Table 2.
Biomarkers for lipid storage disorders.
| Disease | Sample Type | Methodology | Sample Size | Potential Biomarkers Upregulated (↑) or Downregulated (↓) |
Statistical Models and Analysis | References |
|---|---|---|---|---|---|---|
| (1) NPC | Mouse Brain Mouse Liver Mouse Spleen |
LC-MS-MS | ~20 specimens | ↑ Sphingolipid species in the liver ↑ GM2 and GM3 gangliosides in the brain and spleen |
Results expressed as mean±SEM values; two-tailed single factor ANOVA model (p ≤ 0.05 was considered significant). Bonferroni correction applied. Biomarker heatmap. |
[9] |
| Human Plasma Human CSF |
LC-MS-MS | 56 NPC1 56 Controls |
↑ Monohexosylceramide and ceramide ↓ Sphingoid bases |
|||
| Human Plasma | LC-MS-MS | 109 NPC1 88 Controls 45 Heterozygous carriers |
↑ 7-Ketocholestrol and cholestane-3β5α6β-triol | Results expressed as mean ±SEM values; two-tailed single factor ANOVA model (p ≤ 0.05 was considered significant). | [10] | |
| Human Plasma | LC-MS-MS | 5 NPC1 7 Controls |
↑ Lysophingomyelin-509 and sphingosylphosphorylcholine | Results expressed as mean ± SD; Student’s t-test | [11] | |
| Human Plasma | LC-ESI-MS-MS | 13 NPC1 60 Controls |
↑ Cholestane-3β5α6β-triol and 7-ketocholestrol | Preliminary Kolmogorov-Smirnov test. Student’s t-test for normally distributed variables and a Mann-Whitney test for non-normally distributed variables. Pearson correlation coefficients were employed to evaluate relationships (p < 0.05 was considered significant). ROC curves were generated (0.998 for cholestane-3β5α6β-triol): specificity of 98.3%; sensitivity of 100%. | [12] | |
| Human Plasma | LC-MS-MS | 148 non-NPC1 6 NPC 11 NPB1 NPA |
↑ Cholestane-3β5α6β-triol | Box and whisker plot, mean values only. | [13] | |
| Human Urine | LC-ESI-MS | 1 NPC1 patient 1 3β-HSD deficiency 1 control infant 1 control adolescent |
↑ Sulphate conjugates of cholesterol and bile acids in NPC and 3β-HSD deficiency | Counts reported in NPC and 3 β-HSD and no counts reported in controls. No statistical analysis performed. | [14] | |
| Human Plasma | GC-MS | 25 fasting NPC1 25 Controls 23 heterozygotes |
↑ Cholestane-3β5α6β-triol and 7-ketocholestrol | Results expressed as mean±SEM values. Two-tail single factor ANOVA (p ≤ 0.05 was considered significant). Bonferroni correction applied. Cholestane-3β5α6β-triol (p < 0.001) ROC curve AUROC value 1.0. 7-ketocholesterol (p = 0.05) ROC Curve AUROC value 0.9984. |
[15] | |
| Mouse Liver | 1H NMR | 28 Wild-Type 31 Heterozygote 30 NPC1 |
↑ Phenylalanine, tyrosine glutamate, lysine/ornithine, valine threonine, hypotaurine/methionine ↓ Inosine nicotinate/niacinamide, phosphoenolpyruvate, 3-hydryphenylacetate |
Analysis of covariance (ANCOVA) model incorporating 3 first-order interaction components of variance; ANOVA-simultaneous component analysis (ASCA); Random Forest (RF) model (out-of-the-bag error value 0.175 and 0.19); AUROC value 0.94; Metabolic pathway topological analysis (MPTA). | [16] | |
| Human Plasma | 1H NMR | 40 untreated NPC Total 89 from 34 in duplicate treated with Miglustat over 2-year period 30 Control 31 Heterozygote |
