Table 3.
Emerging omics markers and their classification performance for diagnosing NAFLD/NASH.
| Platform | Sample Type | Number of Analytes Quantified in Total | Sample Size (Discovery Cohort) | Sample Size (Validation Cohort) | Classifier | Prediction Target | Markers | AUROC | Reference |
|---|---|---|---|---|---|---|---|---|---|
| Omics | Serum | 1129 proteins (SomaScan), 1 genotype, >200 clinical variables | n = 443 | n = 133 | Logistic regression | Hepatic steatosis in obesity | 8 proteins + 1 genotype + 12 clinical variables: ACY1, SHBG, CTSZ, MET, GSN, LGALS3BP, CHL1, SERPINC1, PNPLA3 variant. | 0.935 (0.914 in validation cohort) | (Wood et al., 2017) [90] |
| Omics | Serum | 860 proteins, 288 metabolites, 108 SNPs, 16,209 protein-coding genes, 58 clinical variables | n = 1049 | No for the omics model | Random forest | Fatty liver | 185 clinical and omics features | 0.84 | (Atabaki-Pasdar et al., 2020) [89] |
| SOMAscan proteomics | serum | 1305 proteins | n = 113 | n = 71, n = 32 |
Elastic-Net | Fibrosis F0–1 against F2–4 | serum amyloid P, fibrinogen, olfactomedin, and SHBG | 0.74 (0.52–0.78 in validation cohorts) | (Luo et al., 2021) [91] |
| SOMAscan proteomics | serum | 1305 proteins | n = 113 | n = 71, n = 32 |
Elastic-Net | Fibrosis (F3–4 against F0–2) | latent transforming growth factor beta binding protein 4, IGF-1, vascular cell adhesion molecule 1, interleukin-1 soluble receptor type-1, IL18BP, thrombospondin-2, collectin kidney 1, SHBG, interleukin-27 receptor subunit alpha, leukemia inhibitory factor receptor, soluble, fibulin-3, and plexin-B2 | 0.83 (0.74–0.78 in validation cohorts) | (Luo et al., 2021) [91] |
| MS-based proteomics | Plasma | 235–277 proteins | n = 19 | NA | Unclear | Fibrosis F2–4 against F0–1 | Complement component C7, α-2-macroglobulin, Fibulin-1, Complement component C8 γ chain; α-1-antichymotrypsin | 0.79–1 for each individual protein | (Hou et al., 2020) [93] |
| Metabolomics | serum | 365 lipids, 61 glycans and 23 fatty acids | n = 31 | NA | support vector machine | Fibrosis F2–4 against F0–1 | 10 lipids: DG(36:3), LPC(18:0), PC(36:2), PC(37:2), PC(40:5), TG(38:0), TG(50:0), TG(51:1), TG(57:1), TG(60:2) | 1 | (Perakakis et al., 2019) [105] |
| Metabolomics | Serum | 365 lipids, 61 glycans and 23 fatty acids | n = 80 | NA | Support vector machine | NASH vs. NAFL vs. Healthy | 29 lipids: AcCa(10:0), Cer(d34:2), DG(34:1), DG(36:4), LPC(20:0e), LPC(22:5), LPE(16:0), PC(32:0), PC(32:1e), PC(34:0), PC(34:2e), PC(35:3), PC(36:4), PC(36:5e), PC(37:2), PC(40:6e), PC(40:7), PC(40:8), PC(42:6), PE(38:1), PE(38:6), PI(36:1), SM(d32:0), SM(d32:2), SM(d40:1), TG(38:0), TG(38:2), TG(43:1), TG(53:5) |
0.94–0.99 (one vs. rest) | (Perakakis et al., 2019) [105]. |
| Metabolomics | Plasma | 13,008 metabolic features | n = 559 | NA | Random forest | NAFLD vs. non-NAFLD | 11 metabolite features + 3 clinical variables: serine, leucine/isoleucine, tryptophan, three putatively annotated compounds, two unknowns, lysoPE(20:0), lysoPC(18:1), WC, WBISI, and triglycerides | 0.94 | (Khusial et al., 2019) [106] |
| Metabolomics | Serum | 652 metabolites | n = 156 | n = 142 | Logistic regression | Fibrosis F3–4 vs. F0–2 in NAFLD | 8 lipids + 1 amino acid + 1 carbohydrate: 5alpha-androstan-3beta monosulfate, pregnanediol-3-glucuronide, androsterone sulfate, epiandrosterone sulfate, palmitoleate, dehydroisoandrosterone sulfate, 5alpha-androstan-3beta disulfate, glycocholate, taurine, fucose | 0.94 (0.84–0.94 in validation cohort) | (Caussy et al., 2019) [100] |
| Metabolomics | Serum | 540 lipids and amino acids | n = 467 | n = 192 | Logistic regression | NAFLD vs. Healthy | 11 triglycerides | 0.9 (0.88 in validation cohort) | (Mayo et al., 2018) [102] |
| Metabolomics | Serum | 540 lipids and amino acids | n = 467 | n = 192 | Logistic regression | NASH against NAFL | 20 triglycerides | 0.95 (0.79 in validation cohort) | (Mayo et al., 2018) [102] |
| Metabolomics | Serum | Sphingolipids and branched fatty acid esters of hydroxy fatty acids | n = 1479 | NA | Logistic regression | oleic acid-hydroxy oleic acid | 0.61 | (Hu et al., 2018) [120] | |
| Metabolomics | Serum | 1761 metabolic features | n = 59 | NA | Unclear | NASH against NAFL | pyroglutamate | 0.846 | (Qi et al., 2017) [117] |
| Metabolomics | Urine | Unclear | n = 78 | NA | Unclear | NASH against NAFL | Pyroglutamic acid | 0.65 | (Dong et al., 2017) [114] |
| Metabolomics | Serum | Unclear | n = 223 | n = 95 | Logistic regression | NASH against non-NASH | glutamate, isoleucine, glycine, lysophosphatidylcholine 16:0, phosphoethanolamine 40:6, AST, and fasting insulin | 0.882 (0.856 in validation cohort) | (Zhou et al., 2016) [112] |
| Lipidomics | Serum | 239 lipids | n = 42 | n = 22 | Logistic regression | NASH in NAFLD | Monounsaturated triglycerol | 0.83 in both discovery and validation cohorts | (Yang et al., 2017) [119] |