Role of Urinary Biomarkers in the Diagnosis of Adenoma and Colorectal Cancer: A Systematic Review and Meta-Analysis.

The growing interest in enhancing and spreading colorectal cancer (CRC) screening has been stimulating the exploration of novel biomarkers with greater sensitivity and specificity than immunochemical faecal occult blood test (iFOBT). The present study provides i) a systematic review of the urinary biomarkers that have been tested to achieve early CRC diagnosis and assess the risk of colorectal adenoma and adenocarcinoma, and ii) a meta-analysis of the data regarding the urinary prostaglandin (PG) metabolite PGE-M. As regard to gene markers, we found significantly different percent methylation of the vimentin gene in CRC patients and healthy controls (HC) (p<0.0001). Respect to metabolism of nitrogenous bases, cytidine, 1-methyladenosine, and adenosine, have higher concentrations in CRC patients than in HC (respectively, p<0.01, p=0.01, and p<0.01). As regard to spermine we found that N1,N12 diacetyl spermine (DiAcSpm) and N1, N8 diacetylspermidine (DiAcSpd) were significantly higher in CRC than in HC (respectively p=0.01 and p<0.01). Respect to PGE-M, levels were higher in CRC than in those with multiple polyposis (p<0.006) and HC subjects (p<0.0004). PGE-M seems to be the most interesting and promising urinary marker for CRC and adenoma risk assessment and for CRC screening. In conclusion, evidence suggests that urinary biomarker could have a potential role as urinary biomarkers in the diagnosis of colorectal cancer. Particularly, PGE-M seems to be the most promising urinary marker for CRC early detection.

Introduction

Colorectal cancer (CRC) is a highly common malignant tumour; it has been estimated that 1.13 million new CRC cases are diagnosed every year and that CRC causes about 694,000 deaths per year 1,2. CRC has become a major public health concern, because despite treatment as many as half of patients die from the disease 3. One reason for such high mortality rate is that CRC is a silent disease, since specific signs such as bleeding and abdominal pain often arise only in patients with advanced tumour stages 4; as a result, Dukes' stage C is found at diagnosis in half of the patients, whose 5 year survival is <50% 5-7. CRC is sporadic in 90% of patients; in <10% it is inherited 8,9 or is a complication of inflammatory bowel disease, either ulcerative colitis or Crohn's disease 10-12. Even though treatment approaches have considerably improved over the past decade, they have not exerted commensurate effects on outcomes or survival of patients with advanced disease. Moreover, they have induced a steep rise in treatment cost. Several modelling studies have demonstrated that a number of screening strategies are cost-saving 13. In the majority of cases, CRC develops from a preclinical benign precursor, adenoma; progression from early adenoma to invasive cancer takes years 14, 15. High incidence, protracted preclinical phase, recognisable and treatable precursor, high cost of treatment, and correlation of mortality with disease stage make CRC highly suitable for population screening 16, 17.

Faecal occult blood test (FOBT) has shown several limitations as an early diagnosis tool 18. The growing interest in enhancing and spreading CRC screening has been stimulating the exploration of novel biomarkers with greater sensitivity (SE) and specificity (SP) than the gold standard (immunochemical FOB test [iFOBT]) and capable of being detected in blood 19, stools and urine 20-22. Some studies have compared the values obtained with urinary and serum markers 23 and others have compared these with the FOBT 24, thus increasing their diffusion and the interest in developing laboratory techniques capable of detecting specific molecules. Several markers have been vetted, including nucleosides 25, the methylation level of specific genes 23,24, species deriving from oxidative stress 26, spermine concentration 27-29, and enzyme activities 30,31. In the past decade, urinary prostaglandins (PG) have been examined as possible early markers of high-risk adenoma (HRA) 32-35and adenocarcinoma 35, 36.

The present study provides: i) a systematic review of the urinary biomarkers that have been tested to achieve early CRC diagnosis and assess the risk of colorectal adenoma and adenocarcinoma, and ii) a meta-analysis of the data regarding the urinary prostaglandin (PG) metabolite PGE-M.

Material and methods

Relevant studies were identified using MEDLINE, EMBASE, Scopus, Clinicaltrial.gov, Web of Science, and Cochrane Library searches up to August 2015. The search strategy used the following terms: Urinary Biomarkers AND Colorectal Cancer; Urinary Biomarkers AND Colorectal Tumour; Urinary Biomarkers AND Colorectal Adenoma; Biomarkers AND Urine AND Colorectal cancer; Biomarkers AND Urine AND Colorectal tumour; Biomarkers AND Urine AND Colorectal neoplasm; Biomarkers AND Urine AND Colorectal adenoma. Papers were selected using the PRISMA checklist 37. A manual search was also made. Only studies in English published in the previous 10 years were included. The references of all studies were screened by two independent reviewers (PMA and GL); a methodologist (EA) resolved any disagreements. Inclusion criteria were adult clinical studies and investigations of urinary biomarkers of CRC, including case-control, nested case-control, and early detection studies. Exclusion criteria were in vitro studies, studies validating laboratory techniques, prognostic studies and studies published as proceedings.

The general characteristics of the papers thus selected are summarised in table 1. The odds ratio (OR), reported in the studies, are listed in table 2. The SE, SP, positive (PPV) and negative predictive value (PNV), and receiver operating characteristic (ROC) curves of all urinary markers examined are shown in table 3.

