DNA methylation detection methods used in colorectal cancer

Table 1 Methods used to study DNA methylation in colorectal cancer cases and controls

No.[Ref.]	Year	Specimen	Gene	Sample number		Method	Sensitivity (95%CI)	Specificity (95%CI)	Conclusion and comments
				Case	Control
1[77]	2014	Tissue	18 genes1	20	20	Illumina2	96% (87%-105%)	100%	These results suggest that methylation biomarkers in this study may be developed that will, at minimum, serve as useful objective and quantitative diagnostic complements to colonoscopy as a cancer-screening tool
2[78]	2015	Tissue	FAM78A, FSTL1, KCNC1, MYOCD, SLC6A4	28	97	Illumina3	100.00%	100%	Four distinct DNA methylation subgroups of CRC identified can provide novel insight regarding the role of CIMP-specific DNA hypermethylation in gene silencing
3[20]	2013	Stool	NDRG4, BMP3	86	794	Automated MT sDNA assay	98% (95%-101%)4	90% (88%-92%)	This automated high-throughput system could be a widely accessible noninvasive approach to general CRC screening
							92% (86%-98%)5
4[4]	2014	Stool	BMP3, NDRG4	65	9167	MT sDNA test	92.3% (83.0%-97.5%)	87% (86%-87%)	This stool test shows a higher single-application sensitivity than a commercial FIT for both colorectal cancer and advanced precancerous lesions, although with lower specificity
5[21]	2012	Stool	Vimentin, NDRG4, BMP3, TFPI2	252	293	QuARTS	89% (83%-93%)	90% (85%-94%)	Next-generation sDNA tests appear capable of detecting early-stage CRC and large adenomas with a high sensitivity and specificity at all sites throughout the colon; Validation of optimized next-generation sDNA tests in the screening setting is now needed
6[19]	2018	Tissues	BMP3	421	44	QuARTS	58% (53%-63%)	100% (90%-102%)	This novel panel is highly specific for colitis-associated neoplasia across geographically diverse patient subsets and among those with varying risk; The marker panel has been optimized by the discovery and incorporation of methylated ZDHHC1, an epithelium-specific marker, as a normalizer for the neoplasm-specific markers, in stool. Further studies are needed to corroborate and extend this finding. Except for DTX1, each of these novel markers is individually more sensitive than the most sensitive marker in the MT-sDNA panel
			NDRG4				78% (74%-82%)	100% (90%-102%)
			SFMBT_895				90% (86%-92%)	100% (90%-102%)
			SFMBT_896				89% (86%-92%)	100% (90%-102%)
			SFMBT_897				89% (85%-92%)	100% (90%-102%)
			CHST2_7890				82% (77%-85%)	100% (90%-102%)
			PDGFD				83% (79%-87%)	100% (90%-102%)
			CHST2_7889				82% (78%-86%)	100% (90%-102%)
			VAV3				82% (78%-85%)	100% (90%-102%)
			DTX1				71% (66%-75%	100% (90%-102%)
		Stool	BMP3, VAV3, ZDHHC1	12	291	MSP	92% (60%-100%)	90% (86%-93%)
7[79]	2018	Tissue	BCAT1	36	36	MSP	92% (83% -101%)	69% (54%-85%)	This study has shown that highly methylated BCAT1 and IKZF1 exist in all stages of CRC; A positive correlation was observed between hypermethylated IKZF1 and gene activity silencing
			IKZF1				72% (58%-87%)	97% (92%-103%)
8[80]	2016	Plasma	BCAT1/IKZF	28	94	MSP	68% (48%-84%)	87% (79%-95%)	Methylated BCAT1/IKZF1 blood test has a significantly higher sensitivity for recurrent CRC than the CEA test
9[81]	2015	Plasma	BCAT1	129	450	MSP	57% (48%-66%)	95% (93%-97%)	Accuracy of this two-marker blood test approximates that of gFOBT; This two-marker blood test for screening is likely a rescue strategy for those refusing more sensitive RCT-proven methods such as FIT, flexible sigmoidoscopy, or colonoscopy
			IKZF				48% (39%-57%)	99% (98.1%-100%)
			BCAT1/IKZF				66% (57%-74%)	95% (93%-97%)
10[82]	2014	Plasma	BCAT1	74	144	MSP	65% (54%-76%)	96% (93%-99%)	Detection of methylated BCAT1 and/or IKZF1 DNA in plasma may have clinical application as a novel blood test for CRC; Combining the results from the two methylation-specific PCR assays improves CRC detection with a minimal change in specificity
			IKZF				68% (57% -79%)	95% (91%-99%)
			BCAT1/IKZF				77% (64 %-84%)	92% (88%-96%)
11[6]	2018	Tissue	SEPT9	145	50	qMSP	86% (80%-91%)	94% (87%-101%)	Results indicate that MGMT/RASSF1A/SEPT9 gene promoter methylation panel accurately identifies CRC, irrespective of molecular subtype and may have a better performance than currently available epigenetic based biomarkers but requiring assessment of its performance in liquid biopsies
			MGMT/SEPT9	145	50		94% (90%-98%)	82% (71%-93%)
			MGMT/RASSF1A/SEPT9	145	50		97% (94%-100%)	74% (62%-86%)
12[35]	2013	Serum	NPY, PENK, WIF1	32	161	qMSP	94% (86%-102%)6	34% (27%-41%)	This test can be a cost-effective screening tool for detection of asymptomatic cancer patients for colonoscopy, who are at CRC risk that is hard to access and does not necessarily need further examination
							59% (42%-76%)7	95% (86%-98%)
13[36]	2018	Bowel lavage fluid	SDC2	10	54	qMSP	80% (44%-97%)	89% (77%-96%)	The SDC2 DNA methylation in BLF shows a high sensitivity and specificity in patients with CRC and precancerous lesions
14[39]	2017	Stool	SDC2	50	22	qMSP	90% (78%-97%)	91% (71%-99%)	Abnormal SDC2 methylation can be a new potential diagnostic biomarker for noninvasive screening of CRC
15[27]	2017	Tissue	UNC5C	73	28	MSP	78% (69%-88%)	100%	This study found that aberrant methylation of UNC5C gene exists in CRCs and APs; UNC5C protein expression is negatively correlated with methylation status
16[83]	2017	Serum	MGMT	30	40	MSP	90% (79%-101%)	100%	MGMT hypermethylation can be used “as a clinical biomarker” for early diagnosis and prognostic assessment of the disease
