Intake or Blood Levels of n-3 Polyunsaturated Fatty Acids and Risk of Colorectal Cancer: A Systematic Review and Meta-analysis of Prospective Studies

Worldwide, colorectal cancer is the third most frequently diagnosed cancer and the fourth most common cause of death from cancer, accounting for 700,000 cancer-related deaths (1). Considering the high global burden of colorectal cancer, it is important to identify potential risk factors for colorectal cancer and develop preventive strategies against colorectal cancer. The most recent report from the World Cancer Research Fund concluded that physical activity, alcohol consumption, and body fatness are convincingly associated with the risk of colorectal cancer (2). However, most of the foods and nutrients had weak or limited evidence of an association with the risk of colorectal cancer, except for processed meat. n-3 polyunsaturated fatty acid (PUFA) intake was also categorized as “limited-no conclusion.”

n-3 PUFAs include α-linolenic acid (ALA), docosahexaenoic acid (DHA), docosapentaenoic acid (DPA), and eicosapentaenoic acid (EPA). Marine n-3 PUFAs, which are derived from marine sources and include DHA, DPA, and EPA, have been observed to exert potent anti-inflammatory effects and inhibit colorectal carcinogenesis (3–5). To date, many prospective studies have examined the association between the intake or blood levels of n-3 PUFAs and colorectal cancer risk (6–24). A previous meta-analysis including 8,775 colorectal cancer cases reported no association between n-3 PUFA intake and the risk of colorectal cancer (25). Since then, additional prospective studies including a large cohort study that included 6,291 cases among 476,160 subjects have investigated the association between n-3 PUFAs and the risk of colorectal cancer (18–21, 24). Dietary fat intake was mostly assessed with self-report questionnaires, which have a limitation of potential misclassification (26); n-3 PUFAs are essential fatty acids that humans cannot synthesize and should obtain through dietary intake. Thus, n-3 PUFA biomarkers, such as blood levels, may provide a more precise assessment of n-3 PUFA intake (27, 28).

Therefore, we performed a systematic review and meta-analysis of prospective studies to quantitatively assess the evidence of the association between the intake or blood levels of n-3 PUFAs and the risk of colorectal cancer.

We searched the PubMed and Embase databases from inception to July 10, 2019, to identify eligible studies published in the form of full-length articles. The following keywords were used: “(n-3 OR omega-3 OR marine OR docosahexaenoic OR eicosapentaenoic OR docosapentaenoic OR α-linolenic acid OR DHA or EPA or DPA or ALA) AND (colorectal OR colon OR rectal) AND (cancer OR neoplasms OR carcinoma)” (Supplementary Table S1). In addition, we supplemented the search by reviewing the reference lists from the retrieved articles or published reviews.

Studies were included in this meta-analysis when they met the following criteria: (i) prospective design (cohort or nested case–control); (ii) the exposure of interest was intake of n-3 PUFAs or blood (whole blood, plasma, or serum) levels of n-3 PUFAs; (iii) the outcome of interest was colorectal, colon, or rectal cancer; and (iv) RRs and 95% confidence intervals (CI) were provided. When more than one publication was from the same cohort, the study with the largest population and longest follow-up duration was included in the meta-analysis.

Two investigators (Y. Kim and J. Kim) independently extracted data according to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses statement (29). Any discrepancies were addressed by examining the original studies and through discussion. The information extracted from each study is as follows: first author's last name, year of publication, country or region where the study was conducted, duration of follow-up or dates of study period, sex, baseline age, number of cases and subjects/controls or person-time, types of n-3 PUFAs (ALA, EPA, DHA, DPA), anatomic site of the tumor (colorectal, colon, rectal), adjustment for potential confounders, and RRs with corresponding 95% CIs for each category of n-3 PUFA intake or blood levels of n-3 PUFAs. When a study provided several RRs, we included the RRs that reflected the greatest degree of adjustment for potential confounders. One publication was considered as two prospective cohort studies because it provided RRs from two different cohorts (17).

