IJC International Journal of Cancer
Dietary supplement use and colorectal cancer risk: A systematic review and meta-analyses of prospective cohort studies €ring1, Renate M. Winkels1, Jacoba M.S. Renkema2, Lea Kragt1, Anne-Claire B. van Orten-Luiten1, Renate C. Heine-Bro 1 Ettje F. Tigchelaar , Doris S.M. Chan3, Teresa Norat3 and Ellen Kampman1,4,5 1
Division of Human Nutrition, Wageningen University, Wageningen, The Netherlands Wageningen UR library, Wageningen University, Wageningen, The Netherlands 3 Department of Epidemiology and Biostatistics, School of Public Health, Imperial College London, London, United Kingdom 4 Department of Epidemiology, Biostatistics, and HTA, Radboud University Nijmegen Medical Centre, Nijmegen, The Netherlands 5 Department for Health Sciences, VU University Amsterdam, Amsterdam, The Netherlands 2
Key words: colorectal cancer, dietary supplements, epidemiology, meta-analysis, nutrition, review Abbreviations: AICR: American Institute for Cancer Research; CC: colon cancer; 95% CI: 95% confidence interval; CRC: colorectal cancer; DR: “Dose-response” meta-analysis; H-L: “Highestlowest” meta-analysis; PHS II trial: physicians’ health study II randomized controlled trial; RC: rectal cancer; RR: relative risk; U-NU: “use-no use” meta-analysis; WCRF: World Cancer Research Fund Additional Supporting Information may be found in the online version of this article. Conflict of interest: The authors have declared no conflict of interest. Grant sponsor: Wereld Kanker Onderzoeks Fonds (WCRF NL); Grant sponsor: World Cancer Research Fund International (WCRF International) DOI: 10.1002/ijc.29277 History: Received 18 Apr 2014; Accepted 19 Sep 2014; Online 21 Oct 2014 Correspondence to: Prof. E. Kampman, Division of Human Nutrition, Wageningen University, P.O. Box 8129, 6700 EV Wageningen, The Netherlands, Tel.: 0031 (0)317 483867, Fax: 0031 (0)317 482782, E-mail:
[email protected] C 2014 UICC Int. J. Cancer: 00, 00–00 (2014) V
Diet and lifestyle play a role in colorectal cancer development.1–3 The “Food, Nutrition, Physical activity and the prevention of cancer: a global perspective” Report in 20074 and the Continuous Update Project Report on colorectal cancer in 20115 of the World Cancer Research Fund and the American Institute for Cancer Research (WCRF/AICR) concluded that red and processed meat, alcoholic drinks among men, high body and abdominal fatness and adult attained height increase the risk for colorectal cancer, while foods containing dietary fibre and being physically active decrease the risk for colorectal cancer.5 However, no conclusions about the evidence on dietary supplement use and colorectal cancer risk could be made as too few studies were available to show consistent associations. Since use of dietary supplements is rising in countries where colorectal cancer is prevalent, it is of great relevance to summarize the evidence between dietary supplement use and colorectal cancer risk. Several randomized trials on dietary supplement use and chronic diseases have been conducted. The primary endpoints in those trials were cardiovascular diseases, osteoporosis and (overall) cancer,6–9 but not colorectal cancer. Moreover, in those trials very specific subgroups, for example, postmenopausal women,10,11 heavy smokers,12,13 health professionals,14–16 and populations at high risk for cancer14,17,18 and cardiovascular diseases19,20 were studied, which
Epidemiology
Use of dietary supplements is rising in countries where colorectal cancer is prevalent. We conducted a systematic literature review and meta-analyses of prospective cohort studies on dietary supplement use and colorectal cancer risk. We identified relevant studies in Medline, Embase and Cochrane up to January 2013. Original and peer-reviewed papers on dietary supplement use and colorectal cancer, colon cancer, or rectal cancer incidence were included. “Use-no use”(U-NU), “highest-lowest”(H-L) and “dose-response”(DR) meta-analyses were performed. Random-effects models were used to estimate summary estimates. In total, 24 papers were included in the meta-analyses. We observed inverse associations for colorectal cancer risk and multivitamin (U-NU: RR 5 0.92; 95% CI: 0.87,0.97) and calcium supplements (U-NU: RR 5 0.86; 95% CI: 0.79,0.95; H-L: RR 5 0.80; 95% CI: 0.70,0.92; DR: for an increase of 100 mg/day, RR 5 0.96; 95% CI: 0.94,0.99). Inconsistent associations were found for colon cancer risk and supplemental vitamin A and vitamin C, and for colorectal cancer risk and supplemental vitamin D, vitamin E, garlic and folic acid. Meta-analyses of observational studies suggest a beneficial role for multivitamins and calcium supplements on colorectal cancer risk, while the association with other supplements and colorectal cancer risk is inconsistent. Residual confounding of lifestyle factors might be present. Before recommendations can be made, an extensive assessment of dietary supplement use and a better understanding of underlying mechanisms is needed.
