Excessive alcohol use

Based on solid evidence from observational studies, excessive alcohol use is associated with an increased risk of colorectal cancer (CRC).[1,2]

Magnitude of Effect: A pooled analysis of eight cohort studies estimated an adjusted relative risk (RR) of 1.41 (95% confidence interval [CI], 1.16–1.72) for consumption exceeding 45 g/day.[1]

• Study Design: Cohort studies.
• Internal Validity: Fair.
• Consistency: Fair.
• External Validity: Fair.

Cigarette smoking

Based on solid evidence, cigarette smoking is associated with increased incidence and mortality from CRC.

Magnitude of Effect: A pooled analysis of 106 observational studies estimated an adjusted RR (current smokers vs. never smokers) for developing CRC of 1.18 (95% CI, 1.11–1.25).[3,4]

• Study Design: 106 observational studies.
• Internal Validity: Fair.
• Consistency: Good.
• External Validity: Good.

Obesity

Based on solid evidence, obesity is associated with increased incidence and mortality from CRC.

Magnitude of Effect: In one large cohort study, the adjusted RR for developing colon cancer for women with a body mass index of more than 29 was 1.45 (95% CI, 1.02–2.07).[5,6] A similar increase in CRC mortality was found in another large cohort study.[7,8]

• Study Design: Large cohort studies.
• Internal Validity: Fair.
• Consistency: Good.
• External Validity: Good.

Physical activity

Based on solid evidence, regular physical activity is associated with a decreased incidence of CRC.

Magnitude of Effect: A meta-analysis of 52 observational studies found a statistically significant 24% reduction in CRC incidence (RR, 0.76; 95% CI, 0.72–0.81).[9]

• Study Design: Cohort studies and meta-analysis.
• Internal Validity: Fair.
• Consistency: Good.
• External Validity: Good.

Aspirin

Hormone therapy (estrogen plus progestin)

Based on solid evidence, combined hormone therapy (conjugated equine estrogen and progestin) decreases the incidence of invasive CRC.[16]

Based on fair evidence, combination conjugated equine estrogen and progestin has little or no benefit in reducing mortality from CRC. Data from the Women’s Health Initiative (WHI), a randomized, placebo-controlled trial evaluating estrogen plus progestin, with a mean intervention of 5.6 years and a follow-up of 11.6 years showed that women taking combined hormone therapy had a statistically significant higher stage of cancer (regional and distant) at diagnosis but not a statistically significant number of deaths from CRC compared with women taking the placebo.[16]

Magnitude of Effect: There were fewer CRCs in the combined hormone therapy group than in the placebo group (0.12% vs. 0.16%; HR, 0.72; 95% CI, 0.56–0.94). A meta-analysis of cohort studies observed a RR of 0.86 (95% CI, 0.76–0.97) for incidence of CRC associated with combined hormone therapy.

There were 37 CRC deaths in the combined hormone therapy arm compared with 27 deaths in the placebo arm (0.04% vs. 0.03%; HR, 1.29; 95% CI, 0.78–2.11).

• Study Design: RCT and cohort studies
• Internal Validity: Good.
• Consistency: Good for effect on incidence; not applicable (N/A) for effect on mortality; results were based on one trial.
• External Validity: Good.

Polyp removal

Nonsteroidal anti-inflammatory drugs (NSAIDs)

Calcium supplementation

The evidence is inadequate to determine whether calcium supplementation reduces the risk of CRC.

Dietary factors

There is no reliable evidence that a diet started in adulthood that is low in fat and meat and high in fiber, fruits, and vegetables reduces the risk of CRC by a clinically important degree.