↓ HDL-cholesterol and LDL-cholesterol in NPC vs Control ↑ Triacylglycerol in NPC vs Control ↑ Ca2+ ions in NPC vs HET ↓ HDL-Cholesterol in NPC vs HET |
ANOVA with Bonferroni correction. PCA scores plots (RF model: Out-of-the-bag error 0.089±0.0002 (mean±SEM) Canonical correlation analysis (CCorA). |
[17] | |
| Human Urine | LC-ESI-MS-MS | 2 NPC patients 2 3B-hydroxysteroid dehydrogenase deficient patient 8 controls |
↑ 3β-Sulfooxy-7β-N-acetylglucosaminyl-5-cholen-24-oic acid, and its glycine-amide and taurine-amide derivatives | Mean±SD values only. | [18] | |
| Human Urine | LC-MS-MS | 23 NPC patients 28 Controls 7 Patients with cerebrotendienous xanthomatosis, glycogen storage disease types I and II, citrin deficiency and abetalipoproteinemia |
↑ 3β-Sulfooxy-7β-N-acetylglucosaminyl-5-cholen-24-oic acid, and its glycine-amide and taurine-amide derivatives | Wilcoxon’s t-test applied to all metabolites detectable (p < 0.05) AUROC value > 0.95 for each marker |
[19] | |
| Human Urine | LC-MS-MS | 28 NPC Patients 38 Controls |
↑ 3β-Sulfooxy-7β-hydroxy-5-cholen-24-oic acid, 3β-sulfooxy-7-oxo-5-cholen-24-oic acid, non-amidated 3βsulfooxy-7β-N-acetylglucosaminyl-5-cholen-24-oic acid, glycine-amidated 3β-sulfooxy-7β-N-acetylglucosaminyl-5-cholen-24-oic acid and taurine-amidated 3β-sulfooxy-7β-N-acetylglucosaminyl-5-cholen-24-oic acid | AUROC value > 0.92 for each marker AUROC value 1.0 for 3β-sulfooxy-7β-N-acetylglucosaminyl-5-cholen-24-oic acid |
[20] | |
| Human Plasma | HPLC-MS-MS | 135 NPC1 Patients 66 NPC1 Carriers 241 Other LSD 46 Controls |
↑ Cholestane-3β,5α,6β-triol and lyso-Sphingomyelin-509 | ROC (100% sensitivity and 91% specificity) and AUROC value 0.99 with 95% CI: 0.98–1.00 ANOVA (p < 0.05). |
[21] | |
| Human Urine | 1H NMR | 13 untreated NPC1 47 control heterozygote carriers |
↑ Bile acids, BCAAs, 3-aminoisobutyrate, glutamine, 3-methylhistidine, creatine, quinolinate, succinate, trimethylamine, nicotinate, N-methylnicotinamide, N-methyl-2-pyridone-5-carboxamide, N- methyl-4-pyridone-5-carboxamide and trigonelline |
Cube root-transformed, Pareto-scaling, false discovery rate or Holm step-down Bonferroni correction; normalisation to Cn; ANOVA; Principal component analysis (PCA); Partial least-squares discriminatory analysis (PLS-DA): Q2 = 0.56 and Accuracy 0.93 (p = 5.0 × 10−4); Partial redundancy analysis (P-RDA) p < 10−4; MV ROC AUROC value 0.93 (95% CI 0.78–0.99). UV ROC AUROC value 0.81–0.90; RF model 83% overall mean classification success; Heatmap generated. |
[22] | |
| Human Urine | LC-ESI-MS-MS | 1 NPC | ↑ 3β-Sulfooxy-7β-hydroxy-5-cholen-24-oic acid and 3β-sulfooxy-7-oxo-5-cholen-24-oic acid | No statistical analysis performed. | [23] | |
| Human Plasma | LC-MS-MS | 70 Control 57 NPC |
↑Lysosophingomyelin and glucosylsphingosine | AUC AUROC values±95% CI Lysosophingomyelin 0.9994, glucosylsphingosine 0.7764. | [24] | |
| Human Dried Blood Spot | LC-MS-MS | 27 NPB 20 Control |
↑ Lysosophingomyelin | Mean (p < 0.0001), Student’s t-test. | [25] | |
| Human Plasma | LC-ESI-MS-MS | 38 NPD not ASM Deficient 7 NPD ASM Deficient 7 NPC 34 GD 12 KD >300 Controls |
↑ 7-Ketocholestrol in NPC and NPB ASM deficient patients | Mean ± SEM; ANOVA (p ≤ 0.05). | [26] | |
| Human Dried Blood Spot Human Urine |