To assess the quality of the studies included in the meta-analysis, the Newcastle–Ottawa quality assessment scale (30) was used. The quality of each study was assessed independently by two investigators (Y. Kim and J. Kim) based on three domains: selection (representativeness of the exposed cohort; ascertainment of exposure), comparability of cohorts (adjustment for important confounders), and outcome (ascertainment of outcome; duration of follow-up; adequacy of follow-up). Any disagreements between the investigators were resolved by discussion to reach a consensus. We considered a study with scores of 10 or higher (out of 13), 7 to 9, and 6 or less as being of high, good, and low quality, respectively.

We combined study-specific RRs using a DerSimonian and Laird random-effects model (31), which considers both within- and between-study variations. For studies that reported RRs separately for different types of n-3 PUFAs (8, 24), we combined the RRs using a fixed effects model, and the pooled RRs were included in the meta-analysis. For analysis of intake, we performed an analysis of marine n-3 PUFAs (including EPA, DHA, and DPA) in addition to total n-3 PUFAs (including ALA, EPA, DHA, and DPA). We conducted subgroup analyses by sex, geographic region, cancer sites, duration of follow-up, type of fatty acids, quality score and adjustment for age, body mass index (BMI), smoking, alcohol, and physical activity to explore whether the association differed by characteristics of the study or population. To test for variations across subgroups, a meta-regression analysis was performed.

For the dose–response analysis, the method proposed by Greenland and Longnecker (32–34) was used. For each study, the median value of n-3 PUFA intake or blood levels of n-3 PUFAs in each category was assigned to each corresponding RR. Open-ended categories were assumed to have the same interval of dietary intake or blood level as the adjacent category. When n-3 PUFA intakes were provided in densities (g/1,000 kcal; ref. 11), we converted the values into absolute intakes (g/day) using median daily energy intake of the study population. We presented the RRs for a 0.5 g/day increase in n-3 PUFA intake and as a 1% increase in total fatty acids for blood levels of n-3 PUFAs. A potential nonlinear association between dietary or blood levels of n-3 PUFAs and colorectal cancer risk was also examined using restricted cubic splines with 3 knots at fixed percentiles (10%, 50%, and 90%) of the distribution (35). The P value for nonlinearity was calculated by testing the null hypothesis that the coefficient of the second spline is equal to zero.

Statistical heterogeneity among studies and inconsistencies was evaluated using the Q statistic (36) and I² statistic (37). A sensitivity analysis that removed one study at one time and pooled the remaining studies was also performed to explore the robustness of the results. Potential publication bias was assessed through Begg (38) and Egger (39) tests. For statistical significance, the two-tailed α was set at P = 0.05. All statistical analyses were performed using Stata software version 14.2 (StataCorp).

A total of 20 prospective studies (6–24) including 18,102 colorectal cancer cases among 1,360,046 participants were eligible for the current meta-analysis. The detailed study selection process is shown in Fig. 1. Table 1 shows the general characteristics of the studies included in the present meta-analysis. Fifteen studies provided RRs of n-3 PUFA intake (6–17, 19, 20), 3 studies provided RRs of blood levels of n-3 PUFAs (22–24), and 2 studies provided RRs of both n-3 PUFA intake and blood levels of n-3 PUFAs (18, 21). The follow-up periods ranged from 3.3 years to 26 years, and the median follow-up time was 9.3 years. Eight studies were conducted in the United States (6, 10, 12, 16, 17, 20, 23), 6 studies were conducted in Asia (9, 11, 13, 14, 22, 24), 5 studies were conducted in Europe (7, 8, 15, 19, 21), and one study was conducted in Oceania (18). All studies controlled for age, and most of the studies adjusted for alcohol consumption (n = 17; refs. 7, 9–11, 13–24), BMI (n = 16; refs. 7–17, 20–23), smoking (n = 16; refs. 7, 9–11, 13–18, 20–24), and physical activity (n = 16; refs. 7, 9, 11–18, 20–24). According to the quality assessment, scores ranged from 9 to 12 indicating good or high quality.