Estrogen-only therapy

Statins

Incidence and mortality

Colorectal cancer (CRC) is the third most common malignant neoplasm worldwide [1] and the second leading cause of cancer deaths in men and women in the United States.[2] It is estimated that there will be 132,700 new cases diagnosed in the United States in 2015 and 49,700 deaths due to this disease.[2] Between 2007 and 2011, CRC incidence rates in the United States declined by 4.3% per year among adults aged 50 years and older.[2] For the past 20 years, the mortality rate has been declining in both men and women. Between 2007 and 2011, the mortality rate declined by 2.5% per year. In adults younger than 50 years, CRC incidence rates increased by about 1.8% per year.[2] Incidence and mortality rates are higher in African Americans compared with other races.[3]

The overall 5-year survival rate is 65%. About 5% of Americans are expected to develop CRC within their lifetimes.[2,4] The risk of CRC begins to increase after the age of 40 years and rises sharply at ages 50 to 55 years; the risk doubles with each succeeding decade, and continues to rise exponentially. Despite advances in surgical techniques and adjuvant therapy, there has been only a modest improvement in survival for patients who present with advanced neoplasms.[5,6] Hence, effective primary and secondary preventive approaches must be developed to reduce the morbidity and mortality from CRC.

Definition of prevention

Primary prevention involves the use of medications or other interventions before the clinical appearance of CRC with the intent of preventing clinical CRC and CRC mortality.

Etiology and pathogenesis of colorectal cancer

Genetics,[7,8] experimental,[9,10] and epidemiologic [11-13] studies suggest that CRC results from complex interactions between inherited susceptibility and environmental factors. The exact nature and contribution of these factors to CRC incidence and mortality is the subject of ongoing research.

Excessive alcohol use

There is evidence of an association of CRC with alcoholic beverage consumption. In a meta-analysis of eight cohort studies, the relative risk (RR) for consumption of 45 g/day (i.e., about three standard drinks per day) compared with nondrinkers was 1.41 (95% confidence interval [CI], 1.16–1.72).[14] Case-control studies suggest a modest-to-strong positive relationship between alcohol consumption and large bowel cancers.[15,16] A meta-analysis found that the association did not vary by sex or location within the large bowel.[17]

Five studies have reported a positive association between alcohol intake and colorectal adenomas.[18] A case-control study of diet, genetic factors, and the adenoma-carcinoma sequence was conducted in Burgundy.[19] It separated adenomas smaller than 10.0 mm in diameter from larger adenomas. A positive association between current alcohol intake and adenomas was found to be limited to the larger adenomas, suggesting that alcohol intake could act at the promotional phase of the adenoma-carcinoma sequence.[19]

A large cohort study found a dose-response relationship between alcohol intake and death from CRC, with a RR of 1.2 (95% CI, 1.0–1.5) for four or more drinks per day compared with nondrinkers.[20]

Cigarette smoking

Most case-control studies of cigarette exposure and adenomas have found an elevated risk for smokers.[21] In addition, a significantly increased risk of adenoma recurrence following polypectomy has been associated with smoking in both men and women.[21] In the Nurses’ Health Study, the minimum induction period for cancer appears to be at least 35 years.[22] Similarly, in the Health Professionals Follow-up Study, a history of smoking was associated with both small and large adenomas and with a long induction period of at least 35 years for CRC.[23] In the Cancer Prevention Study II (CPS II), a large nationwide cohort study, multivariate-adjusted CRC mortality rates were highest among current smokers, intermediate among former smokers, and lowest in nonsmokers, with increased risk observed after 20 or more years of smoking in men and women combined.[24] On the basis of CPS II data, it was estimated that 12% of CRC deaths in the U.S. population in 1997 were attributable to smoking. A large population-based cohort study of Swedish twins found that heavy smoking of 35 or more years' duration was associated with a nearly threefold increased risk of developing colon cancer, although subsite analysis found a statistically significant effect only for rectal cancer, but not colon cancer.[25] Another large population-based case-control study supports the view that current tobacco use and tobacco use within the last 10 years is associated with colon cancer. A 50% increase in risk was associated with smoking more than a pack a day relative to never smoking.[26] However, a 28-year follow-up of 57,000 Finns showed no association between the development of CRC and baseline smoking status, although there was a 57% to 71% increased risk in persistent smokers.[27] No relationship was found between cigarette smoking, even smoking of long duration, and recurrence of adenomas in a population followed for 4 years after initial colonoscopy.[28]

A meta-analysis of 106 observational studies found a RR (ever smokers compared with nonsmokers) for CRC incidence of 1.18 (95% CI, 1.11–1.25), with an absolute risk increase of 10.8 cases per 100,000 person-years (95% CI, 7.9–13.6). There was a statistically significant dose-response effect. In 17 studies with data on CRC mortality, cigarette smoking was associated with CRC death, with a RR (ever smokers vs. never smokers) of 1.25 (95% CI, 1.14–1.37), and an absolute increase in the death rate of 6.0 deaths per 100,000 person-years. For both incidence and mortality, the association was stronger for rectal cancer than for colon cancer.[29]

