If you don't remember your password, you can reset it by entering your email address and clicking the Reset Password button. You will then receive an email that contains a secure link for resetting your password
If the address matches a valid account an email will be sent to __email__ with instructions for resetting your password
Fish consumption may protect against colorectal cancer, but results from observational studies are inconsistent; therefore, a systematic review with a meta-analysis was conducted.
Methods
Relevant studies were identified by a search of MEDLINE and EMBASE databases to May 2011, with no restrictions. Reference lists from retrieved articles also were reviewed. Studies that reported odds ratio (OR) or relative risk estimates with 95% confidence intervals (CIs) for the association between the consumption of fish and the risk of colorectal, colon, or rectal cancer were included. Two authors independently extracted data and assessed study quality. The risk estimate (hazard ratio, relative risk, or OR) of the highest and lowest reported categories of fish intake were extracted from each study and analyzed using a random-effects model.
Results
Twenty-two prospective cohort and 19 case-control studies on fish consumption and colorectal cancer risk met the inclusion criteria and were included in the meta-analysis. Our analysis found that fish consumption decreased the risk of colorectal cancer by 12% (summary OR, 0.88; 95% CI, 0.80-0.95). The pooled ORs of colorectal cancer for the highest versus lowest fish consumption in case-control studies and cohort studies were 0.83 (95% CI, 0.72-0.95) and 0.93 (95% CI, 0.86-1.01), respectively. There was heterogeneity among case-control studies (P<.001) but not among cohort studies. A significant inverse association was found between fish intake and rectal cancer (summary OR, 0.79; 95% CI, 0.65-0.97), and there was a modest trend seen between fish consumption and colon cancer (summary OR, 0.96; 95% CI, 0.81-1.14). This study had no publication bias.
Conclusion
Our findings from this meta-analysis suggest that fish consumption is inversely associated with colorectal cancer.
Overall, colorectal cancer has a disease-specific mortality rate of approximately 33% in developed countries, making it a significant death concern and burden in these countries. The role that lifestyle plays in the cause of colorectal cancer remains an active area of research, and includes smoking,
and overall diet in general. In 2007, the World Cancer Research Fund released a report exploring the role of red and processed meat, obesity, and alcohol in the development of colorectal cancer.
World Cancer Research Fund/American Institute for Cancer Research Food, Nutrition, Physical Activity, and the Prevention of Cancer: A Global Perspective.
American Institute for Cancer Research,
Washington DC2007
In addition to these risks, it is estimated that dietary factors contributed to approximately one half of all colorectal cancer cases that were diagnosed.
This meta-analysis was conducted of currently available epidemiologic studies on the association between consumption of fish and colorectal cancer to provide a quantitative evaluation in a standardized format on whether fish consumption may decrease the risk of colorectal cancer by 12%.
•
The relationships between fish consumption and colon or rectal cancer are investigated separately, which may provide better insight into colorectal cancer prevention.
In a population with higher rates of fish consumption, such as that of Finland
the incidence and mortality from colorectal cancer are greatly reduced. Although some studies have demonstrated an inverse relationship between fish consumption and colorectal cancer,
As a result, the nature of the association between dietary fish consumption and risk of colorectal cancer remains controversial.
Opposing roles of omega-3 and omega-6 polyunsaturated fatty acids were hypothesized to be behind the beneficial effect of fish consumption in preventing the pathogenesis of t-colorectal cancer.
An n-3 PUFA-rich microalgal oil diet protects to a similar extent as a fish oil-rich diet against AOM-induced colonic aberrant crypt foci in F344 rats.
Another group showed that fish fatty acids have the capacity to regulate cell proliferation and apoptosis in human colorectal cancer cell lines, but at the same time fish consumption had no impact on apoptosis induction ex vivo.
Fish fatty acids alter markers of apoptosis in colorectal adenoma and adenocarcinoma cell lines but fish consumption has no impact on apoptosis-induction ex vivo.
On the basis of all of the studies available, the relationship between fish consumption and colorectal cancer remains inconsistent, regardless of whether the experimental setup was clinically or laboratory derived. In a meta-analysis on this topic in 2007, Geelen et al
analyzed 18 prospective cohort studies and claimed that the pooled relative risks (RRs) for the highest compared with the lowest fish consumption category were 0.88 (95% CI, 0.78-1.00) for incidence and 1.02 (95% CI, 0.90-1.16) for mortality. More case-control and prospective cohort studies have been performed in recent years,
Development of a quantitative food frequency questionnaire for assessing food, nutrient, and heterocyclic aromatic amines intake in Japanese Brazilians for a colorectal adenoma case-control study.
and as such it is necessary to perform a systematic meta-analysis on fish consumption and colorectal cancer risk to more fully elucidate this relationship.