2
Dietary supplement use and colorectal cancer risk
What’s new? Use of dietary supplements is rising in countries where colorectal cancer is prevalent. In this meta-analysis, the authors analysed the association between dietary supplement use and colorectal cancer risk. The results of the study suggest a beneficial role for multivitamins and calcium supplements on colorectal cancer risk, while the association with other supplements and colorectal cancer risk is inconsistent.
makes it difficult to generalize the results to the general population. According to meta-analyses conducted on the results of these trials, supplemental antioxidants,6,7 folic acid9 and calcium8 did not significantly influence colorectal cancer risk. To the best of our knowledge, no systematic review and meta-analyses of prospective cohort studies have been conducted that focus on colorectal cancer risk and use of dietary supplements specifically. Pooled analyses from 13 cohort studies on total intakes or intakes from foods only, but not solely from supplements, of vitamins A, C, E and folate showed modest inverse associations on colon cancer risk.21,22 A statistically significant inverse association was also found for multivitamin use and colon cancer risk.21 In addition, a pooled analysis of 10 cohort studies showed that total calcium intake was inversely associated with colorectal cancer risk.23 We conducted a systematic review and meta-analyses of prospective cohort studies on dietary supplement use and colorectal cancer risk up to January 2013. Because of the abundant prospective data from cohort studies and potential recall bias from case-control studies, case-control studies were not summarized.
Materials and Methods
Epidemiology
Search strategy
The present systematic literature review has been carried out according to the guidelines of the WCRF.24 The search strategy identified several terms on dietary supplement use and colorectal cancer risk, and yielded studies with outcomes on colorectal adenomas and colorectal carcinomas as it was part of a larger research project. As the current study focuses on colorectal carcinomas only, studies on colorectal adenomas were excluded. We retrieved relevant articles by searching in Medline, Embase and Cochrane from their inception up to January 2013, and hand-searched reference lists for additional studies (Supporting Information 1). No language restrictions were made. Study selection
We included prospective cohort studies if they reported original and peer-reviewed data on the association of dietary supplement use and colorectal, colon, or rectal cancer incidence. To be included in the meta-analyses, information on ascertainment of colorectal cancer cases, and estimates of the relative risk with 95% confidence intervals (95% CI) were required in the publication. When we identified multiple
articles on the same study, we selected the publication with the largest number of cases, a longer follow-up and completeness of the information. Data extraction
From each relevant study, information on study characteristics, number of cases, follow-up period, cancer site, description of dietary supplement use, relative risks (RRs) and 95% CIs, and details of the adjustment for confounders were extracted and stored in a database. The literature search (RHB, RW, LK, AOL, ET), data selection (RHB, RW, LK, AOL, ET) and data extraction (RHB, LK, AOL, ET) were performed independently by several reviewers at Wageningen University, Wageningen, the Netherlands. Statistical analyses
