Table 3.

Results of the metabolomics studies conducted on tissue samples.

Author	Year	Sample Size	Technique	Use Case	Biomarkers Candidates	Validated?	Statistical Details	Comments
Wang [50]	2017	n = 3 subjects Cancer tissue: 3 Benign adjacent tissue: 3	MALDI-FTICR-MS	Diagnosis	differential metabolites were not mentioned	No	students t-test	coated tissue slice
Huan [42]	2016	Training set: n = 16 Pca: 13; Benign: 12 Validation set 1: n = 18 Pca: 19; Benign: 17 Validation set 2: n = 12 Pca: 12; benign: 12	LC QTOF-MS	Diagnosis	adenosine monophosphate, spermidine, uracil, ophthalmic acid + HPO3, 2,3-diaminopropionic acid + HPO3 + 2 unknown metabolites = putative identification	Yes	OPLS-DA volcano plot	Tissue extracts from core biopsies after prostatectomy
McDunn [44]	2013	Pca: 331 Benign: 178 (matched to 178 of the Pca samples)	UHPLC-MS/MS GC-MS	Staging	aggressive Pca:ADP, Glucose, 6-sialyl-N-actyllyctosamine, 2-hydroxypalmitate, 5,6 dihydrouracil, choline, fumarate, kynurenine, phophate, 2-hydrxoystearate, Ac-SDKP, choline phosphate, glycerol-3-phophate, n-acetylaspartate Gleason pattern progression:proline, malate, ADP-ribose, 6-sialyl-N-acetyllactosamineextracapsular extension:NAD+, N-acetylaspartate, putrescine, Glucose Tumor spread (regional lymph nodes/seminal vesicles):choline phosphate, Glycerol3-phophate, putrescine, 6-sialyl-N-acetyllactosamine	No	paired t-testWilcoxon test linear regression	Tissue extract from OCT embedded tissue from prostatectomies
Jung [47]	2013	n = 95 matched cancer and benign adjacent tissue	LC-MS/MS GC-MS	Diagnosis Prognosis Biological recurrence	Diagnosis of Pca:2-hdroxybehenic acid, crebronic acid, tricosanoic acid, glycerophophoethanolamine, isopentenyl pyrophosphate, 7-methylguanine, 2-aminoadipic acid, gluconic acid, maltotriose, tricosanoic acid Prediction of biological recurrence risk:2-aminoadipic acid, gluconic acid, maltotriose	No	Wilcoxon paired test ROC analysis logistic regression Kaplan–Meier curves univariate and multivariate Cox regression	tissue extracts of punch biopsy from cryosections after prostatectomy
Brown [51]	2012	n = 8 matched cancer and benign adjacent tissue	UHPLC-MS/MS GC-MS	Diagnosis	> 40 metabolites not specified further	No	Welch’s two sample t-test hierachical clustering	tissue extracts from core biopsy after prostatectomy
Selnæs [53]	2012	n = 13 subjects 40 tissue samples	HR-MAS-1H-MRS	Staging	CCS/C ratio (choline+creatine+spermine over citrate)	No	Spearman’s rank correlation Jonchheere– Terpstra test	intact tissue from needle biopsy after prostatectomy
Maxeiner [60]	2010	Pca with biological recurrence: 16 Pca without biological recurrence: 32	HR-MAS-1H-MRS	Prognosis	spermine, glutamine, myo-inositol, phophoryl choline, scylloinositol, glutamate	Yes	PCA student’s t-test Canonical analysis ANOVA ROC analysis	intact tissue from needle biopsy after prostatectomy;training and test set with identical case group but distinct control groups
Sreekumar [28]	2009	n = 42 tissue samples benign adjacent: 16 localized Pca: 12 metastatic Pca: 14	UHPLC-MS/MS GC-MS	Diagnosis	sarcosine, uracil, kynurenine, glycerol-3-phophate, leucine, proline	Yes	Wilcoxon rank-sum test	tissue extracts from biopsy samplesonly sarcosine was analyzed in validation set
Cacciatore [49]	2017	matched benign and Pca samples Training set: n = 8 Validation set: n = 4	UHPLC-MS/MS GC-MS	Diagnosis	32 metabolites reported; biomarkers included in the model not specified	Yes	Hierarchical clustering OSC-PLS	tissue extracts from tissue section after prostatectomyalso compared OCT-embedded and FFPE tissue as biospecimen