17[37]	2017	Plasma	RASSF1A	108	78	MSP	44% (35%-54%)8	95% (90%-100%)	Methylation of RASSF1A in the promoter region is independently associated with prognosis in CRC patients treated with oxaliplatin-based chemotherapy, and might be a promising target for improving chemotherapeutic effects
							21% (14%-29%)9
18[84]	2015	Tissue	hMLH1	61	14	MSP	25% (14%-36%)	57% (31%-83%)	The findings suggest that DNA methylation is a useful marker and that promoter methylation in certain genes is associated with more advanced tumor stages, poor differentiation, and metastasis, which can have an application as a risk assessment tool or as a marker of recurrence to help decide on the aggressiveness of the treatment
			MGMT	61	27		66% (54%-78%)	19% (4%-33%)
			APC	61	14		30% (19%-42%)	14% (-4%-32%)
			CDH1	61	31		87% (79%-95%)	25.8% (10%-41%)
			2 OR MORE	61	N/A		71% (60%-82%)	N/A
			ALL	61	N/A		10% (3%-18%)	N/A
19[85]	2014	Tissue	CDKN2A	44	155	MSP	82% (71%-93%)	32% (25%-39%)	The presence of CDKN2A methylation is associated with poorer overall survival in stages B and C combined
20[86]	2017	Tissue	DIRAS1	146	50	MSP, BS-Seq	47% (39%-55%)	100%	Methylation of DIRAS1 is a marker of poor prognosis in human colorectal cancer; Methylation of DIRAS1 may promote colorectal carcinogenesis and progression
21[87]	2017	Tissue	ZNF331	146	10	MSP, BS-Seq	67% (60%-75%)	100 %	Methylation of ZNF331 is a poor prognostic marker in human colorectal cancer; ZNF331 may serve as a tumor suppressor in human colorectal cancer
22[34]	2016	Tissue	RET10	25	184	Direct MSP	11% (-1%-23%)	87% (82%-91%)	The data also show that qualitative techniques such as direct MSP and nested MSP (although showing different methylation frequencies) can, when carefully developed, optimized, and interpreted, yield comparable clinical results as pyrosequencing and MS-HRM and could therefore be used for biomarker detection/validation
				71	169	Nested MSP	33% (22%-44%)	74% (67%-80%)
				62	178	Pyrosequencing	23% (12%- 33%)	73% (67%-80%)
				41	199	Pyrosequencing	18% (6%-30%)	84% (79%-89%)
				37	203	Pyrosequencing	16% (4%-28%)	85% (80%-90%)
				32	208	Pyrosequencing	14% (2%-27%)	88% (83%-92%)
				25	215	Pyrosequencing	11% (-1%-23 %)	90% (86%-94%)
				33	206	MS-HRM	15% (3%-28%)	87.7% (83.2%-92.2%)
23[7]	2017	Tissue	EDNRB location 1	45	45	MS-HRM	91% (83%-99%)	88.9% (80%-98%)	The combination of EDNRB locations 1 and 2 can be used as a powerful biomarker, in which the first location is significantly correlated with tumor stage and grade while the second one is aberrantly methylated independent of any clinicopathological features
			EDNRB location 2				91% (83%-99%)	80% (68%-92%)
			EDNRB location 3				76% (63%-88%)	78% (66%-90%)
			EDNRB location 4				80% (68%-92%)	84% (74%-95%)
			KISS1				58% (43%-72%)	71% (58%-84%)
24[88]	2013	Tissue	AGTR1, WNT2, SLIT2	60	38	Pyrosequencing	95% (89%-101%)	89% (79%-99%)	This study offers a novel panel of specific methylation markers that can be assessed in stools and may complement currently applied protocols for the early detection of sporadic CRC, which may contribute to improving the follow-up and early diagnosis of high-risk patients with IBD when assessed in non-neoplastic tissues obtained by surveillance colonoscopy
			SEPT9	14	26		93% (50%-100%)	100.00%
			VIM	12	26		83% (33%-92%)	86.00% (73%-99%)
		Stool	AGTR1, WNT2, SLIT4	64	N/A		78% (68%-88%)	N/A
			SEPT9	35	N/A		20% (6%-31%)	N/A
			VIM	33	N/A		55% (33%-70%).	N/A
25[60]	2016	Tissue	CMTM3 + SSTR2 + MDF	42	42	Pyrosequencing	81% (69%-93%)	91% (82%-100%)	This study has shown that the combination of CMTM3, SSTR2, and MDFI gene methylation may be an epigenetic biomarker for early stages of CRC but should be studied further to determine their potential role as non-invasive diagnostic biomarkers for CRC
26[89]	2013	Tissue	O6-MGMT	111	53	Pyrosequencing	34% (25%-43%)	100%	p14arf, RASSF1A, APC1A, and O6-MGMT methylation as biomarkers of prognosis in CRC could be utilized as a relevant stratification factor in future prospective and interventional studies on CRC, and might serve as a tool in tailoring treatment for individual patients
			p14ARF				29% (21%-37%)	100%
			p16INK4a				28% (20%-36%)	100%
			RASSF1A				14% (8%-21%)	100%
			APC1A				27% (19%-35%)	100%
27[90]	2016	Plasma	PDX1	20	20	SYBR Green detection	45% (23%-67%)	70% (50%-90%)	These data demonstrate the utility of close consideration of the background levels of DNA methylation in WBC DNA as an important step in the selection of biomarkers suitable for development as plasma-based assays
			SDC2	44	44		59% (45-74%)	84.1% (73%-95%)
			IKZF1	74	144		68% (57-78%)	95% (92%-99%)
			BCAT1	74	144	MethyLight	65% (54%-76%)	97% (94%-100%)
			FGF5	20	40		85% (69%-101%)	83% (71%-94%)
			GRASP	44	44		55% (40%-69%)	93%(86%-101%%)
			IRF4	22	24		59% (39%-80%)	96% (88%-104%)
			SEPT9	44	44		59.1% (45%-74%)	96% (89 %-102%)
			SOX21	20	20		85% (69%-101%)	50% (28%-72%)
28[46]	2017	Plasma	SFRP1	47	37	MethyLight	85% (75%-95%)	81% (69%-94%)	The present study offers the possibility to measure the hypermethylation of the marker panel in cell-free plasma DNA and provides a potential non-invasive, epigenetic diagnostic test. It also shows that the altered methylation pattern might serve as a key for early diagnosis of precancerous stages
			SFRP2	47	37		72% (60%-85%)	89% (79%-99%)
			SDC2	47	37		89% (81%-98%)	97% (92%-103%)
			PRIMA1	47	37		81% (70%-92%)	73% (59%-87%)
			4 genes as a panel	47	37		92% (84%-100%)	97% (92%-102%)