UPLC-MS-MS | DBS 73 NPC1 2 NPC1 Homozygous 2 NPC1 Heterozygous 2 NPC1 Wild Type 84 Controls 70 NPC1 Heterozygotes Urine 14 NPC1 2 NPC1 Homozygous 10 Controls 47 NPC1 Heterozygotes |
↑ 3β-Hydroxy,7β-N-acetylglucosaminyl-5-cholenoic acid 40 times higher in NPC disease than that ofcontrols; however, these values were considered normal in some patients in view of a mutation ↑ 3β,5α,6β-Trihydroxycholanoyl-glycine 10 times higher in NPC compared to controls better biomarker in dried blood spots |
Kruskal-Wallis test; Dunn’s significance level of 0.05. ↑ 3β,5α,6β-Trihydroxycholanoyl-glycine (p < 0.001) when compared to controls and carriers in dried blood spots. |
[27] | |
| Human EDTA Plasma | HPLC-ESI-MS-MS | 107 Controls 16 NPC 91 Other diseases |
7-Ketocholestrol (7-KC) unspecific biomarker ↑ Cholestane-3β,5α,6β-triol (CT) sensitive but not specific |
Mann-Whitney U-test, Spearman’s correlation analysis, ROC Curve AUROC value (p < 0.05 considered statistically significant). NPC vs Controls 95% CI for CT, 62–275 vs. 3.5–4.4 ng/mL, p< 0.0001; 95% CI for 7-KC, 178–795 vs. 11.8–14 ng/mL, p < 0.0001. 100% sensitivity and 88.7% specificity. |
[28] | |
| (2) Fabry | Human Urine/Human Plasma | UPLC-ESI-MS-MS | 111 Urine 129 Plasma |
↑ Sphingolipids significantly in 31 plasma and 26 urine samples. 48% elevation in plasma, 42% in urine over those of controls. Phospholipids also reported to be higher: fold-changes 15% in plasma and 13 in urine. | PCA plot: Fabry patients loaded positively on PC1, Controls loaded negatively on PC1. Sphingolipids also loaded strongly on PC1. | [29] |
| Human Urine | qTOF-MS | 63 Untreated Fabry 59 Controls |
↑ Globtriaosylceramide and globotriosylsphingosine | PCA, OPLS-DA and S-plot: p = 0.05. | [30] | |
| Human Urine | LC-MS | 42 FD 48 Controls |
↑ Lyso-Gb3 plus its analogues | AUROC value 1.0 | [31] | |
| Human Plasma | LC-MS-MS | 38 FD male and female patients 120 controls |
↑ Lyso-Gb3 | Mean ± SD or median. Whitney U test. ROC curve analysis (p < 0.05) | [32] | |
| Human Urine | qTOF-MS | 16 untreated Fabry males 16 healthy Control males |
↑ Galabiosyceramide Analogs | OPLS-DA, Pareto-scaling p = 0.05. |
[33] | |
| Human Urine | UPLC-MS-MS | 52 Fabry paediatric male 108 Fabry adults 52 control paediatric 45 control adults |
Gender effects biomarkers; paediatric females have lower levels ↑ Lyso-Gb3 / related analog profile |
ROC Curve AUROC values; Mann-Whitney U test; Spearman rank correlation coefficient p < 0.05 in Fabry male children Treated females had lower lyso-Gb3 / related analogue profiles. |
[34] | |
| Human Plasma | Nano-LC-MS | 16 Male 10 heterozygous females 5 functional variants 40 controls |
Gender effects biomarkers; females have lower levels ↑ Lyso-Gb3 / related analog profile |
Mean±SD Tukey multiple comparison test (p < 0.0001); no difference in functional variants found (p > 0.05). |
[35] | |
| Human Plasma | UPLC-MS-MS | 178 healthy controls 74 Fabry |
↑ Lyso-Gb3 Lower levels in females |
Specificity 100%; Sensitivity Males 95%, Females 88%; Significance Males (p < 0.005) Females (p < 0.15). | [36] | |
| Human Urine | LC-MS | 164 Fabry 94 controls |