The current study conducted a meta-analysis of currently available epidemiologic reports that explored the association between consumption of fish and colorectal cancer with the goal of providing a quantitative evaluation in a standardized format that permitted a numeric analysis across studies.
Materials and Methods
Data Sources and Searches
Relevant studies were identified through MEDLINE and EMBASE using a combined text word and Medical Subject Headings (MESH) search strategy with the terms colorectal cancer, colorectal neoplasm, colorectal tumor, colorectal carcinoma, colon cancer, colon neoplasm, colon tumor, colon carcinoma, rectal cancer, rectal neoplasm, rectal tumor, and rectal carcinoma combined with fish. Furthermore, we reviewed the reference lists from retrieved articles to search for more studies.
Study Selection
To be included in our meta-analysis, the following criteria had to be met. First, the study had to have a case-control or cohort study design. Second, the exposure of interest was fresh fish consumption. Third, the number of colorectal cancer cases and controls had to be reported. Fourth, the RRs or odds ratios (ORs) with their corresponding 95% confidence interval (CI) for highest versus non/lowest level of fish intake had to be reported.
Data Extraction and Quality Assessment
If data were duplicated in more than 1 study, the most recent study was included in the analysis. We identified 46 potentially relevant articles concerning fish consumption and colorectal cancer risk.
After the duplicated studies were excluded, the remaining publications in the meta-analysis of fish and colorectal cancer included 41 articles: 22 cohort studies and 19 case-control studies. We used a standardized protocol and reporting form to abstract the following data from each publication: reference (first author, year of publication), study design, population used in the case-control study (population-based case-control or hospital-based case-control), gender of subjects and location of cancer (colon cancer, rectal cancer, and colorectal cancer), country in which the study was performed, number of cases and controls, lowest and highest level of fish consumption, RRs or ORs with 95% CIs for colorectal cancer associated with fish intake, and covariate adjustment.
Data Synthesis and Analysis
Study-specific ORs/RRs and corresponding 95% CIs for highest versus non/lowest fish consumption levels were extracted. For case-control studies, the proportions (expressed as percentages) of control subjects in the highest and non/lowest consumption categories were reported. For cohort studies, the percentages of subjects in the highest and non/lowest consumption levels were calculated as the proportion of the number of subjects in these 2 categories over the total number of study subjects or as the proportion of person-years in these 2 categories over the total person-years.
Q and Higgins I2 statistics were used to examine heterogeneity among the studies included in this meta-analysis.
For the Q statistics, a P value less than .10 indicated statistically significant heterogeneity. We defined statistical significance as P<.10 rather than the conventional level of .05 because of the low power of this test.
I2 lies between 0% (no observed heterogeneity) and 100% (maximal heterogeneity), and an I2 value greater than 50% may be considered to represent substantial heterogeneity. We used the fixed effect model to calculate the summary ORs and its 95% CI among studies with homogeneous results, whereas the random-effect model proposed by DerSimonian and Laird was used to calculate summary ORs when a significant heterogeneity was found. We calculated summary estimates for the 2 study types (case-control and cohort) separately and in combination.
To assess the potential for publication bias, we used funnel plots and Egger's regression.
All statistical analyses were performed with STATA (v 10.0; StataCorp, College Station, Tex).
Results
Literature Search
The detailed steps of our literature search are shown in Figure 1. Briefly, we identified 46 potentially relevant articles concerning fish consumption and colorectal cancer. Five articles were excluded because of duplicated reports from the same study population. In the end, 41 studies were included in this meta-analysis.
Figure 1Selection of studies for inclusion in meta-analysis.
The 41 articles that met our inclusion criteria in this meta-analysis were published between 1990 and 2011. There were 22 cohort studies and 19 case-control studies (10 population-based and 9 hospital-based case-control studies).
Sensitive Analysis
According to the highest and lowest level of fish consumption reported, the RRs or ORs for each of the studies, along with their summary ORs, are shown in Table 1 and Figure 2. Significant heterogeneity was found in the results across the 41 studies (Q=131.97, P<.001, I2 = 56.80%). The summary ORs among all studies showed a significant reduction in risk of colorectal cancer in those populations with high fish consumption intake (summary OR, 0.88; 95% CI, 0.80-0.95).
