Statistical strategies to improve the efficiency of molecular studies of colorectal cancer prognosis.

The evaluation of tumour molecular markers may be beneficial in prognosis and predictive in therapy. We develop a stopping rule approach to assist in the efficient utilisation of resources and samples involved in such evaluations. This approach has application in determining whether a specific molecular marker has sufficient variability to yield meaningful results after the evaluation of molecular markers in the first n patients in a study of sample size N (n

The benefit of combination therapy depends on disease phenotype and duration in Crohn's disease.

BACKGROUND: The impact of combination therapy on disease-related morbidity in patients with established Crohn's disease (CD) or ulcerative colitis (UC) remains to be well-defined. AIM: To examine the effect of combination therapy on disease outcomes in CD and UC. METHODS: Using a multicenter prospective cohort, we classified CD and UC patients as being on monotherapy with anti-TNF or on combination with an immunomodulator. The primary outcome was a composite of new IBD-related surgery, hospitalisations, penetrating complications, need for corticosteroids or new biological at 1 year. Multivariable regression models adjusted for potential confounders. RESULTS: We included 707 patients with CD (45% combination therapy) and 164 with UC (38% combination therapy). Combination therapy was not associated with reduction in the composite outcome in either CD (OR: 0.87, 95% CI: 0.63-1.22) or UC (OR: 1.45, 95% CI: 0.63-3.38). However, while no difference was noted in those with nonstricturing, nonpenetrating CD, a significant reduction in the likelihood of the outcome was seen in those with stricturing or penetrating CD (30% vs 39%, OR: 0.58, 95% CI: 0.37-0.90). A stronger effect was also observed in those with disease duration <5 years (OR: 0.35, 95% CI: 0.14-0.87) compared to those with a longer duration (OR: 0.75, 95% CI: 0.45-1.27). A similar reduction in occurrence of composite outcome was noted with infliximab and with other anti-TNF biologics. CONCLUSION: The benefit of combination immunomodulator-biological therapy is stronger in those with complicated Crohn's disease, particularly early on in their disease course.

Cigarette smoking and the risk of colorectal cancer in women.

Colorectal cancer incidence rates for smokers, nonsmokers living with smokers (i.e., passive smokers), and nonsmokers in smoke-free households were compared in a 12-year prospective study of 25,369 women who participated in a private census conducted in Washington County, MD, in 1963. Women who smoked had a decreased relative risk of colorectal cancer compared with the risk for nonsmokers (age-adjusted relative risk, 0.76; 95% confidence interval, 0.52-1.10). The risk for passive smokers was similar to that for smokers. The relative risks were significantly reduced for older women; relative risks were 0.42 for smokers and 0.66 for passive smokers over age 65. The data suggest that older women who smoke have a lower risk of colorectal cancer than nonsmokers. The effect may be mediated by an antiestrogenic effect of smoking.

Diet and risk of colorectal adenomas: macronutrients, cholesterol, and fiber.

BACKGROUND: Diet is thought to be important in the etiology of colorectal cancer. Studies suggest that a diet high in red meat and saturated fat and low in dietary fiber and vegetables may increase cancer risk. Diet may also be important in the development of colorectal adenomas that are precursors of most colorectal cancers, but this hypothesis has not been well studied. PURPOSE: This case-control study was designed to examine the effects of dietary consumption of cholesterol, fiber (vegetables, fruits, beans, and grains), and macronutrients (protein, carbohydrate, and fat) on risk for colorectal adenomas. METHODS: Analyses were based on data from 236 subjects (105 men and 131 women) with histologically confirmed adenomas (cases) and 409 adenoma-free control subjects (165 men and 244 women), all of whom had had colonoscopy. Case and control subjects were similar with respect to gender, body mass, race, marital status, education, and indications for colonoscopy. Using a validated quantitative food-frequency questionnaire, an experienced graduate nutritionist interviewed each subject by telephone. Sex-specific analyses were conducted because the ranges of nutrient intake were substantially different for men and women. Odds ratios (ORs) were calculated according to quintiles of nutrient intake. RESULTS: Carbohydrate intake was inversely related to adenoma risk in women (P for trend = .002). Compared with women in the lowest quintile of carbohydrate consumption, those in the highest quintile were 60% less likely to develop adenomas (OR = 0.39; 95% confidence interval [CI] = 0.19-0.80). Intake of fruit (P for trend = .028) and intake of fiber derived from vegetables and fruits (P for trend = .012) were also inversely related to adenomas in women. Total fat showed a positive association in women (P for trend = .004), with an OR of 2.69 for the highest versus the lowest quintile (95% CI = 1.31-5.50). Results were comparable for saturated fat (P for trend = .027). The risks in men were generally similar in direction and magnitude but were not statistically significant. CONCLUSIONS: These data support the hypothesis that a diet high in fat and low in carbohydrates, fruits, and fruit and vegetable fiber increases risk not only for colorectal cancer but also for precursor colorectal adenomas. IMPLICATIONS: These results, which are consistent with findings of other investigators, suggest that environmental factors, influencing risk for colorectal cancer, such as a high-risk diet, may lead to development of the precursor adenomas.