↑ Lyso-Gb3 and analogs Lower levels in females |
p < 0.001 lyso-Gb3 analogue levels in urine and gender, lyso-Gb3 levels analogue levels correlated well with enzyme replacement therapy (ERT) in males (p< 0.05). Mann-Whitney U-test and Spearman rank correlations were performed. | [37] | |
| Human Plasma | HPLC-ESI-MS-MS | 48 untreated Fabry 79 treated Fabry 36 control |
↑ Lyso-Gb3 Lower levels in females Untreated Fabry males distinguishable from controls; ERT- treated Fabry males less so. |
Box-and-whisker plots only. | [30] | |
| Human Plasma | HPLC | 10 Fabry Male 8 Heterozygote Female |
↑ Lyso-Gb3 increased in males | Mean±SD values. Student’s t-test p < 0.05 considered significant. Males significant (p < 0.05) | [38] | |
| Human Urine | UPLC-MS-MS | 150 Fabry Patients 95 Controls |
↑ Gb3 Isoforms | Mann-Whitney U Test (p < 0.001) established significant differences between Gb3 Isoforms, genders, treatments and concentrations. | [39] | |
| Human Plasma Human Urine |
UPLC-ESI-MS-MS | 10 Fabry Hemizygotes 10 Fabry Heterozygotes 20 Controls |
↑ Lyso-Gb3 in both urine and plasma | Mean±SD values only. | [40] | |
| Mouse Plasma Mouse Spleen Mouse Heart Mouse Liver Mouse Kidney |
LC-MS-MS | Wild Type (n = 3) Fabry (n = 5) |
↓ Significantly lower lyso-Gb3 after administration of 0.5 mg/kg human α-galactosidase A mRNA in plasma, spleen, heart, liver and kidney | Mean±SD/SEM and ANOVA | [41] | |
| Human Plasma Human Urine |
LC-MS-MS | 18 asymptomatic females 18 symptomatic females 27 males 27 control urine 58 control plasmas |
↑ C22:1, C22:0, C22:1-OH, C22:0-OH, C24:2, C24:0, C24:2-OH, C24:1-OH, C24:0-OH, C26:0 Galabiosylceramide in asymptomatic females | PCA and OPLS-DA of urine and plasma samples. Kruskal-Wallis and Mann-Whitney tests used for univariate comparisons. ROC curve analysis. | [42] | |
| (3) Gaucher | Human Serum Human Peritoneal Fluid Human Pericardial Fluid |
MALDI TOF-MS |
2 Gaucher Disease 1 Leukemia 3 Ovarian Tumors 5 Respiratory Infection |
↑ Glycosylceramide in all biofluids examined for GD patients | Mean±SD values only. | [43] |
| Human Plasma | HPLC | 27 Gaucher Disease Type I 15 Control |
↑ Glycosylceramide in disease vs control ↓ Glycosylceramide and Glycosylceramide /Ceramide ratio in disease vs 6 months of treatment with ERT/SRT group |
Median and range values; Mann-Whitney U test; Correlations tested by rank correlation test (Spearman coefficient) p < 0.05 considered significant. Glycosylceramide p = 0.0327 and Glysocylcarmide/ceramide ratio p = 0.0034 after 6 months after receiving ERT/SRT. | [44] | |
| Human Plasma | LC-MS-MS | 148 Controls 98 GD 13 GD Carriers 262 Patients with other LSD including NPC KD FD and Hunters Disease |
↑ Glucosylsphingosine | 100% specificity/100% sensitivity AUC 95% CI in ROC Analysis: AUROC value 1.0. | [45] | |
| Human Plasma | LC-ESI-MS-MS | 64 GD 28 control |
↑ Glucosylsphingosine 200-fold higher than controls Positive correlations between glucosylsphingosine and sphingolpids established in addition to plasma glucosylsphingosine and liver volume, bone marrow fat fraction |
Median and range, Mann-Whitney U test, rank correlation test (Spearman coefficient) statistically significant when 2-tailed p < 0.05 glucosylsphingosine and sphingolipids correlation, with p = 0.002. |
