A recent randomized controlled trial [2] involving 46 551 participants, showed that screening with fecal occult blood testing can lead to decreases in the number of deaths from colorectal cancer. A 33% cumulative decrease was observed at 13 years in the group having annual screening when compared with the control group. However, a large percentage of the screened group had at least one colonoscopy during the follow-up, thereby raising the possibility that colonoscopy, rather than fecal occult blood screening, was the effective intervention. Sigmoidoscopy and colonoscopy, with removal of precursor adenomas, decrease the incidence of colorectal cancer and its mortality [3, 4]. Nevertheless, widespread use of endoscopy remains impractical and expensive; future advances depend on developing methods for targeted screening in identifiable high-risk persons [5].

Primary prevention of this disease is another promising approach. Fairly consistent epidemiologic data [6, 7] suggest links between diet—those high in fat and low in fiber, certain vegetables, and fruit—and an increased incidence of colorectal cancer. Dietary strategies are being tested in ongoing clinical trials of either modified diets or specific nutritional interventions with agents such as calcium, antioxidant vitamins, or fiber supplements. Because of the considerable logistic and practical problems associated with trials that use cancer as the end point, the primary end point for most of these studies is enumeration of adenomatous polyps, the probable precursors of invasive malignancy. Although this is a reasonable approach, this compromise does have theoretical limitations.

Chemoprevention, another approach to primary prevention, involves the use of specific pharmacologic agents to inhibit carcinogenesis. Nonsteroidal anti-inflammatory drugs (NSAIDs), including sulindac, indomethacin, piroxicam, and aspirin, are a group of agents that have shown great promise in recent years. In this issue, Giovannucci and colleagues [8] report data suggesting that regular aspirin use may decrease the incidence of disease and death from colorectal cancer. In animal models, NSAIDs inhibit cancers induced by chemical carcinogens [9, 10]. Although their common property of inhibiting cyclooxygenase activity may be the underlying mechanism, this hypothesis remains to be proven.

Interest in human applications was enhanced by the demonstration in the 1980s of the ability of sulindac, given in the usual therapeutic doses of 150 to 400 mg/d, to produce regression of colorectal polyps in familial polyposis syndromes [11]. This effect has now been confirmed in a randomized, placebo-controlled, double-blind study by Giardiello and colleagues [12].

However, NSAIDs, used in the usual therapeutic doses, are associated with several different and serious organ system toxicities. Although published incidence rates differ, the yearly estimate of the risk for NSAID-induced serious gastrointestinal complications reported to the Food and Drug Administration is 2% to 4% per year [13]. In addition to upper gastrointestinal bleeding and ulceration, intestinal complications (such as membranous rings, strictures, and colitis) have been described. Nephrotoxicity induced by NSAIDs is also well known, comprising a wide spectrum that includes the nephritis syndromes, acute tubular necrosis, and renal failure [14]. A simple cost–benefit analysis suggests that the accumulating rate of serious toxicity outstrips the benefit from colorectal cancer prevention rapidly, generally within 5 to 6 years, particularly in elderly persons in whom NSAIDs would most likely be used [15].

The side effects of NSAIDs are dose related; "low-dose" regimens, such as those equal to 325 mg or less of aspirin daily, are considerably safer. Aspirin is now recommended for primary and secondary prevention of thromboembolic conditions in a broad range of atherosclerotic vascular diseases. It is also recommended for men and women aged 50 years or older because of the substantial prevalence of silent coronary disease after the age of 50 years [16]. The possibility that the same strategy may inhibit development of colorectal cancer is indeed an exciting one. It is for this reason that aspirin, especially in low doses, is particularly appealing.

Since the late 1980s, several large case–control or cohort studies have found statistically and clinically significant decreases of 30% to 50% in the incidence of colorectal cancer among aspirin users [17-22]. Because these studies were not specifically designed as intervention trials, precise data on dosage and duration of use were not generally available. Nevertheless, many participants were consuming low-to-modest doses, compatible with those used for preventing cardiovascular disease.

Giovannucci and colleagues [8] now report their findings from the Health Professionals Follow-up Study, showing that even after controlling for several potential confounding factors, a lower risk for the incidence of total and advanced (metastatic and fatal) colorectal cancer was related to regular aspirin use ( 2 times per week). Data collected in biannual questionnaires administered to 51 529 health professionals (from 1986 through 1992) showed that the lowest risks were found in persons who consistently used aspirin, many of whom, by 1990, had been taking aspirin for at least 10 years. A decrease in the number of colorectal adenomas was observed in a subset of 10 521 men who had endoscopy for reasons other than bleeding. The large data set allowed the authors to correct for numerous potentially confounding factors, including diet, alcohol intake, body mass, physical activity, smoking, and family history. Interestingly, data from a brief supplementary questionnaire (from a randomly selected sample) indicated that low-to-moderate aspirin doses were used rather than the high doses conventionally used for treating pain and arthritis.

Can we conclude that aspirin prevents colorectal cancer? Most epidemiologic evidence supports this association. The only exception is a cohort study [23] done in California, with elderly persons of a median age considerably older than those in the trials that suggested an association. Nevertheless, epidemiologic data cannot by themselves provide conclusive proof. In the only reported randomized, controlled trial (the Physicians Health Study), Gann and colleagues [24] did not find a statistically significant decrease in colorectal cancer or polyp incidence in physicians randomly assigned to receive one aspirin pill (325 mg) every other day for a 4.7-year follow-up. The authors acknowledge the substantial potential limitations of this study, such as a short duration of follow-up and the small number of patients. Given the current hypotheses about colon cancer development, a considerably longer follow-up is likely to be necessary to show an effect. However, it is virtually impossible to do trials of very long duration with large numbers of patients because of logistical difficulties. Further, low-dose aspirin is also recommended for precisely the same group of patients for preventing cardiovascular disease; this raises ethical concerns related to a placebo-controlled design of long duration. Shorter trials using surrogate end points such as adenoma recurrence are more practical and are in progress.

Unlike many other putative chemopreventive agents, the tremendous appeal of aspirin, particularly low-dose aspirin, as a viable option for widespread use makes it imperative that further research continue. In addition to efficacy, better information is needed about duration of use and, especially, dosage. It would be most convenient if optimal dosage was similar to what is recommended for preventing cardiovascular disease. With recent developments proving the efficacy of screening, ongoing advances in the identification of high-risk groups, and the possibility of risk modification by diet and chemoprevention with aspirin, the prospect for decreasing mortality from colorectal cancer by preventive approaches seems bright.