Influence of methyl donor nutrients as epigenetic regulators in colorectal cancer: A systematic review of observational studies

Table 1 Characteristics of the eight case-control studies included in this systematic review examining the interactive effects between single-nucleotide polymorphisms in genes encoding methyl-metabolizing enzymes and one-carbon metabolism-related dietary compounds on colorectal cancer risk

Ref.	Country	Age (yr)	No. cases (M/W), endpoint	No. controls, type	Gene (SNP)	Nutrient/alcohol	Method for measuring nutrition intake	Outcome (OR, 95%CI)			Adjustments to OR	NOS
								SNP	Nutrient/alcohol	Interaction
Chen et al[23]	United States	40-75	144 M, CRC	627 C	MTHFR (677C>T)	Dietary folate, Met, and alcohol	Validated FFQ (self-reported)	No assoc	Alcohol (≥ 5 vs ≤ 1 drinks/wk): 1.61 (1.01-2.58)	677TT (vs CC/CT)-low alcohol consumption (≤ 1 drinks/wk): 0.11 (0.01-0.85) (P-interac = 0.02)	Age, CRC family history	7
Guerreiro et al[24]	Portugal	Cases (64.2 ± 11.3), controls (62.2 ± 12.1)	104/92 CRC	200 C	MTHFR (677C>T), MS (2756A>G), SHMT (1420C>T)	Dietary folate, vitamins B6 and B12, glycine, Met, serine, and alcohol	Validated FFQ (by interview)	677TT (vs CC/CT): 3.01, (1.3-6.7); 1420TT (vs CC/CT): 2.6, (1.1-5.9)	Folate (> 406.7 mcg/d vs < 406.7 mcg/d): 0.67 (0.45-0.99)	677TT (vs CC/CT)– folate (< 406.7 mcg/d): 14.0, 1.8-108.5 (P = 0.05)	Age, CRC history, and sex	5
Kim et al[25]	Korea	30-79	465/322 CRC (363 CCa, 330 RCa)	656 H	MTHFR(677C>T)	Dietary folate and alcohol	Validated FFQ	677TT (vs CC/CT): 0.60 (0.46-0.78)	Folate (high vs low intake): 0.64 (0.49-0.84); alcohol (high vs low intake): 1.76 (1.26-2.46)	677CC/CT–Low-methyl diet (folate < 209.69 mcg/d and alcohol ≥ 30 g/d): 2.32 (1.18-4.56) (P-interac = no assoc)	Age, BMI, CRC family history, energy intake, multivitamin use, sex, smoking status	7
Ma et al[26]	United States	40-84	202 M, CRC	326 C	MTHFR (677C>T)	Plasma folate, and alcohol consumption	FFQ (self-reported)	677TT (vs CC): 0.45 (0.24-0.86)	Folate (plasma deficiency vs adequate levels): No assoc	677TT (vs TC/CC)–folate (adequate levels): 0.32 (0.15-0.68)	Age, alcohol consumption, aspirin use, BMI, exercise, multivitamin use, and smoking status	7
									Alcohol: Unk	677TT (vs CC)–alcohol (0-0.14 drinks/d): 0.12 (0.03-0.57)	Age
Murtaugh et al[27]	United States	30-79	446/305 RCa	979 C	MTHFR (677C>T, 1298A>C)	Dietary folate, riboflavin, vitamins B6 and B12, Met, and alcohol	FFQ (by interview)	W, 677TT (vs CC): 0.54 (0.30-0.98). M&W, 1298CC (vs AA): 0.67 (0.46-0.98)	Dietary folate (> 475 mcg/d vs < = 322 mcg/d): 0.66 (0.48-0.92). High methyl donor status (vs low status): 0.79 (0.66-0.95)	No assoc	Age, BMI, ibuprofen use, intake of energy, fibre and calcium, PA, sex, and smoking status	6
Pufulete et al[28]	United Kingdom	38-90	13/15 CRC	76 C	MTHFR (677C>T, 1298A>C), MS (2756A>G), CBS (844ins68)	Plasma folate, vitamin B12, and homocysteine, alcohol and folate intake	Validated FFQ (by interview)	677TT (vs CC): 5.98 (0.92-38.66), P = 0.06; 1298CC (vs AA): 12.6 (1.12-143.70), P = 0.04	Folate status score (T3 vs T1): 0.09 (0.01-0.57), P-trend = 0.01	Unk	Age, alcohol consumption, BMI, sex, and smoking status	7
Sharp et al[29]	United Kingdom		150/114 (189 CCa, 75 RCa)	408C	MTHFR (677C>T, 1298A>C)	Dietary folate, riboflavin, vitamins B6 and B12, and alcohol	Validated FFQ (self-reported)	No assoc	No assoc	677CT/TT (vs CC)–folate (> mean): P-interac = 0.029. 677CT/TT (vs CC)–vitamin B6 (> mean): P-interac = 0.016	Age, CRC family history, energy intake, NSAID use, PA, and sex	6
Slattery et al[30]	United States	30-79	824/849 CC (DCCa 405/303; PCCa 395/327)	1816 C	MTHFR (677C>T)	Dietary folate, Met, vitamins B6 and B12, and alcohol	Validated CARDIA diet questionnaire	677TT: No assoc	Unk	TT-low risk diet (high in folate and Met and without alcohol): 0.4 (0.1-0.9)	Age, BMI, intake of energy and fibre, PA, and smoking intensity	7

