[0117] Table 12

[0118]

[0119]

[0120]

[0121] (12) Combined diagnosis of non-alcoholic fatty liver disease using urinary protein biomarkers, metabolite biomarkers, and lipid biomarkers.

[0122] Using urinary protein biomarkers, metabolite biomarkers, and lipid biomarkers in a combined diagnostic approach for non-alcoholic fatty liver disease (NAFLD), we found that a group containing four metabolites (alpha-Methylene-gamma-butyrolactone, (2S)-3-Hydroxypropane-1,2-diyl didecanoate, Genipin, Ophiobolin F, and Ophiobolene) could accurately diagnose NAFLD. The area under the ROC curve (AUC) in the training cohort was 1.0, and the AUC in the test cohort was also 1.0. Figure 3 D).

[0123] (13) Urinary protein biomarkers for diagnosing the presence or absence of fibrosis and significant fibrosis (≥F2) in non-alcoholic fatty liver disease.

[0124] This invention utilizes proteomics to discover that, compared to patients with non-alcoholic fatty liver disease (NAFLD) without fibrosis, patients with fibrotic NAFLD showed differential expression of 30 proteins in their urine (Welch t-test, pvalue < 0.05, fold change > 1.5 or < 2 / 3), with 22 proteins upregulated and 8 downregulated (Table 13). Based on this data, a set of two urinary proteins, FLRT1 and NOTCH3, can be used to diagnose whether NAFLD patients have liver fibrosis. The area under the ROC curve (AUC) in the training cohort was 0.740, and the AUC in the testing cohort was 0.655. Figure 4 A).

[0125] Table 13

[0126]

[0127] We found that compared with patients with non-alcoholic fatty liver disease (NAFLD) without significant liver fibrosis, patients with NAFLD with significant fibrosis (≥F2) showed differences in the expression of 25 proteins in their urine (Welch t-test, p value <0.05, fold change >1.5 or <2 / 3), with 4 proteins upregulated and 21 proteins downregulated (Table 14). Based on this data, a group containing 7 urinary proteins—AMY2A, PECAM1, ENG, ARSF, FLRT1, GMPR2, and MAN1B1—could distinguish significant liver fibrosis (≥F2). The area under the ROC curve (AUC) in the training cohort was 0.902, and the AUC in the test cohort was 0.485. Figure 4 B).

[0128] Table 14

[0129]

[0130]

[0131] (14) Urinary metabolite biomarkers for diagnosing non-alcoholic fatty liver disease with or without liver fibrosis and significant liver fibrosis (≥F2).

[0132] In this invention, metabolomics was used to find that compared with patients with non-alcoholic fatty liver disease without fibrosis, patients with non-alcoholic fatty liver disease with fibrosis had 190 different metabolites in their urine (Welch t test, p value < 0.05, fold change > 1.5 or < 2 / 3), with 109 upregulated and 81 downregulated (Table 15). Based on this data, a group containing two urinary metabolites, (4R)-4-[(3R,5R,8R,9S,10S,13R,14S,17R)-10,13-dimethyl-3-sulfooxy-2,3,4,5,6,7,8,9,11,12,14,15,16,17-tetradecahydro-1H-cyclopenta[a]phenanthren-17-yl]pentanoic acid and [(2S)-2-pentadecanoyloxy-3-tetradecanoyloxypropyl](7Z,10Z,13Z,16Z,19Z)-docosa-7,10,13,16,19-pentaenoate, can diagnose non-alcoholic fatty liver disease with or without liver fibrosis. The area under the ROC curve (AUC) in the training cohort was 0.772, and the AUC in the test cohort was 0.736. Figure 4 C).

[0133] Table 15

[0134]

[0135]

[0136]

[0137]

[0138] In this invention, metabolomics was used to find that compared with non-alcoholic fatty liver disease patients without significant liver fibrosis, patients with significant liver fibrosis (≥F2) had 82 different metabolites in their urine (Welch t test, p value <0.05, fold change >1.5 or <2 / 3), with 24 upregulated and 58 downregulated (Table 16). Based on this data, a group containing 24 urinary metabolites includes 5-Chloro-1,3-dihydro-1-(4-piperidinyl)-2H-benzimidazol-2-one, 3-Methoxybenzoic acid, 17-phenyl trinor PGF2 cyclopropyl methyl amide, Hexaethylene glycol, Leu-Gly-Leu, 3-O-Coumaroylquinic acid, Gossypin; Gossypetin 8-O-glucoside, Palmitoyl Serotonin, Phellamurin, Isorhamnetin 3-glucoside, Lys Trp Lys, Meclofenamate sodium, Lunarine, Asp Gln Asn Asp, PA (16:0 / 22:6(4Z,7Z,10Z,13Z,16Z,19Z)), Canaliculatol, and 2-Ethyl-4-methyl Thiazole, Morllinol, 4-(dimethylamino)azobenzene n-oxide, 1-Aminopropan-2-ol, Kalkitoxinthioamide alc ohol, 2,3,5-Triphenyltetrazolium chloride, morpholin-4-ium-4-methoxyphenyl-(morpholino)-phosphinodithioate, and Triamifos can diagnose non-alcoholic fatty liver disease with or without significant liver fibrosis (≥F2). The area under the ROC curve (AUC) in the training cohort was 1.000, and the AUC in the test cohort was 0.605. Figure 4 D).

