Inconsistent results regarding the association between red and processed meat intake and the risk of colorectal adenoma (CRA), the precursor of colorectal cancer (CRC), have been reported. To provide a quantitative assessment of this association, we summarized the evidence from observational studies. Relevant studies were identified in MEDLINE and EMBASE until December 31, 2011. Summary relative risks (SRRs) with 95% confidence intervals (CIs) were pooled with a random-effects model. Between-study heterogeneity was assessed using the Cochran's Q and I2 statistics. A total of 21 studies (16 case–control studies and five cohort/nested case–control studies) were included in this meta-analysis. The SRRs of CRA were 1.36 (95% CI = 1.17–1.58) for every 100 g/day increase in red meat intake, and 1.24 (95% CI = 1.12–1.36) for the highest versus the lowest level of red meat intake. Nonlinear dose-response meta-analysis indicated that CRA risk increased approximately linearly with increasing intake of red meat up to ∼ 90 g/day, where the curve reached its plateau. Subgrouped analyses revealed that the increased risk of CRA with intake of red meat was independent of geographic locations, design and confounders. The SRRs of CRA was 1.28 (95% CI = 1.03–1.60) for per 50 g/day increase in processed meat intake, and 1.17 (95% CI = 1.08–1.26) for the highest versus the lowest level of processed meat intake. Increased intake of red and processed meat is associated with significantly increased risk of CRA.
The incidence of and mortality from colorectal cancer (CRC) have been on the rise worldwide.1 Annually, ∼1 million new cases of CRC are diagnosed, and nearly 530,000 individuals may die from this disease, equivalent to ∼8% of all cancer-related deaths worldwide.1 The World Cancer Research Fund (WCRF)/American Institute for Cancer Research (AICR) concluded in a report published in 2007 that high consumption of red and processed meat convincingly increases the risk of CRC.2 In this regard, a more recent meta-analysis of 24 prospective studies of CRC showed an increased risk of 17% (95% confidence interval [CI] = 5–31%) per 100 g/day increased intake of red meat and of 18% (95% CI = 10–28) per 50 g/day increased intake of processed meat.3 One underlying cause might be the formation of heterocyclic amines (HCAs) and polycyclic aromatic hydrocarbons (PAHs) during cooking or processing of meat, both of which are potent mutagens in rodents and humans.4–7 Additionally, heme iron from red meat might be another potential risk factor for CRC development.8
It has been established that most CRCs arise from colorectal adenoma (CRA) by a process referred to as the adenoma–carcinoma sequence.9 On the basis of this theory, we may assume that adenoma and carcinoma share a common etiology and similar epidemiological features. Therefore, better information about risk factors for adenomas might permit more rational development of intervention studies with adenomas as end points.10
Based on three case–control studies, an early quantitative review published in 2000 found that red meat probably increased colorectal polyps risk, whereas processed meat was not the case.11 Since then, a number of epidemiologic studies have evaluated the association between red and processed meat intake and CRA risk and have yielded inconsistent results.12–22 Therefore, we conducted an updated and comprehensive meta-analysis of the current epidemiological literature to better characterize this issue.
Material and Methods
Data sources and searches
A computerized literature search was conducted in MEDLINE (from January 1, 1966) and EMBASE (from January 1, 1974), through December 31, 2011, by two independent investigators (X.X.D. and G.X.H.). We searched the relevant studies with the following text word and/or Medical Subject Heading terms: (i) red meat OR processed meat OR preserved meat OR beef OR pork OR veal OR mutton OR lamb OR ham OR sausage OR bacon; (ii) adenoma OR polyp OR polyps and (iii) colorectal OR colon OR rectal OR large bowel, following the Meta-analysis Of Observational Studies in Epidemiology guidelines.23 Furthermore, we reviewed the reference lists of the relevant articles to identify additional studies. No language restrictions were imposed.
