Different types of fruit intake and colorectal cancer risk: A meta-analysis of observational studies

Table 1 Characteristic of eligible studies included in the meta-analysis assessing the relationship between different types of fruit intake and the risk of colorectal cancers

Ref.	Country	No. of cases/controls (age)	Dietary assessment	Comparison of exposure level	Category, OR/RR (95%CI)	Confounding factors
Lee et al[39], 2017	-	923 (625 males, 298 females)/1846 (1250 males, 596 females)	SQFFQ, 106 food items	Orange/yellow fruits (g/d); males: T3 (≥ 47.9) vs T1 (< 15.9); females: T3 (≥ 90.6) vs T1 (< 32.5); proximal colon/distal colon/rectum: T3 (≥ 90.6) vs T1 (< 32.5)	Males: 0.98 (0.75-1.28); females: 0.64 (0.43-0.97); total: 0.85 (0.69-1.06); proximal colon: 0.79 (0.37-1.70); distal colon: 0.77 (0.44-1.35); rectum: 0.44 (0.25-0.80)	Age, education, alcohol consumption, BMI, regular exercise, red meat, processed meat, total EI
Leenders et al[40], 2015	Ten European countries (Denmark, France, Germany, Greece, Italy, The Netherlands, Norway, Spain, Sweden, and United Kingdom	442961 cohort; 3082 incident cases (2128 colon cancer (954 proximal, 965 distal), 1242 rectal cancer cases); 51.2 (38.3-63.0) years; follow-up 8 years	Center-specific dietary questionnaire	Medians of consumption per quartiles; berries: 21 g/d vs 1 g/d; citrus fruits: 110 g/d vs 7 g/d; grapes: 32 g/d vs 1 g/d; hard fruits: 153 g/d vs 10 g/d; stone fruits: 83 g/d vs 2 g/d	Colon cancer; berries: 1.04 (0.88-1.24); citrus fruits: 1.02 (0.88-1.17); grapes: 1.15 (0.97-1.37); stone fruits: 0.97 (0.81-1.15); rectal cancer: Berries: 1.04 (0.83-1.30); citrus fruits: 1.15 (0.95-1.38); grapes: 0.98 (0.78-1.25); stone fruits: 1.19 (0.94-1.50)	All other fruit and vegetable consumption, height, weight, dietary calcium consumption, dietary alcohol consumption, dietary cereal fiber consumption, smoking status, time since stopped smoking, duration of smoking, number of cigarettes smoked per day and PA
Abu Mweis et al[41], 2015	Jordan	167/240, NA	FFQ, 109 food items	≥ 3 times/wk (high) vs ≥ 2 times/wk (low)	Apples: 0.915 (0.545-1.535); bananas: 1.167 (0.670-2.033); oranges: 0.999 (0.581-1.715)	Age, sex, total EI, metabolic equivalent, smoking, education level, marital status, work, income, and family history of CRC
Tayyem et al[42], 2014	Jordan	220/281 (248 males, 253 females); males: mean age 55.27 years; females: mean age 48.67 years	NA; 42 food items	Daily (high) vs ≤ rarely (low)	Apple: 0.73 (0.27-1.96); banana: 1.12 (0.34-3.67); orange: 0.90 (0.44-1.82); pear: 1.13 (0.56-2.29); peach: 0.64 (0.32-1.25); grape: 0.62 (0.27-1.40); melon: 0.82 (0.38-1.78); watermelon: 0.54 (0.26-1.11); strawberry: 0.75 (0.26-2.13); fig: 0.51 (0.28-0.92); kiwi: 1.14 (0.25-5.06); dried Fruit: 1.42 (0.55-3.67)	Age, sex, total EI, MET minutes/week, tobacco use, education level, marital status, work, income, PA, marital status, family history of CRC
Rosato et al[43], 2013	Italia and Swiss	329/1361, median age 40 yr	FFQ; 78 food items	High vs Low	Citrus fruit: 0.61 (0.45-0.84)	Age, sex, center, study, year of interview, education, family history, alcohol consumption, EI