Table 1

Characteristics of included studies

ReferenceYear	Biomarkers and Laboratory		Study Population	Differences between groups	Epidemiological analysis
		Sample Size	Characteristics of groups
GENETIC MARKERS
Amiot et al(2014)23	VimentinMethylation	247	Diagnostic Early detection: two groups	Vimentin % OF METHYLATION HC 0% vs CRC 8%(p=0.57)	Specificity
	ALX-4 Methylation		No. of patients by group, sex, mean age and SD (y)	ALX 4 % OF METHYLATIONHC 0% vs CRC 15%(p=0.33)	Sensibility
	Wif-1 Methylation		Group 1: CRC: No.= 90, 53 M, 37 F, 56.6 ± 11.3 (y)	Wif 1% OF METHYLATIONHC 1.3% vs CRC 10.5(p<0.001)	ROC curve
	Laboratory:PCR		Group 2: HC: No.= 157, 76 M, 81 F, 68.3 ± 10.5 (y)		(see table 3 for details)
Song et al(2012)75	VimentinMethylation	40	Case control: two groups	Vimentin % OF METHYLATION	-
	Laboratory:MethyLight PCR		No. of patients by group, sex, mean age (y)	HC 10% vs CRC 75%(p<0.0001)
			Group 1: CRC: No.= 20, 14 M, 6 F, 60.45 (y)
			Group 2: HC No.= 20, 8 M, 12 F, 73.8 (y)
Xiao et al(2014)22	NDRG4Methylation	103	Diagnostic Early detection: two groups	NDRG4% OF METHYLATION	Specificity
	Laboratory:Nested methylation-specific polymerase chain reaction		No. of patients by group, sex, median age and range (y)	72.4% of positive methylated gene	Sensibility
			Group 1: CRC: No.= 87, 52 M, 35 F, 56 (38-73) y		ROC curve
			Group 2: HC: No.= 16, 9 M, 6 F, 55.6 (40-74) (y)		(see table 3 for detail)
METABOLITES OF NITROGENOUS BASES
Rozalski et al(2015)76		143	Diagnostic Early detection: three groups		Specificity
			No. of patients by group, sex, median age (y)		Sensibility
	8-oxo-7,8-dihydro-2'-deoxyguanosine		Group 1: AdenomasNo.=15, 8 M 7 F, 66 (y)	Mean levels of 8-oxoGua ± SD(nmol/mmol creatinine)CRC: 8.25 ± 10.22HC: 4.58 ± 3.55 (p=0.00017)	ROC curve
	8-oxo-7,8-dihydroguanine		Group 2: CRCNo.= 56, 32 M 24 F, 65 (y)	Mean levels of 8-oxoGuo ± SDCRC: 2.47 ± 1.31HC: 1.79 ± 0.58 (p=0.008)	(see table 3 for detail)
	5-hydroxymethyluracil		Group 3: HCNo. 72, 30 M 42F, 54 (y)	Mean levels of 5-hmUra± SDCRC: 5.61± 1.82HC: 6.65 ± 1.75 (p=0.005)
	Laboratory:HPLCGC/MS
Hsu et al(2013)24		71	Diagnostic Early detection: two groups	Nucleoside concentration in urine (nmol/ creatinine) mean and SD
			No. of patients by group, sex, mean age and SD (y)		Specificity
	Cytidine		Group 1: CRC: No.= 26, 12 M, 14 F, 65.30 ± 14.04 (y)	CytidineHC 1.25 ±0.83 vs CRC 2.08 ± 1.35 (p<0.01)	Sensibility
	3-Methylcytidine		Group 2: Breast cancer: No.= 36, all females, age not reported	3-MethylcytidineHC 0.81 ±0.27 vs CRC 0.86 ± 0.55 (p=0.55)	(see table 2 for detail)
	1-Methyladenosine		Group 3: Lung cancer: No.= 31, 16 M, 11 F, 60.68 ± 10.58 (y)	1-methyladenosineHC 1.25 ±0.83 vs CRC 9.71 ± 5.42 (p=0.01)
	2-Deoxyguanosine		Group 4: HC No.= 18, sex not reported, age not reported	2-DeoxyguanosineHC 0.15 ±0.13 vs CRC 0.15 ± 0.19 (p=0.45)
	Adenosine			AdenosineHC 1.25 ±0.83 vs CRC 5.97 ± 3.4(p<0.01)
	Inosine			InosineHC 0.3 ±0.32 vs CRC 0.3 ± 0.33 (p=0.78)
	Laboratory:HPLC MS
Cheng et al(2012)105	Citrate	204	Diagnostic Early detection: two groups	Not reported	ROC curve
	Hippurate		No. of patients by sex, median age and range (y)		(see table 3 for detail)
	P-cresol		Group 1: CRC: No.= 101, 58 M , 43 F, 60 (24-83) (y)
	2-aminobutyrate		Group 2 : HC: No.=103, 31 M, 72 F, 58 (31-76) (y)
	Myristate
	Putrescine
	Kynurenate
	Laboratory:GT-TOFL
Feng et al(2005)77		114	Diagnostic Early detection: two groups	Nucleoside concentration in urine (nmol nucleoside/ μmol creatinine) median [IQR]
	Adenosine		No. of patients for group, sex, median age and range (y)	AdenosineHC 0.51 [0.20] vs CRC 0.66 [0.72)](p<0.01)	Sensivity
	Pseudouridine			PseudouridineHC 22.74 [6.86] vs CRC 39.63 [31.09] (p<0.01)	Sensibility
	N4-acetylcytidine		Group 1: CRC: No.= 52, 27 M, 25 F, 63 (26-87) (y)	N4-acetylcytidineHC 0.66 [0.23] vs CRC 0.85 [0.41](p<0.01)	ROC curve
	Cytidine		Group 2: HC: No.= 62, 33 M, 29 F, 59 (24-78) (y)	CytidineHC 0.14 [0.14] vs CRC 0.3 [0.25](p<0.01)	(see table 3 for detail)
	Guanosine			GuanosineHC 0.09 [0.04] vs CRC 1.37 [0.06](p=0.268)
	Inosine			InosineHC 0.29 [0.18] vs CRC 0.43 [0.27](p<0.01)
	1-methylinosine			1-methylinosineHC 1.19 [0.37] vs CRC 2.16 [1.57](p<0.01)
	1-methladenosine			1-methyladenosineHC 2.03 [6.86] vs CRC 2.76 [1.18](p<0.01)
	1-methylguanosine			1-methylguanosineHC 0.79 [0.29] vs CRC 1.37 [0.70](p<0.01)
	2,2-methylguanosine			2,2-methylguanosine HC 1.23 [0.37] vs CRC 1.71 [0.77](p<0.01)
	2-methylguanosine			2--methylguanosineHC 1.25 [0.23] vs CRC 1.11 [0.55](p<0.01)
	N6-methyladenosine			N6-methylguanosineHC 0.03 [0.03] vs CRC 0.04 [0.77](p=0.023)
	3 methyluridine +5-methyluridine,			3-methyluridine +5-methyluridineHC 0.01 [0.07] vs CRC 0.11 [0.09](p<0.01)
	Uridine			UridineHC 0.89 [0.18] vs CRC 0.29 [0.18](p=0.869)
	(**)
Zheng et al(2005)78		112	Two groups	Nucleoside concentration in urine (nmol nucleoside/ μmol creatinine)Mean, SD
			No. of patients for group, sex, median age and range (y)
	Adenosine		Group 1: CRC: No.= 52, 29 M, 23 F, 60 (26-87) (y)	AdenosineHC 0.52 ± 0.16 vs CRC 0.66 ± 0.30(p<0.05)
	Pseudouridine		Group 2: HC: No.= 60, 31 M, 29 F, 52 (21-71) (y)	PseudouridineHC 22.08 ± 5.11 vs CRC 42.19 ± 22.25 (p<0.05)
	N4-acetylcytidine			N4-acetylcytidineHC 0.69 ± 0.20 vs CRC 0.84 ± 0.30(p<0.05)
	Cytidine			CytidineHC 0.15 ± 0.12 vs CRC 0.43 ± 0.49(p<0.05)
	Guanosine			GuanosineHC 0.09± 0.03 vs CRC 0.10 ± 0.04p= Not statistically significant
	Inosine			InosineHC 0.28 ± 0.11 vs CRC 0.50 ± 0.35(p<0.05)
	1-Methladenosine			1-MethyladenosineHC 2.04 ± 0.53 vs CRC 2.74 ± 0.80(p<0.05)
	1-Methylguanosine			1-MethylguanosineHC 0.82 ± 0.24 vs CRC 1.44 ± 0.51(p<0.05)
	1-Methylinosine			1-MethylinosineHC 1.25 ± 0.28 vs CRC 2.76 ± 1.94(p<0.05)
	N2-methylguanosine			N2-methylguanosineHC 0.55 ± 0.14 vs CRC 0.63 ± 0.26(p<0.05)
	N6-Methyladenosine			N6-MethyladenosineHC 0.04 ± 0.02 vs CRC 0.07 ± 0.05(p<0.05)
	2,2-Methylguanosine			2,2-methylguanosineHC 1.25 ± 0.23 vs CRC 1.81 ± 0.55(p<0.05)
	3 methyluridine +5-methyluridine			3 methyluridine +5-methyluridineHC 0.04 ± 0.06 vs CRC 0.13 ± 0.08(p<0.05)
	Uridine			UridineHC 0.30 ± 0.15 vs CRC 0.31 ± 0.23p= Not statistically significant
	Laboratory:HPLC
EZIMATIC ACTIVITY MARKERS
Niu et al(2012)29	Arylsulfatase	516	Diagnostic Early detection: three groups	Arylsulfatase activity (U/L, 1U=1μmol/min)mean, SD	Specificity
	Laboratory:Enzimatic Activity		No. of patients by sex, median age, range (y)	Male Group:	Sensibility
			Group 1: CRC: No.= 119, 71 M, 63 (30-82) (y), 48 F, 66 (34-82) (y)	HC 0.17 ± 0.09	VVP
			Group 2: Benign tumor: No.= 97, 49 M, 57 (34-88) y, 48 F, 53 (30-82) (y)	Benign tumor 0.18 ± 0.11	VPN
			Group 3: HC: No.= 300, 137 M, 53 (30-85) (y), 163 F, 54 (30-85) (y)	CRC 0.45 ± 0.25	ROC curve
				comparison between groups(p <0.0001)	(see table 3 for detail)
				Female Group
				HC 0.12 ± 0.07
				Benign tumor 0.19 ± 0.11
				CRC 0.45 ± 0.35(p <0.0001 for both groups)