Sandsmark [54]	2017	n = 41 subjects Pca: 95 benign adjacent: 34	HR-MAS-1H-MRS	Diagnosis	Pca with high vs. Pca with low/intermediate NCWP-EMT score: spermine and citratePca with low vs. Pca with high NCWP-EMT score taurine, phosphoethanolamine	No	t-test	intact tissue from prostatectomiesmain focus: alterations in non-canonical WNT signaling pathway (NCWP) and EMT in Pcajoint gene expression and metabolomic analyses; targeted analysis of 23 metabolites; metabolomics was performed only on the main cohort
Hansen [55]	2016	n = 41 subjects Pca: 95 benign adjacent: 34	HR-MAS MRS	Staging	citrate, spermine (correlated with presence ERG translocation)	Yes	unsupervised multivariate analysis PLS-DA	gene expression analysis and TMPRSS2-ERG as marker for disease aggressiveness;intact tissue from tissue slices collected from prostatectomies; analyzed metabolic alterations in PCA patients positive for TMPRSS2-ERG/high ERG gene fusion; targeted analysis of 23 metabolites
Meller [57]	2016	n = 106 subjects matched cancer and benign adjacent tissue ERG-positive Pca: 27 ERG-negative Pca: 23	GC-MS LC-MS	Staging Prognosis Biological recurrence	Gleason score:pantothenic acid, maltose, fructose-6-phosphate, gluconic acid, cholesterol ERG status:maltotriose, gluconic acid, citrate, cis-aconitate, spermine, putrescine, cerebronic acid, 2-hydroxybehenic acid, tricosanoic acid, Biological relapse:tyrosine and tryptophan	No	ANOVA PCA	tissue extracts of punch biopsy from cryosections after prostatectomy
Ren [43]	2016	Training set:25 paired PCa and adjacent benign Validation set:51 paired Pca and adjacent benign 16 BPH	LC-MS	Diagnosis	sphingosine, citicoline, choline, pantothenic acid, carnitine C4-OH, GPC, NAD, phenylacetyl-glutamine, carnitine C14:2	Yes	PCA PLS-DA Signrank Wilcoxon signed rank two-sides test (biomarker analysis)	joint transciptomics and metabolomics to identify altered metabolic pathways in PCA tissue;tissue extracts from prostatectomies
Liu [46]	2015	n = 42 tissue samples benign adjacent: 16 localized Pca: 12 metastatic Pca: 14	n.a.	Diagnosis	Proline, Cholesterol, sarcosine, spermidine, spermine, Putrescine, 4-Acetamidobutanoate	Yes	DRW-GM + logistic regression	joint analysis of genomic and metabolomic data and pathway topology using directed random walk on a global gene-metabolite pathway graph;used dataset established by Sreekumar et al. 2009
Priolo [48]	2014	Discovery set: Pca: 61; benign: 25 Validation set: Pca: 40; benign: 16	UHPLC-MS/MS GC-MS	Diagnosis Tumour subtyping	MYC-driven Pca:Oleic acid, arachidonic acid, docosahexaenoic acidsAKT1-driven Pca:2-aminoadipic acid, creatine	Yes	Mann–Whitney test	metabolomic profiling of tumors driven by MYC and AKT1 oncogenes;extracts of frozen tissue from prostatectomy;metabolomic profiling in cell lines, mice and human tissue; validation of selected markers in human tissue samples
Keshari [58]	2011	n = 49 tissue samples high-grade Pca: 13 low-grade Pca: 22 benign: 14	1-D and 2-D HR-MAS Spectroscopy	Staging	Benign vs. Pca tissue: choline, phosphocholine, glycerophosphocholine, phosphoethanolamine, glycerophosphoethanolamine, citrate, polyamineslow-grade vs. high-grade Pca:phosphocholine, glycerophosphocholine	No	Student’s t-test	intact tissue from core biopsies after prostatectomy;targeted analysis of phospholipid metabolites