Vogtmann et al[50], 2013	China	61274 male’s cohort (40-74 years); 398 incident cases (236 colon, 162 rectal); follow-up 2002-2006 to 2010	Validated FFQ; 46 food items	Citrus fruit intake g/day: ≥ 12.61 (high) vs < 2.70 (low); watermelon intake g/day: ≥ 93.33 (high) vs < 33.33 (low)	Citrus fruit: Colorectal cancer: 0.82 (0.64-1.06); colon cancer: 0.86 (0.62-1.19); rectal cancer: 0.76 (0.51-1.14); watermelon: Colorectal cancer: 0.77 (0.59-0.99); colon cancer: 0.76 (0.55-1.06); rectal cancer 0.77 (0.51-1.15)	Age, total EI, red meat intake, total meat intake, education, income, occupation, smoking status, alcohol consumption, BMI, MET hours of exercise participation, history of diabetes mellitus, family history of CRC
Annema et al[44], 2011	Western Australia	834 (64.9 yr ± 8.9 yr)/939 (64.6 yr ± 9.4 yr)	FFQ; 74 food items	Servings/d) ≥ 0.50 (high) vs < 0.07 (low)	Total: Citrus fruit: 0.95 (0.72-1.25); apples: 0.74 (0.56-0.99); fruit juice: 1.38 (1.08-1.75); citrus fruit: Proximal Colon: 0.97 (0.65-1.45); distal Colon: 0.81 (0.53-1.24); rectum: 1.03 (0.71-1.49); apples: Proximal Colon: 1.13 (0.72-1.77); distal colon: 0.51 (0.34-0.77); rectum: 0.73 (0.49-1.08); fruit juice: Proximal Colon: 1.06 (0.74-1.49); distal colon: 1.41 (0.99-2.01); rectum: 1.74 (1.24-2.45)	Sex, age, BMI at age 20 yr, EI, multivitamin use, alcohol consumption, PA, smoking, diabetes, socioeconomic status
Foschi et al[45], 2010	Italy and Switzerland	3634 (median age 62 yr)/6804 (median age 57 yr)	Validated FFQ; 78 food items	Citrus fruit or citrus fruit juice intake: ≥ 4 portions/wk vs < 1 portion/wk	Citrus: 0.82 (0.72-0.93)	Age, sex, study center, tobacco smoking, alcohol, education, BMI, PA, EI
Li et al[51], 2010	Japan	42470 cohort (40-79 yr) (20222 males, 22248 females); 665 incident cases; follow-up 9 years	FFQ; 40 food items	Citrus consumption daily vs ≤ 2 times/wk	Citrus: 0.80 (0.61-1.06)	Age, sex, job status, years of education, BMI, time engaging in sports or exercise, time spent walking, cigarette smoking, alcohol drinking, history of hypertension, diabetes mellitus and gastric ulcer, family history of cancer, daily total EI, consumption of rice/miso soup/soybean products/total meat/total fish/dairy products/other fruits/total vegetables/oolong tea/black tea/coffee /green tea
Jedrychowski et al[32], 2010	Poland	592/765; NA	EPIC-FFQ148 food items	Apples, servings/d: > 1.50 (Q5; high) vs < 0.18 (Q1; low)	Apples: 0.53 (0.35-0.79)	Age, gender, place of residency, marital status, tobacco smoking, total EI, intake of vegetables, fruits excluding apples
Williams et al[46], 2009	North Carolina	945/959; 40-79 yr; whites (n = 1520); African-Americans (n = 384)	Diet history questionnaire; 124 food items	Citrus fruit (servings/wk): White: 16.4 Q4 (high) vs 1.89 Q1 (low); African-Americans: 21.7 Q4 (high) vs 2.3 Q1 (low)	Whites: 0.61 (0.43-0.86); African-Americans: 1.54 (0.71-3.35)	Age, sex, education, income, BMI 1 yr ago, PA, family history, nonsteroidal anti-inflammatory drug use, total EI
Nomura et al[52], 2008	Hawaii and Los Angeles	191011 cohort (85903 males, 105108 females); 2110 incident cases (1138 males, 972 females) (1571 of the colon, 515 of the rectum, 24 cases both sites) (45-75 yr); follow-up 7.3 years	Self-administered quantitative FFQ (QFFQ); 180 food items	Citrus fruit were quantified as g × 1000 kcal-1 × d-1; Q5 (high) vs Q1 (low)	Citrus fruit: Male: 0.85 (0.70-1.04); female: 1.04 (0.83-1.30)	Ethnicity, age, family history of CRC, history of colorectal polyp, pack-years of cigarette smoking, BMI, hours of vigorous activity, aspirin use, multivitamin use, replacement hormone use (women), log EI, alcohol, red meat, folate, vitamin D, calcium