Obesity

At least three large cohort studies have found an association between obesity and CRC incidence or mortality.[30-32] The Nurses’ Health Study found that women with a body mass index (BMI) of more than 29, compared with women with a BMI of less than 21, had an adjusted RR for CRC incidence of 1.45 (95% CI,1.02–2.07).[30] In the CPS II [32], men and women with a BMI of 30 to 34.9 had an adjusted RR for CRC mortality (compared with people with a BMI of 18.5–24.9) of 1.47 (95% CI, 1.30–1.66), with a statistically significant dose-response effect.[32] The effects were similar in men and women.

Physical activity

A sedentary lifestyle has been associated with an increased risk of CRC in some [33,34] but not all [35] studies. Numerous observational studies have examined the relationship between physical activity and colon cancer risk.[36] Most of these studies have shown an inverse relationship between level of physical activity and colon cancer incidence. The average RR reduction is reportedly 40% to 50%. Large U.S. cohort studies have found statistically significant adjusted RRs of 0.54 (95% CI, 0.33–0.90) [30] and 0.53 (95% CI, 0.32–0.88) [31] when comparing people with high versus low average energy expenditure. A meta-analysis of 52 observational studies found an overall adjusted RR of 0.76 (95% CI, 0.72–0.81), with similar results for men and women.[37]

Aspirin

The preponderance of evidence from both observational studies and long-term follow-up of RCTs indicates that daily acetylsalicylic acid (ASA) for at least 5 years reduces the incidence of CRC. Among a group of more than 600,000 adults enrolled in an American Cancer Society study, mortality in regular users of ASA was about 40% lower for cancers of the colon and rectum.[38,39] In a report from the Health Professionals Follow-up Study of 47,000 males, regular use of ASA (at least 2 times per week) was associated with a 30% overall reduction in CRC, including a 50% reduction in advanced cases.[40] In the Women's Health Study (WHS), a randomized 2 x 2 factorial trial of 100 mg of ASA every other day for an average of 10 years, similar rates of breast, colorectal, or other site-specific cancers were observed in both the ASA and placebo arms.[41] In a report from the Nurses’ Health Study involving 82,911 women followed for 20 years, the multivariate RR for colon cancer was 0.77 (95% CI, 0.67–0.88) among women who regularly used ASA (≥2 standard 325-mg tablets per week) compared with nonregular use. Significant RR was not observed, however, until more than 10 years of use. The benefit appeared to be dose-related (e.g., women who used more than 14 ASA per week for longer than 10 years had a multivariate RR for cancer of 0.47 [95% CI, 0.31–0.71]).

A systematic review of 46 observational studies of ASA and CRC in 2007 found a reduction in CRC (odds ratio [OR] for any use 0.80 [0.73–0.87]).[42] A large cohort study (301,240 people with 3,894 CRC cases) published after this systematic review found an association between weekly or daily ASA use and reduced 10-year incidence of distal and rectal (but not proximal) CRC, with a hazard ratio (HR) of 0.76 (95% CI, 0.64–0.90) for rectal cancer for daily use. However, use was assessed at only one time, and there is no information about dose or duration of use.[43]

In the Physicians’ Health Study, 22,000 men aged 40 to 84 years were randomly assigned to receive placebo or ASA (325 mg every other day) for 5 years. There was no reduction in invasive cancers or adenomas at a median follow-up of 4.5 years.[44] In a subsequent analysis of more than 12 years, both randomized and observational analyses indicated that there was no association between the use of ASA and the incidence of CRC. The low dose of ASA and the short treatment period may account for the null findings.[45]