Table 1Characteristics of Studies on Fish Consumption and Colorectal Cancer
Study
Design
Cases
Study Population/Controls
Gender of Subjects
Location of Cancer
Population
Range of Exposure
OR or RR (95% CI)
Adjustments
Bostick et al 1994
Cohort
212
35,215
Female
Colon
USA
<1 time/wk vs >2.5 times/wk, Q1 vs Q5
0.76 (0.49-1.18)
Age, energy intake, height, parity, vitamin E, vitamin E*age interaction, and vitamin A
Engeset et al 2007
Cohort
254
63,914
Female
Colon
Norway
<70.8 g/d vs >117 g/d, T1 vs T3
1.28 (0.9-1.81)
Age, daily intake of energy, smoking, fish liver, fruit and vegetables, fiber, fats, and sauces
English et al 2004
Cohort
451
37,112
Both
Colorectal
Australia
<1 times/wk vs ≥2.5 times/wk,Q1 vs Q4
0.9 (0.7-1.2)
Age, energy intake, country of birth, gender, fat, and cereal intake
Gaard et al 1996
Cohort
56
24,897
Female
Colorectal
Norway
<2 meals/wk vs ≥5 meals/wk
0.81 (0.32-2.06)
Age
Gaard et al 1996
Cohort
87
25,638
Male
Colorectal
Norway
<2 meals/wk vs ≥5 meals/wk
0.46 (0.19-1.11)
Age
Giovannucci et al 1994
Cohort
205
47,949
Male
Colon
USA
8.4 g/d vs 83.4 g/d
1.06 (0.7-1.6)
Age and total energy intake
Hall et al 2008
Cohort
500
21,406
Male
Colorectal
USA
<1 time/wk vs ≥5 times/wk
0.63 (0.42-0.95)
Age, smoking, BMI, multivitamin use, history of diabetes, random assignment to aspirin or placebo, vigorous exercise, alcohol, and red meat intake
Hsing et al 1998
Cohort
145
17,633
Male
Colorectal
USA
<0.8 time/mo vs >4 times/mo, Q1 vs Q4
1.5 (0.9-2.6)
Age, calories, smoking, alcohol intake, and total energy
Kato et al 1997
Cohort
100
14,727
Female
Colorectal
USA
Q1 vs Q4
0.49 (0.27-0.89)
Age, total energy intake, education, and place of residence
Khan et al 2004
Cohort
14
1634
Female
Colorectal
Japan
Never+several times/year+several times/mo vs several times/wk+daily
1.2 (0.4-3.7)
Age, health status, health education, health screening, and smoking
Khan et al 2004
Cohort
15
1524
Male
Colorectal
Japan
Never+several times/y+several times/mo vs several times/wk+daily
0.5 (0.2-1.4)
Age and smoking
Knekt et al 1999
Cohort
73
9985
Both
Colorectal
Finland
Q1 vs Q4
1.11 (0.55-2.28)
Age, energy intake, gender, municipality, and smoking
Kojima et al 2004
Cohort
116
45,181
Male
Rectum
Japan
3-7 times/wk vs 0-2 times/mo
0.95 (0.6-1.51)
Age, family history, BMI, smoking, physical activity, education, alcohol intake, and region
Kojima et al 2004
Cohort
138
45,181
Male
Colon
Japan
3-7 times/wk vs 0-2 times/mo
1.04 (0.65-1.66)
Age, family history, BMI, smoking, physical activity, education, alcohol intake, and region
Kojima et al 2004
Cohort
146
62,643
Female
Colon
Japan
3-7 times/wk vs 0-2 times/mo
0.97 (0.62-1.5)
Age, family history, BMI, smoking, physical activity, education, alcohol intake, and region
Kojima et al 2004
Cohort
57
62,643
Female
Rectum
Japan
3-7 times/wk vs 0-2 times/mo
0.9 (0.44-1.84)
Age, family history, BMI, smoking, physical activity, education, alcohol intake, and region
Larsson et al 2005
Cohort
733
61,433
Female
Colorectal
Sweden
<0.5 servings/wk vs ≥2 servings/wk
1.08 (0.81-1.43)
Age, energy, education, BMI, alcohol, saturated fat, calcium, fruits and vegetables, whole-grain foods, red meat, and poultry
Lee et al 2009
Cohort
394
73,224
Female
Colorectal
China
<20 g/d vs ≥74 g/d, Q1 vs Q5
1.3 (0.9-1.9)
Age, education, income, survey season, tea consumption, NSAID use, energy intake, and fiber intake
Lüchtenborg et al 2005
Cohort
154
2948
Both
Rectum
The Netherlands
0 g/d vs 28.2 g/d, Q1 vs Q4
0.94 (0.59-1.52)
Age, energy intake, gender, family history of colorectal cancer, smoking, and BMI
Lüchtenborg et al 2005
Cohort
434
2948
Both
Colon
The Netherlands
0 g/d vs 28.2 g/d, Q1 vs Q4
1.03 (0.76-1.4)
Age, energy intake, gender, family history of colorectal cancer, smoking, and BMI
Ma et al 2001
Cohort
193
14,916
Male
Colorectal
USA
≤0.14 servings/d vs 0.35-2.03 servings/d, T1 vs T3
0.92 (0.56-1.51)
Age, smoking, BMI, alcohol, multivitamin, aspirin, exercise, and molar ratio of IGF-I to IGFBP-3
Norat et al 2005
Cohort
1329
478,040
Both
Colorectal
10 European countries
<10 g/d vs ≥80 g/d
0.69 (0.54-0.88)
Age, energy intake, gender, height, weight, occupational physical activity, smoking, dietary fiber, alcohol, and center
Pietinen et al 1999
Cohort
185
27,111
Male
Colorectal
Finland
13 g/d vs 68 g/d, Q1 vs Q4
0.9 (0.6-1.4)
Age, education, smoking, BMI, alcohol, physical activity, and calcium intake
