Table 1 Summary of observational studies on the association between ARA and risk of colorectal cancer

Study design: cohort study

Exposure assessment: dietary intake

Murff et al. 2009 [30]

SWHS, China, 1996-2007, prospective cohort design (7-year biennial follow-up, follow-up rate = 96.7%)

73,243 women aged 40-70, no prior history of cancer

SWHS's FFQ, 77 items, previously validated against 24 x 24-HDR

Self-reported physician diagnosis, combined with annual record linkage with the Shanghai Cancer Registry and Shanghai Vital Statistics database

Age at baseline, total energy intake, smoking status, alcohol intake, physical activity, energy-adjusted total red meat consumption, menopausal status, use of HRT, multivitamin, aspirin, total n-3 PUFA intake, n-6 to n-3 PUFA ratio

18

Dietary ARA intake, g/day, quintile, median

Ptrend

Q1: 0.02

0.03

Q2: 0.03

Q3: 0.05

Q4: 0.06

Q5: 0.09

Lin et al. 2004 [28]

WHS, USA, 1993–2003, prospective cohort design nested randomized, double-blind, placebo-controlled 2 × 2 factorial aspirin and vitamin A trial (average 8.7 years follow-up)

37,547 female health professionals aged ≥45, free of heart disease and cancer except NMSC

FFQ, 131 items, validated against 2 x 7-day WR

Self-reported physician diagnosis, reviewed and confirmed medical diagnoses

Age, treatment assignment, BMI, family history of CRC, colorectal polyps, physical activity, smoking status, alcohol intake, use of HRT, total energy intake

15

Dietary ARA intake, %energy, quintile, median

Ptrend

Q1: 0.04

0.55

Q2: 0.06

Q3: 0.07

Q4: 0.09

Q5: 0.12

Study design: nested case-control study

Exposure assessment: blood ARA level

Hall et al. 2007 [34]

PHS, USA, 1982-1995, nested case-control design within a randomized, double-blind, placebo-controlled factorial aspirin and beta-carotene trial (average 5 and 7 years follow-up)

178 CRC patients, 282 controls, male physicians without history of cancer aged 40-84 years at baseline, 1 case matched with 1-2 controls by age, smoking status

Whole blood fatty acids, GC analysis blinded to case-control status at a time, precision indicated

Self-report, combined with review of medical records

None

23

ARA composition%, geometric mean(95%CI) Case:

P

9.77(9.57-9.99)

Not significant

Kojima et al. 2005 [35]

JACC Study, Japan, 1988-1997, nested case-control design (7 years follow-up)

169 primary CRC patients, 481 controls without previous history of cancer, aged 40-79 years at baseline, 1 case matched with 2-3 controls by age, sex, resident area

Serum fatty acids, GC analysis blinded to case-control status, precision not indicated

Population-based cancer registries, supplemented by death certificates

Age at completing final education, family history of CRC, BMI, smoking status, alcohol intake, intake of green leafy vegetables, physical activity

23

ARA composition, weight % of total serum lipids, quartile

P trend

Men:

Men:

Q1: <3.71

0.99

Q2: 3.71-4.619

Q3: 4.62-5.269

Q4: ≥5.27

Women:

Women:

Q1: <4.20

0.40

Q2: 4.20-4.879

Q3: 4.88-5.634

Q4: ≥5.635

Study design: case-control study (temporal relationship among exposure and outcome is demonstrated)

Exposure assessment: dietary intake

Theodoratou et al. 2007 [36]

Survey, UK, 1999-2006, case-control design

1,455 primary CRC patients aged 16-79, 1,455 controls (eligibility criteria not shown), matched by age, sex, resident area

Scotish FFQ, 150 items, validated against 4-day WR, (response rate = case 82%, control 97%)

Not shown

Family history of CRC, total energy intake, total fiber intake, alcohol intake, NSAIDs use, smoking status, BMI, physical activity, total fatty acid intake

20

Dietary ARA intake, mg/day, quartile

Ptrend

Q1: 0-5.82

0.163

Q2: 5.83-8.40

Q3: 8.41-11.34

Q4:≥11.35

Nkondjock et al. 2003 [31]

Survey, Canada, 1989-1993, case-control design

402 CRC patients aged 35-79, 688 controls, matched by age, language, place of residence

FFQ, 132 items, validated against 7-day Food Record

Histological diagnosis

Age, BMI, family history of CRC, marital status, physical activity

20

Dietary ARA intake, g/day, quartile

Ptrend

Q1:<0.06

0.001

Q2:0.06-0.09

Q3:0.10-0.14

Q4:>0.14

Slattery et al. 1997 [37]

Survey, USA, 1991-1994

1993 CRC patients aged 30-79, 2410 controls without history of CRC (population characteristic partially not shown), matched by age, sex, resident state

CARDIA Diet History Questionnaire, validated against 7 x 24-HDR

Cancer registries

Total energy intake, age at selection, BMI, family history of CRC, physical activity, dietary cholesterol, calcium, fiber, NSAIDs use

19

Dietary ARA intake, g/MJ, quintile

Ptrend

Men:

Men:

Q1:<0.17

Not shown

Q2:0.17-0.22

Q3:0.23-0.26

Q4:0.27-0.33

Q5:>0.33

Women:

Women:

Q1:<0.039

Not shown

Q2:0.039-0.051

Q3:0.052-0.063

Q4:0.064-0.077

Q5:>0.077

Exposure assessment: blood ARA level

Kuriki et al. 2006 [38]

Survey, Japan, 2002-2004, case-control design

74 CRC patients, 221 controls, aged 20-80 without history of cancer or current diseases, 1 case matched with 3 controls by age, sex, season of blood collection

Erythrocyte phospholipids, GC analysis blinded to case-control status, precision indicated