Szajda et al(2009)30	Hex	58	Diagnostic Early detection: three groups	Enzymatic activity (pKat/mL)Mean and DS	Specificity
	HexA		No. of patients by sex, mean age ,range and SD (y)		Sensibility
	HexB		Group 1: Total CRC subjects: No.= 38, 16 M, 22 F, 68.18 ± 14.63 (y) divided into two groups		ROC curve
			Adenocarcinoma: No.= 32, sex not reported, age not reported	Control Group	(see table 3 for detail)
				HEX 143.26 ± 58.86
				HEX A 51.54 ± 34.45
				HEX B 91.73 ± 34.09
			Adenocarcinoma mucinosum: No.= 6, sex not reported, age not reported	(p=0.00643)
			Group 2: HC: No.= 20, 11 M, 9 F, 63.35 (51-78) ± 10.74 (y)
				Adenocarcinoma group
				HEX 376.55 ± 277.52
				HEX A 185.55 ± 133.58
				HEX B 191.70 ± 173.05
				(p=0.86005)
				Adenocarcinoma mucinosum
				HEX 187.30 ± 136.14
				HEX A 94.90 86.85
				HEX B 92.40 ± 43.95
				(p=0.87334)
CEA
El-Masry et al(2007)79	CEA	298	Eight groups	Not Reported differences between groups	Sensibility
			No. of patients by group, sex, mean,SD (y)		PPV
			Group 1: Liver: No.= 20, 12 M, 8 F, 56.55 ± 6.85 (y)		NPV
			Group 2: Ampulla of Vater: No.= 11 M, 12 F, 55.83 ± 7.67 (y)		(see table 3 for detail)
			Group 3: Biliary tract (Hilar): No.= 9 M, 6 F, 51.67 ± 8.18 (y)
			Group 4: Pancreatic: No.= 38 M, 20 F, 54.33 ± 8.6 (y)
			Group 5: Gastric: No.= 17 M, 11 F, 53.04 ± 11.01 (y)
			Group 6: Esophagus: No.= 14, 9 M, 5 F, 57.79 ± 12.01 (y)
			Group 7: Colorectal: No.= 90, 55 M, 35 F, 50.8 ± 11.36 (y)
			Group 8: HC: No.= 50, 38 M, 12 F, 52 ± 8.1 (y)
Umemori et al(2010)26		107	Diagnostic Early detection: three groups
			No. of patients by group, sex, median age and [IQR](y)	concentration (nmol/ g creatinine) mean and SD	Sensibility
	N¹,N¹² - diacetylspermine		Group 1: CRC: No.= 33, 17 M, 16 F, 69.2 [56-84] (y)	N¹,N¹² - diacetylspermineHC 185 ±056 significant lower than cancer groups (p<0.01)	(see table 3 for detail)
	N¹,N8 - diacetylspermidine		Group 2: Breast cancer : No.= 28, 1 M, 27 F, 61.3 [43-84] (y)	N¹,N8 - diacetylspermidineHC 522 ±.167 significant lower than cancer groups (p=0.01)
	Laboratory:ELISA		Group 3: HC: No.= 46, 24 M, 22 F, 68.8 [46-86] (y)
Hiramatsu et al(2005)27	N¹,N¹² - diacetylspermine	434	Diagnostic Early detection : four groups	Not Reported differences between groups	Specificity
	Laboratory:ELISA		No. of patients by group, sex, range (y)		Sensibility
			Group1: CRC: No.= 248, sex not reported, age not reported		(see table 3 for detail)
			Group 2: Breast cancer: No.= 83, sex not reported, age not reported
			Group 3: Bening gastrointestinal disease: No.= 51, sex not reported, age not reported
			Group 4: HC: No.= 52, 27 M, 25 F, (22-52) (y)
OXIGEN REACTIVE SPECIES
			Three groups
			No. of patients by group, sex, mean,SD	Metabolites µmol/l mean, SD
Chandramathi et al(2008)25	AOPP	245	Group 1: CRC: No.= 49, 30 M, 19 F, 60 ± 11 (y)	AOPPMalignant colorectal cancer 179.39±5.43 vs breast cancer 136.90±5.43 (p<0.05) vs HC 106.34±4.13 (p<0.001)
	MDA		Group 2: Breast cancer : No.= 101, 0 M, 101 F, 56 ± 12 (y)	MDAMalignant colorectal cancer 0.781± 0.0072 vs HC 0.579±0.029 (p<0.01)
	H2O2		Group 3: HC: No.= 95, 37 M, 58 F, 57 ± 11 (y)	H2O2Malignant colorectal cancer 23.23± 2.16 vs breast cancer 17.65±1.18 (p<0.05) vs HC 17.28±0.83 (p<0.01)
	FRAP			FRAPMalignant colorectal cancer 1530.20± 97.66 vs HC 17.28±0.83 (p<0.01)
PROSTAGLANDINES
Davenportet al(2015)31	PGE M	1102	Case control: three groups, all females		OR with C-reactive protein(see table 2 for detail)
	Laboratory:Liquid ChromatographyMass spectrometry		No. of patients by group, mean age and SD (y)	Mean ± SD levels of PGE-M (ng/g creatinine)
			Total subjects: No.= 420, divided into two groups:
			Group 1: SSTANo.= 226, 57.5 ± 7.1 (y)	SSTA 2.4 ± 0.8
			Group 2: MSTANo.= 198, 60 ± 6.5 (y)	MSTA 2.5 ± 0.7
			Group 2: AANo.= 283, 58 ± 7.3 (y)	AA 2,6 ± 0,7
			Group 4: HC: No.= 395 57,6 ±7,1 (y)	HC: 2.3 ± 0,8
Bezawadaet al(2014)33	PGE M	840	Case control: three groups, all females	PGE-M (ng/g creatinine)	OR(See table 2 for detail)
	Laboratory:Liquid ChromatographyMass spectrometry		No. of patients by group, mean age and SD (y)
			Total Colorectal Adenoma subjects:No.= 420, divided into two groups:	Other findings:PGE-M rises with years (p=0.002)And smoking (p=0.02)
			Group 1: Low risk adenoma (<1 cm, tubular):No.= 130, 66.2 ± 6.5 (y)	Low Risk: 5.01 [3.71-7-09]
			Group 2: High risk adenoma (≥1 cm, tubulovillous, villous or high grade dysplasia):No.= 290, 66.8 ± 6.6 (y)	High Risk: 6.26 [3.89-8-61]
			Group 3: HC: No.= 420, 66.7± 6.6 (y)	HC: 5.57 [3.50-7-64]
				Differences among groups: p<0.001
Shrubsoleet al(2012)32	PGE M	1040	Case Control: four groups	PGE-M (PGE-M ng/g creatinine)Median,[IQR]	OR
	Laboratory:Liquid ChromatographyMass spectrometry		No. of patients by group, sex, mean age, range and SD (y)		(See table 2 for detail)
			Group 1: AA: No.= 224, 159 M, 65 F59.2 (40-75) ± 7.1 (y)	AA: 12.6 [8.4-21.4] (p<0.001)
			Group 2: MSTA No.=152,120 M, 32 F, 58.9 (40-75) ± 6.7 (y)	MSTA: 13.2[7.9-21.9] (p<0.001)
			Group 3: SST: No.=300,209 M, 91 F, 57.8 (40-75) ± 7.3 (y)	MSTA: 10.1[5.7-17.1] (p=39)
			Group 4: HC: No.= 364, 236 M, 128 F57.6 (40-75) ± 7.1 (y)	HC: 10.1 [5.7-17.1]
				Other findings:Levels of PGE-M are associated with:level of education (p=0.001) caucasian race (p=0.02), academic medical center (p<0.001), current use of NSAID (p=0.03). current alcohol drinking (p=0.002), cigarette smoking (p<0.001), BMI (p=0.03), red meat intake (p<0.001), physically activity in last 10 years (p=0.002)
Johnsonet al(2006)34	PGE M	228	Diagnostic Early detection: five groups	PGE-M (PGE-M ng/g creatinine)Median, [IQR]	Specificity
	Laboratory:Liquid ChromatographyMass spectrometry		No. of patients by group, sex, mean age, SD (y)		Sensibility
			Group 1: CRC: No.= 58, M 32, F 26, 60.9 ± 10 (y)	CRC 15 [9.11 - 26.9]	ROC curve
			Group 2: Large or multiple polyps: No.= 44, 33 M, 11 F, 59.6 ± 11.5 (y)	Large or multiple polyps 15.6 [7.69-22.9]	(see table 3 for detail)
			Group 3: Small polyps: No.= 26,18 M, 8 F, 60.5 ± 7.2 (y)	Small polyps 9.69 [6.54-20.4]
			Group 4: Crohn disease No.= 28, 13 M, 15 F, 38.8 ± 11.7 (y)	Crohn disease 21.9 [17.4-49.1]
			Group 5: HC: No.= 72, 53 M, 19 F, 60.9 ± 7.5 (y)	HC 7.17 [4.69-15.9]
				In CRC PGE-M higher than large or multiple polyps (p<0.006) and HC (p<0.0004)
				In CRC and large or multiple polyps PGE-M higher than small polyps and HC (p<0.0001)
Cai et al(2006)35	PGE M	300	Case Control (nested): three groups, all females	PGE-M (PGE-M ng/g creatinine)Mean, [IQR]	RR(see table 2 for detail)
	Laboratory:Liquid ChromatographyMass spectrometry		No. of patients by group, mean age, range and SD (y)	CRC (all groups) : 11.6 [6.4-18.9]
			Total CRC subjects : No.= 150, 60.3 (40-70) ± 8.3 (y), divided into	HC: 7.0[3.9-14.0] (p<0.001)
			Group 1: CC: No.= 88, age not reported	Colon cancer: 12.0[7.1-19.1] CI 95% 9.7-13.7Vs HC (p=0.002)
			Group 2: RR No.= 62, age not reported	Rectal cancer: 11.0 [5.6-18.3] CI 95% 7.4-12.6Vs HC (p=0.49)
			Group 3: HC: No.= 150, 60.1 (40-70) ± 8.5 (y)