Sanjoaquin et al 2004
Cohort
95
10,998
Both
Colorectal
UK
0 times/wk vs ≥1 time/wk
1.17 (0.71-1.92)
Age, gender, smoking, and alcohol
Spencer et al 2010
Cohort
579
2575
Both
Colorectal
UK
<1 g/d vs ≥30 g/d
0.86 (0.64-1.16)
Age, height, weight, smoking, energy, alcohol, and dietary fiber
Sugawara et al 2009
Cohort
187
18,858
Female
Colorectal
Japan
Q1 vs Q4
0.96 (0.61-1.53)
Age, BMI, family history of cancer, history of stroke, hypertension, myocardial infarction and diabetes mellitus, education, marital status, job status, smoking, alcohol, time spent walking, total calories, and fruit and vegetables
Sugawara et al 2009
Cohort
379
20,640
Male
Colorectal
Japan
Q1 vs Q4
1.07 (0.78-1.46)
Age, BMI, family history of cancer, history of stroke, hypertension, myocardial infarction and diabetes mellitus, education, marital status, job status, smoking, alcohol, time spent walking, total calories, and fruit and vegetables
Tiemersma et al 2002
Cohort
102
537
Both
Colorectal
The Netherlands
0-1 time/mo vs >4 times/mo
0.7 (0.4-1.3)
Age, gender, center, total energy intake, alcohol, and height
Willett et al 1990
Cohort
150
88,751
Female
Colon
USA
<1 times/mo vs ≥17 times/mo
1.06 (0.36-3.12)
Age
Yang et al 2003
HCC
392
32,285
Female
Colon
Japan
<1 time/wk vs >4 times/wk
0.8 (0.52-1.24)
Age, season and year of questionnaire study, family history of colorectal cancer, smoking, alcohol, physical exercise, vegetables, and meat
Yang et al 2003
HCC
212
32,285
Female
Rectum
Japan
<1 time/wk vs >4 times/wk
0.62 (0.33-1.16)
Age, season and year of questionnaire study, family history of colorectal cancer, smoking, alcohol, physical exercise, vegetables, and meat
Yang et al 2003
HCC
536
14,601
Male
Colon
Japan
<1 time/wk vs >4 times/wk
0.68 (0.47-0.99)
Age, season and year of questionnaire study, family history of colorectal cancer, smoking, alcohol, physical exercise, vegetables and meat
Yang et al 2003
HCC
410
14,601
Male
Rectum
Japan
<1 time/wk vs >4 times/wk
1.13 (0.76-1.68)
Age, season and year of questionnaire study, family history of colorectal cancer, smoking, alcohol, physical exercise, vegetables, and meat
Busstra et al 2003
HCC
52
57
Both
Colorectal
The Netherlands
<1 time/mo vs >1 time/wk
0.5 (0.2-1.6)
Energy intake, age, familial background, and gender
Chiu et al 2003
PCC
469
701
Female
Colon
China
Q1 vs Q4
1.2 (0.8-1.7)
Age, total energy, education, BMI, income, and occupational physical activity
Chiu et al 2003
PCC
462
851
Male
Colon
China
Q1 vs Q4
1.7 (1.2-2.4)
Age, total energy, education, BMI, income, and occupational physical activity
Diergaarde et al 2005
PCC
278
414
Both
Colorectal
The Netherlands
<7 g/d vs >19 g/d, T1 vs T3
0.9 (0.6-1.3)
Age, gender, and energy intake
Fernandez et al 1999
HCC
828
7990
Both
Colon
Italy
<1 serving/wk vs ≥2 servings/wk
0.6 (0.5-0.7)
Age, gender, area of residence, education, smoking, alcohol, and BMI
Fernandez et al 1999
HCC
498
7990
Both
Rectum
Italy
<1 serving/wk vs ≥2 servings/wk
0.5 (0.3-0.6)
Age, gender, area of residence, education, smoking, alcohol, and BMI
Franceschi and Favero 1999
HCC
1953
4154
Both
Colorectal
Italy
Q1 vs Q5
0.72 (0.59-0.88)
Age, gender, center, education, physical activity, and total energy intake
Ganesh et al 2009
HCC
203
1628
Both
Colorectal
India
No vs yes
0.6 (0.4-0.9)
Age, place of residence, vegetables intake, dry fish and meat, religion, and occupation
Hu et al 2008
PCC
1727
5039
Both
Colon
Canada
Q1 vs Q4
1 (0.8-1.3)
Age, province, education, BMI, gender, alcohol use, smoking, vegetable and fruit intake, and total energy intake
Hu et al 2008
PCC
1447
5039
Both
Rectum
Canada
Q1 vs Q4
0.9 (0.7-1.2)
Age, province, education, BMI, gender, alcohol use, smoking, vegetable and fruit intake, and total energy intake
Figure 2Consumption of fish in association with colorectal cancer risk. Each study specific point estimate is plotted as a square box. The size of the box is proportional to the precision of the estimate, and its 95% CI is denoted by a horizontal line through the box. The vertical dashed line and the lower vertex of the diamond indicate the combined risk estimate of the analysis, and the left and right vertices of the diamond represent its 95% confidence limits. Results of the analysis using the random-effects model yielded a combined risk estimate of 0.88 (95% CI, 0.80-0.95; Q=131.97, P<.001, I2 = 56.80%). CI=confidence interval; OR=odds ratio.