Folate intake, alcohol consumption, cigarette smoking, and risk of colorectal adenomas.

BACKGROUND: Recent evidence suggests that folic acid (and derivatives) could contribute to the protective effect of fruits and vegetables against the risk of large-bowel cancer. Other evidence indicates that alcohol drinking and cigarette smoking may impair the biologic actions of folate. We used data from an adenoma prevention trial to investigate the occurrence of colorectal adenomas (possible precursors of colorectal cancer) in association with folate intake, alcohol consumption, and cigarette smoking. METHODS: Patients with at least one recent large-bowel adenoma were followed with colonoscopy 1 year and 4 years after their qualifying colon examinations. Adenomas detected after the year 1 examination were used as end points. A food-frequency questionnaire was administered at study entry and at study completion; nutrient intake at study entry was used in this analysis. All statistical tests were two-sided. RESULTS: After adjustment for caloric intake, dietary folate had a significant protective association with the risk of recurrence of large-bowel adenoma (P for trend = .04). However, this inverse association was attenuated by further adjustment for intake of dietary fiber and fat. Use of folate supplements was not associated with a reduction in risk. Alcohol intake (seven or more drinks/week) was associated with increased risk (odds ratio = 2.04; 95% confidence interval = 1.28-3.26). Cigarette smoking, even smoking for long duration, was not related to adenoma recurrence. CONCLUSIONS: These data provide only modest support for previous findings suggesting beneficial effects of folate on colorectal adenoma risk. We find no evidence that cigarette smoking increases risk. These findings do suggest a substantial increase in risk with alcohol consumption.

Histamine receptor antagonists and incident colorectal adenomas.

BACKGROUND: Prior studies suggest that histamines may modulate the development of colorectal neoplasia. AIM: To assess whether histamine receptor antagonist use was associated with adenoma formation. METHODS: Patients (n = 2366) were drawn from three adenoma chemoprevention trials. All underwent baseline colonoscopy with removal of adenoma(s) and were deemed free of remaining lesions; they were followed with surveillance colonoscopy. Medication use was assessed by questionnaire. Adjusted risk ratios for adenoma formation related to histamine receptor antagonist use (histamine H1 and H2 receptor, H1RA and H2RA) were determined using log linear models. RESULTS: In pooled analyses, H1RA exposure was not associated with subsequent adenoma risk (RR = 1.10; 95% CI 0.97-1.25) or multiple adenoma formation (RR = 0.85; 95% CI 0.67-1.07). H2RA use also was not associated with adenoma (RR = 0.90; 95% CI 0.77-1.06), or multiple adenoma (RR = 0.77; 95% CI 0.57-1.04) in the pooled analyses, but H2RA users in the first trial had a decreased risk of adenoma (RR = 0.70; 95% CI 0.48-1.03) and multiple adenoma (RR = 0.31; 95% CI 0.12-0.79). CONCLUSION: H2RA use was associated with reduced risk for adenoma in one trial, but not in the pooled analyses. Further study would be warranted before undertaking randomized trials of H2RAs for adenoma chemoprevention.

Nephrosclerosis postpartum and in women taking oral contraceptives.

The condition of a patient with postpartum nephrosclerosis improved during heparin therapy. Review of the literature disclosed 29 other patients with the same histopathologic characteristics, eight of whom also recovered substantial renal function after anticoagulation therapy. Also reported is a patient in whom renal failure occurred while she was taking oral anovulatory agents. Renal biopsy specimen showed the same histopathologic features, which raises the question of similar factors mediating the expression of this disease. We suggest a uniform terminology for this syndrome, either postpartum nephrosclerosis or nephrosclerosis in women taking oral contraceptives.

A critical appraisal of ticlopidine, a new antiplatelet agent. Effectiveness and clinical indications for prophylaxis of atherosclerotic events.