LEGEND: [IQR] interquartile range, SD standard deviation, HC healthy controls

AA: Advanced adenoma, CRC: colorectal cancer, SSTA: single small tubular adenoma, MMSA: Multiple small tubular adenoma, CC: colon cancer, RC: rectal cancer

Table 2

Early Detection

Reference, Year		Biomarkers					Parameters evaluated					Relation to diagnostic routine assays
Rozalskiet al(2015)76	Urine			SE	SP	PPV	PNV	ROC Curve	8-oxoGua in combination with 8-oxodGuo and 5-hmUra
	8-oxodGuo			33.9	94.4			0.63 (0.53-0.73)	SE	SP	ROC Curve
	8-oxoGua			78.6	62.5			0.69 (0.60-0.78)	78.6	75.0	0.77(0.69-0.89)
	5-hmUra			48.2	79.2			0.66 (0.57-0.76)
	(8-oxoGua+8oxodGuo)/5-hmUra			78.6	75.0			0.77 (0.69-0.86)
Amiotet al(2014)23	Urine								Wif-1 methylation in combination with fecal occult blood test
	Wif-1			27.0	99.0			0.90 (0.84-0.94)			ROC Curve
	ALX-4			15.0	100						0.91 (0.87-0.95)
	Vimentin			8.0	100
	Stool								Wif-1 methylation combined in urine or serum samples
	Wif-1			19.0	99.0				SE	SP	ROC curve
	ALX 4			11.0	99.0				48.8	99	0.90 (0.84-0-94)
	Vimentin			33.0	100
	Blood
	Wif-1			33.0	99.0
	ALX-4			23.0	100.0
	Vimentin			4.0	100
Xiaoet al(2014)22	UrineMethylation NRDG4			72.6			85
	TissueMethylation NRDG4			81.0			91.7
	ParacarcinomaMethylation NRDG4			8.3
	Blood Methylation NRDG4			54.8			78.1
	FecesMethylation NRDG4			76.2			89.1
Hsuet al(2013)24	Urine								In combination with serum CEA
	Adenosine			69.0					81.0
	Cytidine			35.0
	3-methylcytidine
	1-methyladenosine			27.0
	Inosine
	2-deoxyguanine
	Set of six nucleosides			69.0					85.0
Niuet al(2012)29	UrineArylsulfatase		M	63.4	96.4	90.0	83.5	0.89(0.83-0.95)
			F	81.3	95.1	83.0	94.5	0.87 (0.79-0.94)
	Accuracy		M	85.1
			F	91.9
Chenget al(2012)105	Urine
	Citrate							0.72
	Hippurate							0.92
	P-cresol							0.80
	2-aminobutyrate							0.75
	Myristate							0.84
	Putrescine							0.70
	Kynurenate							0.88
	Total panel			97.5	97.6			0.99 (0.97-1.00)
Umemori et al(2010)26	Urine								DiAcSpm in combination with
	N¹,N¹²-diacetylspermine (DiAcspm)			69.6						SE
	N¹,N8-diacetylspermidine (DiAcSpd)			36.3					DiacSpd	75.7
	Blood								CEA	78.7
	CEA			46.8					CA 19-9	78.7
	CA 19-9			15.6
Hiramatsu et al(2005) 27	Urine								DiAcSpm in combination with
	N1,N12 - diacetylspermine (DiAcSpm)Blood			75.8	96				CEA	39.5
	CEA								CA 19-9	14.1
	CA 19-9
Szajdaet al(2009)30	Urine
	Hex			78.9	72.7			0.84
	HexA			73.6	63.9			0.67
	HexB			84.2	90.9			0.89
El-Masry et al(2007)79	Urine
	CEA (ELISA)			64.4		79.5	52.2
	CEA (SB-ELISA)			76.7		86.3	65.0
Johnsonet al(2006)34	PGE-M¹		A	90.0	45.0			0.64
			M	88.0	35.0			0.54
			F	92.0	78.0			0.84
	PGE-M²		A	88.0	53.0			0.70
			M	92.0	46.0			0.67
			F	83.0	86.0			0.84

LEGEND: PGE-M¹: Cancer vs non cancer. PGE-M²: Cancer/large polyps small/no polyps. DiAcSpm: N¹,N¹²-diacetylspermine. DiAcSpd: N¹,N8-diacetylspermidine.

