International ecological studies show that countries where people eat more red meat are also countries where the risk of CRC is high (17
). Analytical studies suggest that this association is also seen at the individual level, but the link is significant in only one study out of three (18
). Three meta-analyses have been published since 2000, and their quantitative risk estimate for fresh red meat and processed meat intake are summarized below and in .
Excess risk of CRC associated with the intake of fresh red meat and of processed meat in three dose-response meta-analyses of analytical studies
Sandhu et al. (2001)
made a meta-analysis gathering 13 cohort studies, selected from 17 studies, according to pre-established quality criteria (4
). All cohorts’ studies with relative risks between meat/processed meat intake and colon/colorectal cancer incidence or mortality were included up to 1999. Prospective studies that did not report the level of exposure (red meat/processed meat consumption) were excluded. Norat et al. (2002)
study derives 18 case-control and 6 cohort studies, selected from 48 (5
). All studies published up to 1999, and providing association between total meat, red meat or processed meat intake and colon, rectal and colorectal incidence or mortality, were included. Sandhu’s and Norat’s meta-analyses were not independent, since eleven studies were common to both articles. Last, Larsson et al. published in 2006 a meta-analysis of 18 prospective studies, selected from 23, gathering a total of more than one million subjects. Norat’s and Larsson’s studies were quite independent, since only 15% of Norat’s subjects were counted again in Larsson’s study (6
These three meta-analyses take all previous studies into account, and bring global and consistent conclusions on the effect of different types of meat: total meat, red meat, processed meat. Briefly:
- Total intake of meat (including white and red meat from all sources) is not associated with CRC risk in Norat’s and in Larsson’s analyses. Sandhu’s study shows a significant moderate risk associated with total meat intake, but the authors did not include white meat (poultry) in total meat.
- A high intake of red meat (usually including beef, veal, lamb, mutton, pork, and offal) is associated with a moderate and significant increased risk of CRC in the three studies:
- In Sandhu’s study (4), the average relative risk (RR) of CRC for a 100 g portion of red meat is 1.17. The 95% confidence interval (CI) is 1.05–1.31. Processed meat was not included in red meat, we think, in this study (4).
- In Norat’s study (5), CRC RR = 1.35 (CI: 1.21–1.51) for the highest quartile of consumption of red meat (including processed meat). A minor difference is observed between results from case-control and cohort studies (RR=1.36 and 1.27 respectively). The intake of 120 g/d of fresh (unprocessed) red meat is associated with a significant risk, but of lower magnitude than when processed meat is included (+ 19% compared with + 35%) (5).
- In Larsson’s study (6), CRC RR = 1.28 (CI: 1.15–1.42) for the highest category of consumption of red meat (including processed meat). Fresh red meat intake (unprocessed meat) was reported in nine studies out of fifteen, and the associated RR was 1.22, a significant value. The risk excess associated with intake of 120 g/d of red meat was + 28%. Larsson’s article does not report the quantitative effect of fresh red meat, and no precision is given on the categories (6).
- Processed meat intake (usually including sausages, meats burgers, ham, bacon, salami, nitrite-treated meat and meat products) is associated with CRC risk in all reports: Global RR are 1.49 (CI: 1.22–1.81), 1.31 (CI: 1.13–1.51) and 1.20 (CI: 1.11–1.31) in the three meta-analyses (4–6). In Norat’s analysis, a minor difference is observed between results from case-control and cohort studies (RR=1.29 and 1.39 respectively).
Thus the estimated excess risk associated with fresh red meat intake was 17%, 19% and 22%, and the risk associated with processed meat was 49%, 31% and 20%, in the three reviews, respectively. The estimates of risk for fresh red meat are within a narrow range, but estimates of risk for processed meat are more dispersed. However, all RRs are significant, and none is larger than 1.5, which shows the consistency of the meta-analyses. As shown in , doses-response meta-analyses suggest that one gram of processed meat is eleven-times, six-times or twice more “promoting” than one gram of fresh red meat in the three meta-analyses, respectively (4
). It is not easy to explain why the processed/fresh meat ratio is higher in Sandhu’s study than in Larsson’s study. However, the three studies indicate that processed meat intake is associated with a higher CRC risk than the intake of other types of meat.
Four cohort study articles dealing with processed meat intake and CRC have been published after Larsson’s 2006 review (one new cohort, and three re-analyses, ), and seven case-control studies shown in were published after Norat’s 2002
review. Let us examine below if they strengthen or weaken the above-reported meta-analyses results.
Prospective studies published between 2003 and 2006, on the association between processed meat intake and colorectal cancer risk.
Case-control studies published between 2003 and 2007, on the relationship between processed meat intake and colorectal cancer risk.