In a randomized study of 635 patients with prior CRC (T1–T2 N0 M0) who had undergone curative resection, ASA intake at 325 mg/day was associated with a decrease in the adjusted RR of any recurrent adenoma as compared with the placebo group (0.65; 95% CI, 0.46–0.91) after a median duration of treatment of 31 months. The time to detection of a first adenoma was longer in the ASA group than in the placebo group (HR for the detection of a new polyp, 0.54; 95% CI, 0.43–0.94, P = .022). Harms of treatment included upper gastrointestinal hemorrhage and hemorrhagic stroke.[46] In a study of 1,121 patients with a recent history of colorectal adenomas, after a mean duration of treatment of 33 months, the unadjusted RRs of any adenoma (as compared with the placebo group) were 0.81 in the 81-mg ASA group (95% CI, 0.69–0.96) and 0.96 in the 325-mg ASA group (95% CI, 0.81–1.13). For advanced neoplasms (adenomas ≥10.0 mm in diameter or with tubulovillous or villous features, severe dysplasia, or invasive cancer), the RRs were 0.59 (95% CI, 0.38–0.92) in the 81-mg ASA group, and 0.83 (95% CI, 0.55–1.23) in the 325-mg ASA group.[47] Harms of treatment were similar in the two groups and included upper gastrointestinal bleeding and hemorrhagic stroke.

Four reports in 2007, 2010, 2011, and 2012 [42,48-50] have analyzed long-term follow-up of RCTs of daily ASA versus the control. The 2007 analysis reported on two RCTs with reliable follow-up of more than 20 years. This report found that the use of 300 mg or more of ASA per day for at least 5 years reduced the incidence of CRC after a latency of 10 years (RR at 10–19 years, 0.60; 95% CI, 0.42–0.87). The 2010 analysis analyzed long-term follow-up data from four RCTs, finding that allocation to ASA for 5 or more years reduced the 20-year incidence and mortality of proximal colon cancer (adjusted incidence HR, 0.35; 95% CI, 0.20–0.63; adjusted mortality HR, 0.24; 95% CI, 0.11–0.52) and also reduced the 20-year risk of rectal cancer (RR, 0.58; 95% CI, 0.36–0.92) but not distal colon cancer. There was no increase in benefit at doses more than 75 mg/day. The absolute 20-year risk reduction in fatal CRC was 1.76% (95% CI, 0.61–2.91).

The 2011 meta-analysis examined data from eight RCTs, seven of which provided individual patient data and three of which provided 20-year follow-up data. In trials with allocation to ASA of at least 5 years, the 20-year HR for CRC mortality was 0.60 (95% CI, 0.45–0.81). Six RCTs, including five from the United Kingdom, were included in a meta-analysis in which patients were randomly assigned to receive either aspirin or placebo and mean scheduled duration of trial treatment was 4 years or more. Individual patient data for all in-trial cancer deaths were obtained. In the three United Kingdom trials, cancer deaths after completion of the trials were obtained via death certification and cancer registration, taking the follow-up to 20 years after randomization. Based on meta-analysis of ORs from each trial rather than on more sensitive actuarial analysis of the individual patient data, allocation to aspirin in the RCTs reduced the 20-year risk of death due to CRC (and esophageal cancer). ORs for maximum aspirin use were 0.55 for colorectal cancer risk (95% CI, 0.41–0.76) and 0.47 for esophageal cancer risk (95% CI, 0.27–0.81) and for any aspirin use were 0.58 for colorectal cancer risk (95% CI, 0.44–0.78) and 0.51 for esophageal cancer (95% CI, 0.31–0.83).

In a large cohort study, an association between recent daily aspirin use and lower-cancer mortality in the gastrointestinal tract (RR, 0.61; 95% CI, 0.47–0.78), liver (RR, 0.52; 95% CI, 0.30–0.93), and bladder (RR, 0.52; 95% CI, 0.28–0.97) were observed among the 100,139 analysis-eligible participants from the CPS II Nutrition Cohort established by the American Cancer Society in 1982. The analysis excluded participants who had a history of cancer in the baseline year or whose records contained incomplete information on aspirin use or smoking, and was based on follow-up questionnaires mailed to participants in 1997 (the baseline year for the analysis), 1999, 2001, and 2003. Mortality follow-up continued through Dec 31, 2008 via automated linkage to the National Death Index vital status and cause of death codes (ICD-10); death certificates were obtained for 99.3% of known deaths.[51]