8-oxodGuo: 8-oxo-7,8-dihydro-2'-deoxyguanosine. 8-oxoGua: 8-oxo-7,8-dihydroguanine. 5-hmUra: 5-hydroxymethyluracil. A: All population, M: male, F: female

Table 3

Prostaglandines: case control and case control nested studies

Reference, year		PGE-M(number of cases vs controls)								p-value
		Low PGE-M (<9.99 mg/mg creatinine)				High PGE-M (>9.99 mg/mg creatinine)
		CRP T1	CRP T2		CRP T3	CRP T1	CRP T2		CRP T3
Davenport et al(2015) 31	Single small tubular	30 vs 34	43 vs 43		35 vs 41	29 vs 38	52 vs 38		54 vs 40	0.08
	OR (95% CI)	1*	0.78 (0.34-1.76)		0.46 (0.18-1.13)	0.66 (0.25-1.69)	0.99 (0.42-2.34)		0.99 (0.44-2.29)
	Multiple small tubular	8 vs 34	13 vs 43		17 vs 41	14 vs 38	24 vs 38		29 vs 40	0.90
	OR (95% CI)	1*	1.48 (0.35-6.37)		1.39 (0.41-4.73)	1.10 (0.27-4.53)	1.98 (0.52-7.56)		3.00 (0.70-12.86)
	Advanced	13 vs 34	21 vs 43		21 vs 41	21 vs 38	39 vs 38		51 vs 40
	OR (95% CI)	1*	1.96 (0.61-6.32)		1.29 (0.43-3.86)	1.35 (0.40-4.51)	4.16 (1.32-13.11)		5.31 (1.55-18.18)	0.08
	Multiple or small tubular or advanced	21 vs 34	34 vs 43		38 vs 41	35 vs 38	63/38		80 vs 40	0.25
	OR (95% CI)	1*	1.67 (0.65-4.29)		1.23 (0.51-2.96)	1.22 (0.47-3.17)	3.04 (1.21-7.64)		3.72 (1.42-9.72)
		Quartiles PGE-M(number of cases vs controls)
Bezawadaet al(2014)33		Q1		Q2		Q3		Q4
	All Adenoma	84 vs 105		117 vs 105		94 vs 103		125 vs 107
	OR (95% CI)	1*		1.38 (0.90-2.10)		1.13 (0.73-1.73)		1.40 (0.92-2.14)		0.26
	Low Risk Adenoma	27 vs 105		48 vs 105		28 vs 103		27 vs 107
	OR (95% CI)	1*		1.63 (0.92-2.89)		1.02 (0.55-1.90)		0.91 (0.48-1.72)		0.30
	High risk adenoma	57 vs 105		69 vs 105		66 vs 103		98 vs 107
	OR (95% CI)	1*		1.23 (0.75-2.0)		1.19 (0.73-1.94)		1.66 (1.04-2.67)		0.04
	Non regular aspirin or NSAID users	21 vs 38		33 vs 42		36 vs 45		48 vs 31
	OR (95% CI)	1*		2.01 (0.90-4.51)		1.69 (0.76-3.76)		2.08 (0.93-4.65)		0.18
	Regular aspirin or NSAID users	36 vs 67		36 vs 63		30 vs 58		50 vs 66
	OR (95% CI)	1*		0.86 (0.45-1.65)		0.89 (0.46-1.73)		1.41 (0.76-2.61)		0.18
	BMI ≥ 25	30 vs 47		42 vs 52		37 vs 51		58 vs 60
	OR (95% CI)	1*		1.21 (0.61-2.41)		1.10 (0.54-2.21)		1.36 (0.69-2.67)		0.43
	BMI ≤ 25	27 vs 58		27 vs 53		29 vs 52		40 vs 47
	OR (95% CI)	1*		1.11 (0.54-2.31)		1.43 (0.70-2.93)		2.07 (1.04-4.13)		0.02
	Past o current smokers	33 vs 52		36 vs 51		35 vs 54		47 vs 62
	OR (95% CI)	1*		1.28 (0.66-2.51)		1.15 (0.58-2.27)		1.35 (0.71-2.57)		0.45
	Never smokers	29 vs 53		32 vs 54		31 vs 49		50 vs 45
	OR (95% CI)	1*		1.22 (0.58-2.56)		1.40 (0.66-3.00)		2.23 (1.08-4.60)		0.02
	High calcium ≥ 1068 mg/dl	29 vs 54		33 vs 52		32 vs 49		48 vs 55
	OR (95% CI)	1*		1.28 (0.66-2.51)		1.15 (0.58-2.27)		1.35 (0.71-2.57)		0.09
	Low calcium ≤ 1068 mg/dl	28 vs 51		36 vs 53		34 vs 54		50 vs 52
	OR (95% CI)	1*		1.43 (0.70-2.91)		1.17 (0.57-2.40)		1.66 (0.82-3.38)		0.24
Shrubsole(2012)32	Any advanced adenoma	26 vs 90		51 vs 89		76 vs 90		69 vs 89
	OR (95% CI)a	1*		1.64 (0.9-2.95)		2.42 (1.37-4-28)		2.17 (1.20-3.92)		0.006
	OR (95% CI)b	1*		1.56 (0.84-2.90)		2.25 (1.23-4.09)		1.84 (0.97-3.48)		0.04
	Multiple small tubular adenoma	16 vs 90		36 vs 89		44 vs 90		52 vs 89
	OR (95% CI)a	1*		2.13 (1.02-4.44)		2.13 (1.03-4.40)		2.57 (1.24-5.34)		0.03
	OR (95% CI)b	1*		2.59 (1.20-5.60)		2.31(1.07-5.00)		2.88(1.32-6.24)		0.04
	Single small tubular adenoma	66 vs 90		80 vs 89		73 vs 90		79 vs 89
	OR (95% CI)a	1*		1.14 (0.72-1.82)		0.99 (0.61-1.61)		1.11 (0.67-1.82)		0.87
	OR (95% CI)b	1*		1.16 (0.72-1.86)		0.93 (0.56-1.53)		1.04 (0.62-1.74)		0.88
	Advanced or multiple adenoma	42 vs 90		87 vs 89		120 vs 90		121 vs 89
	OR (95% CI)a	1*		1.84 (1.11-3.05)		2.32 (1.41-3.81)		2.34 (1.41-3.87)		0.001
	OR (95% CI)	1*		1.95 (1.15-3.30)		2.31 (1.37-3.89)		2.19 (1.28-3.76)		0.008
	Male	24 vs 34		59 vs 55		97 vs 68		95 vs 76
	OR (95% CI)	1*		1.52 (0.75-3.09)		1.96 (1.01-3.83)		1.58 (0.81-3.10)		0.24
	Female	18 vs 56		28 vs 34		23 vs 22		26 vs 13
	OR (95% CI)	1*		2.71 (1.14-6.42)		2.33 (0.87-6.23)		5.40 (1.78-16.41)		0.006
	Never or former NSAID user	23 vs 46		42 vs 41		50 vs 39		62 vs 39
	OR (95% CI)	1*		1.91 (0.90-4.04)		2.19 (1.01-4.75)		2.29 (1.03-5.09)		0.06
	Current NSAID users	15 vs 43		34 vs 47		53 vs 50		46 vs 49
	OR (95% CI)	1*		2.36 (1.06-5.28)		2.96 (1.38-6.36)		2.51 (1.14-5.54)		0.04
	Nevers smokers	15 vs 55		32 vs 50		30 vs 42		30 vs 33
	OR (95% CI)	1*		1.90 (0.86-4.20)		1.91 (0.83-4.38)		2.44 (1.01-4.20)		0.03
	Former/ currentsmokers	23 vs 34		43 vs 38		73 vs 47		80 vs 56
	OR (95% CI)	1*		1.98 (0.94-4.16)		2.74 (1.36-5.54)		2.09 (1.04-4.20)		0.06
Cai et al(2006)35	All colorectal cancer	14 vs 37		28 vs 38		47 vs 37		61 vs 38
	RR (95% CI)	1*		2.5 (1.1-5.8)		4.5 (1.9-10.9)		5.6 (2.4-13.5)		<0.001
	Colon cancer	7 vs 21		15 vs 22		29 vs 20		37 vs 25
	RR (95% CI)	1*		2.1 (0.7-6.5)		4.8 (1.6-14.8)		4.9 (1.7-14.7)		0.009
	Rectalcancer	7 vs 16		13 vs 16		18 vs 17		24 vs 13
	RR (95% CI)	1*		3.1 (0.8-11.6)		4.1 (1.0-17.3)		7.2 (1.7-3.7)		0.048
	≤ 30 months	6 vs 16		12 vs 23		26 vs 18		31 vs 18
	RR (95% CI)	1*		2.3 (0.6-8.9)		5.4 (1.3-21.8)		7.6 (1.8-32.0)		0.035
	> 30 months	8 vs 21		16 vs 15		21 vs 19		30 vs 20
	RR (95% CI)	1*		2.9 (1.0-9.1)		3.6 (1.1-11.4)		4.4 (1.5-13.3)		0.012

*reference category

a: adjusted for age, gender, race, educational attainment and study site

b: additionally adjusted for cigarette smoking, alcohol consumption, BMI, red meat intake and NSAID use.

