Red meat consumption and risk of cancers of the proximal colon, distal colon and rectum: The Swedish Mammography Cohort

Abstract

Although there is considerable evidence that high consumption of red meat may increase the risk of colorectal cancer, data by subsite within the colon are sparse. The objective of our study was to prospectively examine whether the association of red meat consumption with cancer risk varies by subsite within the large bowel. We analyzed data from the Swedish Mammography Cohort of 61,433 women aged 40–75 years and free from diagnosed cancer at baseline in 1987–1990. Diet was assessed at baseline using a self-administered food-frequency questionnaire. Over a mean follow-up of 13.9 years, we identified 234 proximal colon cancers, 155 distal colon cancers and 230 rectal cancers. We observed a significant positive association between red meat consumption and risk of distal colon cancer (p for trend = 0.001) but not of cancers of the proximal colon (p for trend = 0.95) or rectum (p for trend = 0.32). The multivariate rate ratio for women who consumed 94 or more g/day of red meat compared to those who consumed less than 50 g/day was 2.22 (95% confidence interval [CI] 1.34–3.68) for distal colon, 1.03 (95% CI 0.67–1.60) for proximal colon and 1.28 (95% CI 0.83–1.98) for rectum. Although there was no association between consumption of fish and risk of cancer at any subsite, poultry consumption was weakly inversely related to risk of total colorectal cancer (p for trend = 0.04). These findings suggest that high consumption of red meat may substantially increase the risk of distal colon cancer. Future investigations on red meat and colorectal cancer risk should consider cancer subsites separately. © 2004 Wiley-Liss, Inc.

Although there is mounting support in the literature of an increased risk of colon or colorectal cancer associated with high consumption of red meat,1, 2 data for subsites within the colon are limited and conflicting.3, 4, 5 Multiple lines of evidence indicate that proximal and distal colon cancers may have distinct etiologies.6, 7, 8, 9, 10 Cancers arising in the proximal and distal colon exhibit differences in incidence according to geographic region, age and sex.7, 10 Furthermore, there are various molecular and clinical differences between proximal and distal sites of the colon that may influence the susceptibility to environmental factors.10 For example, differences between subsites exist in pH,11 bacterial composition and bacterial metabolic capacity,12, 13 apoptotic index,14 activity of ornithine decarboxylase15 and expression of carcinogen metabolizing enzymes.16 These differences underscore the need for separate analyses by subsite within the colon.

Components in red meat or which are formed during the cooking of meat that may increase colorectal cancer risk include animal fat,17, 18 heme iron19, 20, 21 and heterocyclic amines (HCAs).22, 23 In addition, human experimental studies have demonstrated that high red meat diets enhance the endogenous formation of N-nitroso compounds (NOCs),24, 25, 26, 27 which have been found to induce the formation of DNA adducts in human colonocytes.28 The concentrations of the pro-mutagenic lesion O⁶-methyldeoxyguanosine, a marker of exposure to many NOCs, have been shown to be significantly greater in tissues from the distal colon and rectum than from the proximal colon.29 A direct association of meat consumption with body iron stores has been found in many studies.30, 31, 32 In a nested case-control study, an increased colorectal cancer risk associated with high serum iron levels was restricted to the distal colon and rectum.33 Thus, given these findings, a potential effect of diets high in red meat might be limited to cancers of the distal colon and rectum.

In our study, we analyzed data from a large population-based prospective cohort of women to examine whether the association of red meat consumption with colorectal cancer risk differs according to subsite within the large bowel.

Material and methods

Study population

The Swedish Mammography Cohort is a population-based prospective study including 66,651 women from Uppsala and Västmanland Counties (central Sweden) who were 40–75 years of age at recruitment in 1987–1990.34 The women responded to a mailed questionnaire at baseline that elicited information about diet, weight, height and educational level. In 1997, a follow-up questionnaire was sent to all surviving participants (70% response rate). This questionnaire was expanded to include information on cigarette smoking status and history, hormone replacement therapy use and aspirin use.