The definitions of red and processed meat varied across studies. In our meta-analysis, red meat was defined as the intake of beef, veal, pork, lamb or a combination thereof,2 and processed meat was generally defined as meat made largely from pork, beef or poultry that undergoes methods of preservation, such as curing, smoking or drying.2
AICR: American Institute for Cancer Research; CI: confidence interval; CRA: colorectal adenoma; CRC: colorectal cancer; FFQ: Food Frequency Questionnaire; HCA: heterocyclic amine; NOC: N-nitroso compound; OR: odd ratio; PAH: polycyclic aromatic hydrocarbon; SRR: summary relative risk; WCRF: World Cancer Research Fund
Studies were included in this meta-analysis if: (i) published as an original article; (ii) using a case–control, nested case–control or cohort design and (iii) reported relative risk (RR) estimates with corresponding 95% CIs for the association between red and/or processed meat intake and the risk of CRA at least adjusted for age. Non-peer-reviewed articles, ecologic assessments, correlation studies, experimental animal studies and mechanistic studies were excluded. Studies were excluded if there were no data specific for CRA or concerning of recurrence or growth of adenomatous polyps. If data were duplicated in more than one study, the most recent or complete studies were included in this analysis.
Two researchers (X.X.D. and G.X.H.) independently extracted the following data from each study that met the criteria for inclusion: the first author's last name, year of publication, geographic locations, sample source, methods of ascertainment of red and processed meat dietary variables, cases size, duration of follow-up, adjustments for confounders and the RR estimates with corresponding 95% CIs for the highest versus the lowest level. From each study, we extracted the risk estimates that reflected the greatest degree of control for potential confounders.
We used the method of a random-effects models to calculate summary relative risks (SRRs) and 95% CIs of CRA for the highest versus the lowest level of red and processed meat intake and for the linear dose-response analysis. This method of a random-effects models was developed by DerSimonian and Laird, which accounts for heterogeneity among studies.24
We performed linear dose-response meta-analysis of CRA risk per increment of intake of 100 g/day for red meat and 50 g/day for processed meat using generalized least-squares trend (GLST) estimation analysis based on the methods developed by Greenland and Orsini.25, 26 These methods require the number of cases and person-time or controls for at least three quantitative exposure categories are known. When this information was not available, we estimated the dose-response slopes by using variance-weighted least squares (vwls) regression analysis. Both the methods, GLST and vwls, require medians for categories of intake levels. If medians were not reported, we estimated the midpoint of the upper and lower boundaries in each category as the average intake level. If the highest category was open ended, the open-ended boundary was calculated using as interval length of the width of the closest interval. When the lowest category was open ended, the lowest boundary was considered as zero. When the exposures were expressed in “times” or “servings,” we converted it into grams (g) using 120 g as a standard portion size for red meat, and 50 g for processed meat, as in the WCRF/AICR report.2 For studies reporting intakes in g/1,000 kcal/day, the intake in g/day was estimated using the average energy intake reported in the article. When sex-specific estimates were available, we first analyzed together (as RR estimates for CRA) and then separately.
A potential nonlinear dose-response relationship was examined by using fractional polynomial models.27 We determined the best-fitting second-order fractional polynomial regression model, defined as the one with the lowest deviance. A likelihood ratio test was used to assess the difference between the nonlinear and linear models to test for nonlinearity.27
In assessing heterogeneity among studies, we used the Cochran's Q and I2 statistics. For I2, a value >50% is considered a measure of severe heterogeneity. Sources of heterogeneity were explored in stratified analysis and by linear meta-regression, according to study design, gender, geographic locations, cases size and confounders, including smoking, BMI, NSAIDs use and dietary energy intake. We also conducted sensitivity analysis to estimate the influence of each individual study on the summary results by repeating the random-effects meta-analysis after omitting one study at a time.
Publication bias was assessed by using funnel plots and the further Egger's regression asymmetry test.28 All statistical analyses were performed using STATA, version 11.0 (STATA, College Station, TX) and R-package statistical software (version 2.11.0 beta). A two-tailed p value <0.05 was considered to be significant.