Random effects models25 were used to calculate summary RRs and 95% CIs for the associations of colorectal, colon or rectal cancer with use of multivitamins, vitamin A, vitamin C, vitamin D, vitamin E, calcium, folic acid and garlic supplements. We performed a meta-analysis if at least two cohort studies were available. We used the most fully adjusted relative risk in the analysis. “Use-no use” meta-analyses were done for the association between multivitamins and supplemental vitamin A, vitamin C, vitamin D, vitamin E, calcium and garlic with colorectal cancer risk. In those meta-analyses, “no use” incorporates either “never use,” “no current use” and/or “no past use.” In addition, “use” was defined as “current use,” “past use” and/ or “any use” in those meta-analyses. Studies that focussed on current use of dietary supplements only, and reported on specified categories of dietary supplement use were included in the “highest-lowest” meta-analyses. In those analyses, we always compared the highest versus the lowest category of intake, and did not include the middle categories, and we included the association as reported in the original publication, which could be tertiles, quartiles, or quintiles. Details about the contrasts between categories in the original publications can be found in Table 1. For the association between supplemental vitamin A, vitamin C, vitamin D, vitamin E, calcium and folic acid and colorectal cancer risk, “highestlowest” meta-analyses were conducted. In the “doseresponse” meta-analyses, we tested whether there was a linear association between the dosage of a supplement and colorectal cancer risk: thus, in those analyses we could only include studies that provided information on the dosage of C 2014 UICC Int. J. Cancer: 00, 00–00 (2014) V
Study characteristics
Cohort (N 537,916) 10 years Women 220 CRC cases Age: 45
Cohort (N 5 145,260) 5 years Men and women 797 CRC cases Age: 30
Cohort (N 5 161,808) 8 years Women 1,590 CRC cases Age: 50–79
Cohort (N 5 10,998) 17 years Men and women 95 CRC cases Age: 16–89
Cohort (N 5 35,197) 14 years Women 954 CRC cases Age: 55–69
Cohort (N 5 182,099) 11 years Men and women 1,494 CRC cases (men) 1,292 CRC cases (women) Age: 45–75
Cohort (N 5 135,151) 24 years Men and women 2,295 CRC cases Age: 30–75
Author, year, study, country
Zhang, 2006, Women’s Health Study, USA26
Jacobs, 2003, Cancer Prevention Study II Nutrition Cohort, USA,27
Neuhouser, 2009, Women’s Health Initiative, USA,28
Sanjoaquin, 2004, Oxford Vegetarian Study, UK,29
C 2014 UICC Int. J. Cancer: 00, 00–00 (2014) V
McCarl, 2006, Iowa Women’s Health Study, USA,30
Park, 2011, Multiethnic Cohort Study, USA,31
Lee, 2011, Nurses’ Health Study2 Health Professionals Follow-up Study, USA,32
Any use at least 1x per week, CRC
0.99
0.96 0.89 0.89
use use use use
Never vs. past use
Never vs. current 1–5 y Never vs. current 6–9 y Never vs. current 10–15 y Never vs. current 16–19 y Never vs. current use 20 y
Confounders1
1, 3, 4, 5, 7, 8, 10, 11, 12, 13, 14
X X
0.53, 0.96 0.64, 0.94
0.71 0.77
X
X
X
X
X
X
X
X
X
X
X
X
X
Use-no DoseHighestuse response lowest
X
0.75, 1.06
0.81, 1.13
0.82, 1.21 1, 3, 4, 5, 6, 7, 8, 9, 10, 13
0.66, 1.75 1, 3, 4, 5, 7, 8, 9, 0.43, 1.18 10, 13, 14
0.71, 0.94 1
0.63, 1.59 1, 2, 7, 8
0.88, 1.11
0.87, 1.23
10, 14
0.44, 1.19 1, 2, 4, 5,
0.72, 1.61
7, 8, 9, 10, 13
0.64, 1.37 1, 3, 4, 5, 6,
95% CI
0.76, 1.05
0.71
Multivitamins Duration of use,3 CRC
1.08
0.82
1.00
0.99
1.04
0.73
No use vs. use at both time points (men) No use vs. use at both time points (women)
No vs. yes
No use vs. use
No vs. yes
No past & recent use vs. Occasional use (1–3x/week) No past & recent use vs. Regular use ( 4x/week)
Multivitamins Use at least 1x per week over the last year at baseline and during 5yrs FU, CRC
Multivitamins Use over the past year, CRC
Multivitamins Use,2 CRC
Multivitamins
Recent use in 1992–1993, CRC
1.07
Never vs. current use
