Reviewers: Kenneth Blank, MD
Source: The New England Journal of the Medicine, January 21, 1999, Vol. 340 No. 3 p.169
As we enter the final year of the 20th century, a look back reveals amazing strides in medicine: the advent of antibiotics, vaccination, the safe use of general anesthesia, the effective treatment of many cancers with radiotherapy and chemotherapy. Unfortunately, similar advances have yet to be made into understanding the causes of cancer. In 1971 Burkitt and other epidemiologists found the risk of colorectal cancer to be reduced in countries with diets high in fiber. Many theories were postulated to explain this inverse relationship from the ability of fiber to act as a colonic broom and push carcinogens quickly out of the gastrointestinal tract to their ability to alter the colonic milieu thereby deactivating carcinogens.
Case control and cohort studies examining the role of dietary fiber's effect on colorectal cancer have been inconclusive. A meta-analysis of case-control studies found a significant decrease in colon cancer when comparing those in the highest and lowest quintile of dietary fiber intake. Researchers at Harvard University examined this issue in the largest prospective study to date, the results of which appear in the January 21, 1999 issue of The New England Journal of Medicine.
In 1976, over 120,000 nurses who were between 30 and 55 years of age began participation in the Nurses' Health Study. The study participants completed questionnaires every two years on risk factors for cancer and coronary heart disease. Among the items in this questionnaire was the food frequency survey that originally included 61 items and was expanded to 136 items in 1986. For each food listed, participants provided information on how often they consumed a specified portion of food per year. In addition, the brand of breakfast cereal was recorded. The main contributors to dietary fiber intake were bread, cereal, apples, bananas and green vegetables.
On each questionnaire, women reported whether colorectal cancer had been diagnosed. For each report of colorectal cancer, a study physician reviewed the medical records and recorded pathologic and anatomical data. Colorectal cancer in-situ and cancers other than adenocarcinomas were excluded.
The incidence of colorectal adenomas was investigated as well. In total, 25,530 women who reported undergoing a sigmoidoscopy or colonoscopy, and all reports of colon polyps were investigated by study physicians with respect to pathology and anatomical location. Only polyps of the distal colon and rectum were studied because the majority of patients underwent a sigmoidoscopy, not a full colonoscopy.
A total of 787 cases of colorectal cancer were reported in the sixteen-year follow-up period. Women were divided into five groups based on the amount of dietary fiber ingested (measured in grams/day). The quintile with the lowest intake averaged 9.8 grams per day compared to 24.9 grams per day in the highest group. Each group contained roughly 17,500 women.
Multivariate analysis did not find dietary fiber intake to be associated with colorectal cancer. Even when controlling for age, established risk factors and total energy, no association was found. In addition, no evidence between dietary fiber intake and colorectal cancer was found in any subgroup defined by age, family history of colorectal cancer, aspirin use, physical activity, body-mass index, cigarette smoking, total fat intake or alcohol consumption.
The epidemiological evidence which first hinted at a relationship between fiber intake and colon cancer risk has not been supported by prospective case control and cohort studies. This study is among the largest and most comprehensive prospective trials to examine this issue and the results clearly fail to demonstrate an association between the amount of fiber in one's diet and the risk of colorectal cancer in women.