Introduction
Cancer represents a major challenge for modern medicine. A relatively rare disease (or group of diseases) until the late 19th century, it has evolved into a much-feared modern scourge. Historically, prevention has been the preferred approach of medicine to disease eradication. Several infectious diseases serve as paradigms of a unique medical triumph since Edward Jenner’s epic experiment with smallpox vaccination in 1796, when he inoculated the arm of James Phipps with material from a pustule from Sarah Nelmes, his fellow villager with cowpox. To place cancer prevention in its appropriate perspective, one has to consider the various ways, real or conceptual, in which this disease can be approached. If we take colon cancer as an example, we can list the following options in ascending order of complexity:

1) elimination of all causative factors,
2) prevention with a single vaccination that is devoid of side effects,
3) a simple, effective, safe and inexpensive treatment; a clinical analogy would be the present-day treatment of streptococcal pharyngitis with a brief course of antibiotics,
4) chemoprevention,
5) prevention by detection and treatment during its premalignant or early malignant stages, and
6) surgery and/or chemotherapy.

Of these six options, the first three do not even exist in a realistic sense, the fifth is employed with excellent results but at significant financial cost and inconvenience to the patient, while the last one is disappointing for advanced colon cancer. This is the reason why the fourth option, chemoprevention, has emerged as a realistic possibility that can impact colon cancer morbidity and mortality. In this paper I present an overview of the concept of cancer prevention, summarize the various approaches to chemoprevention against colon cancer and discuss some recent developments that could revolutionize our approach to one of the deadliest cancers in the western world. Cancer prevention is classified as primary, secondary and tertiary. Primary prevention is concerned with preventable etiological factors of cancer. Cessation of smoking is an excellent example of an effective intervention for the primary prevention of lung cancer. In secondary prevention premalignant conditions are identified and treated in subjects at risk. Screening colonoscopy with removal of benign colonic polyps (the premalignant stage in colon carcinogenesis) represents such an intervention. In tertiary prevention or chemoprevention a naturally occurring or pharmacological agent is administered to individuals at risk to prevent the development or the recurrence of cancer. In the last two decades there has been an extensive effort to develop effective strategies for the chemoprevention of colon cancer.

Tumor biology - Biomarkers
The two pillars on which our efforts to prevent colon cancer rest are the biology of colon cancer and biomarkers. There has been significant progress in the first area. Although all the answers are not in yet, it is now clear that colon carcinogenesis is a long process requiring years (sometimes decades) between the initial cellular change and the development of metastatic disease that kills the host. The “adenoma-carcinoma sequence”, i.e., colon cancer is preceded by an adenoma stage, is now standard knowledge. It is also clear that colon carcinogenesis is characterized by a succession of molecular changes involving basic cellular processes such as cell proliferation, cell signaling and DNA integrity. There is also a broad understanding that there is in place an important interplay between genes, diet and drugs. It is, however, poorly understood what shifts the balance between these three “forces” so that a colonic cell fails to maintain its normal phenotype. Such knowledge could be crucial, as it would point out the first step towards the primary prevention of colon cancer. Since colon cancer is a lengthy process, it became apparent that for progress to be made within a practical time frame, prevention studies should not use the development of colon cancer as an end point. Consequently, it was reasoned that intermediate biomarkers of colon cancer should be used instead. In a broad sense a biomarker is the property or analyte one assays to monitor relevant tumor biology. A surrogate end point biomarker or intermediate end point is a property that correlates with cancer incidence and/or indicates likely progression to cancer. Colon polyps, for example, have been used successfully as biomarkers in several intervention trials. Aberrant crypt foci, a constellation of morphologically abnormal crypts, have been recently recognized to precede the development of adenoma. The evidence from animal and some human studies that they may be an excellent biomarker of colon cancer, if not yet conclusive is quite persuasive. Additional markers, derived from the study of colon cell kinetics such as proliferation and apoptosis or related to tumor biology such as mutations of genes, e.g. APC, have been sought assiduously in the last decade but none are formalized to date.

