The authors tested the hypothesis that the metabolic syndrome (≥3 of the following components: high blood pressure, increased waist circumference, hypertriglyceridemia, low levels of high-density lipoprotein cholesterol, or diabetes/hyperglycemia) is a risk factor for colorectal cancer.
Data from the Atherosclerosis Risk in Communities (ARIC) multicenter prospective cohort study were used. Metabolic syndrome components and other risk factors were collected during 1987 to 1989 from the 14,109 men and women in these analyses. One hundred ninety-four incident colorectal cancers were identified through the Year 2000. Multivariate Cox proportional hazards regression analyses were used to examine associations.
Baseline metabolic syndrome (≥3 components vs. 0 components) had a positive association with age-adjusted and gender-adjusted colorectal cancer incidence (relative risk [RR], 1.49; 95% confidence interval [95%CI], 1.0-2.4); this association was attenuated after multivariate adjustment (RR, 1.39; 95%CI, 0.9-2.2). There was a dose-response association between colorectal cancer incidence and the number of metabolic syndrome components present at baseline (P for trend = .006) after multivariate adjustment. Analysis of gender revealed that the multivariate-adjusted association of metabolic syndrome with colorectal cancer was stronger in men (RR, 1.78; 95%CI, 1.0-3.6) and weaker in women (RR, 1.16; 95%CI, 0.6-2.2).
Colorectal cancer is the 3rd most common incident cancer in the United States, with an estimated 146,940 diagnoses in 2004, and it is the third leading cause of cancer mortality.1 Understanding risk factors for colorectal cancer may guide the development of strategies targeted at its prevention.
There is evidence that body composition and hormonal factors contribute to colorectal cancer etiology. Excess body mass and abdominal obesity have been associated positively with colorectal cancer in prospective human research.2 Abdominal obesity may be a stronger risk factor than generalized obesity.3 Weight gain in middle age that results in abdominal obesity increases the risk of colonic adenomas4 and colorectal carcinoma.5 Abdominal obesity is associated with hyperinsulinemia6; in vitro, animal, and human epidemiologic studies support a relation between insulin and colorectal carcinogenesis.7–11
Hyperinsulinemia,11 markers of insulin resistance (such as hypertriglyceridemia or hyperglycemia12), diabetes, obesity, and other factors associated with obesity, such as elevated levels of inflammatory factors, growth hormones, and gender hormones,13 all have been hypothesized as factors that increase the risk of colorectal cancer. These hypothesized risk factors may interact to increase risk; therefore, a collection of metabolic factors may be an appropriate model in which to examine colorectal cancer risk. The metabolic syndrome comprises several risk factors that predispose for cardiovascular disease, Type II diabetes, and all-cause and cardiovascular disease mortality14, 15; it also is hypothesized that the metabolic syndrome increases the risk of colorectal cancer.11 The prevalence of the metabolic syndrome16 has been estimated at 24% and 23% among adult males and females, respectively, in the United States.17 Three prospective studies have examined associations between the insulin-resistance syndrome, which often is used synonymously with the metabolic syndrome, and colorectal cancer incidence18 and mortality.19, 20 To our knowledge, no study has examined whether or not the metabolic syndrome, as defined by the National Cholesterol Education Program Adult Treatment Panel III (ATP III),16 is a risk factor for colorectal cancer. To examine this question, we used data from the Atherosclerosis Risk in Communities (ARIC) Study. Associations between colorectal cancer and individual components of the metabolic syndrome, other than obesity, have not been studied extensively; therefore, we used these data to examine whether or not individual components of the metabolic syndrome altered colorectal cancer risk.
MATERIALS AND METHODS
The ARIC study is a prospective investigation that was designed originally to study the etiology and natural history of atherosclerotic disease in 4 communities in the United States: Forsyth County, NC; Jackson, MS (African Americans only); suburban Minneapolis, MN; and Washington County, MD.21 The ARIC Study and its cancer follow-up were approved by each institution's Institutional Review Board. Between 1987 and 1989, a cohort of 15,792 men and women, ages 45 to 64 years, was established from the 4 communities. Of those who were recruited, there was a 46% response in the Jackson cohort and a 65% to 67% response in the other cohorts. After participants provided written informed consent, completion of the baseline clinical examination (Visit 1) marked study enrollment. Individuals participated in up to 3 follow-up examinations that were conducted at 3-year intervals. Of the original cohort, 93% of participants attended the 2nd examination (Visit 2; 1990-1992), 86% attended the 3rd examination (Visit 3; 1993-1995), and 80% attended the 4th examination (Visit 4; 1996-1998). Between examinations and after Visit 4, participants completed annual telephone interviews. Between 1994 and 1996, the telephone interview, which was completed by 90% of the cohort, included information about family history of any cancer (including site and age of onset) in first-degree relatives; history of nonsteroidal antiinflammatory drug (NSAIDs) use; history of sigmoidoscopy, colonoscopy, or test for blood in the stools; and history of polyps or noncancerous tumors of the colon or rectum. Participants self-reported all hospitalizations during annual follow-up interviews.