Search results and study characteristics
The search strategy generated 240 citations, of which 35 were considered of potential value, and the full text was retrieved for detailed evaluation (Fig. 1). Seventeen of these 35 articles were subsequently excluded from the meta-analysis for various reasons. Additional three articles were included from reference review. Therefore, a total of 21 articles (five cohort/nested case–control studies and 16 case–control studies), including 13,757 subjects with CRA, were used in this meta-analysis (Tables 1 and 2). The continents or countries where the studies were conducted in were the United States (n = 13) and Europe (n = 8). The methods of assessment of exposure were based on food items semiquantitative Food Frequency Questionnaire (FFQ) in all, except for one, studies.19 The ascertainment of CRA outcome was based on histological findings in all studies.
Table 1. Characteristics of case–control and cross-sectional studies of red meat and processed meat and colorectal adenoma risk
Table 2. Characteristics of cohort and nested case–control studies of intake of red meat and processed meat and colorectal adenoma risk
Nineteen studies were included in the dose-response analysis of red meat intake and CRA risk. The SRR of CRA per 100 g/day increment in red meat intake was 1.36 (95% CI = 1.17–1.58), with evidence of severe heterogeneity (Q = 58.46, p < 0.001, I2 = 69.2%; Fig. 2a).
In stratified analysis by gender, the SRRs for CRA risk per 100 g/day increment in red meat intake were 1.48 in males (95% CI = 1.02–2.15; n = 4) and 1.71 in females (95% CI = 0.88–3.31; n = 2). The difference in SRRs across gender strata was not significant (p = 0.709). The SRRs were similar for studies conducted in the United States (SRR = 1.27; 95% CI = 1.07–1.51; pheterogeneity = 0.001; I2 = 66.8%) and Europe (SRR = 1.51; 95% CI = 1.24–1.84; pheterogeneity = 0.234; I2 = 24.4%). The SRRs for the CRA risk per 100 g/day increment in red meat intake were 1.44 (95% CI = 1.18–1.76) for case–control studies and 1.22 (95% CI = 1.04–1.42) for cohort/nested case control studies (Table 3 and Fig. 2a).
Table 3. Stratified dose-response meta-analyses of intake of red meat and processed meat and colorectal adenoma risk
Furthermore, when stratified analysis for type of adenoma, summary estimates of per 100 g/day increment in red meat intake and risk of nonadvanced adenoma were 1.26 (95% CI = 0.87–1.83; n = 3) and were 1.49 (95% CI = 0.89–2.48; n = 4) for advanced adenoma. The summary estimates for studies with case size larger than 400 were 1.24 (95% CI = 1.06–1.45), which were significantly less than that for studies with case size less than 400 (SRR = 1.57; 95% CI = 1.28–1.94; p for difference = 0.084). Studies that required conversion from times or servings to grams per day (10, 13, 14, 16, 17, 19, 21, 22 and 29–31) had similar SRRs of CRA with studies that did not require the conversion (Table 3).
In subgroup analyses by confounders, the SRRs for CRA risk per 100 g/day increment in red meat intake were positive in all strata. For example, when combining results from studies adjusted for BMI, the pooled RRs were positive (SRR = 1.43; 95% CI = 1.25–1.64). Similarly, when pooling estimates for studies were controlled for smoking, NSAIDs use, and total energy intake, the pooled RRs were not significantly changed (smoking: SRR = 1.30; 95% CI = 1.09–1.55; NSAIDs use: SRR = 1.32; 95% CI = 1.06–1.65; total energy intake: SRR = 1.37; 95% CI = 1.15–1.62; Table 3).
There was evidence of a nonlinear association of red intake and CRA risk (p = 0.03). Visual inspection of the curve suggests that the risk increases linearly up to ∼90 g/day of intake. Above that intake level, the risk increase is less pronounced.
In univariate meta-regression analysis, only case size was found to be a significant factor for the association between red meat intake and CRA risk. The between-study variance was reduced from 0.03467 to 0.0267 based on REML estimate, and the heterogeneity explained by case size was 23%.
