Table 2 Brief information about metabolomics studies on urinary biomarkers for hepatocellular carcinoma
From: Urinary biomarkers for hepatocellular carcinoma: current knowledge for clinicians
Author/Year | Nations* | Samples | Platform | Modeling | Validation | Biomarkers | Applications a | Results | Evidence from Serum/tissue | AFP Comparison |
|---|---|---|---|---|---|---|---|---|---|---|
Shariff/ 2010 | Nigeria | HCC, 18 LC, 10 Healthy, 15 | [1] H-NMR | PCA PLS-DA | Cross-validation External validation (30% samples) | carnitine↑, creatine↑, creatinine↓, acetone↓ | Diagnosis | HCC vs. Healthy Sensitivity 100% Specificity 93.3% HCC vs. LC Sensitivity 89.5% Specificity 88.9% |  | Cutoff at 20 IU/mL Sensitivity 88.9% Specificity 77.8% |
Shariff/ 2011 | Egypt | HCC, 18 LC, 20 Healthy, 20 | [1] H-NMR | PCA PLS-DA | Cross-validation External validation (30% samples) | carnitine↑, creatine↑, TMAO↓ | Diagnosis | HCC vs. LC Sensitivity 81% Specificity71% |  | Cutoff at 20 IU/mL HCC vs. LC Specificity 0% |
Ladep/ 2014 | Nigeria Gambia | Training set HCC, 63 LC, 32 Non-cirrhotic liver disease, 107 Healthy, 88 Validation set HCC, 141 LC, 56 Non-cirrhotic liver disease, 178 Healthy, 88 | [1] H-NMR | PCA PLS-DA LR | Cross-validation External validation (independent cohort) | inosine↓, indole-3-acetate↑, NAA↑, galactose↑ | Diagnosis | HCC vs. LC Training set AUROC 0.90 Sensitivity 86.9% Specificity90.3% Validation set AUROC 0.72 Sensitivity 77.1% Specificity 63.5% |  | HCC vs. LC Training set AUROC 0.68 Sensitivity 49.2% Specificity77.4% Validation set AUROC 0.58 Sensitivity 60% Specificity 66% BCLC-D vs. BCLC A-C not significant |
Prognosis (Clinical stages) | BCLC D vs. BCLC A-C (methionine, acetylcarnitine, indole-3-acetate, NAA, dimethylglycine, 1-methylnicotinamide, creatine) significant | |||||||||
Shariff/ 2016 | UK | HCC, 13 LC, 25 | [1] H-NMR | PCA PLS-DA | Cross-validation | carnitine↑, formate↑, citrate doublet↓, hippurate↓, p-cresol sulfate↓, creatinine methyl↓, creatinine methylene↓ | Diagnosis | HCC vs. LC Sensitivity 53.6% Specificity 96% |  | Cutoff at 20 IU/mL Sensitivity 45% Specificity 95% |
Cox/ 2016 | Bangladesh | HCC, 46 LC, 50 CH, 48 Healthy, 8 | [1] H-NMR | PCA PLS-DA | Cross-validation | carnitine↑, creatine↑, TMAO↓, hippurate↓ | Diagnosis | HCC vs. non-HCC carnitine↑, creatine↑, TMAO↓, hippurate↓ (P < 0.05) |  | HCC vs. non-HCC AFP↑ (P < 0.05) |
Wang/ 2022 | Animal (Rats) | HCC model, 18 Control, 18 | [1] H-NMR | PCA |  | choline↑, taurine↑, creatinine↑, hippurate↓, PUT↑ | Diagnosis | HCC vs. Control AUROC Hippurate: 0.812 (0.667–0.957) b creatinine: 0.701 (0.527–0.874) PUT: 0.738 (0.561–0.914) choline: 0.722 (0.547–0.897) taurine: 0.722 (0.551–0.894) |  |  |