- A cohort of 30,000 men and women in Japan was studied by Oba et al. (2006), with 231 CRC cases. Processed meats were ham, sausage, bacon, and yakibuta (Chinese roasted pork). In men, there was a positive association between CRC and the highest tertile of processed meat consumption (RR=1.98, CI: 1.24–3.16). No association was seen in women (RR=0.85, CI: 0.5–1.43) (19). Three other articles made use of already published cohort studies, but they analyzed prospective data by dietary patterns, instead of type of foods. Fung et al. (2003) used data from the Nurses’ Health Study (20). The highest quintile of women eating a “western pattern”, defined by a high intake of red and processed meats, sweets and desserts, French fries, and refined grains, had a marginally significant increase in colon cancer risk, consistent with meta-analyses result (RR= 1.46, CI: 0.97–2.19). No association was found with rectal cancer (20). Dixon et al. (2004) analyzed three prospective studies: the Alpha-Tocopherol Beta-Carotene Study (ATBC), the Netherlands Cohort (NLC), and the Swedish Mammography Cohort (SMC) (21). Exploratory factor analysis identified a dietary pattern that includes processed meat in the three cohorts: the Processed meat, Pork, and Potatoes pattern. This pattern was associated with an increased risk of colon cancer in the SMC women (RR=1.62, CI: 1.12–2.34), and of rectal cancer in the ATBC men (RR=2.21, CI: 1.07–4.57), but not in the NLC study (RRs=0.9) (21). Kesse et al. (2006) studied food patterns in a French cohort of women, already reported in the EPIC study. The “Western” diet pattern included: processed meat, potatoes, pizzas and pies, sandwiches, sweets, cakes, cheese, cereal products, eggs, and butter. The three other diets were: “Healthy” diet (vegetables, fruit, yogurt, sea products, and olive oil, “Drinker” diet (sandwiches, snacks, processed meat, and alcoholic beverages) and “Meat eaters” diet (meat, poultry, and margarine). “The” Western pattern increased adenoma risk, but not CRC risk (RR= 1.39, CI: 1.00–1.94 and RR = 1.09, CI: 0.60–2.00 respectively). “The” Drinker and the Meat eaters diets increased the adenoma risk and the CRC risk (see RRs on ) (22). To sum up these recent prospective studies, they bring some support to the conclusions of Larsson’s metaanalysis that processed meat intake is associated with increased risk, and the RR is in the range 1.5–2. However, the link was not found in all sub-groups (male/female, colon/rectum), and the risk associated with dietary patterns cannot be attributed to processed meat alone.
- Seven case-control studies dealing with processed meat have been published after Norat’s meta-analysis. All studies report OR above 1.15, but only three studies out of six found a significant association between processed meat intake and CRC risk. In Shangai, China, Chiu et al. (2003) found that a high intake of preserved foods (whether animal or plant source) was associated with an increased risk of colon cancer (OR= 2.0, CI: 1.5–2.9 in men, and OR=2.7, CI: 1.9–3.8 in women). Preserved vegetables was more strongly associated with cancer risk than preserved animal foods (23). In the U.S.A., Chiu and Gapstur (2004) investigated the effect of dietary changes during adult life. They showed that risk was higher for people who did not reduce their consumption of red meat and processed meat after the age of 30 years, and risk was particularly high for pork chops/ham steaks eaters (OR= 3.7, CI: 1.6–8.7) (24). In Canada, Nkondjock et al. established dietary patterns, as reported above for cohorts. The “pork and processed meat” pattern, characterized by a high consumption of processed meat, pork, and white bread, increased colon cancer risk nearly significantly (RR=1.6, CI: 0.9–2.8) (25). In Utah and Northern California, Murtaugh et al. (2004) found no association between processed meat intake and the risk of rectal cancer (RR=1.2, CI: 0.85–1.7) (26). In Japan, Kimura et al. found that processed meat intake (and red meat intake) was not related to CRC risk (OR=1.15, CI: 0.83–1.60) (27). A Maryland case-control study of colorectal adenoma found a two-fold increased risk in the highest, compared to the lowest, quartile of processed meat intake (95% CI = 1.0–4.0). This OR was mostly explained by nitrate/nitrite intake, and marginally attenuated by MeIQx intake (a heterocyclic amine formed by cooking). In addition, ham steak/pork chops, hot dogs/other sausages, and liverwurst intake each were associated with a two-fold risk of adenoma, while bacon, breakfast sausages, ham, bologna, salami, and other luncheon meats intake were not associated with the risk (16). Lastly, In Canada, Hu et al. (28) found that consumption of processed meat increase risk of both proximal and distal colon cancer in men and women (all four OR were between 1.4 and 1.6, all CI:1.0–2.0, 2.2 or 2.4). Bacon intake was particularly associated with risk of colon cancer (proximal and distal) in women.
The estimation of cancer risk associated with meat may be influenced by other dietary factors, as shown clearly in the “dietary pattern” studies cited above (20
). In those studies, the intake of processed meat was associated with intake of French fries (or potatoes), sweets, cakes, desserts, snacks and alcoholic beverages: These high glycemic index diets, and alcohol intake, may be risk factors for colorectal cancer. In addition, high-meat diets have been negatively associated with food groups rich in antioxidants and fiber, components which have been associated with a reduced risk of colorectal cancer (4
). Thus, the effect of processed meat consumption on the risk of colorectal cancer may be confounded by other foods, as discussed further in the “Indirect mechanisms” section below. However, red meat intake is more consistently associated with risk than any other dietary factor, except the total energy intake (3
In summary, the results of these meta-analyses support the hypothesis that high consumption of red and processed meat may increase the risk of CRC. The few studies published after the metaanalyses also support the evidence, although individual studies are seldom significant. In addition, the risk associated with consumption of one gram of processed meat was two to ten times higher than the risk associated with one gram of fresh red meat. It is thus likely that processed meat contains some components that are more potent than fresh red meat components.