[0139] Table 16

[0140]

[0141]

[0142]

[0143] (15) Urinary lipid biomarkers for diagnosing non-alcoholic fatty liver disease with or without liver fibrosis and significant liver fibrosis (≥F2).

[0144] Lipidomics analysis revealed differences in 14 lipids in the urine of patients with non-alcoholic fatty liver disease (NAFLD) and those without liver fibrosis (Welch t-test, p value < 0.05), with 11 lipids upregulated and 3 downregulated (Table 17). We found that combinations of four urinary lipid markers—MG (20:0), BMP (18:1–22:6), PGF2α, and FFA (24:4)—could diagnose the presence or absence of liver fibrosis in NAFLD. The area under the ROC curve (AUC) was 0.904 in the training cohort and 0.643 in the testing cohort. Figure 4 E).

[0145] Table 17

[0146]

[0147] Lipomics analysis revealed significant differences in five urinary lipids compared to patients with non-alcoholic fatty liver disease (NAFLD) without significant liver fibrosis (≥F2) (Welch t-test, p value < 0.05), with all five being upregulated (Table 18). Based on these data, we found that three urinary lipid markers—DG (14:0–16:1), Glycoursodeoxycholic acid, and Taurocholic acid—can diagnose the presence or absence of significant liver fibrosis (≥F2) in NAFLD. The area under the ROC curve (AUC) in the training cohort was 0.803, and in the testing cohort, it was 0.265. Figure 4 F).

[0148] Table 18

[0149]

[0150] (16) Combined diagnosis of non-alcoholic fatty liver disease using urine protein biomarkers, metabolite biomarkers, and lipid biomarkers, including the presence or absence of liver fibrosis and significant liver fibrosis (≥F2).

[0151] Using a combination of urinary protein markers, metabolite markers, and lipid markers to diagnose the presence of liver fibrosis in non-alcoholic fatty liver disease (NAFLD), we found that a combination of one protein (FLRT1) and two lipids (MG (20:0) and FFA (24:4)) can diagnose the presence of liver fibrosis in NAFLD patients. The area under the ROC curve (AUC) in the training cohort was 0.904, and in the testing cohort it was 0.708. Figure 4 G).

[0152] Using a combination of urinary protein markers, metabolite markers, and lipid markers to diagnose whether significant liver fibrosis (≥F2) is present in non-alcoholic fatty liver disease (NAFLD), we found that a combination containing one protein (ENG) and four metabolites (3-Methoxybenzoic acid, Leu-Gly-Leu, Isorhamnetin 3-glucoside, and 1-Aminopropan-2-ol) can diagnose whether NAFLD patients have significant liver fibrosis (≥F2). The area under the ROC curve (AUC) for the training cohort was 0.964, and for the testing cohort it was 0.577. Figure 4 H).

[0153] (17) Fecal metabolite biomarkers for the diagnosis of non-alcoholic fatty liver disease

[0154] Metabolomics analysis revealed 1293 different metabolites in the feces of patients with non-alcoholic fatty liver disease (NAFLD) compared to the control group (Welch t-test, p-value < 0.01, fold change > 2.0 or < 1 / 2), with 828 upregulated and 465 downregulated (Table 19). Based on these data, we found that a combination of five fecal metabolite markers (6-(3,4-dihydroxy-6-methyl-5-oxooxan-2-yl)-5,7-dihydroxy-2-(4-hydroxy-3-methoxyphenyl)-4H-chromen-4-one, Tris(2-chloroethyl)phosphate; LC-ESI-ITFT; MS2; CE, 8-Epi-prostaglandin F2alpha, Benzanthrone, and PYRROLIDINE) demonstrated good diagnostic ability for NAFLD. The area under the ROC curve (AUC) in the training cohort was 0.997, and the AUC in the test cohort was 0.945. Figure 5 A).