The WHS, the largest randomized trial of aspirin to date (N = 39,876), found no reduction in the incidence of colon or other cancers during the 10-year active intervention. However, among women who voluntarily participated in extended follow-up (N = 33,682; 16,913 from the intervention group and 16,769 from the placebo group), there was a significant reduction in CRC incidence (HR, 0.58; 95% CI, 0.42–0.8, P < .001) with a median follow-up of 8 years. Calculated from the beginning of the intervention period through the extended follow-up (median, 18 years) an overall reduction in CRC incidence was observed for the WHS (HR, 0.80; 95% CI, 0.67–0.97; P = .021; intervention period versus extended follow-up, P = .012). During the intervention phase, women were randomly assigned to receive either an annual supply of aspirin (100 mg) or placebo, taken every other day. During the extended follow-up, the intervention was discontinued. Protocol compliance and medical incidences were tracked via identical annual questionnaires throughout the 18-year period. Medical record reviews by a panel of experts blinded to random assignment confirmed the endpoints reported. Whereas previously reported meta-analyses of randomized trials of daily aspirin use demonstrated a reduction in colon cancer incidence with extended follow-up, these findings from the WHS demonstrate a similar effect from aspirin taken every other day.[52]

Hormone therapy (estrogen plus progestin)

Several observational studies have suggested a decreased risk of colon cancer among users of postmenopausal female hormone supplements.[53-56] For rectal cancer, most studies have observed no association or a slightly elevated risk.[57-59]

The Women’s Health Initiative (WHI) trial examined, as a secondary endpoint, the effect of combined estrogen-plus-progestin therapy and estrogen-only therapy on CRC incidence and mortality. Among women in the combined estrogen-plus-progestin group of the WHI, an extended follow-up (mean, 11.6 years) confirmed that fewer CRC were diagnosed in the combined hormone therapy group than in the placebo group (HR, 0.72; 95% CI, 0.56–0.94); the CRCs in women in the combined group were more likely to have lymph node involvement than the CRCs in women in the placebo group (50.5% vs. 28.6%; P < .001) and were classified at higher stages (regional and distant) (68.8% vs. 51.4%; P = .003). The number of CRC deaths in the combined group was higher than in the placebo group (37 vs. 27 deaths), but the difference was not statistically significant (HR, 1 .29; 95% CI, 0.78–2.11).[60]

Polyp removal

An analysis of data from the National Polyp Study (NPS), with external, historical controls, has commonly been cited to show a reduction of 76% to 90% in the subsequent incidence of CRC after colonoscopic polypectomy compared with three nonconcurrent, historical control groups.[61] This study may be biased in several ways that inflate the apparent efficacy of polyp removal; the main problem is that potential enrollees in the NPS were excluded if they had CRC at their baseline examination. Because no such exclusions (or baseline colonoscopy examinations) were done in the three comparison groups, persons who had CRC at baseline would be counted as having incident CRC in subsequent follow-up. Although adjustments were attempted, it is not possible to know the magnitude of the impact of this problem on the result because it is not known how long CRC may be present without causing symptoms.

An additional long-term follow-up study (median follow-up, 15.8 years; maximum, 23 years) of the NPS cohort suggested an approximately 53% reduction in CRC mortality due to polypectomy (not just exclusion of persons with CRC at initial exam). However, the degree of reduction must be viewed with caution because this study did not have a direct comparison group, relying mainly on comparison to expected data from the National Cancer Institute's Surveillance, Epidemiology and End Results Program. Further, details are not clear regarding factors that may have led to decreased mortality. Patients in the NPS were assigned to colonoscopy at years 1 and 3; colonoscopy was also offered to one of the two comparison groups at year 1; all participants were offered colonoscopy at year 6. However, following year 6, the exact surveillance that patients may have undergone and how that surveillance might have been associated with decreased CRC mortality were not well described.[62]

It is expected that further follow-up in the United Kingdom Flexible Sigmoidoscopy Screening Trial will be able to provide more detail about the long-term effect of polypectomy, at least on the left side of the colon.[62]