For the present analyses, we excluded women with an erroneous National Registration Number, women with an extreme energy intake (3 SD below or above the log_e transformed energy intake) and women with a previously diagnosed cancer (except nonmelanoma skin cancer) at baseline. After these exclusions, 61,433 women remained for these analyses. This study was approved by the Ethics Committees at the Uppsala University Hospital (Uppsala, Sweden) and the Karolinska Institutet (Stockholm, Sweden).

Dietary assessment

A self-administered food-frequency questionnaire was used to assess dietary intake. Women were asked about their average frequency of consumption of 67 food items (including 9 items about red and processed meats) during the past 6 months. The 8 prespecified response categories ranged from never/seldom to 4 or more times per day. We used age-specific (<53, 53–65, ≥66 years) portion sizes that were based on mean values obtained from 213 randomly chosen women from the study area whose food intake for 5,922 days was weighed and recorded. Nutrient intake was calculated by multiplying the consumption frequency of each food item by the nutrient content per serving, using composition values obtained from the Swedish National Food Administration Database.35

Consumption of red meat and processed meats was computed by multiplying the consumption frequency of each meat item by its age-specific portion size and then summing up those items. Total red meat consisted of the following items: whole beef, chopped meat, minced meat, bacon, hot dogs, ham or other lunch meat, blood pudding, kidney or liver, and liver pate. Processed meats consisted of bacon, hot dogs, ham or other lunch meat and blood pudding (high content of heme iron). Fish was considered the sum of lean fish and fatty fish (i.e., salmon, mackerel, sardines and herring). In a validation study in a random sample of 129 women from the cohort, the Pearson correlation coefficients between intakes from the mean of 4 1-week diet records (obtained 3–4 months apart) and the questionnaire ranged from 0.2 to 0.5 for individual red and processed meat items and was 0.3 for poultry, 0.3 for lean fish and 0.4 for fatty fish.

Ascertainment of colon and rectal cancer cases and follow-up of the cohort

Incident cases of colon and rectal cancers that occurred in the cohort were identified by computerized record linkages to the National Swedish Cancer Registry (from March 1987 through December 31, 2002) and the Regional Cancer Registry in the study area (from January 1, 2003 through June 30, 2003). Follow-up of the cohort through the Swedish Cancer Registries is estimated to be more than 98% complete.36 Date of deaths and the date of leaving the study area were ascertained through the Swedish Death Registry and the Swedish Population Registry.

Colon cancers were defined as those occurring above the peritoneal delineation of the abdominal cavity, and rectal cancers were those occurring below this delineation. Tumors originating proximal to the splenic flexure (cecum, ascending colon and transverse colon) were considered proximal colon cancers, whereas those tumors arising in the descending or sigmoid colon were considered distal colon cancers.

Statistical analysis

Each woman contributed follow-up time from the date of her entry to the cohort to the date of a colon or rectal cancer diagnosis, date of death from any cause, date of moving out from the study area (only for the period from January 1, 2003 through June 30, 2003) or June 30, 2003, whichever occurred first. Women were categorized into quartiles according to consumption of red meat and fish. Women were categorized into 2 groups according to blood pudding consumption and into 3 groups according to poultry consumption. We computed incidence rates by dividing the number of cases by person-years of follow-up in each category. Rate ratios (RRs) were calculated by dividing the incidence rate in a specific category of meat consumption by the rate in the lowest category. The proportional hazards assumptions were satisfied, and we used Cox proportional hazards regression to estimate RRs with 95% confidence intervals (CIs). All multivariate analyses were adjusted for age, body mass index (the weight in kilograms divided by the square of height in meters), educational level and intakes of total energy, alcohol, saturated fat, calcium, folate, fruits, vegetables and whole grain foods. We adjusted nutrient values for total energy intake using regression analysis.37 For tests of linear trend, we used the median value of each category of intake analyzed as a continuous variable. We tested for differences in associations with cancers of the proximal colon, distal colon and rectum using a χ² test. Analyses were performed using the SAS statistical software (version 8.2; SAS Institute, Cary, NS). All p-values were 2-sided.