Gallus et al[47], 2005	Italy	1953 (1225 of the colon, 728 of the rectum)/4154	Validated FFQ 78 food items	Average consumption of apples per day ≥ 1 (high) vs < 1 (low)	Apples: 0.70 (0.62-0.79)	Age, sex, study center, education, BMI, tobacco smoking, alcohol drinking, total EI, vegetable consumption, PA, other fruit
Lin et al[55], 2005	United States	39876 female cohort (mean age 45 years); 240 incident cases; follow-up 10 years	FFQ; 131 food items	Citrus fruit (serving/day) Median intake; 1.6 (Q5) high vs 0.1 (Q1) low	Citrus fruit: 1.11 (0.71-1.74)	Age, randomized treatment assignment, BMI, family history of CRC in a first-degree relative, history of colon polyps, PA, smoking status, baseline aspirin use, red meat intake, alcohol consumption, total EI, menopausal status and baseline post-menopausal HT use, folate intake, multivitamin use
Satia-Abouta et al[48], 2004	United States	613 (337 Caucasians, 276 African-Americans)/996(596 Caucasians, 400 African-Americans) (40-80 years)	Validated FFQ; 100 food items	Citrus fruits and juices: Median g/d: Caucasians: 4th quartile (high) 168 g vs 1st quartile 0 g (low); African-Americans: 4th quartile (high) 173 g vs 1st quartile 0 g (low)	Caucasians: 1.0 (0.7-1.6); African-Americans: 1.0 (0.6-1.6)	Age, gender, total EI, education, BMI, smoking history, PA, family history of colon cancer, NSAID use, fat, carbohydrates, dietary fiber, vitamin C, vitamin E, beta-carotene, calcium, folate, fruits, vegetables
Voorrips et al[53], 2000	Netherlands	4087 cohort; 620 colon cancer cases (332 males, 288 females), 344 rectal cancer cases (217 males, 127 females); follow-up 6.3 years	Validated FFQ; 150 food items	Citrus fruit Median intake (g/d); male: Q5 (167 g/dk) (high) vs Q1 (0g/dk) (low); female: Q5 (187 g/dk) (high) vs Q1 (8 g/dk) (low)	Male: Colon cancer: 1.09 (0.75-1.59) Rectal cancer: 0.77 (0.49-1.20); female: Colon cancer: 1.00 (0.66-1.52); rectal cancer: 1.16 (0.63-2.12)	Age, family history of CRC, category of alcohol intake
Michels et al[54], 2000	United States	136089 cohort (88764 females (30-55 years), 47325 males (40-75years); 1181 incident cases (937 colon cancer, 244 rectal cancer); follow-up 16 years	Validated FFQs; 61 food items	Citrus fruit: Frequencies of intake ≥ 2 servings/d vs 1 serving/wk or fewer	Colon cancer: 1.05 (0.80-1.39); rectal cancer: 0.97 (0.58-1.64)	Age, family history of CRC, sigmoidoscopy, height, BMI, pack-years of smoking, alcohol intake, PA, menopausal status, postmenopausal hormone use, aspirin use, vitamin supplement intake, total caloric intake, red meat consumption
Franceschi et al[29], 1998	Italy	1953 (1225 colon cancer, 728 rectal cancer)/4154 (2073 males, 2081 females)	Validated FFQ; 78 food items	Mean weekly servings: Citrus fruit Q5 7.5 (high)/Q1 1.0 (low); apples/pears: Q5 15.0 (high)/Q1 3.0 (low); bananas: Q5 3.0 (high)/Q1 0.5 (low); kiwi: Q5 4.0 (high)/Q1 0.5 (low); peaches/apricots/prunes: Q5 5.0 (high)/Q1 0.8 (low); melon: Q5 0.5 (high)/Q1 0.1 (low); grapes: Q5 1.0 (high)/Q1 0.2 (low); Strawberries/cherries: Q5 0.4 (high)/Q1 0.1 (low)	Citrus: Total: 1.02 (0.85-1.22), colon: 1.0 (0.9-1.1), rectal: 0.8 (0.7-1.0); apples/pears: Colon: 0.9 (0.8-1.1) rectal: 0.8 (0.7-1.0); bananas: Colon 1.0 (0.9-1.1), rectal 1.0 (0.8-1.1); kiwi: colon 0.9 (0.8-1.0) rectal 0.8 (0.7-1.0); peaches/apricots/prunes: Colon 1.0 (0.9-1.1) rectal 0.8 (0.7-0.9); melon: Colon 1.0 (0.9-1.0) rectal 0.9 (0.8-1.0); grapes: colon 1.0 (0.9-1.0) rectal 0.9 (0.8-1.0); strawberries/cherries: Colon 1.0 (0.9-1.0) rectal 0.9 (0.9-1.0)	Age, sex, centre, education, PA, total EI