Asso: Significant association; BMI: Body mass index; C: Community controls; CBS: Cystathionine synthase; CCa: Colon cancer; CI: Confidence interval; CRC: Colorectal cancer; DCCa: Distal colon cancer; FFQ: Food frequency questionnaire; H: Hospital controls; interac: Interaction; M: Men; Met: Methionine; MS: Methionine synthase; MTHFR: Methylenetetrahydrofolate reductase; NSAID: Nonsteroidal anti-inflammatory drugs; NOS: Quality Newcastle-Ottawa Scale; OR: Odds ratio; PA: Physical activity; PCCa: Proximal colon cancer; RCa: Rectal cancer; SHMT: Serine hydroxymethyltransferase; SNP: Single-nucleotide polymorphism; unk: Unknown; W: Women.

Table 2 Characteristics of the cohort study included in this systematic review examining the interactive effects between single-nucleotide polymorphisms in genes encoding methyl-metabolizing enzymes and one-carbon metabolism-related dietary compounds on colorectal cancer risk

Ref.	Country	Study cohort (age, yr)	No. participants (M/W)	No. incident cases	Follow-up length, y	Gene (SNP)	Nutrient/alcohol	Method for measuring nutrition intake	Outcome (RR, 95%CI)			Adjustments to RR	NOS
									SNP	Nutrient/alcohol	Interaction
de Vogel et al[9]	Netherlands	Netherlands Cohort Study on diet and cancer (55-69)	58279/62573	734 CRC	7.3	MTHFR (rs1801133, rs1801131), MTR (rs1805087), MTRR (rs1801394), DNMT3B (rs2424913, rs406193), EHMT1 (rs4634736), EHMT2 (rs535586), PRDM2 (rs2235515)	Dietary folate, Met, vitamins B2 and B6, alcohol	Validated FFQ (self-reported)	Unk	Unk	≤ 1 rare allele in folate metabolizing enzymes–vitamin B2 (T3 vs T1): 0.30, (0.11-0.81), P-trend = 0.005. Rare allele of DNMT3B C>T (rs406193)–vitamin B6 (T3 vs T1): 1.90 (1.00-3.60), P-trend = 0.04. Common allele of PRDM2 G>A (rs2235515)–vitamin B6 (T3 vs T1): 1.49 (1.00-2.22), P-trend = 0.03. No assoc	Age, alcohol consumption, BMI, CRC family history, intake of energy and alcohol, sex, and smoking status	9

Asso: Significant association; BMI: Body mass index; CI: Confidence interval; CRC: Colorectal cancer; DNMT3B: ADN (cytosine-5-)-methyltransferase 3 ; EHMT1: Euchromatin histone methyltransferase 1; FFQ: Food frequency questionnaire; M: Men; MTHFR: Methylenetetrahydrofolate reductase; MTR: Methionine synthase; MTRR: Methionine synthase reductase; NOS: Quality Newcastle-Ottawa Scale; PRDM2: PR domain zinc finger protein 2; RR: Relative risk; SNP: Single-nucleotide polymorphism; T: Tertile; unk: Unknown; W: Women.