BACKGROUND: Recommendations are broadening for the prophylaxis of atherosclerotic disorders, but aspirin is the only widely used agent. Ticlopidine hydrochloride, a new antiplatelet medication, has recently been approved for prescription in North America. We reviewed the major clinical trials of ticlopidine and derived guidelines for its use. METHODS: Studies of ticlopidine were sought through MEDLINE for 1980 to 1990 and through bibliographies of retrieved articles. All published, randomized trials of ticlopidine were appraised if they reported major morbidity and mortality as primary end points. All eligible studies were formally reviewed by an expert panel according to published principles for critical appraisal of the medical literature. Both benefits and risks were quantified. RESULTS: Four randomized trials reported major clinical end points. In these, ticlopidine was more effective than placebo for preventing recurrences after completed stroke; was more effective than aspirin for patients with transient ischemic attacks and partial strokes; and reduced vascular death and nonfatal myocardial infarction in an open trial among patients with unstable angina. For patients with intermittent claudication ticlopidine, was not significantly better than placebo for preventing myocardial infarction or stroke. Side effects were more common with ticlopidine than with aspirin or placebo. CONCLUSIONS: Ticlopidine should be prescribed in place of aspirin for stroke prophylaxis or unstable angina if the patient is unable to tolerate aspirin. Ticlopidine may also benefit patients who experience new ischemic events while taking aspirin or, probably, patients with peripheral vascular disease. A complete blood cell count should be performed every 2 weeks during the first 3 months of therapy to check for leukopenia.

Reproductive history, use of exogenous hormones, and risk of colorectal adenomas.

Evidence suggests a role for reproductive and hormonal factors in the etiology of colorectal cancer. Investigation of adenomas, the presumed precursors of cancer, and reproductive characteristics may place such associations within a particular stage of carcinogenesis. We examined parity, age at first birth, age at menopause, and age at menarche as well as contraceptive and noncontraceptive hormone use in a case-control study of 347 women (115 cases with adenomas and 232 controls) conducted in North Carolina. Using unconditional logistic regression analysis, increasing age at menopause was found to be associated with a reduction in the risk of adenomas [odds ratio (OR), 0.26; 95% confidence interval (CI), 0.10-0.68]. An increased risk for adenomas was found for women undergoing surgical menopause as compared with women undergoing natural menopause (OR, 2.10; 95% CI, 1.05-4.21). Our results suggest a reduced risk of adenomas associated with noncontraceptive hormone use that was limited to a subgroup of women with natural menopause or bilateral oophorectomy (OR, 0.39; 95% CI, 0.15-0.97). No associations were seen between other reproductive characteristics and adenomas. These results suggest protective effects for both endogenous and exogenous female hormones that operate early in the process of carcinogenesis. Alternatively, lifestyle factors or other correlates of exogenous hormone use and delayed menopause could play a role in reduced adenoma risk.

Disposable versus reusable biopsy forceps for colorectal epithelial cell proliferation in humans.

The performance of various measures of rectal mucosal proliferation has been evaluated in the literature, but the performance of the forceps used to obtain the tissue has received little attention. We used data from two large studies of proliferation at a single institution to compare reusable and disposable endoscopic forceps. Endoscopic pinch biopsies were taken 10 cm from the anal verge using either reusable or disposable, oval-cupped, sheathed forceps. The specimens were fixed, embedded, and sectioned, taking care to orient the specimens longitudinally. Five sections were placed on each slide. We determined how many slides did not contain eight scorable crypts (inadequate) and how many sections were necessary to identify eight complete crypts. There were 395 subjects who had biopsies taken with reusable forceps and 185 subjects who had biopsies taken with disposable forceps. The specimens were inadequate in 27.6% of the reusable forceps specimens versus 2.7% of the disposable forceps (P < 0.0001). The mean number of tissue sections necessary to identify eight scorable crypts for the reusable forceps was 3.82 (SD, 0.87) compared with 3.17 (SD, 0.83) for disposable forceps (P = 0.0001). The specimens taken with the disposable forceps were better, probably because the forceps were sharper. We believe that the better quality of the specimens and the sterility justify the higher cost of disposable forceps. We would urge investigators in proliferation studies to evaluate the biopsy equipment as carefully as they evaluate other aspects of their methods.

Coffee and tea and the risk of recurrent colorectal adenomas.