Multivitamins
Multivitamins
0.94
Never use vs. past use
Status,2 CRC
Supplement RR
Exposure categories/increment
Exposure details, outcome
Epidemiology
Self-report Medical records
Cancer registry Population registry
Cancer registry Population registry
Population registry
Self-report Medical records
Self-report Medical records
Self-report Medical records
Case ascertainment
Table 1. Publications on dietary supplement use and colorectal cancer risk included in the meta-analyses
€ring et al. Heine-Bro
3
Cohort (N 535,216) 10 years Women 241 CC cases without family history of CRC Age:55–69
Sellers, 1998, Iowa Womens’s Health Study, USA,34
Cohort (N 556,332) 10.6 years Men and women 465 CC cases 283 RC cases Age: 50–64
Cohort (N 511,580) 8 years Men and women 97 CC cases (men) 105 CC cases (women) Age: 65–84
Shibata, 1992, Leisure World Study, USA,33
Roswall, 2010, Diet Cancer and Health Study, Denmark,35
Study characteristics
Author, year, study, country
Cancer registry
Cancer registry
Medical records
Case ascertainment
Folic acid
Vitamin E
Vitamin C
Calcium
Vitamin D
Vitamin E
Vitamin C
X
0.81, 1.06
0.93
0 0 0 0 0 0 0 0 0 0
vs. 5,000 IU/day vs. > 5,000 IU/day vs. 180 mg/day vs. > 180 mg/day vs. 30 mg/day vs. > 30 mg/day vs. 400 IU/day vs. > 400 IU/day vs. 500 mg/day vs. > 500 mg/day
0.73, 1.00, 0.64, 0.70, 0.73, 0.96, 0.55, 0.61, 0.58, 0.97, 0.34, 0.51, 0.36,
1.05 1.01 0.93 1.05 1.06 1.01 0.79 0.90 0.83 0.98 0.56 0.82 0.60
1.49 1.03 1.36 1.57 1.56 1.06 1.13 1.30 1.20 1.06 0.91 1.33 0.99
0.95, 1.04 4, 5, 7, 8, 0.82, 1.70 10, 13, 14 0.86, 1.87
1.10 1, 9 1.30 1.10 1.40 1.20 0.90 0.90 1.30 0.90 0.90 1.00 1.18 1.27
0.50, 0.50, 0.50, 0.50, 0.10, 0.40, 0.40, 0.50, 0.50, 0.40,
X X X
X X
X X X
X
X X
X X X X X X X X X X
X
X
X
X
X
X
X
X
X
X
X
X X
X X 0.52, 1.12 0.76 0.70 0.80 0.70 0.70 0.80 0.60 0.60 0.80 0.70 0.60
X X
X X
0.45, 0.99
0.67
0.68, 1.51
X X
0.62, 1.38
0.92
1.01
X X
0.42, 0.94
0.63
X X X
0.66, 1.49 1, 7
0.99
X X X X X X X X
X
0.78, 1.00
0.89
No vs. yes (10,000 IU/ day, men) No vs. yes (10,000 IU/ day, women) No vs. yes (500 mg/day, men) No vs. yes (500 mg/day, women) No vs. yes (200 IU/day, men) No vs. yes (200 IU/day, women)
X
0.86, 1.10
X
Use-no DoseHighestuse response lowest
0.97
Confounders1
< 50 vs. 100– 6.66– 10 mg 0 vs. > 10 mg Supplemental folic Per 100 mcg folate acid during the last 0 vs. >0–83.2 mcg year, CC 0 vs. >83.2–142.8 mcg 0 vs. > 142.8 mcg Supplemental folic Per 100 mcg folate acid during the last 0 vs. >0–83.2 mcg year, RC 0 vs. >83.2–142.8 mcg 0 vs. > 142.8 mcg
the past
the past
the past
the past
the past
Current use at least 1x per week,2 CC
Vitamin E
Use over year, CC Use over year, CC Use over year, CC Use over year, CC Use over year, CC
Current use at least 1x per week,2 CC
Vitamin C
Vitamin A
Current use at least 1x per week, CC
Cumulative average of past and recent use of synthetic folic acid from fortification and supplementation, CRC
Vitamin A
Folic acid
Supplement
Exposure details, outcome
Table 1. Publications on dietary supplement use and colorectal cancer risk included in the meta-analyses (Continued)
Epidemiology
4 Dietary supplement use and colorectal cancer risk
C 2014 UICC Int. J. Cancer: 00, 00–00 (2014) V
Study characteristics
Cohort NHS (N 5 87,998) 18 years Women 626 CC cases Age: 30–55 Cohort HPFS (n 5 47,344) 8 years Men 399 CC cases Age: 40–75
Cohort (N 5 36,976) 10 years Women 223 CRC cases Age: 45
Cohort (N 589,448) 12 years Women 501 CRC cases Age: 30–55
Cohort (N 547,935) 6 years Men 203 CC cases Age: 40–75
Cohort (N5191,011) 7.3 years Men and women 2,110 CRC cases Age: 45–75
Cohort (N 5 9,959) 24 years Men and women 72 CRC cases Age: 15
Author, year, study, country