[0155] Table 19

[0156]

[0157]

[0158]

[0159]

[0160]

[0161]

[0162]

[0163]

[0164]

[0165]

[0166]

[0167]

[0168]

[0169]

[0170]

[0171]

[0172]

[0173]

[0174]

[0175]

[0176] (18) Fecal lipid biomarkers for the diagnosis of non-alcoholic fatty liver disease

[0177] Lipidomics analysis revealed that 169 lipids were differentially expressed in the feces of patients with non-alcoholic fatty liver disease (NAFLD) compared to the control group (Welch t-test, p-value < 0.05, fold change > 1.5 or < 2 / 3), with 91 lipids upregulated and 78 downregulated (Table 20). A combination of seven lipids—LPE (0:0 / 18:1), PE (P-16:0_18:1), PE (P-18:0_18:1), PG (18:1_20:2), PA (18:1_22:6), LNAPE (18:1 / N-17:0), and CE (20:5)—was found to accurately diagnose NAFLD. The area under the ROC curve (AUC) in the training cohort was 0.978, and in the testing cohort, it was 0.852. Figure 5 B).

[0178] Table 20

[0179]

[0180]

[0181]

[0182]

[0183] (19) Combined diagnosis of non-alcoholic fatty liver disease using fecal metabolite biomarkers and lipid biomarkers.

[0184] Using fecal metabolite markers and lipid markers for the combined diagnosis of non-alcoholic fatty liver disease (NAFLD), we found that a group containing five metabolites (6-(3,4-dihydroxy-6-methyl-5-oxooxan-2-yl)-5,7-dihydroxy-2-(4-hydroxy-3-methoxyphenyl)-4H-chromen-4-one, Tris(2-chloroethyl)phosphate, LC-ESI-ITFT, MS2, CE, 8-Epi-prostaglandin F2alpha, Benzanthrone, and PYRROL IDINE) and three lipids (PE (P-16:0-18:1), PA (18:1-22:6), and CE (20:5)) could accurately diagnose NAFLD. The area under the ROC curve (AUC) in the training cohort was 1.000, and in the test cohort it was 0.945. Figure 5 C and Figure 5 D).

[0185] (20) Fecal metabolite biomarkers for diagnosing the presence or absence of liver fibrosis and significant liver fibrosis (≥F2) in patients with non-alcoholic fatty liver disease.

[0186] Metabolomics revealed that, compared with patients with non-alcoholic fatty liver disease (NAFLD) without fibrosis, patients with fibrotic NAFLD showed differential expression of 156 metabolites in their feces (Welch t test, p value < 0.05), with 54 upregulated and 102 downregulated (Table 21). A set of 19 fecal metabolites 6-[4-(1H-Imidazol-1-yl)phenoxy]-N,N-dimethyl-1-hexanamine,dihydrochloride, Phe Leu Asn, Capric Acid(C10:0), (4E)-1-(4-hydroxy-3-methoxyphenyl)dec-4-en-3-one, His Glu His, Rimexolone, 13-OxoODE; 13-Keto-9Z,11E-octadecadienoic acid, 1,2-Dihydroxyheptadec-16-yn-4-yl acetate, Tyr Pr o Ile, LPC(18:1 / 0:0), Fenproporex, 2-Phenylethanol, Ethylbutyrate, Codeine, Leu-Ala-Val, 3,4-Dimethylstyrene, Soyasaponin Bb, His Asn PheLys, NG, and NG-Dimethyl-L-arginine can diagnose whether liver fibrosis is present in patients with non-alcoholic fatty liver disease. The area under the ROC curve (AUC) in the training cohort was 1.000, while the AUC in the test cohort was 0.300. Figure 6 A).

[0187] Table 21

[0188]

[0189]

[0190]

[0191]

[0192] Metabolomics analysis revealed that compared with non-alcoholic fatty liver disease (NAFLD) patients with significant liver fibrosis (≥F2), 64 fecal metabolites showed differential expression (Welch ttest, p value <0.05), with 22 upregulated and 42 downregulated (Table 22). A group containing seven fecal metabolites—Artemisinin, Lys Thr Leu, 1-Dehydro-12-gingerdione, Ciloszolinol, Penicillic acid, Met His, and Sarcoaldesterol A; (3beta,5alpha,6beta,11alpha)-Gorgostane-3,5,6,11-tetrol—could accurately diagnose whether NAFLD patients had significant liver fibrosis (≥F2). The area under the ROC curve (AUC) in the training cohort was 0.981, and in the testing cohort, it was 0.430. Figure 6 B).

[0193] Table 22

[0194]

[0195]

[0196] (21) Fecal lipid biomarkers for diagnosing the presence or absence of liver fibrosis and significant liver fibrosis (≥F2) in patients with non-alcoholic fatty liver disease.