Risk factors that were examined in these analyses were ascertained at Visit 1 or during the 1994 to 1996 medical history interview. Data collection and quality-control methods have been described.22
Anthropometrics were taken with participants wearing scrub suits and no shoes. Body weight was measured by using a calibrated scale (Detecto; model 437). Standing height was measured with a vertical metal ruler. Body mass index (BMI) was calculated in kg/m2. Waist (at the umbilicus) and hips (maximum girth) were measured once with anthropometric tape. The waist:hip ratio (WHR) was calculated. Anthropometric measures in ARIC are reliable.23
Sitting blood pressure was taken 3 times with a random zero sphygmomanometer (standardized Hawksley machines) with 5-minute and 30-second rest periods after the first and second readings, respectively. The second and third readings were averaged.
Participants fasted for 12 hours prior to examination. Fasting blood was drawn from an antecubital vein of seated participants into vacuum tubes that contained ethylenediamine tetraacetic acid (lipids) or a serum separator gel (glucose and insulin). Blood serum and plasma aliquots were stored at − 70 °C and shipped to central laboratories for analyses. Total cholesterol and total triglycerides were measured by enzymatic methods, high-density lipoprotein (HDL) cholesterol was measured after dextran-magnesium precipitation, and low-density lipoprotein (LDL) cholesterol was calculated (Friedewald formula). Serum glucose was assayed (hexokinase/glucose-6-phosphate dehydrogenase method); serum insulin was measured by radioimmunoassay (123Insulin Kit; Cambridge Medical Diagnostics, Inc., Billerica, MA). These blood assays are reliable.24 Fasting hyperglycemia was defined as fasting glucose ≥100 mg/dL; prevalent diabetes mellitus was defined as fasting glucose ≥126 mg/dL,25 nonfasting glucose ≥200 mg/dL, or a self-reported history of or treatment for diabetes.
Information assessed by questionnaire included years of education, smoking status (current, former, never), number of cigarettes per day and duration of smoking, alcohol drinking status (current, former, never), and usual consumption of wine, beer, and hard liquor. Usual food intake during the previous year was collected by using a modified and validated version of the Willett 61-item semiquantitative food frequency questionnaire.26 Physical activity was assessed by using the Baecke Questionnaire.27 Three physical-activity indices were derived: occupational, leisure-time participation in sports, and nonsport leisure activity. Participants were asked to bring to the examination all medications taken during the 2 previous weeks.
Participants were categorized by the number of metabolic syndrome components present at baseline and were classified with the metabolic syndrome if they had ≥3 components: 1) high blood pressure (≥130 mmHg systolic or ≥85 mmHg diastolic), 2) central obesity (waist circumference ≥102 cm in men or ≥88 cm in women), 3) high triglyceride level (>150 mg/dL), 4) low HDL cholesterol (<35 mg/dL in men or <45 mg/dL in women), and 5) diabetes/hyperglycemia.16
Ascertainment of Incident Colorectal Cancers
At examinations, participants self-reported whether they ever had been diagnosed with cancer (yes or no). Incident cancers were ascertained between January 1, 1987 and December 31, 2000. From 1987 to 2000, Minneapolis, Forsyth County, and Washington County were covered by well established state or county cancer registries. After 1995, a state registry also covered Jackson; the Mississippi registry reported 90% completeness in 1999 and 96% completeness in 2000. Cancers that occurred in Jackson prior to 1995 were identified by ARIC hospital surveillance. The Minnesota Cancer Surveillance System had data completeness estimated at 99.7%.28 The Washington County registry had 90% estimated completeness and was supplemented by the Maryland Cancer Registry to ensure coverage. From 1987 to 1989, the North Carolina Central Cancer Registry was complete for the major hospitals in Forsyth County only; statewide data were available beginning in 1990.