Seventeen prospective studies (6–21) that included 17,405 cases among 1,358,236 participants were included in the meta-analysis of the association between n-3 PUFA intake and the risk of colorectal cancer. The pooled RR for the highest versus lowest levels of n-3 PUFA intake was 0.97 (95% CI, 0.90–1.04; I² = 34.6%; P_{heterogeneity} = 0.05), with no significant heterogeneity (Fig. 2; Table 2). The association between n-3 PUFA intake and risk of cancer was not significantly different by sex, geographical region, cancer sites, duration of follow-up, quality score, or adjustment for covariates (P for difference > 0.1 in all comparisons; Table 2). Regarding the type of n-3 PUFA, EPA (RR = 0.89; 95% CI, 0.80–0.99), and DHA (RR = 0.88; 95% CI, 0.81–0.96) intakes were inversely associated with lower risks of colorectal cancer. On the other hand, a nonsignificant positive association was observed between ALA intake and colorectal cancer risk (RR = 1.01; 95% CI, 0.92–1.12). In the dose–response analysis, 13 prospective studies (7, 9, 11–17, 19–21) including 15,838 cases among 1,198,638 participants were included. The pooled RR for a 0.5 g/day increase in n-3 PUFA intake was 1.00 (95% CI, 0.97–1.03; Table 2). We found no significant nonlinear association between n-3 PUFA intake and the risk of colorectal cancer (P for nonlinearity = 0.39; Fig. 2). Regarding the type of n-3 PUFA, an increase of 0.1 g/day of EPA and DHA intake was associated with 5% and 3% lower risks of colorectal cancer, respectively (Table 2). A significant nonlinear association was not observed for EPA (P for nonlinearity = 0.06), DHA (P for nonlinearity = 0.06), DPA (P for nonlinearity = 0.10), or ALA (P for nonlinearity = 0.48) intake (Fig. 3).

Thirteen prospective studies (8–14, 16–18, 20, 21) that included 16,239 cases among 1,288,528 participants were included in the meta-analysis of the association between marine n-3 PUFA intake and risk of colorectal cancer. The pooled RR for the highest versus lowest levels of marine n-3 PUFA intake was 0.98 (95% CI, 0.92–1.05; Supplementary Table S2). We found no significant difference by sex, geographic region, cancer sites, duration of follow-up, or adjustment for covariates (P for difference > 0.1 in all comparisons). Ten prospective studies (9, 11–14, 16, 17, 20, 21) with 14,884 cases and 1,164,145 participants were included in the dose–response analysis. The pooled RR for a 0.5 g/day increase in marine n-3 PUFA intake was 1.01 (95% CI, 0.95–1.07). A nonlinear association was not observed between marine n-3 PUFA intake and colorectal cancer risk (P for nonlinearity = 0.55; Supplementary Fig. S1).

Five prospective studies (18, 21–24) with 1,553 cases, among 6,937 participants were included in the meta-analysis of the association between blood levels of n-3 PUFAs and risk of colorectal cancer. The pooled RR for the highest versus lowest blood levels of n-3 PUFAs was 0.79 (95% CI, 0.64–0.98; I² = 22.8%; P_{heterogeneity} = 0.26), with no significant heterogeneity (Fig. 4). There was no significant difference by sex, geographical region, cancer sites, duration of follow-up, type of fatty acids, quality score, or adjustment for covariates (P for difference > 0.4 in all comparisons; Table 2). Three prospective studies (21–23) with 808 cases among 2,032 participants were included in the dose–response analysis. A 1% increase in blood n-3 PUFA level was associated with a 4% lower risk of colorectal cancer (RR = 0.96; 95% CI, 0.92–1.00; Table 2). There was no evidence of a nonlinear association between blood n-3 PUFA levels and colorectal cancer risk (P for nonlinearity = 0.46).

There was no evidence of publication bias in the literature regarding colorectal cancer risk and n-3 PUFA intake (Begg P = 0.53, Egger P = 0.18), marine n-3 PUFA intake (Begg P = 0.89, Egger P = 0.27), or blood n-3 PUFA levels (Begg P = 0.76, Egger P = 0.23).