Results

Over a mean follow-up of 13.9 years (855,585 person-years) of 61,433 women, 733 incident cases of colorectal cancer were diagnosed. Of these cancers, 234 were located in the proximal colon, 155 in the distal colon and 230 in the rectum; 4 cases were diagnosed with both colon and rectal cancer and information on subsite was not available for 110 cases. The mean age (±SD) at diagnosis of colorectal cancer was 68.3 years (±9.0 years). At baseline, the median consumption of red meat in the cohort was 70 g/day, and the median values in the lower and upper quartiles were 37 g/day and 114 g/day, respectively (Table I). Women with higher red meat consumption were younger and tended to have a slightly higher BMI than women with lower consumption. Intakes of total energy, alcohol, saturated fat, vegetables, whole-grain foods, fish and poultry increased, whereas calcium intake decreased across quartiles of red meat consumption. To account for these differences, we controlled for the corresponding variables in multivariate analyses.

Table I. Age-Standardized Baseline Characteristics of 61,433 Women According to Red Meat Consumption

Characteristics	Quartiles of red meat consumption (g/day)1
	1 <50	2 50–69	3 70–93	4 ≥94
1 Red meat includes whole beef, chopped meat, minced meat, bacon, hot dogs, ham or other lunch meats, blood pudding, kidney or liver and liver pate. 2 Adjusted for total energy intake. 3 Whole-grain foods include crisp bread, dark bread, breakfast cereals and porridge.
Participants (n)	15,442	15,410	15,335	15,246
Median red meat consumption (g/day)	37	60	80	114
Mean age (years)	57.8	54.8	52.3	50.0
Mean body-mass index (kg/m2)	24.5	24.7	24.8	24.9
≥12 years education (%)	11.7	10.6	11.5	11.7
Mean intake
Total energy (kcal/day)	1,139	1,266	1,368	1,555
Alcohol (g/day)	2.8	3.0	3.2	3.3
Saturated fat (g/day)2	17.1	17.7	18.0	18.3
Folate (μg/day)2	187	181	182	184
Calcium (mg/day)2	745	711	692	648
Fruits (servings/day)	1.5	1.4	1.5	1.6
Vegetables (servings/day)	1.7	1.6	1.8	2.0
Whole-grain foods (servings/day)3	2.5	2.6	2.7	2.8
Fish (servings/week)	1.6	1.6	1.7	1.9
Poultry (servings/week)	0.3	0.3	0.4	0.5
Processed meat (g/day)	9.6	19.1	26.2	38.3
Blood pudding (servings/month)	0.5	1.0	1.4	1.9

The association between meat consumption and risk of colorectal cancer, both overall and by subsites, is shown in Table II. High consumption of red meat was associated with a statistically significant increased risk of colorectal cancer after adjustment for age and other potential risk factors. Compared to women who consumed less than 50 g/day of red meat, the multivariate rate ratio (RR) for those who consumed 94 or more g/day was 1.32 (95% confidence interval [CI] 1.03–1.68; p for trend = 0.03). When red meat consumption was analyzed as a continuous trend, risk of colorectal cancer increased by 20% (RR = 1.20, 95% CI 0.99–1.45) for a daily increment of 100 g of red meat consumption. We observed no statistically significant association between consumption of processed meats or fish and risk of colorectal cancer. By contrast, there was a weak inverse association of poultry consumption with colorectal cancer risk; the multivariate RR of colorectal cancer for women who consumed on average 1 serving of poultry per week was 0.75 (95% CI 0.55–1.02) compared to women who rarely or never consumed poultry.

Table II. Multivariate Rate Ratio of Colorectal Cancer Among 61,433 Women From 1987 to 2003 According to Meat Consumption