In communities that were covered by cancer registries, cancer occurrence, primary site, and diagnosis date were identified by linkage of cohort identifiers with that registry's data base. Additional cancer-related hospital discharge codes (International Classification of Diseases-9th Revision codes) that were identified by ARIC were retrieved. Hospital information related to the cancer diagnosis—primary site, date of diagnosis, and source of diagnosis information (e.g., pathology report)—was copied from medical records. Study investigators reviewed these records and verified cases.
Analysis and Statistical Methods
From the original ARIC cohort (n = 15,792 individuals), participants who did not fast for at least 8 hours (n = 589 individuals), had missing physical activity data (n = 160 individuals), did not provide sufficient data to determine baseline cancer status (n = 184 individuals), or had any prevalent cancer (n = 750 individuals) were excluded. Thus, 14,109 participants were included in the analyses.
SAS software (version 8.02; SAS Institute Inc., Cary, NC) was used for analyses. Person-years (PYs) at risk were calculated as the time from the baseline examination to December 31, 2000, the date of colorectal cancer diagnosis, death, or loss to follow-up, whichever occurred first. Crude and/or age-adjusted incidence rates (per 1000 PYs) were calculated for several variables that were collected at Visit 1 or in the 1994 to 1996 telephone questionnaire. Adjusted relative risks were calculated using by Cox proportional hazards regression (SAS PHREG). Tests for trend of relative risks (RRs) across categories of exposure were obtained by modeling the categories as continuous, ordinal variables. Potential confounders were examined individually, added to individual models in a stepwise fashion, and retained if they changed the RR estimate by ≥10%. Multivariate models included adjustment for age, gender, family history of colorectal cancer, physical activity, NSAID use, aspirin use, pack-years of smoking, and grams of alcohol consumed per week; analyses in women included adjustment for current hormone-replacement therapy use. Additional variables that were tested for confounding but were not added to the final models included race, energy intake, education level, and consumption levels of red meat, animal fat, whole grains, dietary fiber, vitamin D, and calcium. Neither hyperinsulinemia24 nor fasting insulin added to the colorectal cancer risk prediction beyond the metabolic syndrome or its individual components and, thus, were not included in the final models. In the models that examined associations between individual components of the metabolic syndrome and colorectal cancer, adjustment for other metabolic syndrome components did not alter the associations by >10% and, thus, were not included in the final models. Effect modification by gender with the metabolic syndrome components was evaluated by modeling a set of cross-product terms for the categorical risk factors.
Among 14,109 eligible male and female participants, 194 incident colorectal cancers, including 139 colon cancers, were verified over 165,319 total PYs of follow-up (mean, 11.5 years). The crude incidence rates of colorectal and colon cancer were 1.2 and 0.8 per 1000 PYs, respectively. The crude incidence rate of colorectal cancer was 1.5-fold higher in men than in women and increased 4-fold with age (Table 1).
Table 1. Crude and Age-Adjusted Incidence Rates of Colorectal Carcinoma in Relation to Selected Baseline Characteristics: Atherosclerosis Risk in Communities Study, 1987-2000
At baseline, the prevalence of the metabolic syndrome (≥3 components) was 34.5% in men and 34.2% in women; these percentages are greater than those reported previously in the ARIC Study29 because of the revised criteria for hyperglycemia.16 Cohort baseline characteristics according to the number of metabolic syndrome components are shown in Table 2. In 1994 to 1996, 35% of men and 33% of women had ever received a sigmoidoscopy or colonoscopy; 55% of men and 52% of women had ever received a test for blood in the stool; and 11% of men and 8% of women had a history of a polyp or a noncancerous tumor of the colorectum.
Table 2. Prevalence (%) or Mean ± Standard Deviation of Selected Characteristics by Baseline Number of Metabolic Syndrome Components: The Atherosclerosis Risk in Communities Study, 1987 to 1989 or 1994 to 1996*
No. of Metabolic Syndrome Components
0 (n = 2223)
1 or 2 (n = 6896)
3 to 5 (n = 4774)
BMI indicates body mass index; NSAID, nonsteroidal antiinflammatory drugs.