[46] | |
| Dried Blood Spots | LC-MS | 35 Mild Type I GD 34 Severe Type I GD 12 Type III GD |
↑ Glucosylsphingosine | Median and range reported, non-parametric Spearman’s and parametric Pearson’s correlations employed to observe relationships. | [47] | |
| (4) Krabbe | Mouse Hindbrain Tissue | GC-MS/LC-MS | 8 Wild Type 8 ‘Twitcher’ |
↑ Hypoxathinein Twitcher ↓ Glucose usage in Twitcher, phospholipid and membrane turnover in Twitcher, cholesterol lanosterol and lathosterol in Twitcher ↓ N-acetylaspartate |
ANOVA, Random forest model Student’s t-test p < 0.05 considered significant P ≤ 0.05 12 metabolites significantly modified RF 31%-100% accuracy dependent on collection point. |
[48] |
| Dried Blood Spot (DBS) | LC-MS-MS | 75 Controls Newborns 8 Newborns 65 Krabbe Disease |
↑ Pyschosine in KD | Mean±SEM values only. | [49] | |
| DBS | HPLC-MS-MS | 23 ‘at-risk’ KD newborns 8 KD |
↑ Pyschosine in KD | Mean values only. | [50] | |
| DBS | LC-MS-MS | 220 Controls 26 KD 18 GALC mutation |
↑ Pyschosine in KD | Range of values only. | [51] | |
| (5) GM1 and GM2 | Urine | Preparative TLC-1H NMR Analysis | 10 Controls 50 Diseased samples including GM1, GM2, Salla, Fucosidosis, Tyrosinena Type I and II, Citrullinemia, Canavan Disease, French-type sialuria, α- and β- mannsidosis, aspartylglu-cosaminuria |
↑ N-acetylated biantennary octasaccharide, N-acetylated triantennary decasaccharide/ tetraantennary dodecasaccharide in GM1 ↑ Oligosaccharides in GM2 |
No statistical analysis performed | [52] |
Suggested biomarkers from metabolomics investigations for LSDs using LC-MS or NMR-based methodologies. Abbreviations: 3β-HSD, 3β -hydroxysteroid-∆-5C27-steroid dehydrogenase; ANCOVA, analysis of covariance; ANOVA, analysis of variance; ASCA, analysis of variance simultaneous component analysis; AUC, area under the curve; AUROC, area under receiver operating characteristic; BCAA, branched chain amino acid; CCorA, canonical correlation analysis; CT, cholestane-3β,5α,6β-triol; DBS, dried blood spot; ERT, enzyme replacement therapy; ESI, electrospray ionisation; FD, fabry disease; GC, gas chromatography; GD, gaucher disease; HDL, high density lipoprotein; HET, heterozygous; HPLC, high performance liquid chromatography; KC, 7-ketocholesterol; KD, Krabbe disease; LC, liquid chromatography; LDL, low-density-lipoprotein; LSD, lysosomal storage disorder; lyso-Gb3, globotriaosylsphingosine; MALDI, matrix-assisted laser desorption/ionization; MS, mass spectrometry; MV, multivariate; NMR, nuclear magnetic resonance; NPA, Niemann-Pick Disease Type A; NPB1, Niemann-Pick Disease Type B1; NPC1, Niemann-Pick Disease Type C1; O-PLS-DA, ortho-partial least squares discriminant analysis; PCA, principal component Analysis; PLS-DA, partial least squares discriminatory analysis; P-RDA, partial redundancy analysis; QTOF-MS, quadrupole time of flight-mass spectrometry; RF, random forest; ROC, receiver operating curve; SD, standard deviation; SEM, standard error of the mean; SRT, substrate reduction therapy; TLC, thin-layer chromatography; UPLC, ultra-performance liquid chromatography.