[0197] Lipidomics analysis revealed that, compared with patients with non-alcoholic fatty liver disease (NAFLD) without fibrosis, patients with fibrotic NAFLD showed differential expression of 114 lipids in their feces (Welch t test, p value < 0.05), with 110 upregulated and 4 downregulated (Table 23). A set of 14 fecal lipid biomarkers, including TG (16:0_18:0_20:0), TG (16:0_18:1_20:2), DG (18:3_18:3), Cer (t18:0 / 24:0), MGDG (16:0_18:2), DGDG (16:0_20:1), DGDG (16:0_18:2), Taurocholic acid, LPI (18:0), CerP (d18:1 / 14:1), LPI (17:0), FFA (26:0), FFA (32:0), and FFA (32:1), can diagnose whether patients with non-alcoholic fatty liver disease have liver fibrosis. The area under the ROC curve (AUC) in the training cohort was 1.000, and the AUC in the test cohort was 0.483. Figure 6 C).

[0198] Table 23

[0199]

[0200]

[0201]

[0202] Compared with patients with non-alcoholic fatty liver disease (NAFLD) without significant liver fibrosis, patients with NAFLD with significant liver fibrosis (≥F2) showed differential expression of 39 lipids in their feces (Welch t test, p value < 0.05), and these 39 lipids were downregulated (Table 24). A panel containing four fecal lipid markers—DG (16:0–20:4), LPE (17:0), PE (P-14:0–18:1), and norcholic acid—was used to diagnose whether NAFLD patients had significant liver fibrosis (≥F2). The area under the ROC curve (AUC) in the training cohort was 0.808, and the AUC in the test cohort was 0.523. Figure 6 D).

[0203] Table 24

[0204]

[0205] (22) Combined diagnosis of non-alcoholic fatty liver disease using fecal metabolite biomarkers and lipid biomarkers, including the presence or absence of liver fibrosis and significant liver fibrosis (≥F2).

[0206] Using fecal metabolite markers and lipid markers in the combined diagnosis of non-alcoholic fatty liver disease, we identified a group of 21 biomarkers, including Phe Leu Asn, Capric Acid (C10:0), (4E)-1-(4-hydroxy-3-methoxyphenyl)dec-4-en-3-one, His Glu His, 1,2-Dihydroxyheptadec-16-yn-4-ylacetate, Tyr Pro Ile, LPC (18:1 / 0:0), Fenproporex, Codeine, Leu-Ala-Val, 3,4-Dimethylstyrene, Soyasaponin Bb, and His Asn Phe. Lys, NG, NG-Dimethyl-L-arginine, Cer (t18:0 / 24:0), MGDG (16:0_18:2), DGDG (16:0_20:1), DGDG (16:0_18:2), Taurocholic acid, LPI (18:0), and FFA (32:0) can diagnose whether non-alcoholic fatty liver disease has liver fibrosis. The area under the ROC curve (AUC) in the training cohort was 1.000, and the AUC in the test cohort was 0.586. Figure 6 E).

[0207] Using fecal metabolite markers and lipid markers in the combined diagnosis of non-alcoholic fatty liver disease, we found a combination containing 7 metabolites (Artemisinin, Lys Thr Leu, 1-Dehydro-12-gingerdione, Cilostabol, Penicillic acid, Met His, Sarcoaldesterol A; (3beta,5alpha,6beta,11alpha)-Gorgostane-3,5,6,11-tetrol) and 4 lipids (DG (16:0-20:4), LPE (17:0), PE (P-14:0-18:1), and norcholic acid). The area under the ROC curve (AUC) in the training cohort was 0.500, and the AUC in the test cohort was also 0.500. Figure 6 F).

Claims

1. Diagnostic, prognostic marker of non-alcoholic fatty liver disease with or without liver fibrosis, characterized in that: comprises at least one of the following combination of markers: serum protein biomarkers, serum metabolite biomarkers and serum lipid biomarkers; the serum protein biomarkers are GDI1, ADH1C, SERPINB10, EMILIN1 and IDH1; the serum metabolite biomarkers are indole-3-acetic acid, DL-O-tyrosine, (+)-prosopinine, Phe Thr Thr, succinic acid and acetylhomoserine; the serum lipid biomarkers are carnitine C5:1-2OH, CE(20:2), Cer (d18:2 / 23:0), DG (20:0_18:0), SM (d18:1 / 21:1), SM (d18:2 / 26:1), TG(18:2_18:3_18:3), FFA (14:0), PC (14:0_18:1) and PC (20:2_20:4).

2. The diagnostic, prognostic marker for non-alcoholic fatty liver disease with or without liver fibrosis according to claim 1, characterized in that: The five serum protein biomarkers are GDI1, ADH1C, SERPINB10, EMILIN1 and IDH1, the two serum metabolite markers are indole-3-acetic acid and DL-O-tyrosine, and the four serum lipid markers are Cer (d18:2 / 23:0), SM (d18:1 / 21:1), TG (18:2_18:3_18:3) and PC (20:2_20:4).

3. Use of the biomarkers of claim 1 or 2 in the manufacture of a product for diagnosing the presence or absence of liver fibrosis in non-alcoholic fatty liver disease.

Images