Physically active: Highest quartile of Sports Index§ (%)
Current smoker (%)
No. of pack-years smoked
13.5 ± 21.5
15.9 ± 21.4
17.1 ± 21.2
Current drinker (%)
Ethanol (g/week), current drinkers
44.1 ± 95.9
44.5 ± 95.3
39.7 ± 95.7
Current hormone-replacement therapy use (%)
NSAID use (%)
Aspirin use (%)
Energy intake (kcal/day)
1624 ± 611
1649 ± 613
Red meat (servings/day)
0.51 ± 0.4
0.54 ± 0.4
0.57 ± 0.4
Energy from animal fat (%)
19.3 ± 6.1
19.8 ± 5.7
20.2 ± 6.2
Whole grain (servings/day)
1.3 ± 1.2
1.24 ± 1.2
1.20 ± 1.2
Dietary fiber (g/day)
17.5 ± 8.4
17.1 ± 8.2
17.3 ± 8.2
Vitamin D (IU/day)
221.4 ± 144.5
216.8 ± 143.7
221.5 ± 144.2
257.1 ± 96.4
253.0 ± 95.3
252.9 ± 95.7
668.0 ± 382.7
647.6 ± 377.0
659.4 ± 378.6
The individual components of the metabolic syndrome had modest, positive associations with colorectal cancer incidence (Table 3). Modest associations were observed with other, related variables. In men and women, after multivariate adjustment, colorectal cancer had a weak, positive association with the highest versus the lowest categories of BMI (≥35.0 kg/m2 vs. <25.0 kg/m2; RR, 1.54; 95% confidence interval [95% CI], 0.9-2.8), WHR (≥0.98 vs. <0.88, RR, 1.67; 95%CI, 1.1,2.5), and fasting insulin (≥109 pmol/L vs. <43 pmol/L, RR, 1.38; 95%CI, 0.9,2.2), similar to associations for waist circumference. In men, the multivariate-adjusted RRs (95% CIs) of colorectal cancer for the highest versus the lowest quartiles of BMI (≥29.8 kg/m2 vs. <24.7 kg/m2), WHR (≥1.00 vs. <0.93), and insulin (≥107 pmol/L vs. <43 pmol/L), were 1.52 (95%CI, 0.9-2.7), 2.38 (95%CI, 1.3-4.2), and 1.15 (95%CI, 0.7-2.0), respectively. In women, the multivariate-adjusted RRs (95% CIs) of colorectal cancer for the highest versus the lowest quartiles of BMI (≥31.3 kg/m2 vs. <23.4 kg/m2), WHR (≥0.96 vs. <0.84), and insulin (≥108 pmol/L vs. <36 pmol/L), were 1.26 (95%CI, 0.6-2.6), 1.08 (95%CI, 0.6-2.1), and 2.36 (95%CI, 1.0-4.1), respectively. In general, the associations of individual metabolic syndrome components with colorectal cancer incidence were stronger in men than in women (Fig. 1).
Table 3. Incidence, Relative Risk, and 95% Confidence Interval for Colorectal Cancer by Categories and Number of Metabolic Syndrome Components Measured at Baseline: The Atherosclerosis Risk in Communities Study, 1987 to 2000
RR from the Cox proportional hazards regression model was adjusted for age and gender.
Analyses was adjusted for family history of colorectal cancer, physical activity, nonsteroidal antiinflammatory drug use, aspirin use, pack-years of cigarette use, and grams of alcohol per week. Also adjusted for hormone-replacement therapy in women.
Cut-off points were based on the National Heart, Lung, and Blood Institute definition of the metabolic syndrome (see Grundy et al., 200516).
Low: < 102 cm in men and < 88 cm in women; high: ≥ 102 cm in men and ≥ 88 cm in women.
Although effect modification by gender and the metabolic syndrome was not statistically significant (Pinteraction = .54), subgroup analyses (Table 3) suggested that associations between the metabolic syndrome and colorectal cancer were stronger in men and weaker in women. For the analyses shown in Table 3, blood pressure was based on blood pressure level regardless of hypertension medication. When individuals who had high blood pressure controlled by medication were reclassified as meeting the blood-pressure criterion, the association with colorectal cancer was attenuated: The multivariate adjusted RRs (95% CIs) of colorectal cancer for the metabolic syndrome at baseline (≥3 vs. 0 components) were 1.25 (95% CI, 0.8-2.0) overall and 1.61 (95% CI, 0.8-3.2) in men; and, for ≥3 versus <3 components, the multivariate adjusted RRs (95% CIs) were 1.28 (95% CI, 0.9-1.7) overall and 1.35 (95% CI, 0.9-2.0) in men; associations were not observed in women.