Consumption of coffee has been associated with a reduction in the risk of cancer of the colon, and (less consistently) drinking tea has been associated with a reduction in the risk of rectal cancer. The effect of these beverages on the risk of colorectal adenomas, however, has not been well investigated. We used data from an adenoma prevention trial to investigate these associations. Patients with at least one recent large bowel adenoma were followed with colonoscopy 1 and 4 years after their qualifying examinations. Adenomas detected at the year 4 colonoscopy were used as end points. A food frequency questionnaire was administered at study entry and study completion; average intake over the study period was used to estimate the exposures of interest. There was no apparent association between the intake of regular coffee, decaffeinated coffee, or tea and the risk of recurrent colorectal adenomas. The relative risks and 95% confidence intervals per cup daily were 0.96 (0.87-1.05) for regular coffee, 0.97 (0.84-1.12) for decaffeinated coffee, and 1.02 (0.83-1.25) for tea. These negative findings were present both overall and for adenomas of the right and left large bowel.

Rectal mucosal proliferation and risk of colorectal adenomas: results from a randomized controlled trial.

Although rectal mucosal labeling index is thought to be a useful surrogate biomarker for colorectal cancer, the ability of the index to predict future neoplasia is unclear. We obtained rectal mucosal biopsies from 333 participants of a randomized controlled chemoprevention trial of calcium supplementation to determine whether labeling index was correlated with concurrent or future colorectal neoplasms. Labeling index was measured using proliferating cell nuclear antigen immunohistochemistry. Adenomas were enumerated at the time of the biopsies (cross-sectional) and 3 years later (prospective). We used logistic regression to test for an association of adenoma occurrence with overall labeling index, the mean proliferative height, and labeling index in the upper 40% of colon crypts. In the cross-sectional analysis, we found indications that higher proliferation was associated with an increase in the prevalence of adenomas. The overall adjusted odds ratios (OR) (95% confidence interval) were 1.14 (0.90-1.45) per % crypt labeling index, OR 1.08 (0.99-1.19) for upper crypt proliferation, and OR 1.07 (1.03-1.12) for proliferative height. In contrast, individuals with higher labeling index at baseline were actually less likely to have adenomas in the prospective analyses: OR 0.80 (0.62-1.02) per % crypt labeling index, OR 0.86 (0.73-1.00) for upper crypt index, and OR 0.97 (0.93-1.01) for proliferative height. Proliferative index does not predict future colorectal neoplasia, although it may be weakly associated with the presence of current adenomas. These results have important implications for the design of future intervention studies. Although it may be attractive to include the measurement of intermediate markers in large controlled trials, until we have more confidence in their performance, we should rely on better proven and more reliable intermediates, such as adenomas.

Iron intake and the risk of colorectal cancer.

Prospectively gathered data from the National Health and Nutrition Examination Survey I and the National Health Evaluation Follow-Up Study were analyzed to evaluate the risk of colorectal cancer due to consumption of iron. Morbidity and mortality data due to colorectal cancer were available on 14,407 persons first interviewed in 1971 and followed through 1986. A total of 194 possible colorectal cancers occurred in this group over the 15-year period. Subsite analysis showed that the risk of colon cancer due to iron intake was elevated throughout the colon for both men and women, with the highest adjusted risks for the interquartile range seen in the proximal colon for females (relative risk, 1.51; 95% confidence interval, 1.41-1.60). The risk of rectal cancer was not significantly elevated for men or women. Elevated serum iron was also associated with increased risk; however, this effect was strongest in the distal (rather than proximal) colon and was significant only among females (adjusted relative risk, 1.73; 95% confidence interval, 1.03-2.92). The mean transferrin saturation was higher among cases than controls (30.7 versus 28.7%), but total iron-binding capacity did not seem to predict the occurrence of colorectal cancer. Proportional hazards models confirmed that the effects of iron and serum iron were not confounded by age, gender, energy consumption, fat intake, or other known risk factors for colorectal cancer. These data suggest that iron may confer an increased risk for colorectal cancer, and that the localization of risk may be attributable to the mode of epithelial exposure. It seems that luminal exposure to iron increases risk proximally, whereas humoral exposure increases risk distally. These differences may be due to such factors as oxidation state, binding proteins and the presence of other cofactors such as bile acids, products of bacterial metabolism.

Rectal mucosal proliferation, dietary factors, and the risk of colorectal adenomas.