Other evidence about the benefit of sigmoidoscopy screening (at which time both polyps and early cancer would be removed) suggests that the impact of endoscopic screening, at least on the left side of the colon, is substantial and prolonged. In an RCT, 170,000 persons were randomly assigned to one-time sigmoidoscopy versus usual care. At sigmoidoscopy, polyps were removed, cancer was detected, and patients were referred for treatment. Based on sigmoidoscopy findings, persons were considered to have low risk if they had normal exams or only one or two small (<1 cm) tubular adenomas; such persons were not referred either for colonoscopy workup, or for colonoscopic surveillance. In a follow-up of 10 years, the left-sided CRC incidence in the low-risk group (about 95% of attendees were low risk) was 0.02% to 0.04% per year—a very low risk of CRC compared with average risk. The cause of reduced risk—whether due to detection and removal of large polyps or small ones, or selection of individuals at lower risk—is yet unclear.[63] The natural history of large polyps is not well known, but some evidence suggests that such lesions become clinical CRC at a rate of approximately 1% per year.[64] As a result of the strong data about the impact of endoscopy on the left colon, evidence from multiple studies has raised questions about the ability of endoscopy to reduce CRC mortality in the right colon.[65-67] Thus, it is unclear what the overall impact of endoscopy (e.g., colonoscopy screening) is, and whether there may be a large difference in impact on the left side of the colon compared with the right side.[65]

Other studies suggest that the polyps with the greatest potential to progress to CRC are larger polyps (i.e., >1.0 cm), which include most of those with villous or high-grade histologic features. Retrospective cohort studies also show the harms associated with polypectomy, including bleeding.[68,69]

Nonsteroidal Anti-inflammatory Drugs

One large cohort study (301,240 people with 3,894 CRC cases) found an association between daily or weekly nonaspirin (non-ASA) nonsteroidal anti-inflammatory drug (NSAID) use and reduced 10-year incidence of proximal and distal colon cancer, but not rectal cancer, with an HR of 0.67 (95% CI, 0.58–0.77) for daily use for colon cancer. Because exposure to non-ASA NSAIDs was assessed only once, assessment was by self-report, and there is no information on dose or duration of use, the certainty of this single study must be rated low. Further research is needed before this finding can be accepted.[43]

Although evidence is currently inadequate to determine whether NSAIDs reduce CRC incidence, proponents suggest that any effect of these drugs results from their ability to inhibit the activity of cyclooxygenase (COX). COX is important in the transformation of arachidonic acid into prostanoids, prostaglandins, and thromboxane A2. NSAIDs include not only aspirin (ASA, which is considered separately here) and other, first-generation nonselective inhibitors of the two functional isoforms of COX, termed COX-1 and COX-2, but also newer second-generation drugs that inhibit primarily COX-2. Normally, COX-1 is expressed in most tissues and primarily plays a housekeeping role (e.g., gastrointestinal mucosal protection and platelet aggregation). COX-2 activity is crucial in stress responses and in mediating and propagating the pain and inflammation that are characteristic of arthritis.[70]

Nonselective COX inhibitors include, indomethacin (Indocin); sulindac (Clinoril); piroxicam (Feldene); diflunisal (Dolobid); ibuprofen (Advil, Motrin); ketoprofen (Orudis); naproxen (Naprosyn); and naproxen sodium (Aleve, Anaprox). Selective COX-2 inhibitors include celecoxib (Celebrex), rofecoxib (Vioxx), and valdecoxib (Bextra). Rofecoxib and valdecoxib are no longer marketed because of an associated increased risk of serious cardiovascular events.

Both celecoxib and rofecoxib have been associated with serious cardiovascular events including dose-related death from cardiovascular causes, myocardial infarction, stroke, or heart failure.[71-74] Four trials that demonstrated this increased risk are summarized in the Table 1. In addition, a network meta-analysis of all large scale randomized, controlled trials (RCTs) comparing any NSAID to any other NSAID or placebo found that there is little evidence to suggest that any of the investigated drugs are safe in terms of cardiovascular effects. Naproxen seemed least harmful.[75]

Other major harms from all NSAIDs are gastrointestinal bleeding and renal impairment. The incidence of reported major gastrointestinal bleeding events appears to be dose-related.[76]

Celecoxib reduces the incidence of adenomas; however, celecoxib does not have a clinical role in reducing the risk of sporadic CRC. Its long-term efficacy in preventing CRC has not been shown because of increased risk of cardiovascular events, and because there are other effective ways, such as screening to reduce CRC mortality.[77] A population-based retrospective cohort study of nonaspirin NSAID use among individuals aged 65 years and older was associated with lower risk of CRC, particularly with longer durations of use.[78]