We examined the associations between colorectal cancer and the metabolic syndrome with hyperglycemia set at >110 mg/dL from the original ATP III definition.16 In these analyses, the prevalence of metabolic syndrome was 26.4% in men and 27.6% in women.29 The multivariate-adjusted RRs (95% CIs) of colorectal cancer for the metabolic syndrome (≥3 components vs. 0 components) were 1.49 (95% CI, 0.95-2.4) overall and 2.20 (95% CI, 1.1-4.3) in men, and the multivariate-adjusted RRs (95% CIs) for ≥3 components versus <3 components were 1.31 (95% CI, 0.9-1.8) overall and 1.45 (95% CI, 0.9-2.2) in men; associations were not observed in women.
In this prospective study of African Americans and Caucasians, baseline metabolic syndrome was associated positively with incident colorectal cancer; the association was stronger when the metabolic syndrome (≥3 components) was compared with 0 components versus comparisons with <3 components. There was a significant dose-response relation with the number of components of the metabolic syndrome present. Associations were stronger for men than for women. Individual components of the metabolic syndrome were associated weakly with colorectal cancer, suggesting not only that the sum of the metabolic syndrome may represent a milieu that promotes colorectal cancer but also that any tumor-promoting effects may be enhanced by the presence of more components of the metabolic syndrome.
Our findings are consistent with 2 prospective cohort studies that have examined associations between the insulin-resistance syndrome and colorectal cancer mortality.19, 20 Trevisan et al.19 examined the insulin-resistance syndrome (the highest 25% of the study-specific and gender-specific distributions for glucose and triglycerides; the lowest 25% of the gender-specific distribution for HDL cholesterol, systolic pressure ≥140 mmHg, or diastolic pressure ≥90 mmHg; using medication for diabetes, hyperlipidemia, or hypertension) in a series of pooled Italian cohorts based on 54 colorectal cancer deaths (13 in women). Those authors reported hazard ratios and 95% CIs of 2.99 (95% CI, 1.27-7.02) for men and women, 2.96 (95% CI, 1.05-8.31) for men, and 2.71 (95% CI, 0.59-12.51) for women. Colangelo et al.20 examined 317 colorectal cancer deaths (126 in women) in the Chicago Heart Association Detection Project in Industry Study and reported that the insulin-resistance syndrome (≥3 of 4 risk factors in the upper quartile of gender-specific distributions: 1-hour postload glucose, systolic blood pressure, BMI, and heart rate) was associated with a 1.5-fold increased risk overall (95% CI, 1.03-2.19), a 1.67-fold increased risk in men (95% CI, 1.04-2.70), and 1.29-fold increased risk in women (95% CI, 0.70-2.37) compared with participants who had no components.20 By contrast, in a case-control study nested within the Physicians' Health Study, Ma et al.18 did not observe an association between colorectal cancer incidence and 4 factors of the insulin-resistance syndrome (BMI >25 kg/m2; HDL <40 mg/mL; systolic blood pressure >130 mmHg, diastolic blood pressure >85 mmHg, or use of antihypertensive medication; and triglyceride levels in the upper 20% of the distribution for the control group). Although the findings of those 4 studies have not always been consistent, collectively, they suggest that the metabolic syndrome or related components may increase the risk of colorectal cancer incidence and/or mortality.
There is an extensive body of literature examining the associations between Type II diabetes, which commonly is associated with the metabolic syndrome, and colorectal cancer.11 Although the findings have varied, in prospective cohort studies of general populations, diabetes has carried RRs of colorectal cancer ranging from 1.2 to 2.8.11, 30, 31 Epidemiologic studies have reported that colorectal cancer incidence is associated positively with C-peptide levels (a marker of insulin release),18, 32 2-hour postchallenge insulin and glucose concentrations,3 and hemoglobin A1c levels.33, 34 In one study, there was no association across quartiles of fasting insulin; however, greater than median fasting insulin values (14-400 IU/mL) were associated with 1.6-fold increased risk of colorectal cancer compared with below-median values (3-13 IU/mL).3 In the ARIC Study, there was a weak, positive association between incident colorectal cancer and fasting insulin level; the association was somewhat stronger in women than in men. Overall, the data suggest an association between insulin resistance (and subsequent metabolic abnormalities, e.g., hyperinsulinemia) and colorectal cancer.