When the overall homogeneity and effect size were calculated by removing one study at a time, we confirmed the stability of the positive association between per 100 g/day increment in red meat intake and CRA risk (data not shown). In addition, when we repeated the high versus low analysis with the same studies that were included in the linear dose-response analysis for red meat intake, the results were similar to the original analysis (SRR = 1.25; 95% CI = 1.13–1.39; pheterogeneity = 0.038; I2 = 20.0%).
High versus low analysis
We identified 21 studies that presented results on the highest versus the lowest level of red meat intake and CRA risk. The SRRs were 1.24 (95% CI = 1.12–1.36) in a random-effects model of red meat intake and CRA risk for the highest versus the lowest level. There was low heterogeneity among these studies (pheterogeneity = 0.036; I2 = 38.2%).
Seven studies were included in the dose-response analysis of CRA risk per 50 g/day increase in processed meat intake, and the SRRs of CRA were 1.28 (95% CI = 1.03–1.60), with high heterogeneity across studies (pheterogeneity = 0.004; I2 = 69.1%; Fig. 2b).
In stratified analyses by study design, the SRRs of CRA per 50 g/day increase in processed meat intake were similar in both case–control (SRR = 1.21; 95% CI = 0.94–1.56; pheterogeneity = 0.01; I2 = 70%; n = 5) and cohort/nested case control–studies (SRR = 1.50; 95% CI = 1.10–2.03; pheterogeneity = 0.865; I2 = 0; n = 2). All these seven studies were controlled for smoking use. The association between per 50 g/day increase in processed meat intake and CRA risk was positive in controlling for confounders: BMI, NSAIDs use and dietary energy intake (Table 3).
High versus low analysis
We identified seven studies that presented results on the highest versus the lowest level of processed meat intake and CRA risk. The pooled estimate indicated that individuals in the high category of processed meat intake had 17% greater risk of CRA compared to individuals in the lowest category of processed meat intake (SRR = 1.17; 95% CI = 1.08–1.26). There was no heterogeneity among these studies (pheterogeneity = 0.449; I2 = 0).
The p-value of Egger's regression asymmetry test for studies on the association between red meat intake and CRA risk was 0.978 for dose-response analysis and 0.533 for high versus low analysis. We did not assess publication bias for studies on processed meat intake because of very less relevant studies.
In this comprehensive meta-analysis, red and processed meat intake is found to be associated with an increased risk of CRA. The risk increase in CRA estimated in linear dose-response models is 36% for every 100 g/day increase intake of red meat and 28% for every 50 g/day increase intake of processed meat. These results are consistent with those of high versus low meta-analyses. In nonlinear models, CRA risk appears to increase almost linearly with increasing intake of red meat up to ∼90 g/day. Above this level, the risk increase is less pronounced.
To our knowledge, this is the first comprehensive meta-analysis of red and processed meat intake and CRA risk based on high versus low analysis, linear and nonlinear dose-response meta-analysis. In addition, all of the included studies evaluated multiple confounders including BMI, diet factors, NSAIDs use, alcohol use, smoking and so forth, and the relationships in each study were derived from regression after adjustment at least for age.
However, our meta-analysis has several limitations. First, 16 of 21 studies included in this meta-analysis used a case–control design that was more susceptible to recall and selection biases, especially dietary recall bias, than a cohort design. Despite dietary assessment methods for red and processed meat intake in all, except for one, studies being semiquantitative FFQ, the most comprehensive methods, it is still likely that there is a degree of measurement error associated with them, which can lead to attenuated risk estimates; therefore, the actual risks may be higher.
Second, great heterogeneity was observed across studies. The significant heterogeneity may arise from study design, case size, demographics of participants, methods of exposure measurement and classification, intake unit conversion and confounders. Based on meta-regression analysis, we found that case size might account for only 23% of the great heterogeneity among studies of red meat intake and CRA risk. In addition, analyses of high versus low red and processed meat intake are limited because true differences in the level and range of red and processed meat intake among studies are not taken into account and this may contribute to heterogeneity in the results. However, when we repeated the high versus low analysis with the same studies that were included in the linear dose-response analysis, results were similar to the original analysis.
