Preclinical, observational, and clinical data consistently show that non-steroidal anti-inflammatory drugs (NSAIDs)—particularly aspirin—reduce colorectal carcinogenesis.¹ Scores of animal studies show that NSAIDs inhibit the development of colorectal neoplasia across the spectrum of disease, ranging from aberrant crypt foci (ACF) to cancer.² Human data confirm these findings with dozens of observational studies reporting 40–50% reductions in colorectal adenoma incidence, cancer incidence, and cancer associated mortality among aspirin users. The most compelling data were published earlier this year. Two randomised placebo controlled trials conducted in patients at moderate risk for colorectal cancer reported that aspirin administered at doses as low as 81–325 mg/day reduced the development of adenomas by up to 35% after a few years of use.^3,4 In one trial, greater effects were seen against more advanced lesions.³

Colorectal adenomas are established as common non-obligate precursors of colorectal cancer.⁵ Within the last decade, ACF have been identified in rodent models of carcinogenesis, and have been proposed as precursors of colorectal adenomas and cancers.⁶ With the use of recently developed high resolution/magnifying endoscopes, researchers are now making quantitative (for example, number, size, crypt multiplicity) and qualitative (for example, morphology) real time in vivo assessments of ACF in humans.⁷ ACF—or at least a subset of them—may represent important risk markers for adenoma-carcinoma development in humans. They may also serve as markers of chemopreventive response. Thus although relatively little is known about ACF and their relevance to more advanced stages of colorectal neoplasia in humans, they provide an important and promising focus for additional research.

In this issue of Gut, Shpitz and colleagues⁸ weave together these two investigational threads [see page 1598]. This group describes an association between aspirin use and reduced ACF prevalence and histopathological distribution in ex vivo specimens obtained from 194 patients with colorectal cancer. Among patients who used aspirin for at least one year, they observed a 47% reduction in specimens harbouring ACF, a 64–82% reduction in ACF per cm² of colorectal mucosa, and a 52% reduction in dysplastic ACF. Although ACF reductions were observed in all anatomical sites, the reductions tended to be more dramatic in the distal colorectum. These findings suggest that aspirin taken at 100 mg/day over at least one year may exert chemopreventive effects against the early stages of colorectal carcinogenesis.

While intriguing, these data must be interpreted cautiously. Firstly, the study groups differed with regard to variables that may influence baseline risks for colorectal neoplasia and/or aspirin use, such as gender (males 84% v 52% in the aspirin and control groups, respectively) and age (aspirin users were much more homogeneous than controls). The investigators do not address the potential impact of these imbalances on the study results, nor did they adjust for them in the analysis. Because we have scant information about ACF, these variables may have confounded interpretations of the effects of aspirin. In addition, the investigators did not report study participants’ dietary habits and their use of concomitant medications. Preclinical studies suggest that the latter two exposures may modulate ACF and therefore these limitations may be particularly important.^9–12 Secondly, without power estimates, it is impossible to know whether the lack of statistical significance for certain variables is true or merely reflects the small sample size. Finally, the selection of the study cohort is not described in detail; therefore, the generalisability of these study results is uncertain.

Despite these limitations, the preliminary findings of Shpitz and colleagues⁸ are stimulating and should prompt additional investigations into whether ACF reductions correlate with or predict aspirin’s preventive efficacy against more advanced stages of colorectal neoplasia. These data contribute to a growing body of research suggesting that ACF might be used to identify the preventive efficacy of investigational agents against colorectal carcinogenesis. Takayama et al originally reported marked reductions in in vivo colorectal ACF following 12 months of treatment with sulindac.⁷ Both the Shpitz and Takayama studies show that NSAIDs reduce ACF burden after relatively brief exposures. The data of Shpitz and colleagues⁸ add another link to the investigational chain by suggesting that aspirin exerts greater effects against more advanced (that is, dysplastic) ACF. Evidence that aspirin modulates both early and advanced ACF would represent a major advance for the field of chemoprevention research. Needless to say, ultimate validation requires linking NSAID induced reductions in ACF to reductions in colorectal adenomas, cancer, and/or cancer associated mortality.¹³ Nevertheless, this study moves us closer to a distant but still plausible goal of firmly establishing ACF as meaningful short term markers for cancer prevention research.

The study provides other important insights as well. For example, the effects of aspirin on early colorectal neoplasia appear to be relative, not absolute. While aspirin may have significantly reduced the burden of ACF, all 59 aspirin responsive patients included in the study still developed colorectal cancer. Were ACF reduced to the same extent—or perhaps to a greater extent—among aspirin using patients who did not develop cancer? Do reductions in ACF predict for reductions in adenoma, cancer, or cancer mortality, as suggested by preclinical and observational data? How might we best use or build upon the data generated by Shpitz et al? This provocative study leaves a trail of important questions, the answers to which may pave the way for future successes in chemoprevention research.

Shpitz’s study⁸ also advances chemoprevention research through its novel investigational design, by using pathological specimens originally obtained to diagnose or treat cancer, allowing for parallel insights into earlier stages of neoplasia. More than 75% of controls had ACF, confirming how common these preinvasive lesions are in patients with colorectal cancer. If this situation is typical, ACF may provide another measure by which the preventive efficacy of agents might be quickly, albeit preliminarily, assessed.

For example, ACF evaluations nested within trials testing drug or dietary interventions—including those intended for other indications—might accelerate agent development and prioritisation for cancer prevention.¹⁴ This approach could be readily implemented to efficiently test some of the most commonly used drugs or nutritional supplements in Western populations, such as statins, PPAR agonists, NSAIDs, and fibre, all of which may have anticancer properties.^9–12,15 With the exception of NSAIDs, these have yet to be tested in prospective, randomised, controlled trials. Clearly, observational data derived from ancillary endoscopic assessments in single arm or randomised controlled trials might improve the efficiency and speed of agent identification and testing.

Aspirin has been used in various forms to treat pain and inflammation since as early as 500 BC. The array of indications for aspirin has expanded to include the prevention of myocardial infarction and stroke, clinching its role as the world’s first wonder drug.¹⁶ Aspirin now appears to be effective against colorectal neoplasia; why and how we still do not know. New insights into carcinogenesis, such as those provided by Shpitz et al, will profoundly alter our expectations for aspirin and its potential to improve the public’s health.