Among the urinary markers explored to date, PGE-M has been examined in five studies involving more than 1000 patients. These data were subjected to a meta-analysis. Data reported as medians and interquartile ranges were transformed into mean and standard deviation (SD) as described by Pudar Hozo et al 38. Participants were considered as healthy controls (HC) if adenoma/adenocarcinoma had been excluded by colonoscopy; as patients with low-risk adenoma (LRA) if they had a solitary adenoma <1cm in diameter; or as patients with high risk adenoma (HRA) if they had multiple lesions, adenomas >1cm in diameter, or lesions showing a high degree of dysplasia. Pools standardised mean differences were evaluated according to Cohen 39 and analysed with the random-effects model according to Der Simonian and Laird 40. Quality assessment of case-control studies was conducted with the Newcastle-Ottawa scale 41; the last item of outcome/exposure assessment was not used because it was not appropriate (Supplementary table S1).

Results

Search of the electronic databases according to the above-listed criteria found 282 papers; the manual search found 7 papers, totalling 289 papers. There were 25 duplicates (figure 1); 214 papers were excluded because 28 were conference posters, 54 were reviews, 84 regarded other subjects, 30 were animal, in vitro or laboratory technique studies, and 18 evaluated blood/serum/plasma or other tissues. Examination of the remaining 50 papers in the second stage of the PRISMA flow chart led to the exclusion of 31studies for the following reasons: 5 were reviews 1,19,42-44, one assessed animals 45, 5 involved laboratory technique validation 21, 46-49, one was a molecular biology study 50, 4 were survival analysis studies 51-54, 7 assessed the impact of dietary habits 57-61, 5 did not examine urine 62-66, one reported aggregate data for CRC, gastric cancer and pancreatic cancer 67, and finally 2 did not match the inclusion criteria 68,69.

Figure 1

Flow-chart of the search strategy

The general characteristics of these papers are reported in table 1. Since they assessed multiple biomarkers, they were grouped as follows.

Gene markers

A study of the methylation of vimentin, ALX-4, and Wif-1 genes 24 has found significant differences between CRC patients and HC only for the Wif-1gene (p<0.0001) (table 1). However, despite high SP values, SE was very low (table 2) 24.

Song and co-workers have described significantly different (p<0.0001) percent methylation of the vimentin gene in CRC patients and HC (table 1) 70.

As reported by Xiao and colleagues, NDRG4 methylation provided an SP that ranged from 78% (blood) to 91% (tissue), whereas SE ranged from 8% (paracarcinoma tissue) to 81% (carcinoma tissue) in table 2 23.

Metabolism of nitrogenous bases

Several studies have assessed nitrogenous bases as disease markers. Rozalski and co-workers investigated 8-oxo-7,8-dihydro-2'-deoxyguanosine, 8-oxo-7,8-dihydroguanine, and 5-hydroxymethyluracil. Group data and the main findings are reported in table 1, whereas SP and SE values are listed in table 3 71.

Hsu and colleagues examined cytidine, 1-methyladenosine, and adenosine, among other metabolites, and detected higher concentrations in CRC patients than in HC (respectively, p<0.01, p=0.01, and p<0.01; table 1). They also calculated SP and SE (table 2) 25. Cheng et al found no differences in nitrogen metabolites among the groups, but described differences in the ROC curves (table 2) 72. Feng and colleaguesmeasured13 metabolites in CRC patients and HC (table 1) and reported the ROC curve data (table 2) 73. Zheng and co-workers assessed 13 urinary metabolites in CRC patients and HC and found that they were consistently higher in patients (table 1) 74.

Markers of enzyme activity

Niu and colleagues investigated arylsulphatase (ARS) activity in patients with CRC, patients with adenoma, and HC. Data stratification by gender highlighted significant differences among groups (p<0.0001). The SP, SE, PPV, PNV and ROC curve data are reported in table 2 30.

Szajda and colleagues examined urinary hexosaminidase (HEX) activity in CRC patients and HC. They found higher HEX enzyme (HEX, HEX A and HEX B) activity in relation to Dukes' stages A (p=0.03668), C (p=0.02396) and D (P=0.00552). Moreover, the HEX B complex provided an SE of 84.2% and an SP of 90.9%, whereas the area under the curve (AUC) was 0.89. The other values for HEX A and the HEX complex are reported in table 2 31.

Carcinoembryonic antigen

El-Masry and colleagues measure urinary carcinoembryonic antigen (CEA) in different malignant gastrointestinal diseases. The relevant SE, PPV, and PNV values are listed in table 2 75.

Spermines

There are two papers on spermines 27, 28. Umemori and co-workers examined N1,N12diacetylspermine (DiAcSpm) and N1, N8 diacetylspermidine (DiAcSpd) in various groups of subjects (table 1) and found that both were significantly higher in cancer patients than in HC (respectively p=0.01 and p<0.01). They also calculated the SE of the two markers (table 2) 27.

Hiramatsu et al. investigated HC and patients with CRC, breast cancer (BC), and benign gastrointestinal disease (table 1), but only calculated the SE and SP values of urinary spermines (table 2) 28.

Reactive oxygen species

These were investigated only by Chandramathi and colleagues in four groups of subjects. CRC patients showed significantly higher levels than BC patients and control subjects (table 1) 26.

Prostaglandins

There were five papers assessing urinary PGE-M: four compared patients with adenoma and HC 32-35 and two compared cancer patients and HC 35,36. The general characteristics of the study populations and the differences among groups are reported in table 1; the SE and SP values are showed in table 2 and the risk estimates are reported in table 3.

Johnson and colleagues have assessed the possible role of PGE-M as an early diagnostic marker in five groups of subjects (table 1). PGE-M levels were higher in CRC patients than in those with multiple polyposis (p<0.006) and HC subjects (p<0.0004). Davenport and colleagues (table 3) devised a risk model where they considered urinary PGE-M as well as plasma C-reactive protein (CRP). Although none of the associations achieved statistical significance, the risk tended to increase in patients with poorly differentiated lesions on histopathology and higher levels of the two markers 35.

Bezawada and colleagues (table 3) found significant OR in HRA patients (p=0.04) having higher PGE-M levels compared with controls. In LRA patients, none of the risk estimates were significantly related to PGE-M levels (p=0.30). PGE-M values were not associated with a significant risk in smokers or in those who had stopped smoking(p=0.45). Adenoma patients with a body mass index (BMI) >25 (p=0.43) were not at significantly greater risk in relation to PGE-M levels compared with HC. In contrast, HRA patients with a BMI <25 were at greater risk than controls; the risk was double in case of higher PGE-M levels (p=0.02). Risk differences between cases and controls in relation to PGE-M levels were not affected by consumption of non-steroidal anti-inflammatory drugs (NSAID) (p=0.18) or by high or low calcium values (respectively p=0.09 and p=0.24) (table 3) 34.