The current meta-analysis of 20 prospective studies including 18,102 colorectal cancer cases and 1,360,046 subjects investigated the association between n-3 PUFA intake/blood n-3 PUFA levels and colorectal cancer risk. Our findings indicated that blood levels of n-3 PUFAs were inversely associated with a reduced risk of colorectal cancer. People with high blood levels of n-3 PUFAs had a 21% lower risk of colorectal cancer than those with low blood levels of n-3 PUFAs. A 1% increase in blood n-3 PUFA levels was associated with a 4% lower risk of colorectal cancer. For n-3 PUFA intake, we found a nonsignificant, weak inverse association between total n-3 PUFA intake and colorectal cancer risk, but significant inverse associations between EPA and DHA intakes and colorectal cancer risk were observed in both the highest versus lowest analysis and the dose–response analysis. People with high intakes of EPA and DHA had an 11% and 12% lower risk of colorectal cancer than those with low intakes of EPA and DHA, respectively. Each additional 0.1 g of EPA and DHA intake daily was associated with a 5% and 3% lower risk of colorectal cancer, respectively.

We found a significant inverse association between blood n-3 PUFA levels and colorectal cancer risk but did not find a significant association between n-3 PUFA intake and colorectal cancer risk. Differences in the association between intake and blood levels and the risk of colorectal cancer have also been observed in relation to vitamin B₆. A meta-analysis including 13 prospective studies indicated that blood pyridoxal 5′-phosphate (PLP; the active form of vitamin B₆) levels are inversely associated with a lower risk of colorectal cancer, while vitamin B₆ intake was not significantly associated with colorectal cancer risk (40). In epidemiologic studies, dietary assessment is often accompanied by measurement errors, as most studies assess dietary intake using a food frequency questionnaire. Fatty acids are especially prone to this because similar foods can have different fatty acid compositions not distinguished by the food description provided by the tools used to evaluate diet (41); in addition, it is difficult to calculate specific fatty acid intake (42). Measurement errors are likely to be nondifferential and thus would probably have led to results closer to null findings. There is a possibility of attenuation in the inverse association between n-3 PUFA intake and the risk of colorectal cancer because of measurement error in assessing n-3 PUFA intake. As most studies conducted a single assessment of fatty acid levels, there is also a possibility of measurement error in blood measures. However, previous studies suggested that blood levels of fatty acids can be a reasonably good biomarker of habitual dietary fat intake showing the strong correlation between dietary PUFA intake and blood levels of PUFAs even if it is a single blood sample (28, 43).

In the subgroup analysis by type of n-3 fatty acids, intakes of EPA and DHA were associated with a lower risk of colorectal cancer. DPA intake showed a nonsignificant association, and ALA intake was not significantly associated with a risk of colorectal cancer. These results are in line with the findings from previous meta-analyses indicating null associations between intake of ALA and colorectal cancer (25, 44). Most of the previous studies of n-3 PUFAs and risk of disease have focused on long-chain n-3 fatty acids, and evidence for the association between ALA (short-chain n-3 fatty acids) and health outcomes was limited (45). Our results by type of n-3 fatty acids should also be interpreted with caution due to the relatively small number of studies. We could not find significant differences by sex, geographical region, cancer sites, and duration of follow-up. Although the differences were not significant, observed inverse associations were slightly stronger in men and colon cancer than women and rectal cancer. The previous meta-analyses showed positive associations in women and rectal cancer, which was different from inverse associations in men and colon cancer (25, 44).

Several case–control studies have investigated the association between n-3 PUFA intake and the risk of colorectal cancer and found that high n-3 PUFA intake was inversely associated with a lower risk of colorectal cancer (46–50). A meta-analysis of 8 case–control studies also reported a 24% lower risk of colorectal cancer in people who had high biospecimen n-3 PUFA levels than those who had low biospecimen n-3 PUFA levels (51). Experimental studies showed that colorectal tumor incidence was lower in the group with n-3 PUFA treatment in rats (5) and that n-3 PUFA-rich diets inhibited the growth of colorectal tumors in a mouse model (4). In addition, another animal study suggested that EPA is a candidate to reduce the risk of colorectal cancer, showing a decrease in the number and size of polyps in mice fed EPA diets (52). Given the results of our meta-analysis and previous studies, n-3 PUFAs may have a beneficial effect against the development of colorectal cancer.