	Multivariate RR (95% CI)1
	Colorectum (n = 733)	Proximal colon1 (n = 234)	Distal colon2 (n = 155)	Rectum2 (n = 230)
1 Models adjusted for age at baseline (5-year categories), body mass index (quartiles), educational level (less than high school, high school, university) and quartiles of intakes of total energy, alcohol, saturated fat, calcium, folate, fruits, vegetables and whole-grain foods. 2 The sum of proximal colon, distal colon and rectal cancers does not add to the total number of colorectal cancers because of missing data on subsite (n = 110); cases diagnosed with both colon cancer and rectal cancer were excluded from subsite analyses (n = 4). 3 Red meat (or beef and pork as a main dish), fish and poultry are mutually adjusted. 4 Beef and pork as a main dish includes whole beef, chopped meat and minced meat. 5 Further adjusted for beef and pork as a main dish, fish and poultry.
Red meat3 (g/day)
<50 (median, 37)	1.00	1.00	1.00	1.00
50–69 (60)	1.06 (0.87–1.30)	0.95 (0.68–1.33)	0.92 (0.57–1.47)	1.16 (0.81–1.67)
70–93 (80)	1.12 (0.90–1.39)	0.85 (0.58–1.26)	1.30 (0.80–2.11)	1.01 (0.68–1.52)
≥94 (114)	1.32 (1.03–1.68)	1.03 (0.67–1.60)	2.22 (1.34–3.68)	1.28 (0.83–1.98)
p-value for trend	0.03	0.95	0.001	0.32
Processed meats (g/day)
<12 (median, 6)	1.00	1.00	1.00	1.00
12–21 (16)	0.89 (0.72–1.90)	0.92 (0.66–1.32)	1.05 (0.67–1.64)	0.78 (0.52–1.12)
22–31 (26)	1.01 (0.82–1.24)	0.85 (0.58–1.24)	0.98 (0.61–1.58)	1.02 (0.75–1.55)
≥32 (41)	1.07 (0.85–1.33)	1.02 (0.69–1.52)	1.39 (0.86–2.24)	0.90 (0.60–1.34)
p-value for trend	0.23	0.97	0.20	0.88
Beef and pork34 (servings/week)
<2.0 (median, 1.5)	1.00	1.00	1.00	1.00
2.0–<3.0 (2.5)	1.13 (0.95–1.36)	0.90 (0.65–1.24)	1.26 (0.84–1.90)	1.18 (0.86–1.62)
3.0–<4.0 (4.0)	0.90 (0.70–1.17)	0.78 (0.45–1.17)	0.98 (0.55–1.75)	0.87 (0.55–1.37)
≥4.0 (5.5)	1.22 (0.98–1.53)	1.10 (0.74–1.64)	1.99 (1.26–3.14)	1.08 (0.72–1.62)
p-value for trend	0.32	0.90	0.01	0.98
Blood pudding5 (servings/month)
0	1.00	1.00	1.00	1.00
≥1.0 (median, 2.0)	1.13 (0.97–1.32)	0.90 (0.69–1.18)	1.30 (0.94–1.80)	1.20 (0.92–1.57)
Fish3 (servings/week)
<0.5 (median, 0.5)	1.00	1.00	1.00	1.00
0.5–<1.0 (1.0)	0.94 (0.72–1.22)	0.88 (0.57–1.36)	0.79 (0.46–1.35)	1.11 (0.68–1.81)
1.0–<2.0 (1.5)	1.21 (0.94–1.55)	1.13 (0.75–1.71)	0.92 (0.55–1.52)	1.32 (0.83–2.11)
≥2 (3.0)	1.08 (0.81–1.43)	1.03 (0.63–1.67)	0.83 (0.45–1.51)	1.08 (0.63–1.86)
p-value for trend	0.48	0.70	0.55	0.97
Poultry3 (servings/week)
0	1.00	1.00	1.00	1.00
<0.5 (median, 0.5)	0.89 (0.76–1.04)	0.88 (0.67–1.17)	0.91 (0.64–1.29)	0.97 (0.73–1.28)
≥0.5 (1.0)	0.75 (0.55–1.02)	0.77 (0.44–1.36)	0.86 (0.46–1.62)	0.62 (0.34–1.13)
p-value for trend	0.04	0.26	0.76	0.24

When we examined risk by cancer subsite (Table II), high consumption of red meat was associated with a statistically significant more than 2-fold increased risk of distal colon cancer, whereas there was no apparent association with risks of proximal colon or rectal cancers. The χ² test for the difference between the estimates for proximal colon vs. distal colon cancer was statistically significant (p = 0.02) but not the difference for distal colon vs. rectal cancer (p = 0.11). The increased risk of distal colon cancer associated with red meat consumption was attributable to beef and pork as a main dish and blood pudding (Table II). Consumption of blood pudding was also weakly, albeit not statistically significantly, associated with increased risk of rectal cancer. When we combined cancers of the distal colon and rectum, we found that women who consumed on average 2 servings of blood pudding per month had a significant 26% greater risk (RR = 1.26, 95% CI 1.02–1.55) compared to women who rarely or never consumed blood pudding.