Third, residual confounders are always of concern in observational studies. Persons with high intakes of red meat and processed meat are likely to be associated with other unhealthy lifestyles, for example, smoking and high energy intake, both of which are indicated as risk factors for CRA.38 When we limited the meta-analysis to studies controlled for NSAIDs use, dietary energy intake, BMI and smoking, the positive association between red meat intake and CRA risk was not significantly changed. Although most included studies adjusted for a wide range of potential confounders for CRC, we still could not exclude the possibility that other unmeasured or inadequately measured factors have confounded the true association.
Fourth, given that the results of the current analysis were all based on data from Western populations, additional research in other populations is warranted to generalize these findings.
Finally, as in any meta-analysis, the possibility of publication bias is of concern, because small studies with null results tend not to be published. However, the results obtained from funnel plot analysis and formal statistical tests did not provide evidence for such bias.
There are several mechanisms proposed to explain how red and processed meat intakes enhance colorectal neoplasm risk. Mutagens such as HCAs and PAHs are rich in red meat and processed meat formed during high temperatures or smoked foods.36 HCAs are potent mutagens in rodents5, 39 and have been associated with an increased risk of CRA in several epidemiologic studies.7, 16, 31 A large prospective cohort study, the European Prospective Investigation into Cancer and Nutrition–Heidelberg Study, found that individuals with a high intake of 2-amino-1-methyl-6-phenylimidazo [4,5-b]pyridine (PhIP), the most abundant dietary HCA, had a 47% higher risk of developing CRAs than did those with a low intake (RR = 1.47; 95% CI = 1.13–1.93; quartile 4 compared to quartile 1; p for trend = 0.002).7 Benzo(a)pyrene is one of the most potent PAH carcinogens in animal studies.40 Epidemiologic studies have directly investigated the association between dietary intake of PAHs and colorectal neoplasms and found a positive association between benzo(a)pyrene and CRA.41 The high saturated fat content of red and processed meat is also the proposed culprit for the increased risk of CRC in some studies,36 but not in other studies, including a recent meta-analysis, which showed no effect of saturated fat on colorectal carcinogenesis.42 A third hypothesis concerns heme iron and nonheme iron. Heme iron is primarily found in red meat and is more bioavailable than nonheme iron. Heme iron has been associated with the promotion of CRC in rodents and induction of oxidative DNA damage,43 as well as endogenous formation of carcinogenic N-nitroso compounds (NOCs).44 Additional NOC exposure may occur from intake of nitrate and nitrite, precursors to NOCs that are added to processed meat for both preservation and color.
To date, intake of red and/or processed meat was found to be associated with increased CRC risk in a body of studies, including several meta-analyses.3, 45–49 More recently, data from 24 prospective studies of CRC showed increased risk of 17% per 100 g/day increased intake of red meat and 18% per 50 g/day increased intake of processed meat.3 Our data suggest somewhat stronger associations between red and processed meat intake and adenoma risk. However, the two findings are not comparable because our findings are based on the studies that mostly used a case–control design.
In terms of developing CRC prevention strategies, knowledge of risk factors associated with early stages of malignant transformation is important. Although the magnitude of risk increase reported here is small at the individual level, given the high incidence and large burden of CRC, limiting the intake of red meat and processed meat in high meat consumers is of great public health significance with respect to CRC prevention. According to the report from WCRF/AICR in 2009, the preventability estimates for red meat intake and CRC were 5% in the United States and 7% in China; where 26 and 37% of the respective populations were estimated to consume more than 80 g/day of red meat.50 Dietary and lifestyle factors are usually interrelated, and it is likely that a change in a detrimental habit, such as high intake of red and processed meat, will be accompanied by other healthful changes.
In summary, our meta-analysis supports the hypothesis that high consumption of red and processed meat may increase the risk of CRA. Whether the association with red meat and/or processed meat consumption varies according to subsites or type of adenoma in the colorectum warrants further investigation.