Cancer chemoprevention is defined as the use of specific chemical compounds to prevent, inhibit, or reverse carcinogenesis. As shown in Figure 1, cancer development in humans requires an average of 20 to 40 years,[1-6] and the scope of chemoprevention encompasses all phases of this process--from healthy subjects at normal risk; to populations at intermediate risk from environmental and lifestyle factors, genetic predisposition, and precancerous lesions; to previous cancer patients at high risk for second primary malignancies.
Chemoprevention is a relatively new medical science. It is only during the past 5 years that results of clinical intervention trials with chemopreventive drugs have begun to appear in the literature. For example, a 1990 seminal trial by Hong and colleagues demonstrated that 13-cis-retinoic acid (iso-retinoin [Accutane]) prevented second primary cancers in patients with previous squamous-cell carcinoma of the head and neck. Second primaries were seen in 24% (12 patients) of the placebo group but in only 4% (2 patients) of the treatment group.
Of course, such clinical trials represent only a fraction of current knowledge about human chemoprevention. Epidemiologic studies, ranging from descriptive assessment of dietary influences to prospective case-control analyses with specific micronutrients and drugs, have provided and continue to furnish leads for definitive intervention trials. For example, the protective effects of aspirin(Drug information on aspirin) in the colon were substantiated by such studies.[8-10]
Chemoprevention is not the same as cancer chemotherapy. These two treatment modalities are compared and contrasted in Table 1. A primary distinction is the timing and duration of intervention. Chemoprevention is applied throughout the long process of carcinogenesis depicted in Figure 1, before invasive disease develops. Chemopreventive treatment is intended to be long term, conceivably up to a lifetime in high-risk subjects. A consequence of this long duration of treatment in relatively healthy subjects is a requirement for agents with very low toxicity. In contrast, chemotherapeutic agents are administered after invasive disease is detected. Many such agents are given to cancer patients for short periods or in discrete cycles under conditions in which side effects are expected and carefully monitored and palliative treatment can be administered to lessen their immediate impact.
A second fundamental distinction relates to the goals of the two treatment modalities relative to cancer. Chemotherapy seeks to increase survival and remission of invasive disease and to prevent metastases, whereas chemoprevention tries to prevent or prolong the time to the onset of cancer. Because of these differences, the developmental paths for chemotherapeu- tic and chemopreventive agents are divergent. In particular, cancer incidence generally is not a feasible end point for the evaluation of chemopreventive drugs because of the long time needed for carcinogenesis and the relatively low incidences of cancers, even in high-risk populations.
Despite the differences between chemoprevention and chemotherapy, there are areas of conceptual and practical overlap between the two modalities. For instance, chemopreventive agents can be used as adjuvant therapies to prevent recurrences or new primary tumors in patients who have already been treated for cancer. This use was defined by Hong et al in the trial with isotretinoin(Drug information on isotretinoin) cited above.
Also, some of the mechanisms of chemopreventive and chemotherapeutic action are similar. Cytotoxicity was the primary mechanism of early chemotherapeutic agents. Many newer chemotherapeutic agents are cytostatic and have mechanisms similar to those of some chemopreventive agents; ie, they slow the growth and progression of dysplastic cells (eg, by inducing terminal differentiation or apoptosis). A significant feature of chemoprevention is that it intervenes at the early stages of carcinogenesis when normal cell order and function are partially preserved. Chemoprevention strategies can be designed to target these preserved pathways before they are lost during the accelerating disorder that ends in the uncontrolled cell growth of cancer.
As suggested above, a significant aspect of chemoprevention is the end points selected for intervention trials. Numerous chemoprevention studies have used well-recognized precancerous lesions as end points. For example, patients with familial adenomatous polyposis (FAP) develop large numbers of colorectal adenomas beginning in their teens and, if untreated, have a 90% chance of progressing to colorectal cancer by 50 years of age. In a small phase II trial in patients with FAP, Giardiello et al showed that 9 months of sulindac(Drug information on sulindac) (150 mg bid) caused a mean decrease in polyp number to 44% of baseline (p = .014) and a mean reduction in polyp diameter to 35% of baseline (P less than .001), although no patient had complete resolution of all polyps. During the 3-month follow-up, polyp size and incidence increased but remained significantly lower than baseline. Also, Meyskens et al showed that all-trans-retinoic acid (Vesanoid) was effective in causing regression of cervical intraepithelial neoplasia (CIN), a precursor to cervical carcinoma.