We used restricted cubic spline regression with 5 knots to model the association of red meat consumption with distal colon cancer risk continuously. As shown in Figure 1, the risk of distal colon cancer increased monotonically with increasing red meat consumption. A daily increment of 100 g of red meat consumption corresponded to a 70% increase in distal colon cancer risk (multivariate RR = 1.70, 95% CI 1.31–2.21).

Figure 1. Multivariate rate ratio of distal colon cancer according to red meat consumption. Data were fit by using restricted cubic spline Cox proportional hazards models, adjusted for the same variables as in the multivariate model in Table II and consumption of fish and chicken. The 95% confidence intervals are indicated by the dashed lines.

To eliminate the possibility that preclinical symptoms of colorectal cancer might have influenced dietary intake before the diagnosis, we repeated our analyses after excluding distal colon cancer cases (n = 27) that occurred during the first 3 years of follow-up. Compared to women who consumed less than 50 g/day of red meat, those who consumed 94 or more g/day had a multivariate RR of distal colon cancer of 2.04 (95% CI 1.17–3.54; p for trend = 0.01).

In secondary analyses, we used data from the 1997 questionnaire and repeated our analyses after further controlling for cigarette smoking status and history (never smoker; ever smoker, <30 years; and ever smoker, ≥30 years), hormone replacement therapy use (yes or no) and aspirin use (yes or no). In models further adjusted for these variables in addition to the same variables in the multivariate model in Table II, the RRs comparing the highest with the lowest quartile of red meat consumption were 1.01 (95% CI 0.65–1.57) for proximal colon cancer, 2.20 (95% CI 1.33–3.65) for distal colon cancer and 1.27 (95% CI 0.83–1.97) for rectal cancer.

Discussion

This large population-based prospective cohort study of Swedish women supports the concept that high consumption of red meat increases the risk of colorectal cancer. Here, we show that the increased risk may be confined to distal colon. We found no evidence of an association between fish consumption and risk of cancer at any subsite. Poultry consumption had a modest inverse association with colorectal cancer risk.

Two meta-analyses1, 2 have summarized the epidemiologic data on red meat consumption and risk of total colorectal cancer. In one meta-analysis of 13 cohort studies,1 risk of colorectal cancer increased by 13–17% for a daily increment of 100 g of red meat. The other meta-analysis,2 based on data from both cohort and case-control studies, showed that a daily increment of 120 g of red meat corresponded to a 24% increase in the risk of colorectal cancer. Our present study corroborates and extends the results from those 2 meta-analyses by showing a 20% increase in total colorectal cancer risk and a 70% increase in distal colon cancer risk for each daily increment of 100 g of red meat consumption. Our results on poultry are in accord with some,4, 5, 38, 39 although not all,40, 41, 42 previous studies in which inverse associations were observed between consumption of poultry and risk of colorectal cancer.

Despite persuasive evidence that proximal and distal colon cancers may arise through different molecular pathways,6, 7, 8, 9, 10 only 1 cohort study4 and 2 case-control studies3, 5 have related red meat consumption to specific subsites within the colon. In the Health Professionals Follow-up Study,4 the positive association between red meat consumption and colon cancer risk appeared to be limited to distal colon, although the results did not reach statistical significance, possibly owing to a small number of distal colon cancer cases. A case-control study conducted in Hawaii5 found an association of high red meat consumption with increased risk of proximal colon cancer in men but not in women. Another case-control study in the United Kingdom3 indicated a stronger positive association of red meat with proximal colon cancer than with distal colon and rectal cancers combined.