The potential protective effects of n-3 PUFAs in regard to the risk of colorectal cancer are biologically plausible. The most prominent mechanism for the anticancer activity of n-3 PUFAs is inhibition of the production of arachidonic acid–derived eicosanoids, which are positively associated with inflammation and carcinogenesis (53). This suppressive effect on the biosynthesis of arachidonic acid–derived eicosanoids results in reduced cell proliferation (54) and increased cell apoptosis (55). In addition, n-3 PUFAs may contribute to a decreased risk of colorectal cancer by affecting gene expression or signal transduction and alteration of free radical production (53). Recent experimental animal studies have shown that n-3 PUFAs have anti–colorectal cancer effects by modulating profiles of eicosanoid metabolites (4) and regulating the DNA methylation process (5). An intervention study conducted on humans also reported that high n-3 PUFA intake was associated with elevated apoptosis in the normal sigmoid colon of patients polypectomized for adenomas/tumors (56).

To the best of our knowledge, this is the first comprehensive meta-analysis to investigate the risk of colorectal cancer in relation to both n-3 PUFA intake and blood n-3 PUFA levels. Compared with previous meta-analysis (25) (including 8,775 cases) examining the association between n-3 PUFA intake and colorectal cancer risk, our meta-analysis (including 18,102 cases) included more than twice as many cases of colorectal cancer. The large sample population enhanced the statistical power of the results of the meta-analysis. In addition, this quantitative assessment included only prospective studies, and thus, our results were relatively free from recall or selection bias that could be of concern in retrospective studies. Despite these strengths, our meta-analysis has several limitations. First, our results for n-3 PUFA intake had the possibility of some degree of misclassification because most studies included in the meta-analysis evaluated n-3 PUFA intake by a single assessment with a food frequency questionnaire. We tried to supplement for measurement error in the assessment of n-3 PUFA intake by conducting a meta-analysis for both n-3 PUFA intake and blood n-3 PUFA levels. Second, because the current meta-analysis was based on observational studies, there was the possibility of potential effects of unknown or residual confounding factors on our results. When we conducted a limited analysis of studies adjusted for age, BMI, smoking, alcohol consumption, and physical activity, significant differences were not observed. However, the possibility of residual confounding cannot be completely ruled out. Third, the analysis of blood n-3 PUFA levels included a relatively small number of studies due to the limited published data. Nevertheless, a significant inverse association between blood levels of n-3 PUFAs and risk of colorectal cancer was found. Fourth, the highest and lowest categories of intake or blood levels of n-3 PUFAs varied among the studies, which is an inherent limitation in this meta-analysis. However, we explored the dose–response association between the intake and blood levels of n-3 PUFAs and the risk of colorectal cancer through linear and nonlinear dose–response analyses and performed a highest versus lowest analysis. Finally, when interpreting the results of blood levels of PUFAs, it should be noted that we combined results from different blood matrices (whole blood, serum, and plasma).

In conclusion, the current meta-analysis of 20 prospective studies indicated that high blood levels of n-3 PUFAs are inversely associated with a low risk of colorectal cancer and high n-3 PUFA intake was suggestively associated with a low risk of colorectal cancer. Our findings suggest that high blood levels of n-3 PUFAs may decrease colorectal cancer risk, but it is impossible to identify a causal relationship between n-3 PUFAs and colorectal cancer risk because of the observational nature of prospective studies included in the meta-analysis. Further large randomized controlled trials of n-3 PUFA supplementation, prospective cohort studies using food diaries as a dietary assessment tool, and large prospective cohort or nested case–control studies on blood n-3 PUFA levels are needed to verify the association between n-3 PUFA intake/blood n-3 PUFA levels and the risk of colorectal cancer.

No potential conflicts of interest were disclosed.

This work was supported by grants from the National Cancer Center in Korea (1910330).

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.