Associations between colorectal cancer and individual metabolic syndrome components, other than obesity, have not been studied extensively. Hypertension may increase cancer risk by blocking and subsequently modifying apoptosis, thereby affecting cell turnover.35 Similar to the ARIC results, Colangalo et al.20 and Trevisan et al.19 reported modest increases in the risk of colorectal cancer mortality, from 1.14-fold to 1.36-fold, over quartiles of systolic blood pressure; associations were stronger in men and weaker in women. Hypertriglyceridemia, which is a marker of insulin resistance, has been hypothesized as an underlying cancer risk factor12 by providing fuel for growing cancerous cells.36 Epidemiologic studies have shown either no association34 or a significantly increased risk of colorectal tumors with elevated triglycerides19, 36, 37 or with low HDL cholesterol.38 In the ARIC Study, low HDL was associated with a nonsignificant 20% increased risk of colorectal cancer, and the triglyceride association with colorectal cancer was nearly null. By contrast, Trevisan et al. reported that low HDL cholesterol was associated with a modestly decreased risk of colorectal cancer mortality.19
Prospective epidemiologic studies generally have reported that obesity, which often is associated with hyperinsulinemia,7 is associated positively with colorectal cancer. The data are more conclusive in men than in women and in younger women than in older women.39 In the ARIC Study, colorectal cancer increased 1.5-fold to 2-fold across quartiles of BMI, WHR, and waist circumference for men but not for women. Because >50% of women in the ARIC Study meet the metabolic syndrome criteria of high waist circumference,29 it has poor discriminatory power in this cohort. Our finding that the metabolic syndrome and its components are weaker risk factors for colorectal cancer in women than in men is consistent with the literature. How the metabolic syndrome relates to insulin and glucose by gender requires further study. The metabolic syndrome may serve as a surrogate marker for other risk factors. There may be a complex interaction between obesity, insulin, and gender hormones that increases colorectal cancer risk in men but not in women.39 It is possible that the metabolic syndrome is not a risk factor for colorectal cancer in women. It also is possible that we did not control adequately for confounding factors or did not measure protective factors adequately.
It is plausible biologically that hyperinsulinemia and the metabolic syndrome cause colorectal cancer. Insulin directly stimulates cell growth and DNA synthesis in a dose-dependent manner in normal intestinal epithelial cells and in colon cancer cells.8 Through its receptor, insulin may promote cellular mitogenic stimulation, facilitate the transition of cells through the G1-phase of the cell cycle, and delay apoptosis9 in normal and malignant colon tissue.10 In addition, hyperinsulinemia may decrease hepatic synthesis of insulin-like growth factor (IGF)-binding proteins 1 and 2, translating to increased levels of bioavailable IGF-I,40 which is associated with cellular growth and proliferation, is antiapoptotic, and is associated with several cancers of hormonal etiology.41 Insulin and the IGF-axis are connected further: The IGF-I tyrosine kinase receptor has 60% homology to the insulin receptor; receptor expression levels vary by tissue and are regulated by steroid hormones and growth factors.41 Although the metabolic syndrome and hyperinsulinemia may alter cancer risk through several, still unknown pathways, other mechanisms include possible associations between metabolic syndrome and chronic inflammation16 and between diabetes and altered bowel transit time.31
If the metabolic syndrome or its components are causes of colorectal cancer, then there are potential implications for prevention. Obesity and the metabolic syndrome may be prevented and/or reversed with lifestyle modification, such as diet and exercise. For example, a 20-week, supervised (nonrandomized) aerobic exercise intervention reversed the metabolic syndrome in 18 of 55 men and in 14 of 50 women.42 Although associations between physical activity and colorectal cancer were weak in the ARIC Study, a majority of studies have reported that physical activity is associated negatively with colorectal cancer incidence.43
Strengths of the current study are that the sample included 4 populations in distinct areas of the United States, metabolic syndrome components were carefully measured, and follow-up was relatively complete. We were powered at 80% to detect a relative risk of colorectal cancer of 1.5 with the metabolic syndrome (yes or no); however, power for subgroup analyses was limited. We examined the potential of diagnostic suspicion bias and did not observe a difference in baseline screening for colorectal cancer by gender or by metabolic syndrome.
In summary, the metabolic syndrome had a modest, positive association with colorectal cancer incidence in the ARIC cohort among men, but not among women; there was a dose response according to the number of components present. Evidence is growing that the metabolic syndrome may be a marker for a physiologic milieu of growth that encourages tumor initiation, promotion, and/or progression.
We thank the staff and participants of the Atherosclerosis Risk in Communities Study for their important contributions.