In the paper by Shrubsole and co-workers, the risk assessment model, based on urinary PGE-M and histological type, was adjusted for age, gender, education, ethnicity, smoking habits, BMI, and consumption of alcohol, NSAID, and red meat (table 3). In patients with "any advanced adenoma", adjustment for gender, ethnicity, age, and education yielded increased and significant OR in Q3 and Q4, whereas the adjustment for cigarette smoking, BMI, and alcohol, red meat, and NSAID consumption involved an approximately twofold greater risk in the third quartile for cases compared with controls. Notably, in the group of "multiple small tubular adenoma" (MSTA) all risks were increased and significant after adjustment for all the confounding variables included in the model. The same was true of patients with "advanced or multiple adenoma "when they were not stratified by gender. Gender stratification involved very high OR in women with high PGE-M compared with HC in the fourth quartile (table 3). As regards NSAID consumption, a twofold greater risk was found among those who had never taken them in the third and fourth quartile of PGE-M, whereas among users the risk was increased in all PGE-M quartiles for cases compared with controls 33.

Cai and co-workers reported an increased risk with the increase in PGE-M levels in cases compared with controls for all CRC patients (p<0.001). Remarkably, in CRC patients with a disease duration >30 months, higher PGE-M levels were associated with an increased and significant risk (<30 months, p=0.35; >30 months, p=0.012) (table 3) 36.

PGE-M and meta-analysis

The present review highlights that PGE-M is the most frequently assessed marker. The studies published to date have examined the whole CRC spectrum, from benign adenoma to advanced cancer stages. Despite their limited amount, the samples analysed in the various papers were found to be sufficient for a meta-analysis, except for the study by Johnson and colleagues 35. The present meta-analysis aims to assess whether differences in urinary PGE-M among patient groups are statistically significant. The following groups were compared: patients with a single small tubular adenoma (LRA); patients with advanced adenoma or adenoma showing a high degree of dysplasia (HRA); and patients with MSTA, though not all papers included the latter. The results of the comparisons are reported below.

LRA vs. HC. In this comparison the standard mean difference (SMD) was not significantly different (z=0.21; p=0.836); heterogeneity chi-squared was 2.90 (d.f.=3), p=0.408, and I-squared (change in SMD attributable to heterogeneity) was 0.0% and estimate of between-study variance tau-squared <0.0001 (Figure S2).

HRA vs. HC. This comparison yielded z=5.12; p<0.001, heterogeneity chi-squared=4.61 (d.f.=3), p=0.203, I-squared =34.9%, and estimate of between-study variance tau-squared=0.0049 (Figure S3).

HRA vs. LRA. The data found for this comparison were as follows: SMD z=5.83; p<0.001, heterogeneity chi-squared=0.48 (d.f.=3), p=0.924, I-squared =0.0%, and estimate of between-study variance tau-squared<0.001 (Figure S4).

MSTA vs. HC. The following data were obtained for this comparison: SMD z=1.00, p=0.316; heterogeneity chi-squared =6.71, p = 0.010, I-squared=85.1%; and estimate of between-study variance tau-squared=0.0565 (Figure S5).

CRC vs. HC.SMD z=5.09, p<0.001 and heterogeneity chi-squared=1.13 (d.f.=1), p=0.287, I-squared =34.2%, and estimate of between-study variance tau-squared=0.0031 (Figure S6).

Discussion

Research into diagnostic tests has provided a number of non-invasive assays for use in CRC screening that include DNA, RNA and protein biomarkers in faeces, blood, and urine 32,73-77.

Several urinary markers have been investigated, including epigenetic modifications, oxidatively modified DNA bases, nucleosides, enzyme activities, various metabolites (glycolysis, tricarboxylic acid[TCA] cycle, urea cycle, pyrimidine, polyamine and gut microbiota metabolism), oxidative indices, non-enzymatic antioxidants, CEA, and PG metabolites (especially PGE-M).

As regards epigenetic modifications, abnormal DNA methylation is an early event in carcinogenesis, and several abnormally hypermethylated genes have been identified in CRC patients 78-80. Hypermethylation of CpG-rich sequences (CpG islands) in tumour suppressor genes has been reported in CRC neoplastic tissue and in premalignant lesions 81. Since aberrant DNA methylation is found in stool, serum, urine, and other body fluids, such DNA could be used as a biomarker for cancer screening 82-84.

With reference to epigenetic markers, significant hypermethylation of the vimentin gene has been found in CRC patients compared with controls 70. In a cross-sectional study, Amiot and colleagues measured the methylation levels of Wif1, ALX-4, and vimentin genes, selected from a panel of 63 genes, in urine, stool, and serum 24. These markers showed a good specificity but a low sensitivity for early cancer detection (table 2).

Since oxidative stress related to chronic inflammation is associated with CRC etiology, Rozalski and colleagues assessed urinary excretion of oxidatively modified DNA bases/nucleoside (8-oxo-7,8-dihydroguanine, 8-oxo-7,8-dihydro-2'-deoxyguanosine and 5-hydroxymethyluracil) in HC, adenoma patients and CRC patients 71. Although urinary DNA changes may reflect the oxidative stress/chronic inflammation found in CRC 85,86, its performance in early diagnosis is moderate 71.

The levels of advanced oxidative protein products (AOPP), hydrogen peroxide (H2O2), malondialdehyde (MDA; a marker of lipid peroxidation), and ferric reducing antioxidant power were determined in urine from BC and CRC patients 26, and AOPP, H2O2 and MDA were found to be significantly higher in CRC patients compared with control subjects and BC patients.

Urinary nucleosides have been tested as biological markers of malignancy in a variety of cancers 87-90, but very few studies have addressed CRC. An investigation into urinary nucleosides from CRC patients and HC 24 has found higher mean levels of adenosine, cytidine, 1-methyladenosine in patients, even though they provided low SE (range, 27-69%).

Feng and co-workers reported that the levels of 11 of 14 urinary nucleosides were higher in CRC than in HC subjects 73. The SE of the urinary nucleosides (76.9%) was higher than those of CEA (38.5%), CA19-9 (40.4%), CA125 (15.4%), and AFP (17.3%).

A comparison of the urinary concentrations of 14 nucleosides in healthy subjects, patients with intestinal villous adenoma, and patients with CRC found that the concentrations of 12 nucleosides were significantly elevated in CRC patients, whereas only the concentrations of four nucleosides were higher in patients with intestinal villous adenoma compared with healthy subjects 25. In CRC patients, the SE of urinary nucleosides was 71% compared with the 29% of CEA. Mean nucleoside concentrations were significantly lower in patients with intestinal villous adenoma than in CRC patients. Moreover, the level of 1-methylguanosine, pseudo-uridine, and 1-methyladenosinepositively correlated with tumour size and Dukes' stage.

The activity of several enzymes is altered in a number of cancers. Changes in the structure of membrane glycoconjugates and glycosidase and protease activity are important in tumorigenesis 91. Significantly higher activities of lysosomal exoglycosidase shave been reported in human glioma and kidney cancer tissue and in serum and urine from patients with pancreatic cancer and colon cancer 69, 91-95. Β-HEX, a lysosomal exoglycosidase, releases terminal N-acetyl hexosamines from GM2 gangliosides, oligosaccharides of glycoproteins, glycosaminoglycans, and other carbohydrate-containing macromolecules 93. There are two major HEX isoenzymes: A (αβ), and B (ββ) 93. Szajda and colleagues have described a significant increase in the concentration of HEX and in the activity of isoenzymes HEX A and HEX B in serum and urine from CRC patients 31 compared with HC. The SE of HEX, HEX A and HEX B was 78.9%, 73.6%, and 84.9%, respectively. No correlation was found between the activity of HEX, HEX A or HEX B and histological type, Dukes' stage, degree of cellular differentiation, tumour size, or gender.