Levi et al[49], 1999	Swiss	223 (males 142, females 81) (119 colon cancer, 104 rectal cancer, median age 63 yr)/491 (211 males, 280 females, median age 58 yr)	FFQ; 79 food items	Citrus fruits (Servings per week): Q3 (> 3.5/wk) vs Q1 (1.5/wk)	Citrus fruits: 0.65 (0.40-1.05)	Age, sex, education, smoking, alcohol, BMI, PA, meat and vegetable consumption, total EI
Le Marchand et al[56], 1997	Hawaii	1192 (698 males, 494 females) (mean age 66 yr)/1192 (698 males, 494 females) (mean age 66 yr)	Validated FFQ; 282 food items	Bananas: Male ≥ 55 g/d (Q4 high) vs ≤ 9 g/d (Q1 low), female: ≥ 54 g/d (Q4 high) vs ≤ 11 g/d (Q1 low); citrus fruits: Male ≥ 52 g/d (Q4 high) vs ≤ 4 g/d (Q1 low), female: ≥ 58 g/d (Q4 high) vs ≤ 8 g/d (Q1 low)	Bananas: Male: 0.7 (0.5-1.1), female: 0.6 (0.4-0.9); citrus fruits: Male: 0.9 (0.6-1.3), female: 0.9 (0.6-1.4)	Age, family history of CRC, alcoholic drinks per week, pack-years of cigarette smoking, lifetime recreational activity, Quetelet index 5 years earlier, total calories, egg, and calcium
Deneo-Pellegrini et al[57], 1996	Uruguay	160 (71 rectal cancer, 89 colon cancer)/221	FFQ; 61 food items	(T3; high) vs (T1; low)	Orange: 0.76 (0.47-1.19); apple: 0.40 (0.25-0.66); peach: 1.05 (0.65-1.69); pear: 1.06 (0.65-1.74); grape: 1.61 (0.94-2.74); fig: 1.36 (0.73-2.54); banana: 0.28 (0.16-0.50)	Age, sex, residence, education, BMI, total EI, alcohol intake
Lin et al[36], 2006	United States (NHS and HPFS)	71976 female cohort; 498 incident cases (30-55 yr); 35425 male cohort; 380 incident cases (40-75 yr); follow-up 10 yr	Validated FFQ; 131 food items	Apple: ≥ 2 servings/d (Q5; high) vs 0-2 servings/wk (Q1; low)	Total: 0.75 (0.52-1.08); NHS females: 0.64 (0.35-1.17); HPFS males: 0.82 (0.51-1.30)	Age, BMI, PA, history of CRC, previous colorectal polyps, prior screening sigmoidoscopy or colonoscopy, smoking, multivitamin use, current aspirin use, alcohol, EI, red meat, total Ca, total folate, total fibre
Theodoratou et al[37], 2007	United Kingdom	1456 (mean 63.9 yr ± 9.6 yr) yr)/1456 (64.7 yr ± 9.5 yr)	Validated FFQ; 150 food items	Apples: Q4 (high) vs Q1 (low)	Apples: 0.96 (0.62-1.50)	Age, sex, residence area, family history of CRC, total EI, fibre, alcohol, NSAID, smoking, BMI, PA
Deneo-Pellegrini et al[38], 2002	Uruguay	484 (260 colon cancer, 224 rectal cancer)/1452	FFQ; 64 food items	Citrus fruits estimate: Q4 (high) vs Q1 (low); banana estimate: Q4 (high) vs Q1 (low)	Total: Citrus fruits: 0.8 (0.6-1.1), banana: 0.6 (0.4-0.8); citrus fruits: Male: 0.5 (0.3-0.8), female: 1.5 (0.9-2.5); colon: 0.9 (0.9-1.1); rectum: 0.9 (0.7-0.9); banana: male: 0.6 (0.4-0.9), female: 0.6 (0.3-0.9); colon: 0.8 (0.7-0.9); rectum: 0.9 (0.8-1.1)	Age, residence, urban/rural status, education, family history of colon cancer for first-degree relatives, BMI, total EI and red meat intakes

NA: Not available; EI: Energy intake; PA: Physical activity; CRC: Colorectal cancer; BMI: Body mass index; MET: Metabolic equivalentr; NSAID: Non-steroidal anti-inflammatory drugs; NHS: Nurses’ Health Study; HPFS: Health professionals follow-up study.