Chemoprevention agents
There has been intense activity to identify agents, natural or pharmaceutical, which could prevent colon cancer. Table 1, by no means exhaustive, lists representative groups of agents that have been studied. Although conceptually attractive, fiber seems not to withstand the scrutiny of recent studies. However, animal model assays have demonstrated that the protective effects of dietary fiber on colon cancer development depend on the nature and source of the fiber. Wheat bran appears to inhibit colon tumorigenesis more consistently than do oat bran or corn bran .The so-called lipid fraction of wheat bran has strong colon tumor inhibitor properties, but its biologically active constituents are not known. Of the naturally occurring agents, curcumin, for centuries an innocuous component of Asian diets holds significant promise. Curcumin and related compounds affect colon cell kinetics (1) and have worked well against colon cancer in animal models of colon cancer.

The era of NSAIDs
Kune and colleagues reported a seminal observation in 1988 (2). While investigating factors that may modulate the incidence of colon cancer in Melbourne Australia, they recognized that the use of non-steroidal antiinflammatory drugs (NSAIDs) was associated with a reduction by about half in both the incidence of and mortality from colon cancer. Confirmed by almost two dozen studies involving over a million subjects, this observation has spurred a major effort to exploit NSAIDs as chemopreventive agents against colon cancer. These epidemiological studies have provided the “human relevance” to a plethora of animal and clinical studies on the role of various NSAIDs in colon cancer. In 1973, working at Brandeis University, Lawrence Levine made the first observation that a NSAID modulates colon cancer (3). This talented investigator showed that indomethacin reduced by about half the size of fibrosarcomas in mice. In 1980 Pollard et al reported more extensive work demonstrating that indomethacin prevented colon cancer in an animal model (4). Numerous studies have elaborated these observations and documented conclusively that various NSAIDs, notably including aspirin, protect against colon cancer when administered during its initiation or even its post-initiation stages. At about the same time, Waddell observed that indomethacin regressed completely a mediastinal desmoid tumor in a patient with familial adenomatous polyposis (FAP); indomethacin was administered to control the pain of pericarditis caused by the tumor impinging on the pericardium (5-7). Sulindac, used on other patients as a safer substitute of indomethacin, was also effective. A second important observation by Waddell and his colleagues was that sulindac also regressed colonic polyps in FAP patients. These observations, made with exemplary clinical acumen, have led to a series of formal studies that have demonstrated two remarkable properties of sulindac in regards to FAP: it can prevent the development of polyps and can also regress already established polyps (8). The extent to which some of the many NSAIDs can modulate colon carcinogenesis is currently the subject of several intervention studies. Several laboratories around the world including my own have studied how a small molecule like aspirin, the prototypical NSAID, can affect drastically the complex process of colon carcinogenesis (9, 10). It is evident that NSAIDs affect several pathways, all leading to a reduction in tumor cell mass. After all, any tumor is in essence a mass of cells, which exist in violation of the homeostatic mechanisms that maintain the exquisite balance between cell renewal and cell death. Thus NSAIDs: inhibit cell proliferation (directly and via inhibition of prostaglandin synthesis); induce apoptosis (directly and to a small degree via an effect on COX); inhibit carcinogen activation; and perhaps enhance immune surveillance (via an effect on HLA antigen expression). Several other mechanisms have been described and the next few years are expected to clarify which of these putative pathways are really critical for the remarkable effect of NSAIDs on colon cancer. Whether there is any redundancy in the effects of NSAIDs on the multiple tributaries to colon carcinogenesis remains uncertain. Having evaluated in detail the effect of NSAIDs on COX, we have proposed a model postulating that NSAIDs inhibit more than one pathway, each capable by itself of blocking colon carcinogenesis (11). Such redundancy ensures the high rate of success of NSAIDs and may, in fact, be required given the phenotypic variation among the cells of a colon neoplasm. This model was proposed out of the need to reconcile a host of antithetic, if not downright conflicting, observations on the mechanism of action of NSAIDs on colonocyte kinetics. Since COX is the best-known target of NSAIDs, it was generally assumed that COX inhibition accounted for their effects. Because of some incongruous observations, we pursued this question systematically. We identified a colon cancer cell line that lacked the expression of both COX-1 and COX-2 isozymes and showed that NSAIDs brought about the same antiproliferative and proapoptotic effect on these cells that we had first demonstrated a year earlier on cells expressing COX(12). Subsequent studies by several groups have confirmed our observation.