The biologic mechanisms relating high consumption of red meat to an increased risk of colon cancer remain speculative. Much attention has been focused on heterocyclic amines (HCAs) formed during high-temperature cooking of meat and which are potent mutagens and carcinogens in animal models.23 However, the fact that the levels of HCAs in cooked white meat (i.e., fish and poultry) are as high or exceed the levels in cooked red meat43, 44 suggests that this potential mechanism is not likely to explain the observed positive association of red meat, but not of white meat, with risk of cancer at any subsite within the colon. Nevertheless, it is possible that other constituents of white meat (e.g., long-chain n-3 polyunsaturated fatty acids present in fish) outweigh a possible detrimental influence of HCAs on colon cancer risk.

Red meat is a rich source of heme iron. Intakes of heme iron and red meat have consistently been shown to be positively associated with body iron status.30, 31, 32 Iron is a prooxidant that may increase colon cancer risk by enhancing the production of free oxygen radicals.45, 46 In addition, iron appears to be essential for the proliferation of tumor cells.47 Our study provided a unique opportunity to evaluate the association between consumption of blood pudding, which contains very high amounts of heme iron (about 15 mg/serving compared to beef, which contains about 1.5 mg/serving) and risk of colorectal cancer. We found that consumption of blood pudding was associated with a significant 26% greater risk of cancers of the distal colon and rectum combined. Noteworthy, a nested case-control study33 showed that an increased risk of cancer associated with high serum iron levels was confined to distal colon and rectum. One might speculate that a very high concentration of heme iron in the colon and rectum after consumption of blood pudding increases the production of free oxygen radicals in the colon and rectum, thereby increasing cancer risk.

Alternatively, high red meat consumption might increase the risk of colon cancer by enhancing the endogenous formation of NOCs,48 most of which are known carcinogens.24, 25, 26, 27 Human experimental studies have demonstrated that consumption of red meat, but not of white meat, significantly increases in a dose-dependent manner fecal levels of NOCs.24, 25, 26, 27 Recently, a controlled study of 21 healthy men showed that the fecal levels of NOCs were significantly higher on a high red meat diet than on a low red meat diet.27 The authors also reported that supplementation with 8-mg heme iron (as liver pate and blood sausage) significantly increased fecal NOCs output, whereas inorganic iron (35 mg ferrous iron) had no effect. Animal studies have shown that endogenous N-nitrosation in the colon is dependent on a gut flora49 and is most efficient at neutral pH (optimum at pH 7.5).50 Given the higher proportion of bacteria and pH in the distal colon (pH 7.0 ± 0.7) than in the proximal colon (pH 6.4 ± 0.4),11 the production of NOCs might be higher in the distal colon. In this regard, it is noteworthy that the concentration of the promutagenic lesion O⁶-methyldeoxyguanosine, a marker of exposure to many NOCs, has been shown to be significantly greater in tissues from the distal colon and rectum than from the proximal colon.29 The colon carcinogen 1,2-dimethylhydrazine alkylates DNA and the promutagenic lesion O⁶-methylguanine has been detected in DNA from rodent tissues after exposure to this agent.51 Interestingly, experimental studies have demonstrated that 1,2-dimethylhydrazine almost exclusively produces O⁶-methylguanine adducts and tumors in the distal colon.51, 52, 53 Thus, an increased cancer risk associated with high red meat consumption is expected to be stronger or limited to the distal colon, as was found in our present study.

Our study has several important strengths: the population-based design enhances the generalizability of our results; specific subsites within the colon were examined separately; and the Swedish Cancer Register system made it possible to identify virtually all incident colorectal cancer cases that occurred in the cohort. In addition, because exposure information was collected before the cancer diagnosis, any measurement error would have been nondifferential between cases and noncases and would most likely weaken any true association rather than causing an overestimation. Our study further included considerably more colorectal cancer cases than previous prospective cohort studies of red meat consumption and colorectal cancer risk.

Because meat consumption was self-reported by questionnaire, some misclassification of exposure is inevitable. Such misclassification in prospective studies is expected to be random and would generally bias results towards the null. Another concern is our inability to control for physical activity. However, adjustment for body mass index and total energy intake, as rough indicators of energy expenditure and physical activity, did not change our findings.

In summary, this prospective study provides evidence that high consumption of red meat may substantially increase the risk of distal colon cancer. Future investigations on the association between red meat consumption and colorectal cancer risk should consider subsites within the colon separately.