Several rigorous studies have demonstrated the effectiveness of sulindac in reducing the size and number of adenomas in familial polyposis.[79,80] In a randomized, double-blind, placebo-controlled study of 77 patients with familial adenomatous polyposis, patients receiving 400 mg of celecoxib twice a day had a 28.0% reduction in the mean number of colorectal adenomas (P = .003 for the comparison with placebo) and a 30.7% reduction in the polyp burden (sum of polyp diameters; P = .001) as compared with reductions of 4.5% and 4.9%, respectively, in the placebo group. The reductions in the group receiving 100 mg of celecoxib twice a day were 11.9% (P = .33 for the comparison with placebo) and 14.6% (P = .09), respectively. The incidence of adverse events was similar among the groups.[81]

The NSAID piroxicam, at a dose of 20 mg/day, reduced mean rectal prostaglandin concentration by 50% in individuals with a history of adenomas.[82] Several studies assessing the effect of ASA or other nonsteroidals on polyp recurrence following polypectomy are in progress.[83] In several of these studies, mucosal prostaglandin concentration is being measured.

The potential for use of NSAIDs as a primary prevention measure is being studied. There are, however, several unresolved issues that preclude making general recommendations for their use. These include a paucity of knowledge about the proper dose and duration for these agents, and concern about whether the potential preventive benefits such as a reduction in the frequency or intensity of screening or surveillance could counterbalance long-term risks such as gastrointestinal ulceration and hemorrhagic stroke for the average-risk individual.[84]

Calcium Supplements

A randomized placebo-controlled trial tested the effect of calcium supplementation (3 g calcium carbonate daily [1,200 mg elemental calcium]) on the risk of recurrent adenoma.[85] The primary endpoint was the proportion of patients (72% of whom were male) in whom at least one adenoma was detected following a first and/or second follow-up endoscopy. A modest decrease in risk was found for both developing at least one recurrent adenoma (adjusted risk ratio [ARR], 0.81; 95% CI, 0.67–0.99) and in the average number of adenomas (ARR, 0.76; 95% CI, 0.60–0.96). The investigators found the effect of calcium was similar across age, sex, and baseline dietary intake categories of calcium, fat, or fiber. The study was limited to individuals with a recent history of colorectal adenomas and could not determine the effect of calcium on risk of the first adenoma, nor was it large enough or of sufficient duration to examine the risk of invasive CRC. After calcium supplementation is stopped, the lower risk may persist up to 5 years.[86] The results of other ongoing adenoma recurrence studies are awaited with interest. It is important to note that the dose of calcium salt administered may be important; the usual daily doses in trials have ranged from 1,250 to 2,000 mg of calcium.

In a randomized, double-blind, placebo-controlled trial involving 36,282 postmenopausal women, the administration of 500 mg of elemental calcium and 200 IU of vitamin D3 twice daily for an average of 7.0 years was not associated with a reduction in invasive CRC (HR, 1.08; 95% CI, 0.86–1.34; P = .051).[87] The relatively short duration of follow-up, considering the latency period of CRC of 10 to 15 years, and suboptimal doses of calcium and vitamin D, may account for the negative effects of this trial, although other factors may also be responsible.[88]

Dietary Factors

Estrogen-only therapy

The estrogen-only intervention component of the WHI was conducted among women who had a hysterectomy, with CRC incidence included as a secondary trial endpoint. CRC incidence was not decreased among women who had taken estrogens; after a median of 7.1 years of follow-up, 58 invasive cancers occurred in the estrogen arm compared with 53 invasive cancers in the placebo arm (HR, 1.12; 95% CI, 0.77–1.63). Tumor stage and grade were similar in the two groups; deaths after CRC were 34% in the hormone group compared with 30% in the placebo group (HR, 1.34; 95% CI, 0.58–3.19).[112]

Statins

Overall, evidence indicates that statin use neither increases nor decreases the incidence or mortality of CRC. Although some case-control studies have shown a reduction in risk, neither a large cohort study [113] nor a meta-analysis of four RCTs [114] found any effect of statin use.