By stratifying by case-control and prospective cohort studies, significant heterogeneity was found within the 19 case-control studies (Q=89.69, P<.001, I2 = 68.8%), and a significant association between fish intake and colorectal cancer risk also was observed (summary OR, 0.83; 95% CI, 0.72-0.95) (Supplemental Figure 1A). In cohort studies, there was no significant heterogeneity (Q=33.49, P=.22, I2 = 16.4%), and a slight association was found between fish consumption and colorectal cancer (summary OR, 0.93; 95% CI, 0.86-1.01) (Supplemental Figure 1B). When we stratified the various studies by design (case-control vs cohort), there was no significant heterogeneity between the 2 types of study designs (case-control vs cohort studies, Q=1.94, P=.16, I2 = 48.40%) (Table 2).
Table 2Results of Stratified Analysis on the Basis of Study Types, Study Populations, Location of Colorectal Cancer, and Incidence/Mortality of Cancer Risk
In case-control studies, the design could be further divided into hospital-based and population-based studies. By stratifying the 2 subgroups of case-control studies, significant heterogeneity was found within hospital-based (Q=33.91, P=.001, I2 = 61.70%) and population-based case-control studies (Q=31.65, P=.004, I2 = 55.8%). At the same time, significant heterogeneity was found between the 2 subgroups of case-control studies (Q=6.51, P=.01, I2 = 84.60%) (Table 2). In the hospital-based case-control studies, the summary OR among hospital-based case-control studies showed a significant reduction in risk of colorectal cancer with high fish consumption (summary OR, 0.70; 95% CI, 0.59-0.84) (Supplemental Figure 2A). Although the reduction in risk among population-based case-control studies was less dramatic than that seen above. There was still a trend toward an inverse relationship between fish intake and colorectal carcinogenesis (summary OR, 0.96; 95% CI, 0.81-1.13) (Supplemental Figure 2B).
To explore the source of the heterogeneity seen in the case-control studies, we further conducted analyses stratified by type of fish, area of residence (seaside or not), level of adjustment of overall diet or lifestyle, and other factors.
In the results presented above, we selected only the data of fresh fish. We explored the association between processed fish and colorectal cancer. The results indicated no significant heterogeneity (Q=4.89, P=.56, I2 = 0), whereas at the same time, there was a marginal increase in risk between intake of processed fish and colorectal cancer (summary OR, 1.12; 95% CI, 0.98-1.30).
The case-control studies were further stratified by the area of residence. Among 4 studies in which the population lived seaside, no significant heterogeneity was found (Q=5.36, P=.50, I2 = 0), and there was a significant protective effect of fish consumption on colorectal cancer noted (summary OR, 0.82; 95% CI, 0.69-0.96). Among the 9 studies in which the population did not live seaside, significant heterogeneity was observed (Q=50.04, P<.001, I2 = 78.0%), and the association between fish use and colorectal cancer was not statistically significant (summary OR, 0.87; 95% CI, 0.67-1.14).
These case-control studies were stratified by adjustment of overall diet to reflect total energy input and fruit, vegetables, or dietary fiber intake. Among the 9 studies that were adjusted for energy intake, significant heterogeneity was found among them (Q=33.77, P=.001, I2 = 64.5%), and no significant association was found between fish consumption and colorectal cancer (summary OR, 0.96; 95% CI, 0.80-1.15). Among the 6 studies that were adjusted for intake of fruit, vegetables, or dietary fiber, no significant heterogeneity was found (Q=12.30, P=.27, I2 = 18.7%), and a significant association between fish consumption and colorectal cancer was observed (summary OR, 0.89; 95% CI, 0.80-0.99).