Colorectal cancer arises from a series of precursor stages, the so called adenoma-carcinoma sequence. Increased rectal mucosal proliferation may be an early step in this sequence. Because dietary factors are implicated in the etiology of colorectal cancer, one might predict that diet would also be associated with proliferation. We conducted this study to examine the association of diet with rectal mucosal proliferation. Rectal mucosal proliferation was measured in endoscopic biopsy specimens by proliferating cell nuclear antigen (PCNA) immunohistochemistry and whole crypt mitotic counts (WCMCs). Diet was evaluated using a validated quantitative food frequency questionnaire. The correlation between PCNA labeling index (LI) and WCMCs was determined using Kendall's tau, a nonparametric measure of correlation. Logistic regression was used to examine the effect of proliferation on adenoma status, controlling for confounders. The relationship between proliferation and dietary and demographic factors was examined using linear regression. There were 308 patients who had one or both measures of proliferation. There was no significant correlation between PCNA LI and WCMCs (Kendall's tau = 0.04; P = 0.35). Neither measure of proliferation was predictive of adenoma status, even after adjusting for potential confounders. Body mass index and calories per day were significant predictors of WCMC (P = 0.01 and P = 0.03, respectively). PCNA labeling index was not associated with any dietary variables, although its association with dietary fat nearly reached statistical significance (P = 0.09). The association between proliferation and diet were generally inconsistent. There appears to be no simple relationship between colorectal cancer risk factors, colorectal adenomas, and these two measures of rectal mucosal proliferation. We need simpler, more reliable intermediate markers for use in etiological and intervention studies.

Reproducibility and variability of the rectal mucosal proliferation index using proliferating cell nuclear antigen immunohistochemistry.

Rectal mucosal proliferation has been shown to be increased in patients with neoplastic lesions of the large bowel and may serve as a marker of risk for colorectal malignancy. We conducted analyses to determine reliability and components of variability that might suggest optimal analysis strategies for studies of proliferation. Endoscopic pinch biopsies were obtained from 17 adult patients, labeled using proliferating cell nuclear antigen, scored using strict rules, and then rescored. Labeling index, defined as the proportion of labeled cells in a crypt, was calculated for each crypt, biopsy, subject, and group. There was excellent reproducibility. The technician was able to select previously scored crypts 95% of the time. The overall labeling index was identical on repeat. There was considerable variability in labeling index among crypts from a single biopsy and between biopsies of a single subject. Variance component estimates suggested that 20% of the variability of labeling index was due to subject, 30% due to the biopsy within a subject, and 50% due to crypts within a biopsy. There were substantial gains in statistical power by scoring two biopsies rather than one. There was less gain from further increases in biopsy number. There was little statistical advantage for counting more than 8 crypts/biopsy. Demonstrating a decrease of 25% in the mean labeling index with 90% power could require more than 100 subjects/group. We conclude that proliferating cell nuclear antigen is an extremely reproducible method to determine proliferation index. There is considerable variability among subjects, biopsies, and crypts.(ABSTRACT TRUNCATED AT 250 WORDS)

Serum ferritin concentration and recurrence of colorectal adenoma.

Both body iron stores and dietary iron intake have been reported to increase risk of colorectal neoplasms. We assessed whether serum ferritin concentration was associated with recurrence of colorectal adenomas among 733 individuals with baseline determinations of ferritin as part of a multicenter clinical trial of antioxidant supplements for adenoma prevention. All study participants had at least one adenoma removed within 3 months before enrollment, and 269 of them developed one or more adenomas between follow-up colonoscopies conducted 1 and 4 years after enrollment. Baseline serum ferritin concentrations were analyzed both as a log-transformed continuous variable and as a categorical variable, defined as whether iron stores were nonreplete and low (ferritin < or =30 microg/liter), nonreplete and borderline (31-70 microg/liter), replete and adequate (71-160 microg/liter), or replete and high (>160 microg/liter). Analyses were based on multiple logistic regression models, including age, sex, study center, energy, alcohol, fiber, folate, and total fat intake, number of months between colonoscopic examinations, smoking status, and aspirin use. Overall, there was no statistically significant linear association between log ferritin concentration and adenoma recurrence (P = 0.33). Risk of adenoma recurrence was modestly increased among participants with ferritin concentrations >70 microg/liter relative to those with lower ferritin (odds ratio, 1.39; 95% confidence interval, 0.96-2.02). This result seemed more pronounced among women than men. Dietary intake of iron and red meat was inversely associated with adenoma recurrence among participants with replete iron stores but not consistently associated among those with nonreplete stores. Our findings suggest that any role of iron stores and dietary iron in influencing risk of colorectal adenoma recurrence is likely complex.

Correlation between biomarkers of omega-3 fatty acid consumption and questionnaire data in African American and Caucasian United States males with and without prostatic carcinoma.