Table 3 Characteristics of the six case-control studies included in this systematic review examining the interactive effects between single-nucleotide polymorphisms in genes encoding methyl-metabolizing enzymes and/or mutations in oncogenes, CpG island methylator phenotype and/or microsatellite instability, and one-carbon metabolism-related dietary compounds on colorectal cancer risk

Ref.	Country	Age (yr)	No. cases (M/W), endpoint	No. controls and type	Gene (SNP)	CIMP markers	MSI	Nutrient/alcohol	Method for measuring nutrition intake	Outcome (OR, 95%CI), interaction			NOS
										CIMP markers/MSI–nutrient/alcohol	CIMP markers-/MSI–SNP–nutrient/alcohol	Adjustments to OR
Busch et al[31]	United States	40-80 (AAs vs EAs)	244/241 CRC	Analyses were only performed in tumour tissue		CACNA1G, hMLH1, NEUROG1, RUNX3, SOCS1	Unk	Dietary folate and alcohol	Unk	EAs: High CACNA1G methylation tumour (cut point of 5%)–high folate intake: 0.3 (0.14-0.66); high SOCS1 methylation tumour (cut point of 3%)–high folate intake: 0.3 (0.11-0.80)	Unk	-	4
Curtin et al[32]	United States	30-79	518/398 CCa	1972 C	MTHFR (677C>T, 1298A>C), TS variants (TSER, TTAAAG in 3´-UTRs 1494), MTR (919D>G), RFC (80G>A), MTHFD1 (R134K, R653Q), ADH3 (1045A>G)	MINT1, MINT2, MINT31, p16, hMLH1	Unk	Dietary folate, Met, vitamin B12, and alcohol	Adaptation of the CARDIA diet history	Unk	MTHFR 1298AA–alcohol (high vs none): CIMP+, 0.5 (0.3-0.97), P < 0.01; ADH3 (1 or 2 variant, slow catabolizing*2 vs homozygous for the common allele)–folate (low): CIMP+, 1.6 (1.03-2.6), P = 0.02. MTHFR 1298AC or CC-high-risk dietary pattern (low in folate or Met intake, high in alcohol): CIMP+, 2.1 (1.3-3.4), P = 0.03	Age, centre, other SNPs, sex, smoking intensity, and race	9
Curtin et al[33]	United States	30-79	559/392	1205 C	MTHFR (1298A>C), TP53, KRAS2,	CDKN2A, hMLH1, MINT 1, 2 and 31		Folate, riboflavin, vitamins B6, B12, and Met	Adaptation of the CARDIA diet history (by interview)	M: Folate (T3 vs T1)–CIMP+, 3.2 (1.5-6.7), P < 0.01	1298 AC/CC (vs AA)–folate (T3 vs T1): 0.4 (0.2-1.0), P = 0.04, for CIMP+	Age, centre, intake of energy and fibre, NSAID use, oestrogen use (W), PA, race, referent year, sex, screening, and smoking	8
Kim et al[34]	Korea	30-79	465/322 CRC (363 CCa, 330 RCa)	656 H	MTHFR (677C>T)	Unk	2 mononucleotide markers (Bat25 and Bat26) and 3 dinucleotide markers (D2S123, D5S346, and D17S250)	Folate, vitamins B2, B6, B12, niacin, Met, and choline	Validated FFQ	Unk	DCCa: hMSH3 (rs41097) AG/GG (vs AA)–niacin (> 14.00 mg/d vs < 14.00 mg/d)–MSI–MMR status: 0.49 (0.28-0.84), P-interac = 0.008	Age, intake of energy and alcohol, BMI, CRC family history, educational level, occupation, income, PA, sex, and smoking status	7
Slattery et al[35]	United States	30-79	821/689 CRC	2410 C	Unk	Unk	10 tetranucleotide repeats, 3 Bat-26 and TGFbRII	Dietary folate, and alcohol	Validated CARDIA diet questionnaire	Alcohol–MSI+ (vs MSI-): 1.6 (1.0-2.5), P-trend = 0.03; liquor–MSI+ (vs MSI-): 1.6 (1.1-2.4), P-trend = 0.02		Age, BMI, intake of energy, fibre and calcium, intake, PA, sex	7
Slattery et al[36]	United States	30-79	638/516 CRC	2410 C	BRAF (V600E)	MINT1, MINT2, MINT31, p16 and hMLH1	Unk	Folate, vitamins B6 and B12, Met, and alcohol	Diet history questionnaire	No assoc	MSI tumour–alcohol (high vs none): 1.6 (0.9-2.9), P-trend = 0.04, for p16 unmethylated; 1.7 (0.7-4.3), P-trend = 0.06, for CIMPlow (< 2 markers); 2.2 (1.2-3.7), P-trend = 0.01, for BRAF wildtype	Age, alcohol intake, BMI, intake of energy and folate, density of calcium and fibre, NSAIDs use, PA, sex, smoking intensity	7