We then stratified these case-control studies to adjust for lifestyle including the level of physical activity, smoking, and alcohol use. Of the 9 studies that were adjusted for physical activity, significant heterogeneity was found among them (Q=38.72, P<.001, I2 = 63.8%), and a slight protective effect was found between fish consumption and colorectal cancer (summary OR, 0.90; 95% CI, 0.76-1.07). Seven studies were adjusted for smoking, and significant heterogeneity was found among them (Q=27.35, P=.004, I2 = 59.8%). Moreover, the inverse association was found between fish consumption and colorectal cancer in these studies adjusted for smoking (summary OR, 0.78; 95% CI, 0.67-0.92). Within the 7 studies adjusted for alcohol, significant heterogeneity was noted (Q=34.03, P=.001, I2 = 61.8%), and a significant relationship between fish consumption and colorectal cancer was found (summary OR, 0.83; 95% CI, 0.70-0.98).
Last, these case-control studies were stratified to adjust for potential risk factors, including gender, body mass index (BMI), and family history of cancer. Ten studies were adjusted for gender within which significant heterogeneity (Q=28.12, P=.009, I2 = 53.8%) and a significant inverse association were found between fish consumption and colorectal cancer (summary OR, 0.75; 95% CI, 0.66-0.86). Among the 8 studies that were adjusted for BMI, significant heterogeneity was found among them (Q=55.99, P<.001, I2 = 78.6%), and no significant association was found between fish consumption and colorectal cancer (summary OR, 0.87; 95% CI, 0.71-1.07). For the 4 studies that were adjusted for family history of cancer, significant heterogeneity was found among them (Q=14.43, P=.071, I2 = 44.5%), and a marginal therapeutic benefit was noted for the effect of fish consumption on the carcinogenesis of colorectal cancer (summary OR, 0.82; 95% CI, 0.68-1.00).
Despite the fact that colon and rectal cancers share many features and are often referred to as “colorectal cancer,” these 2 cancer types tend to demonstrate many different characteristics.
As a result, it might be worthwhile to investigate whether there is a relationship between fish consumption and colon or rectal cancer, respectively. For the studies stated “colorectal cancer,” some combined both the colon cancer and rectal cancer, whereas there were studies that solely referred to colorectal cancer, so these studies were excluded. Significant heterogeneity was found within 14 colon cancer studies (Q=53.53, P<.001, I2 = 68.20%), which showed a slight reduction of colon cancer risk with fish intake (summary OR, 0.96; 95% CI, 0.81-1.14) (Supplemental Figure 3A). In 7 rectal cancer studies, there was significant heterogeneity (Q=18.89, P=.03, I2 = 52.30%), and a significant decrease was found between fish consumption and risk of rectal cancer (summary OR, 0.79; 95% CI, 0.65-0.97) (Supplemental Figure 3B) (Table 2).
We then stratified the studies by cancer incidence and mortality. Among the 3 mortality studies, no significant heterogeneity was found (Q=4.36, P=.50, I2 = 0), and no significant association between fish intake and colorectal cancer was found (summary OR, 1.01; 95% CI, 0.82-1.26) (Supplemental Figure 4A). Within the 38 incidence studies, significant heterogeneity was found (Q=125.03, P<.001, I2 = 59.20%) and a significant association between fish consumption and colorectal cancer was observed (summary OR, 0.86; 95% CI, 0.79-0.95) (Supplemental Figure 4B) (Table 2).
Publication Bias Analysis
As depicted in a symmetric Begg's funnel plot (Figure 3), no publication bias was found in this study (P=.55). Furthermore, there was no evidence of bias detected by Egger's test (intercept=0.37, P=.45).
Figure 3Forest plot of the RRs of colorectal cancer incidence or mortality and fish consumption (highest compared with the lowest category) in population-based case-control studies. As depicted in a symmetric Begg's funnel plot with intercept=0.37 and P=.45, no evidence of bias was detected in this meta-analysis by Egger's test.
The role that diet plays in preventing cancer has drawn more and more attention in recent years. This meta-analysis, which includes 22 prospective cohort and 19 case-control studies, evaluated the strength of current evidence that suggests that fish consumption may reduce the risk of colorectal cancer by as much as 12%. The strength of the present meta-analysis is that detailed information is provided in the study design, including selection criteria for cases and controls and methods of data collection, and there also is the considerable number of studies and subjects. The relationship between fish consumption and risk of colorectal cancer within subtypes of the studies (eg, prospective cohort and case-control studies, hospital-based and population-based control studies, colon cancer and rectal cancer, colorectal cancer incidence and mortality) were thoroughly analyzed in this meta-analysis, which might provide specific evidence on this topic.
Among the 41 epidemiologic studies we included, one study showed an extremely significant protective effect of fish consumption in colon cancer incidence.
Because the study design met our criteria of analysis, we did not remove the report from our investigation. However, we thought about the potential effect that this study might have on the bias of our meta-analysis, and as such, we performed another meta-analysis without this study. The results from this additional analysis also demonstrated a significant relationship between the consumption of fish and the reduction of risk in colon cancer (summary OR, 0.88; 95% CI, 0.82-0.96; Q=116.38, P<.001, I2 = 51.90%).