Results from animal studies suggest that omega-3 fatty acids from marine sources are protective against cancer. To determine whether adipose tissue and erythrocyte membrane fatty acid composition could serve as biomarkers of essential fatty acid consumption in subjects with prostate cancer, we compared fish consumption, which was estimated using a food frequency survey, to the omega-3 fatty acid content of adipose tissue and erythrocyte membranes. The study was conducted using 127 men who had undergone a prostate biopsy. All subjects were recruited from a university hospital urology clinic. African Americans comprised 23% of the subjects, and 70% were diagnosed with prostate cancer. We found a correlation of 0.44 with 95% confidence intervals (CIs) = 0.29-0.57 between reported fish consumption and the omega-3 fatty acid eicosapentaenoic acid composition in erythrocyte membranes and 0.38 with 95% CI = 0.21-0.53 when the dietary survey was compared to eicosapentaenoic acid in adipose tissue. The survey/biomarker correlations in cases were not significantly different from the correlations in controls. The study had 90% power to detect a 0.35 difference between correlations. These results suggest that the presence of prostate cancer does not affect the adipose tissue or erythrocyte membrane biomarkers of fatty acid consumption, and that erythrocyte membranes are as useful as biomarkers as is adipose tissue. Our findings corroborate previous studies that found that tissue biomarkers can reflect past fatty acid consumption and support the use of biomarkers in case-control studies using cancer patients.

Biomarkers of essential fatty acid consumption and risk of prostatic carcinoma.

Animal studies have suggested that omega-6 fatty acids found in vegetable oils may promote prostate cancer. Our goal was to use erythrocyte membrane and adipose tissue fatty acid composition as biomarkers to investigate whether essential fatty acids modulated prostate cancer risk. An outpatient clinic-based study of 89 cases and 38 controls was conducted in North Carolina between July 1989 and December 1991. Cases were recruited from a university-based urology outpatient clinic. Eligible cases were more than 45 years of age and had histological confirmation of a prostate cancer diagnosis within 1 year of entry into the study. Controls were histologically confirmed free of prostate cancer. Erythrocyte membranes from venous blood and adipose tissue fatty acids from s.c. fat samples were analyzed in batches using capillary gas chromatography. Unconditional logistic regression analysis was used to calculate odds ratios for the association of each fatty acid with prostate cancer while controlling for potential confounders. Linoleic acid consumption was positively associated with prostate cancer risk. The odds ratios comparing the first and fourth quartiles of linoleic acid consumption were 3.54 (95% confidence interval, 1.0-12.53) with P trend < 0.04 for erythrocyte membranes, and 2.47 (95% confidence interval, 0.66-9.26) with P trend < 0.08 for adipose tissue. These data suggest that linoleic acid consumption may increase prostate cancer risk, which is consistent with results from animal experiments. Linoleic acid is found in vegetable oils used in cooking and in cereals, snack foods, and baked goods. Our data failed to demonstrate consistently a protective effect of marine omega-3 fatty acids on prostate cancer.

Dietary iron and recurrence of colorectal adenomas.

Previous research suggests that iron acts as a prooxidant to increase the risk of colorectal neoplasia. This study examined effects of dietary intake of iron on colorectal adenoma recurrence using data from an antioxidant clinical trial. All subjects were free of polyps at study entry but had at least one adenoma removed within the 3 months before enrollment. Follow-up colonoscopies were conducted after 1 and 4 years. Patients who developed one or more adenomatous polyps between years 1 and 4 were classified as cases; all others were controls. Dietary iron intake at baseline and at the end of the study was estimated from self-administered food frequency questionnaires and averaged together for each subject, energy-adjusted, and categorized into quartiles. Odds ratios were adjusted for age, center, sex, calories, treatment group, and alcohol, fiber, folate, and fat intakes in unconditional logistic regression analysis. Dietary iron was inversely associated with adenoma risk, although risk did not decrease monotonically with increasing intake. Odds ratios comparing second, third, and fourth quartiles to the lowest quartile were 0.61 [95% confidence interval (CI), 0.37-1.02], 0.80 (95% CI, 0.45-1.44), and 0.37 (95% CI, 0.19-0.73), respectively. A limited examination showed no clear evidence that use of iron supplements affected risk of recurrence in this study population. This study provides evidence against the hypothesis that recent dietary intake of iron increases risk for colorectal adenomas. However, these results may reflect the presence of other dietary factors found in combination with iron.