Table 2 The main adjusted factors of studies included in the meta-analysis

Ref.	Adjusted confounders
	Age	Sex	Energy intake	BMI	Family history of CRC	Alcohol use	Smoking status	Physical activity	Education level	Red meat
Lee et al[39], 2017	√		√	√		√		√	√	√
Leenders et al[40], 2015				√		√	√	√
Abu Mweis et al[41], 2015	√	√	√		√		√		√
Tayyem et al[42], 2014	√	√	√		√		√	√	√
Rosato et al[43], 2013	√	√	√		√	√			√
Vogtmann et al[50], 2013	√		√	√	√	√	√	√	√	√
Annema et al[44], 2011	√	√	√	√		√	√	√
Foschi et al[45], 2010	√	√	√	√		√	√	√	√
Li et al[51], 2010	√	√	√	√	√	√	√	√	√
Jedrychowski et al[32], 2010	√	√	√				√
Williams et al[46], 2009	√	√	√	√	√			√	√
Nomura et al[52], 2008	√		√	√	√	√	√	√		√
Gallus et al[47], 2005	√	√	√	√		√	√	√	√
Lin et al[55], 2005	√		√	√	√	√	√	√		√
Satia-Abouta et al[48], 2004	√	√	√	√	√		√	√	√
Voorrips et al[53], 2000	√				√	√
Michels et al[54], 2000	√		√	√	√	√	√	√		√
Franceschi et al[29], 1998	√	√	√					√	√
Levi et al[49], 1999	√	√	√	√		√	√	√	√
Le Marchand et al[56], 1997	√		√		√	√	√
Deneo-Pellegrini et al[57], 1996	√	√	√	√		√			√
Lin et al[55], 2005	√		√	√	√	√	√	√		√
Theodoratou et al[37], 2007	√	√	√	√	√	√	√	√
Deneo-Pellegrini et al[38], 2002	√		√	√	√				√	√

BMI: Body mass index; CRC: Colorectal cancer.

Table 3 Risk of bias of 24 included studies, based on the Risk of Bias In Non-randomized Studies of Interventions-I tool

Ref.	Confounding	Selection of participants	Classification of interventions	Deviations from intended interventions	Bias due to missing data	Measurement of outcomes	Selection of reported result	Overall rating
Lee et al[39], 2017	Moderate	Low	Low	Low	Moderate	Low	Low	Moderate
Abu Mweis et al[41], 2015	Moderate	Low	Moderate	Low	Moderate	Low	Low	Moderate
Leenders et al[40], 2015	Moderate	Moderate	Low	Low	Moderate	Low	Low	Moderate
Tayyem et al[42], 2014	Moderate	Low	Low	Low	Low	Low	Low	Moderate
Rosato et al[43], 2013	Moderate	Low	Low	Low	Moderate	Low	Low	Moderate
Vogtmann et al[50], 2013	Low	Low	Low	Low	Moderate	Low	Low	Moderate
Annema et al[44], 2011	Moderate	Low	Low	Low	Moderate	Low	Low	Moderate
Foschi et al[45], 2010	Moderate	Moderate	Low	Low	Moderate	Low	Low	Moderate
Jedrychowski et al[32], 2010	Moderate	Low	Moderate	Low	Moderate	Low	Low	Moderate
Williams et al[46], 2009	Moderate	Low	Low	Low	Moderate	Low	Low	Moderate
Li et al[51], 2010	Moderate	Low	Low	Low	Moderate	Low	Low	Moderate
Gallus et al[47], 2005	Moderate	Low	Moderate	Low	Moderate	Low	Low	Moderate
Lin et al[36], 2006	Low	Moderate	Low	Low	Moderate	Low	Low	Moderate
Satia-Abouta et al[48], 2004	Moderate	Low	Low	Low	Moderate	Low	Low	Moderate
Voorrips et al[53], 2000	Moderate	Moderate	Low	Low	Moderate	Low	Low	Moderate
Franceschi et al[29], 1998	Moderate	Low	Moderate	Low	Moderate	Low	Low	Moderate
Levi et al[49], 1999	Moderate	Low	Low	Low	Moderate	Low	Low	Moderate
Le Marchand et al[56], 1997	Moderate	Low	Moderate	Low	Moderate	Low	Low	Moderate
Deneo-Pellegrini et al[57], 1996	Moderate	Low	Low	Low	Moderate	Low	Low	Moderate
Theodoratou et al[37], 2007	Moderate	Low	Low	Low	Moderate	Low	Low	Moderate
Deneo-Pellegrini et al[38], 2002	Moderate	Low	Low	Low	Moderate	Low	Low	Moderate
Nomura et al[52], 2008	Moderate	Low	Low	Low	Moderate	Low	Low	Moderate
Lin et al[55], 2005	Moderate	Low	Low	Low	Moderate	Low	Low	Moderate
Michels et al[54], 2000	Moderate	Moderate	Low	Low	Moderate	Low	Low	Moderate