Histological diagnosis

BMI, habitual exercise, alcohol intake, smoking status, green-yellow vegetable intake, family history of CRC

22

ARA composition, mol%, tertile

Ptrend

T1: <8.625

<0.05

T2: 8.625-10.178

T3: >10.178

Study design: case-control study (temporal relationship among exposure and outcome is unclear)

Exposure assessment: dietary intake

Busstra et al. 2003 [39]

Survey, Netherlands, 1995-1998, case-control design

52 CRC patients, 57 controls, aged under 75 without history of CRC, colon resection, polyposis coli, inflammatory bowel disease, included subjects with familial HNPCC, matching not indicated

FFQ developed for the Dutch cohorts of the EPIC study, 178 items, validated against 12 x 24-HDR

Histological diagnosis

Age, total energy intake, sex, familial background of HNPCC

13

Dietary ARA intake, g/day, tertile

Ptrend

T1: <0.02

0.37

T2: 0.02-0.04

T3: ≥0.04

Exposure assessment: blood ARA level

Ghadimi et al. 2008 [40]

Survey, Japan, 1997-2003, case-control design

203 CRA patients, 179 controls (negative faecal occult blood test), matching not indicated

Serum fatty acids (fasting blood), GC analysis, precision indicated

Histological diagnosis

Age, BMI, family history of CRA or CRC, history of diabetes, smoking status, alcohol intake, physical activity, season of data collection

18

ARA concentration, mg/dl, quartile

Ptrend

Men:

Men:

Q1:<17.40

0.104

Q2:17.40-19.90

Women:

Q3:19.91-22.50

0.001

Q4:>22.50

Women:

Women:

Q1:<18.05

0.001

Q2:18.05-20.50

Q3:20.51-22.38

Q4:>22.38

Baró et al. 1998 [41]

Survey, Spain

17 CRC patients aged 35-82, 12 controls aged 33-81 with no malignant diseases, matched by age, resident area

Plasma and erythrocyte fatty acids (fasting blood), GC analysis, precision not indicated

Not shown

None

12

Plasma ARA concentration, mg/dl, mean(SEM)

P

Case:

Plasma:

18.59(1.31)

Not significant

Erythrocyte ARA composition%, mean(SEM)

Erythrocyte:

Case:

Not significant

14.61(0.24)

Neoptolemos et al. 1990 [42]

Survey, UK

32 CRC patients, 42 controls admitted for elective operations for benign without DM, metabolic disorders, blood transfusions, matched by age, sex, admittance period

Erythrocyte phospholipids (fasting blood), GC analysis, precision not indicated

Not shown

None

13

ARA composition%, median(range)

P

Case:

20.7(12.8-48.9)

Not significant

Neoptolemos et al. 1988 [43]

Survey, UK

49 CRC patients aged 49-92, 49 controls with benign diaseases aged 48-90, matched by age, sex

Erythrocyte phospholipids (fasting blood), GC analysis, precision not indicated

Not shown

None

12

ARA composition%, median(range)

P

Case:

21.8 (15.3-28.4)

0.043

Exposure assessment: tissue ARA level

Busstra et al. 2003 [39]

Survey, Netherlands, 1995-1998, case-control design

52 CRC patients, 57 controls, aged under 75 without history of CRC, colon resection, polyposis coli, inflammatory bowel disease, included subjects with familial HNPCC, matching not indicated

Buttock adipose tissue fatty acids, GC analysis, precision not indicated

Histological diagnosis

Age, total energy intake, sex, familial background of HNPCC

13

ARA composition mass%, tertile

Ptrend

T1: <0.35

0.42

T2: 0.35-0.45

T3: ≥0.45

Study design: cross-sectional study

Exposure assessment: blood ARA level

Almendingen et al. 2006 [44]

Survey, Norway

38 FAP patients aged 24-70 (all colectomized), 160 healthy controls aged 21-66

Serum phospholipids (fasting blood), GC analysis, precision indicated

Diagnosis by endoscopy and histology

None

17

ARA composition weight%, mean(SD)

P

Case:

10.96(1.85)

≤0.0001

Fernández-Bañares et al. 1996 [45]

Survey, Spain

22 colonic cancer patients, 27 colonic adenoma patients, 12 controls with benign diseases, no significant differences in sex and age

Plasma phospholipids (fasting blood), GC analysis, precision not indicated

Total fibreoptic colonoscopy

None

13

ARA composition%, mean(SEM) Carcinoma:

P

9.38(0.37)

Not significant

Adenoma:

9.95(0.49)

Hietanen et al. 1994 [46]

Survey, UK, cross-sectional design

20 colon cancer patients aged 38-84, controls, matched by age, sex, smoking status

Erythrocyte phospholipids (fasting blood), GC analysis, precision not indicated

Not shown

None

8

ARA concentration, mg/dl, mean(SD)

P

Case:

18.5(0.6)

<0.05

Exposure assessment: tissue ARA level

Fernández-Bañares et al. 1996 [45]

Survey, Spain

15 colonic cancer patients, 21 colonic adenoma patients, 8 controls with benign diseases

Normal colon mucosa fatty acids, GC analysis, precision not indicated

Total fibreoptic colonoscopy

None

13

ARA composition%, mean(SEM) Carcinoma:

P

10.9(0.57)

Not significant

Adenoma:

12.3(0.55)

Berry et al. 1986 [47]

Survey, Israel, 1982-1985

155 consecutive colonoscopies (53 carcinoma, 34 benign neoplastic polyps, 68 controls)

Buttock adipose tissue fatty acids, GC analysis, precision indicated

Histological diagnosis

None

13

ARA composition%, mean (SD) Carcinoma:

P

0.54 (0.2)

Not significant

Benign neoplastic polyps:

0.52 (0.2)