¹18 genes include: ANKRD15, CASP8, EDA2R, ENPEP, GRB10, IGFBP5, INS, ITGB4, LGALS2, MGC9712, NMUR1, RASSF5, SLC16A3, SULT1C2, TIMP4, VAV1, VHL, and VMD2;

²Illumina high throughput “Veracode” array;

³Illumina Infinium HM27 DNA methylation assay;

⁴The data were analyzed by using logistic regression algorithm;

⁵The data were analyzed by using marker-specific cutoff and scoring;

⁶The researchers used the cumulative methylation index (CMI) as a threshold. The first sensitivity was calculated when the CMI was 0.05;

⁷The sensitivity was calculated when the CMI was 2.1;

⁸The plasma samples used were collected before chemotherapy;

⁹The plasma samples used were collected after chemotherapy;

¹⁰To compare the clinical sensitivity and specificity of direct MSP, nested MSP, pyrosequencing, and MS HRM, in this study, they computed clinical sensitivities and specificities among deceased patients and subjects still alive, respectively, using overall mortality for the total follow-up period of up to 8 years as the standard. When using the pyrosequencing, the researchers compared different thresholds for positivity which RET promoter CpG island methylation was dependent on. From the up to down, the thresholds for positivity were, respectively, 5%, 10%, 15%, 20%, and 25%. In this table, we summarize the methods used in the 28 selected articles. The characteristics of each marker or method are described.