AA: African Americans; ADH3: Alcohol dehydrogenase 3; asso: Significant association; Bat: Mononucleotide microsatellite with quasi-monomorphic allele length distribution in healthy controls but unstable; BMI: Body mass index; BRAF: B-Raf proto-oncogene; C: community controls; CACNA1G: Calcium voltage-gated channel subunit a1 G; CCa: Colon cancer; CDKN2A: Cyclin-dependent kinase inhibitor 2A; CI: Confidence interval; CIMP: CpG island methylator phenotype; CRC: Colorectal cancer; EA: European Americans; FFQ: Food frequency questionnaire; H: Hospital controls; hMLH1: Human MutL homolog 1; hMSH3: Human MutS homolog 3; interact: Interaction; KRAS: Ki-ras2 Kirsten rat sarcoma viral oncogene homolog; M: Men; Met: Methionine; MINT: Methylated in tumours; MMR: Mismatch repair; MSI: Microsatellite instability; MTHFD: Metilentetrahidrofolate dehydrogenase; MTHFR: Methylenetetrahydrofolate reductase; MTR: Methionine synthase; NEUROG1: Neurogenin1; NOS: Quality Newcastle-Ottawa Scale; NSAID: Nonsteroidal anti-inflammatory drugs; OR: Odds ratio; PA: Physical activity; RCa: Rectal cancer; RFC: Reduced folate carrier; RUNX3: Runt-related transcription factor 3; SNP: Single-nucleotide polymorphism; SOCS1: Suppressor of cytokine signalling 1; T: Tertile; TGFbRII: Transforming growth factor β receptor type II; TS: Thymidylate synthase; TSER: Thymidylate synthase enhancer region; TP53: Tumour protein p53; unk: Unknown; UTR: Untranslated region; W: Women.

Table 4 Characteristics of the five cohort studies included in this systematic review examining the interactive effects between single-nucleotide polymorphisms in genes encoding methyl-metabolizing enzymes and/or mutations in oncogenes, CpG island methylator phenotype and/or microsatellite instability, and one-carbon metabolism-related dietary compounds on colorectal cancer risk

Ref.	Country	Study cohort (age, yr)	No. participants (M/W)	No. incident cases	Follow-up length, yr	Gene (SNP)	CIMP markers	MSI	Nutrient/alcohol	Method for measuring nutrition intake	Outcome (RR, 95%CI) interaction			NOS
											CIMP markers/MSI–nutrient/alcohol	CIMP markers–/MSI–SNP–nutrient/alcohol	Adjustments to RR
de Vogel et al[9]	Netherlands	The Netherlands Cohort Study on diet and cancer (55-69)	58279/62573	734 CRC	7.3	MTHFR (rs1801133, rs1801131), MTR (rs1805087), MTRR (rs1801394), DNMT3B (rs2424913, rs406193), EHMT1 (rs4634736), EHMT2 (rs535586), PRDM2 (rs2235515)	CACNA1G, IGF2, NEUROG1, RUNX3, SOCS1	Bat-26, Bat-25, NR-21, NR-22, NR-24	Dietary folate, Met, vitamins B2 and B6, alcohol	Validated FFQ (self-reported)	No assoc	Unk	BMI, CRC family history, intake of energy and alcohol, sex, and smoking status	9
de Vogel et al[37]	Netherlands	The Netherlands Cohort Study on diet and cancer (55-69)	58279/62573	734 CRC	7.3	BRAF (V600E)			Dietary folate, Met, vitamins B2 and B6, alcohol	Validated FFQ (self-reported)	M: BRAF mut–folate (T3 vs T1): 3.04 (1.13-8.20), P-trend = 0.03; BRAF mut–Met (T3 vs T1): 0.28 (0.09-0.86), P-trend = 0.02); hMLH1 hypermethylation–vitamin B6 (T3 vs T1): 3.23 (1.15-9.06), P- trend = 0.03	Unk	Age, BMI, CRC family history, intake of energy, meat, total fat, fibre, vitamin C, total iron and calcium, smoking status	9
Schernhammer et al[38]	United States	The Nurses' Health Study (W) (30-55) and the Health Professional Follow-up Study (M) (40–75)	47371/88691	669 CCa	22	KRAS	Unk	D2S123, D5S346, D17S250, Bat25, Bat26 (14), Bat40, D18S55, D18S56, D18S67, D18S487	Folate, vitamins B6 and B12, Met, and alcohol	Validated FFQ (self-reported)	Unk	MSI/KRAS–folate: No assoc for CCa. MSI/KRAS–vitamins B6 or B12: No assoc for CCa	Age, aspirin use, smoking, BMI, colon polyps, CRC family history, intake of alcohol, energy, beef, calcium, vitamins B6 and B12, and Met, multivitamin use, PA, sex, screening sigmoidoscopy	9
Schernhammer et al[39]	United States	The Nurses’ Health Study (30- 55)	88691 W	375 CCa		BRAF	CHFR, MGMT, p14, WRN, HTC1, MINT1, MINT31, IGFBP3	Unk			Folate (Q4 vs Q1: No assoc with CIMP–high tumour risk; and no assoc with BRAF status	Unk
van Engeland et al[40]	Netherlands	Netherlands Cohort Study on Diet and Cancer (55-69)	58279/62573	122 CRC	7.3	Unk	APC-1A, p14ARF, p16INK4A, hMLH1, O6-MGMT, and RASSF1A	Unk	Dietary folate and alcohol	Validated FFQ (self-reported)	Low vs high-methyl donor intake–promoter methylation (> 1 gene methylated): No assoc	Unk	Age, CRC family history, intake of energy, fibre, vitamin C, and iron, sex	9