The need for better agents
For a chemopreventive agent to be clinically useful, it must meet several criteria. At a minimum, it must be: effective, devoid of significant side effects, inexpensive and convenient to administer. The criterion of safety assumes particular significance for a chemopreventive agent. Chemoprevention is quite different from chemotherapy, when even significant toxicity is acceptable in order to save the life of a patient with an already developed cancer. In chemoprevention, an individual at risk for colon cancer will receive for many years, if not for the rest of his/her life, a compound for a cancer that he/she may never develop. NSAIDs have a wide spectrum of side effects, some of them dose-dependent and some life threatening. Indeed, in 1998 as many people died in the US from NSAID-induced gastrointestinal complications as from AIDS. Since low-dose aspirin, as used for cardiac prophylaxis, seems not to protect against colon cancer, there has been an apparent need for safer agents (13). An additional consideration is that NSAIDs confer an about 50% reduction of colon cancer incidence and mortality. Clearly, higher rates of efficacy will be desirable for a disease that claims about 56,000 lives annually in the US alone, where the direct costs for the care of patients with colorectal cancer were $5.2 billion in 1998.

COX-2 inhibitors, NO-NSAIDs and other agents
The severity and range of NSAID side effects combined with the discovery of COX-2, which unlike COX-1 is not constitutively expressed, propelled the discovery of selective COX-2 inhibitors to serve as safer alternatives to traditional NSAIDs. Indeed, the initial expectation seems to have been fulfilled, at least partially, and COX-2 inhibitors enjoy wide clinical application. It terms of colon cancer prevention, the verdict on COX-2 inhibitors is not out yet. Studies in animal models of colon cancer showed a stunning >90% inhibition of colon cancer development by such compounds. A thorough study by Steinbach and colleagues demonstrated that in FAP patients celecoxib, a COX-2 inhibitor, achieved modest results in clinically safe doses: the number and size of colon polyps were each reduced by about one third (14). Perhaps this reflects the need to inhibit more that one of the pathways that operate during colon carcinogenesis. The results of ongoing clinical trials evaluating the efficacy of COX-2 inhibitors against sporadic colon cancer are eagerly awaited. A parallel development has been the synthesis and development of NO-releasing NSAIDs (NO-NSAIDs) by P. del Soldato(15). These compounds consist of a traditional NSAID to which is bound covalently a nitric oxide releasing moiety. Thus, they combine the effects of NSAIDs with the properties of NO, a pivotal molecule not only in cardiovascular physiology but also in other systems including the colon. Furthermore, early work indicates a superior safety profile compared to the corresponding traditional NSAIDs. Our studies, using a cell culture system, indicate that NO-NSAIDs are up to several thousand-fold more effective than their parent compounds in favorably altering colon cancer cell kinetics (16). Our ongoing animal studies seem to confirm these results, but we have to await their completion. Whether NO-NSAIDs will meet all criteria required of an effective chemopreventive agent against colon cancer awaits further studies and certainly human clinical trials. Other, less developed efforts in this area include the use of trefoil peptides, zwitterionic NSAIDs and enantiomers of some NSAIDs. All aim to overcome the (significant) toxicity of traditional NSAIDs.

The Future
Chemoprevention of colon cancer appears to be a very realistic possibility. In fact, it will be surprising if the intense efforts currently going on, which are building on about 30 years of studies on cancer biology, do not culminate to a viable and practical scheme for its prevention. At this stage, there is enough background knowledge to formulate mechanism-driven approaches. In addition, validation of newer biomarkers should simplify clinical testing, a requisite step in the process of assessing candidate compounds. We should, of course, anticipate that once such agents are identified societal issues such as allocation of resources and identification of optimal target groups will have to be addressed. Parallel, but no less important, issues will be the impact of such information on social aspects of the individuals receiving chemoprevention treatment as, for example, their insurability and employment prospects. In contemplating the future of chemoprevention, it is instructive to recall the initial course of cervical cancer screening, introduced by G. N. Papanicolaou around the middle of last century. Eventually accepted, it is now the single most effective cancer prevention method in history, having saved untold millions of lives. Given the current advances in the field of chemoprevention, I believe we are justified in being optimistic about the eventual eradication of colon cancer. Translated by Interpres sas

Basil Rigas
Professor of Medicine
New York Medical College - Valhalla (NY)