Shuster [45]	2011	Pca: 14 benign: 14	GC-MSLC-MS/MS	Diagnosis	cysteine, dihomo-linoleate, docosapentaenoate, N-acetylaspartate, N-acetylglucosamine, uracil, xanthine, and 1-stearoylglycerophophoinositol;uracil, kynurenine, glycerol-3-phosphate, leucine, proline; choline, lactate, alanine citrate, putrescine, spermidine, spermine	No	matched paired t-test	tissue extracts from needle biopsies after prostatectomy;description of the mPREF methodreplicated metabolites previously published by Sreekumar et al. 2009 and various in vivo studies
Zhang [59]	2014	untreated patients: benign: 58; indolent Pca: 5; aggressive Pca: 8 radiation-treated patients: benign: 32; indolent Pca: 7 (relapse); aggressive Pca: 12 (relapse)	1-D and 2-D HR-MAS Spectroscopy	Staging Diagnosis	aggressive vs. indolent Pca:choline, phosphocholine, glycerophosphocholine, [choline + phosphocholine + glycerophosphocholine] to creatine ratio, (lactate; only in untreated)benign vs. Pca (untreated):citrate, polyamines, lactate, glutamate, alanine	No	linear mixed-effects model Wilcoxon Rank Sum Test Kruskal–Wallis Test	intact tissue from core biopsies
Giskeødegård [52]	2013	n = 158 tissue samples from 48 subjects benign: 47v low-grade Pca: 30 high-grade Pca: 81	HR-MAS 1H MRS	Staging	Pca vs. benign: citrate, taurine, creatine, glycerohpophocholine, phosphocholine, choline, glycinelow-grade vs. high-grade Pca:spermine, citrate, CCP/C ratio	No	Linear mixed models PLS-DA models	intact tissue from biopsies
Jentzmik [56]	2011	n = 92 Matched PCa and adjacent benign tissue	GC-MS	Staging	sarcosine	No	Mann–Whitney U test Wilcoxon test Spearman rank correlation Kaplan–Meier curve Cox proportional hazards regression analysis log rank test ROC analysis	target analysis of sarcosine as biomarker for disease progression tissue extracts from punch biopsies of tissue sections collected after prostatectomy
Thysell [29]	2010	Discovery set: bone metastases: 14 (hormone-naive Pca: 7; CRPC 7) adjacent normal bone: 10 Validation set:bone metastases: 13(6 Pca, 7 other cancers) normal bone: 11 Primary tumour: with metastases: 7 w/o metastases: 6benign: 17	GC-TOFMS	Staging	Bone tissue: metastases vs. normal: Cholesterol, myo-inositol-1-phosphate, citrate, fumarate, glycerol-3-phosphate, amino aicds Primary tumour: metastatic PCa vs. benign tissue and non-metastatic PCa:malate, dehydroascrobic acid, urea, hypoxanthine, asparagine, threonine, fumarate, linoleic acid	Yes	OPLS-DA Mann–Whitney U-test	tissue extracts from fresh-frozen biopsies of bone metastases and from biopsies of primary Pca and benign prostateall patients were selected to have hihg-risk tumours (i.e., presence of bone metastases, locally advanced tumour or poorly differnitated cancer)validation set available only for bone metastatic tissueblood plasma samples from men who underwent prostate biopsies were also analyzed

ADT: androgen deprivation therapy; ANOVA: one-way analysis of variance; BPH: benign prostate hyperplasia; GC: gas chromatography; (HP/UP)LC: (high performance/ultra performance) liquid chromatography; MALDI-FTICR: matrix-assisted laser desorption/ionization Fourier-transform ion cyclotron resonance mass spectrometry imaging; MRS: magnetic resonance spectroscopy; MS: mass spectrometry; MS/MS tandem mass spectrometry; (HR-MAS) NMR: (high resolution - magic angle spinning) nuclear magnetic resonance; OPLS-DA: orthogonal projections to latent structures-discriminant analysis; (Q)TOF: (quadrupole) time of flight; PCa: prostate cancer; PCA: principal component analysis; PLS(R)-DA: partial least squares (regression)-discriminant analysis; ROC: receiver-operating characteristic.