Wu/ 2009 | China | HCC, 20 Healthy, 20 | GC-MS/MS | PCA | Cross-validation | octanedioic acid↑, glycine↑, L-tyrosine↑, L-threonine↑, butanedioic acid↑, other 13 metabolites↓ | Diagnosis | HCC vs. Healthy PCA model of 18 metabolites AUROC 0.8825 |  | HCC vs. Healthy AFP alone Cutoff at 20 ng/mL Sensitivity 75% [AFP + urinary metabolites] AUROC 0.9725 |
Li/ 2010 | Animal (Rats) | HCC model, 5 HLM model, 5 Normal, 5 | GC-TOF-MS/MS | PLS-DA |  | Serine↓, Glycine↓, 5-oxyproline↓, Malate↓, 2-methylsuccinic acid↑ | Prognosis (Lung metastasis) | Completely separate HLM from HCC by PLS-DA | HLM vs. HCC Serum: serine, ornithine, phenylalanine, asparaginase, threitol, 5-hydroxyproline, 2,3,4-trihydroxybutyric acid↓; Lactic acid↑ |  |
Chen/ 2011 | China | Training set HCC, 55 BT, 16 Healthy, 47 Validation set HCC, 27 BT, 8 Healthy, 24 | GC-TOF-MS/MS UPLC-QTOF-MS/MS | PCA PLS-DA | Cross-validation External validation | Not specified c | Diagnosis | HCC vs. Healthy Accuracy 100% HCC (AFP < 20ng/mL) vs. Healthy Accuracy 100% |  |  |
Ye/ 2012 | China | HCC, 19 Recurrent, 7 Non-recurrent, 11 Healthy, 20 | LC-TOF-MS/MS | Binary LR |  | Ethanolamine↑, Lactic acid↑, Acotinic acid↑, Phenylalanine↑, Ribose↑ | Prognosis (1-year recurrence) | Recurrent vs. non-recurrent accuracy 100% |  |  |
Osman/ 2017 | Egypt | HCC, 55 LC, 40 Healthy, 45 | GC-MS/MS | PCA |  | glycine↑, serine↑, threonine↑, proline↑, citric acid↑, urea↓, phosphate↓, pyrimidine↓, arabinose↓, xylitol↓, hippuric acid↓, xylonic acid↓, glycerol↓ | Diagnosis | HCC vs. Healthy PCA model of 13 markers AUROC 1.00 |  |  |
Shao/ 2015 | China | Training set HCC, 33 LC, 27 Healthy, 26 Validation set HCC, 33 LC, 21 | LC-QTRAP-MS/MS | PLS-DA Binary LR | External validation | carnitine C4:0↑, hydantoin-5-propionic acid↑ | Diagnosis | Training set HCC vs. LC AUROC 0.786 Small HCC vs. LC AUROC 0.840 Validation set HCC vs. LC AUROC 0.773 |  | Training set HCC vs. LC AUROC 0.778 Small HCC vs. LC AUROC 0.675 Validation set HCC vs. LC AUROC 0.528 Small HCC vs. LC Sensitivity 0% |
Liang/ 2016 | China | Training set HCC, 25 Healthy, 12 Validation set HCC, 15 Validation set, 10 | LC-QTOF-MS/MS | PCA PLS-DA SAM | External validation | palmitic acid, alpha-N-Phenylacetyl-L-glutamine, phytosphingosine, indoleacetyl glutamine, and glycocholic acid ↓/↑ (not specified) | Diagnosis | HCC vs. Healthy AUROC 0.903 Sensitivity 96.5% Specificity 83.0% |  |  |
Dawuti/ 2022 | China | HCC, 55 LC, 49 Healthy, 50 | SERS | SVM | Cross-validation | adenine↓, guanine↓, deoxyribose↓, uric acid↓, uracil↓, proline, Urea, histidine, serine, tryptophan, alanine, creatinine: ↓/↑ (not specified) | Diagnosis | HCC vs. LC Sensitivity 79.6% Specificity 76.0% Accuracy 77.9% HCC or LC vs. Healthy Sensitivity 92.0% Specificity 77.8% Accuracy 87.0% |  | Serum AFP Sensitivity for HCC 34.5% |