APC: Adenomatous polyposis coli; asso: Significant association; Bat: Mononucleotide microsatellite with quasi-monomorphic allele length distribution in healthy controls but unstable; BMI: Body mass index; BRAF: B-Raf proto-oncogene; CACNA1G: Calcium voltage-gated channel subunit a1 G; CCa: Colon cancer; CHFR: RING finger domain protein; CI: Confidence interval; CIMP: CpG island methylator phenotype; CRC: Colorectal cancer; DNMT3B: DNA methyltransferase 3 EHMT: Euchromatin histone methyltransferase; FFQ: Food frequency questionnaire; hMLH1: Human MutL homolog 1; HTC: Histidine triad with channel; IGF2: Insulin-like growth factor 2; IGFBP: Insulin-like growth factor-binding protein; KRAS: Ki-ras2 Kirsten rat sarcoma viral oncogene homolog; M: Men; Met: Methionine; MGMT: DNA repair enzyme O(6)-methylguanine-DNA methyltransferase; MINT: Methylated in tumours; MSI: Microsatellite instability; MTHFR: Methylenetetrahydrofolate reductase; MTR: Methionine synthase; MTRR: Methionine synthase reductase; mut: Mutation; NEUROG1: Neurogenin1; NOS: Quality Newcastle-Ottawa Scale; PA: Physical activity; PRDM2: PR domain zinc finger protein 2; Q: Quartile; RASSF1A: Ras association domain family 1 isoform A; RR: Relative risk; RUNX3: Runt-related transcription factor 3; SNP: Single-nucleotide polymorphism; SOCS1: Suppressor of cytokine signalling 1; T: Tertile; unk: Unknown; W: Women; WRN: Werner syndrome gene.

Table 5 Summary of results of the studies included in this systematic review

Gen, SNP/CIMP/MSI	Nutrients/alcohol	CRC risk/CIMP+	Ref.
MTHFR 677 TT	Folate; Adequate folate	CRC risk; CRC risk	Guerreiro et al[24]; Sharp et al[29]; Ma et al[26]
	Alcohol	CRC risk	Chen et al[23]; Ma et al[26]
	Folate/Met, and without alcohol	CRC risk	Slattery et al[30]
	Vitamin B6	CRC risk	Sharp et al[29]
BRAF mutation	Folate	CRC risk (M)	de Vogel et al[37]
	Met	CRC risk (M)
hMLH1 hypermethylation	Vitamin B6	CRC risk (M)
MTHFR 1298 AC/CC	Folate/Met, and alcohol	CIMP+	Curtin et al[32]
MTHFR 1298 AA	Alcohol	CIMP+ (CCa)
p16 unmethylated, CIMPlow or BRAF mut	Alcohol	CRC risk	Slattery et al[36]
hMSH3, MSI or MMR status	Niacin	CRC risk (DCCa)	Kim et al[34]

BRAF: B-Raf proto-oncogene; CCa: Colon cancer; CRC: Colorectal cancer; CIMP: CpG island methylator phenotype; DCCa: Distal colon cancer; hMLH1: Human MutL homolog 1; hMSH3: Human MutS homolog 3; M: Men; Met: Methionine; MMR: Mismatch repair; MSI: Microsatellite instability; MTHFR: Methylenetetrahydrofolate reductase; mut: Mutation; SNP: Single-nucleotide polymorphism.

↑: High intake or increased risk or high probability.

↓: Low intake or increased risk.