Wu, 2002, Nurses’ Health Study & Health Professionals Follow-up Study, USA,36
Lin, 2005, Women’s Health Study, USA,37
C 2014 UICC Int. J. Cancer: 00, 00–00 (2014) V
Martinez, 1996, Nurses’ Health Study, USA,38
Kearney, 1996, Health Professionals Follow-up Study, USA,39
Park, 2007, Multiethnic Cohort Study, USA,40
€rvinen, 2001, FinJa nish prospective cohort study, Finland,41
Vitamin D
Calcium
Vitamin D
Vitamin D
Vitamin D
Calcium
Vitamin D
Vitamin E
Supplement
Epidemiology
Cancer registry
Cancer registry
Self-report Medical records
Self-report Medical records
Self-report Medical records
Self-report Medical records
Case ascertainment
0.95 0.90 0.78 0.79 1.09 0.73 0.70
vs. 250 IU vs. 300–500 vs. 600 IU vs. vitamin E vs. 250 IU vs. 300–500 vs. 600 IU
< 4.0 vs. 4.0–86 IU/day < 4.0 vs. 87–342 IU/day < 4.0 vs. 343–447 IU/day < 4.0 vs. 448 IU/day
Use, CRC
2
No use vs. Use
Confounders1
0.60, 0.90 0.84, 1.20 0.69, 0.98
0.74 1.00 0.82
0.08, 4.22 1, 2, 4, 7, 11, 9, 10,
0.75, 1.26 0.81, 1.09
0.65 0.98 0.97 0.94
0.22, 1.02 0.80, 1.06 1, 3, 4, 5, 6, 7, 9, 10, 12 0.49, 0.84 0.84, 1.15
0.48 0.92
0.70, 1.79 1, 3, 4, 5, 6, 0.26, 1.18 7, 8, 9, 10 0.59, 1.27
0.79, 1.15 1, 3, 4, 5, 6, 7, 8, 10
0.90, 1.87
0.95, 1.95 1, 3, 4, 5, 7, 8, 9, 10, 13 0.70, 1.38
0.38, 1.29
0.72, 1.64 0.52, 1.03
0.43, 1.42 0.46, 1.34
0.54, 1.19 1, 3, 5, 7, 6, 8, 10, 13 0.65, 1.39 0.66, 1.22
95% CI
1.12 0.55 0.87
0.95
1.30
0 vs. 500 mg/day No use vs. Use
0.98
0 vs. >0–499 mg/day
1.36
0.80
RR
vs. vitamin E
0 vs. > 0–400 IU/day
Never use pills only Never use Never use IU Never use Never use pills only Never use Never use IU Never use
Exposure categories/increment
0 vs. 1– 400 IU/day (men) 0 vs. > 400 IU/day 0 vs. 1– 400 IU/day (women) 0 vs. > 400 IU/day Supplemental cal0 vs. 1–< 200 mg/day cium at least weekly (men) during the last year, 0 vs. 200 mg/day CRC 0 vs. 1–< 200 mg/day (women) 0 vs. 200 mg/day
Supplemental vitamin D at least weekly during the last year, CRC
Supplementary vitamin D during the last year, CC
Use during the last year, CRC
Use during the last year, CRC Use during the last year, CRC
Use during the last year, CC Vitamin E supplementation from vitamin E pills only, CC
Use during the last year, CC Vitamin E supplementation from vitamin E pills only, CC
Exposure details, outcome
Table 1. Publications on dietary supplement use and colorectal cancer risk included in the meta-analyses (Continued)
X
X
X X
X X
X X
X
X
X
X
X
X
X
X
X X
X X
X X
X
X
X X X
X
X
X
X X
X
X X
X
X
X
X
X
X
X
X
X
Use-no DoseHighestuse response lowest
€ring et al. Heine-Bro
5
Study characteristics
Cohort (N 545,354) 8.5 years Women 482 CRC cases Average age: 61.9
Cohort (N 5 492,810) 7 years Men and women 5,098 CRC cases Age: 50–71
Cohort (N 5 127,749) 5 years Men and women 683 CRC cases Age: 50–74
Casecohort (N 5 120,852) 3.3 years Men and women 326 CRC cases Age: 55–69
Cohort (N 5 135,342) 16 years Men and women 1,025 CRC cases Age: 30–75
Cohort (N 5 99,523) 8 years Men and women 1,023 CRC cases Age: 50–74
Author, year, study, country
Flood, 2005, Breast Cancer Detection Demonstration Project, USA,42
Park, 2009, NIH-AARP Diet and Health Study, USA,43
McCullough, 2003, Cancer Prevention Study II Nutrition Cohort, USA,44
Kampman, 1994, Netherlands Cohort Study, the Netherlands,45
Wu, 2002, Nurses’ Health Study & Health Professionals Follow-up Study, USA,46
Stevens, 2011, Cancer Prevention Study II Nutrition Cohort, USA,47
Self-report Medical records
Self-report Medical records
Cancer registry Pathology registry
Self-report Medical records
Cancer registry
Cancer registry Population registry
Case ascertainment
Folic acid
Calcium
Calcium
Calcium
Calcium
Calcium
Supplement
Folic acid from fortification and supplementation during the last year, CRC
< 101 vs. 101–