In our meta-analysis, a more significant association between fish intake and colorectal cancer risk abatement was observed (summary OR, 0.83; 95% CI, 0.72-0.95) from the 19 case-control studies, whereas less of a decrease in risk was found from the 22 prospective cohort studies (summary OR, 0.93; 95% CI, 0.86-1.01). Generally, prospective cohort studies are thought to provide more of a quantitative assessment because they have the potential of minimizing the possibility of bias due to recall or selection that is typically noted in case-control studies.
This study was based on both case-control and prospective cohort studies, which might provide more practical results for meta-analysis. Because results from case-control studies demonstrated significant heterogeneity among them, we conducted further stratified analyses to explore the source of this heterogeneity.
It is well known that therapeutic strategies for colon cancer and rectal cancer differ,
even though colon cancer and rectal cancer share some similarities. For instance, the colon originates embryologically from the midgut and hindgut, whereas the rectum originates from the cloaca.
Because of the differences between colon and rectal tissue and the differences between these 2 cancer types, a meta-analysis was conducted to explore the effect of fish intake on colon and rectal cancer, respectively. We noted that the protective effect of fish consumption is more prominent in rectal cancer than that in colon cancer, which might be accounted for by the different characteristics between colon and rectal cancer. One possible reasoning for this difference may be because colon cancers are generally molecularly heterogeneous, whereas rectal cancers mostly arise via a single neoplastic pathway.
Despite this, further exploration is needed to elucidate the exact mechanism at play. Overall, there is statistically significant heterogeneity among the 41 studies on fish consumption and colorectal cancer risk. The summary ORs among all studies did show a significant association between fish consumption and the risk of colorectal cancer. Of note, no statistically significant funnel plot asymmetry was seen when the effects of fish intake on risk of colorectal cancer were assessed via the Egger method.
Study Limitations
Like all meta-analyses, this meta-analysis has some potential limitations resulting from the availability, quality, and heterogeneity of the published data. There are several limitations that should be considered when interpreting our results.
First, the methods and units of measuring fish intake varied across studies. For example, in some studies the definitive volume of fish consumption was not clearly defined, and in these cases only the lowest and highest categories were reported. Because statistical tests showed heterogeneity among some studies, we used the random-effects model, which considers both within- and between-study variation for the pooled RR estimates and the dose-response analyses.
Second, we included only those studies that were published in English. This is mainly because it is difficult for the authors to interpret all the data that are available in different languages (eg, Japanese
). Only studies with the full text published in English were included in this meta-analysis.
Third, although we set up “fresh fish” as one of the including criteria in this meta-analysis, in an attempt to avoid such confounding factors as “fish oil,” “salted fish,” or “fried fish,” we cannot determine the exact kind of fish or the manner in which fish was prepared in our study. This represents a limitation because a recent article showed that cooking temperature might affect the risk of colorectal cancer.
Fourth, our study is limited because of the small sample size. Only 41 studies met our criteria to examine the association between fish and risk of colorectal cancer. Moreover, further sensitivity analysis restriction led to loss of statistical significance for pooled RRs, even though the risk estimates changed only slightly. Because of the small sample size, we had limited power to conclusively reject the null hypothesis of no publication bias. To account for this, we set statistical significance for publication bias at P < .1, and there was no publication bias visually found or via Egger's test. Because of these limitations, the results of the meta-analysis must be interpreted with caution.
Conclusions
The present meta-analysis suggests that fish consumption has preventive effects on colorectal cancer. Because of the relatively large health impact and burden from colorectal cancer, the findings demonstrated should be further investigated and confirmed in the setting of a large randomized clinical trial.
The authors thank Jeremy Allen Meier for editing the English.