Sulphatase family members share structural and functional similarities 96. They catalyse the hydrolysis of sulpha bonds from different types of substrates, e.g. glycosaminoglycans, sulpholipids, and steroid sulphates. The two main ARS isoenzymes, A (ARSA) and B (ARSB), are soluble and are located in lysosomes. In a retrospective study, Niu and colleagues assessed morning urinary ARS activity as a marker of tumour progression in normal controls, subjects with benign CR tumours and CRC patients 30 and found significantly higher values in the latter group. Moreover, morning urinary ARS activity was fairly effective in distinguishing patients with CRC from those with benign CR tumour.

Several metabolic aberrations are associated with CRC, and significant metabolic changes have been described in CRC tissue 97. Serum and urinary metabolomic studies have found that CRC is associated with dysregulated glycolysis, TCA cycle, urea cycle, and pyrimidine, polyamine and gut microbiota metabolism 98-100. Cheng and co-workers have described a distinct urinary metabolic profile in CRC patients with altered levels of several metabolites derived from gut microbial-host co-metabolism as well as metabolites involved in the TCA cycle and tryptophan and polyamine metabolism. A panel of seven metabolite markers (citrate, hippurate, p-cresol,2-aminobutyrate, myristate, putrescine, and kynurenate), was able to discriminate CRC patients from their healthy counterparts, with an AUC 0.998 72.

Polyamine compounds are low-molecular-weight organic cations involved in several biological processes including carcinogenesis. Among natural polyamines, the concentrations of diacetylated spermine and spermidine increase in urine from cancer patients 101-103. Urinary DiAcSpm and DiAcSpd have been evaluated in BC and CRC patients in two studies 27, 28. In Hiramatsu's study 28, urinary DiAcSpm was found to have a markedly higher SE (75.8%) compared with serum CEA (39.5%) and CA19-9 (14.1%) in CRC patients 26; in Umemori's study 27, the SE of DiAcSpm and DiAcSpd in CRC patients was respectively 69.6% and 36.3%. DiAcSpm and DiAcSpd were higher than in controls even in early-stage CRC, and were highest in patients with stage IV disease 27.

CEA comprises a large family of cell-surface glycoproteins with common antigenic determinants residing in the protein portion of the molecule and is the most widely used tumour marker for CRC monitoring 104, 105. El-Masry and co-workers have developed a Slot-Blot Enzyme Linked Immunosorbent Assay (SB-ELISA) based on anti-CEA monoclonal antibody as a simple, fast, inexpensive, and non-invasive immunodiagnostic technique for CEA detection in urine from patients with gastrointestinal tumours 75. In CRC patients it provided higher SE, PPV and PNV (respectively 76.7%, 86.3%, and 65%) than the ELISA serum test, which showed 64.4% SE, 79.5% PPV, and 52.2 % PNV.

PGE-M seems to be the most interesting and promising urinary marker for CRC and adenoma risk assessment and for CRC screening. PGE-M is the main urinary metabolite of PGE2, probably the major mediator of the effects of cyclooxygenase-2 (COX-2) in colorectal carcinogenesis. COX-2 is abnormally expressed in most CR cancers and adenomas 105,106. In addition, use of NSAID, such as COX-2 inhibitors, reduces CR adenoma recurrence and enhances its regression, confirming the role of the COX-2-related pathways in CR carcinogenesis 107-110.

Epidemiological studies have consistently shown a 40% to 50% reduction in CRC risk associated with NSAID use 111-113. Clinical trials have demonstrated that anti-inflammatory agents can reduce the risk of developing colon polyps 114-116. The chemo preventive effects of NSAID are thought to be largely mediated through their role in the inhibition of COX-2 and PG production 117-119.

As illustrated in figure 2, COX-2 catalyses the conversion of arachidonic acid to PG H2, which is the precursor for several PG including PGE2. Most of the effects of COX-2 on tumour genesis appear to be related to PG2 overproduction. PGE2 is the most abundant PG detected in CR neoplasms and has been shown to inhibit apoptosis, stimulate angiogenesis, and increase cellular proliferation, cyclin, and migration 120-127. PGE2 levels are elevated in CR neoplasms, and reduction in the expression of 15-Hydroxyprostaglandin dehydrogenase (15-PGDH), which degrades PGE2, is commonly found in CRC 120, 128-130. PGE2 exerts its cellular effects by binding to its cognate receptor Prostaglandin E2 receptor 1-4 (EP1-4), which is overexpressed in all CRC tissues 123,130. Given the critical role of PGE2 in CR carcinogenesis, it is conceivable that the main urinary metabolite of PGE2, PGE-M, can be a valuable risk assessment tool for CRC.

Figure 2

Diagram of the possible mechanism of PGE-M in CRC. PLA2: Phospholipases A2; COX-1 cyclooxygenase 1; COX-2: cyclooxygenase 2; PGH2: prostaglandine H2; PGE2 prostaglandine E2; PGE-M urinary metabolite of prostaglandin E2; 15-PGDH: 15-Hydroxyprostaglandin dehydrogenase; EP1-4: Prostaglandin E2 receptors 1-4

Increased urinary PGE2-M levels are associated with MSTA and advanced adenoma as well as CRC 32-36, 110-113. Johnson and co-workers found significantly increased urinary PGE-M in patients with CRC and large CR polyps compared with subjects with normal colonoscopies or small polyps 35. They also found that treatment with a selective COX-2 inhibitor, celecoxib, reduced PGE-M levels among rectal cancer patients.

In a prospective cohort study of Chinese women, baseline urinary PGE-M was more than 50% higher in CRC patients than in controls 36 and was associated with a strong risk for later CRC diagnosis.

In a case-control study of CR adenoma, patients with advanced adenoma or MSTA had more than 25% higher levels of PGE-M than controls, whereas differences in PGE-M between controls and cases with a single small tubular adenoma were not significant 35. Subjects with the highest quartile level of PGE-M shared an approximately 2.5 greater risk of a diagnosis of advanced adenoma or MSTA compared with those with the lowest PGE-M values (OR=2.53, 95% confidence interval=1.54-4.14, p< 0.001).

In a recent, prospective, case-control study, Bezawada and co-workers have shown that urinary PGE-M is associated with an increased risk for advanced, large, and multiple adenoma. In addition, aspirin or NSAID use seemed to be more strongly associated with a lower risk of adenoma among individuals with high PGE-M levels than in those with low PGE-M 34.

In a matched case-control study, Davenport and colleagues evaluated both plasma CRP, a pro-inflammatory protein, and urinary PGE-M in four subjects groups: patients with a single small tubular adenoma, MSTA patients, patients with at least one advanced adenoma, and polyp-free controls. There was no apparent association between CRP level and the risk of single small tubular adenoma. A dose-response relationship with CRP was observed for risk of either MSTA or advanced adenoma. Combined analysis of CRP and PGE-M indicated that the risk of MSTA or advanced adenoma was greatest among subjects with high levels of both CRP and PGE-M than in those with low CRP and PGE-M.

Meta-analysis of the data from the five studies confirmed that PGE-M levels are strongly associated with an increased risk for MSTA, advanced adenoma, and CRC. The increase in urinary PGE-M in patients with CRC and large adenomas suggests that urinary PGE-M is a potentially useful biomarker for the detection of advanced CR neoplasms.

Subjects with MSTA or advanced adenoma are at greater risk of a recurrence compared with individuals with single small tubular adenoma 131,132. The finding that PGE-M was related to MSTA or advanced adenoma and not to simple adenoma may reflect a more substantial effect of PGE2 on tumour progression than on carcinogenesis.

In conclusion, although colonoscopy remains the most effective diagnostic modality for CRC, it generally suffers from poor patient compliance, high cost, and possible complications. The development of more effective methods for early diagnosis, assessment of disease severity, and prognosis would meet all these problems. Growing evidence suggests that urinary screening tests have the potential to respond to these expectations. Particularly, PGE-M seems to be the most promising urinary marker for CRC early detection.

Supplementary tables and figures.

Click here for additional data file.