Appendix
Tabled
1
MOOSE Statement
Content
Y/N
Page(s)
Problem definition
Y
2-4
Hypothesis statement
Y
2, 4
Description of study outcome(s)
Y
2
Type of exposure or intervention used
Y
2
Type of study designs used
Y
2
Study population
Y
2
Qualifications of searchers (eg, librarians and investigators)
Y
2, 4, 5
Search strategy, including time period included in the synthesis and keywords
Y
2, 5
Effort to include all available studies, including contact with authors
Y
5
Databases and registries searched
Y
5-6
Search software used, name and version, including special features used (eg, explosion)
Y
2, 5
Use of hand searching (eg, reference lists of obtained articles)
N
/
List of citations located and those excluded, including justification
Y
8-11
Method of addressing articles published in languages other than English
Y
14
Method of handling abstracts and unpublished studies
N
/
Description of any contact with authors
Y
15
Description of relevance or appropriateness of studies assembled for assessing the hypothesis to be tested
Y
4-6
Rationale for the selection and coding of data (eg, sound clinical principles or convenience)
Y
4-6
Documentation of how data were classified and coded (eg, multiple raters, blinding, and interrater reliability)
Y
4-6
Assessment of confounding (eg, comparability of cases and controls in studies where appropriate)
Y
2-6
Assessment of study quality, including blinding of quality assessors; stratification or regression on possible predictors of study results
Y
5-6
Assessment of heterogeneity
Y
6
Description of statistical methods (eg, complete description of fixed or random-effects models, justification of whether the chosen models account for predictors of study results, dose-response models, or cumulative meta-analysis) in sufficient detail to be replicated
Y
7-11
Provision of appropriate tables and graphics
Y
7-11
Graphic summarizing individual study estimates and overall estimate
Y
7
Table giving descriptive information for each study included
Y
7-11
Results of sensitivity testing (eg, subgroup analysis)
Y
7-11
Indication of statistical uncertainty of findings
Y
7-11
Quantitative assessment of bias (eg, publication bias)
Y
11-12
Justification for exclusion (eg, exclusion of non-English language citations)
Y
14
Assessment of quality of included studies
Y
7-12
Consideration of alternative explanations for observed results
Y
12-15
Generalization of the conclusions (ie, appropriate for the data presented and within the domain of the literature review)
Supplemental Figure 1Consumption of fish in association with colorectal cancer in case-control and prospective cohort studies. A, Consumption of fish in association with colorectal cancer risk in case-control studies, with a combined risk estimate of 0.83 (95% CI, 0.72-0.95; Q=89.69, P<.001, I2 = 68.8%). B, Consumption of fish in association with colorectal cancer risk in prospective cohort studies, with a combined risk estimate of 0.93 (95% CI, 0.86-1.01; Q=33.49, P=.22, I2 = 16.4%). Both analyses were conducted using the random-effects model. CI=confidence interval; OR=odds ratio.
Supplemental Figure 2Consumption of fish in association with colorectal cancer in hospital-based and population-based case-control studies. A, Consumption of fish in association with colorectal cancer risk in hospital-based case-control studies, with a combined risk estimate of 0.70 (95% CI, 0.59-0.84; Q=33.91, P=.001, I2 = 61.70%). B, Consumption of fish in association with colorectal cancer risk in population-based case-control studies, with a combined risk estimate of 0.96 (95% CI, 0.81-1.13; Q=31.65, P=.004, I2 = 55.8%). Both analyses were conducted using the random-effects model. CI=confidence interval; OR=odds ratio.
Supplemental Figure 3Consumption of fish in association with colon or rectal cancer, respectively. A, Consumption of fish in association with colon cancer risk, with a combined risk estimate of 0.96 (95% CI, 0.81-1.14; Q=53.53, P<.001, I2 = 68.20%). B, Consumption of fish in association with rectal cancer risk, with a combined risk estimate of 0.79 (95% CI, 0.65-0.97; Q=18.89, P=.03, I2 = 52.30%). Both analyses were conducted using the random-effects model. CI=confidence interval; OR=odds ratio.
Supplemental Figure 4Consumption of fish in association with colorectal cancer incidence and mortality. A, Consumption of fish in association with colorectal cancer incidence, with a combined risk estimate of 0.86 (95% CI, 0.79-0.95; Q=125.03, P<.001, I2 = 59.20%). B, Consumption of fish not significantly in association with colorectal cancer mortality, with a combined risk estimate of 1.01 (95% CI, 0.82-1.26); Q=33.49, P=.22, I2 = 16.4%). Both analyses were conducted using the random-effects model. CI=confidence interval; OR=odds ratio.
An n-3 PUFA-rich microalgal oil diet protects to a similar extent as a fish oil-rich diet against AOM-induced colonic aberrant crypt foci in F344 rats.
Fish fatty acids alter markers of apoptosis in colorectal adenoma and adenocarcinoma cell lines but fish consumption has no impact on apoptosis-induction ex vivo.
Development of a quantitative food frequency questionnaire for assessing food, nutrient, and heterocyclic aromatic amines intake in Japanese Brazilians for a colorectal adenoma case-control study.
Funding: This study was partially sponsored by the National Natural Science Foundation of China (No. 30900551 to Dr Jie Liang) and a Foundation for the Author of National Excellent Doctoral Dissertation of PR China (No. 201182 to Dr Jie Liang).
Conflict of Interest: None.
Authorship: All authors had access to the data and played a role in writing this manuscript.
00123-4&id=gr1.jpg){kind=link}
00123-4&id=gr2.jpg){kind=link}
00123-4&id=gr3.jpg){kind=link}
00123-4&id=grsu1.jpg){kind=link}
00123-4&id=grsu2.jpg){kind=link}
00123-4&id=grsu3.jpg){kind=link}
00123-4&id=grsu4.jpg){kind=link}