These studies represent the vanguard effort in clinical trial design and evaluation of precancerous lesions and earlier biomarkers as surrogate end points for chemoprevention. Hong and Lippman[16,17] and Lipkin[13,18,19] were early contributors to the conceptualization of intermediate biomarkers as end points for chemoprevention. Our group and others have continued to expand this approach and formalize its application to chemopreventive drug development.[12,20-26] In the comparison between chemoprevention and chemotherapy above, we noted the importance of the safety of chemopreventive drugs during long-term administration. Approaches to eliminate toxicities that would preclude the use of highly efficacious agents is another significant component of chemopreventive drug development. The objective of this paper is to describe these current strategies and current progress in the development of chemopreventive drugs.
We have previously presented a multidisciplinary approach to the development of chemopreventive drugs[12,23-27] and have collaborated with the FDA to provide guidance for applying this approach. The strategy considers experimental and epidemiologic evidence that defines cancer risk at major target sites, as well as the underlying molecular, cellular, and tissue level mechanisms that contribute to the development and progression of human cancers. As previously described, we have identified three critical components to the successful development of chemopreventive drugs:
- Well-characterized agents with the potential for inhibiting the target cancer;
- Biomarkers correlating with cancer incidence for measuring chemopreventive effect; and
- Suitable cohorts for clinical efficacy studies.
The most important criterion for a chemopreventive agent is evidence of chemopreventive efficacy, in particular, a high likelihood that the agent will be active in preventing cancer at the target site. We have already discussed the requirement for low toxicity, and thus, evidence of a high margin of safety is also necessary to warrant consideration of an agent for clinical evaluation. This second criterion implies that sufficient prior clinical use or preclinical efficacy, toxicity, and pharmacodynamics data are available to allow estimation of an efficacy/safety ratio. Often, studies to determine the optimal dose and dosing regimen are performed as part of early clinical efficacy trials.
A third criterion is that there is a logical, presumed mechanism of chemopreventive activity of the agent. Such mechanisms guide the selection of both cohorts and end points for clinical trials. For example, an antiproliferative agent such as 2-difluoromethylornithine (DFMO [Eflornithine]) may be most effective against cancers with a pronounced proliferative component, such as the development of colon cancers from hyperproliferative tissue and adenomas. An antimutagenic agent such as oltipraz may be more effectively evaluated in a cohort such as smokers, who are constantly exposed to the DNA-damaging effects of carcinogens. Likewise, indicators of proliferation, such as S-phase fraction and proliferating cell nuclear antigen (PCNA), may prove to be more reliable and easily quantified measures of the effects of antiproliferative agents than would be the identification of specific mutations.
Generally, biomarkers causally related to subsequent cancer are more easily validated as end points than biomarkers for which the relationship to cancer is indirect. Carcinogenesis is progressive, and therefore, biomarkers appearing close in time to the cancer and those exhibiting increasing or decreasing incidence or potency during carcinogenesis are likely to be the most reliable surrogate end points for cancer.
We have previously identified intraepithelial neoplasia (IEN) as the intermediate biomarker that most closely meets these requirements,[12,20-22,24,27] and histopathologic changes associated with the progression of IEN should prove to be reliable end points for chemoprevention trials. Including colorectal adenomas and CIN, which were cited above, IEN lesions that have been and are currently being evaluated in chemoprevention trials are listed in Table 2. Furthermore, within IEN, other types of biomarkers--direct measures of proliferation and differentiation, mutations, changes in expression and activity of cell growth regulators, and, possibly, less direct biochemical indicators of growth and proliferation (eg, hormone levels)--may be evaluated as potential end points.
As has been elegantly described by Fearon and Vogelstein for the development of colorectal cancer, carcinogenesis is a multipath, stochastic process. Except for IEN, which essentially comprises all the varying changes that occur, this multiplicity of pathways complicates the validation of single intermediate biomarkers as surrogate end points, since they may appear on only one or a few of the many possible causal pathways. Consequently, panels of biomarkers, particularly those representing the range of carcinogenesis pathways, may prove more useful as surrogate end points.
Used alone, biomarkers representing isolated events that may or may not be on the causal pathway or otherwise associated with carcinogenesis are unlikely to be reliable. These include markers of normal cellular processes that may be increased or expressed during carcinogenesis. Examples are expression of receptors and growth factors and activities of such enzymes as glutathione S-transferase (GST) and ornithine decarboxylase (ODC). However, while these biomarkers have the disadvantage of nonspecificity for cancer, they may be useful as part of a panel of biomarkers and provide important mechanistic measurements for identifying new chemopreventive drugs. We have previously described a strategy for selecting and prioritizing biomarkers to evaluate in chemoprevention trials; this strategy is summarized in Table 3.
Clearly, it must be possible to change the expression of an intermediate biomarker/surrogate end point with chemopreventive agents. Intraepithelial neoplasia can be modulated, as demonstrated by the chemoprevention trials cited above that used colorectal adenomas and CIN as end points. While they will more often be risk biomarkers used primarily for identifying cohorts for chemoprevention studies, genetic lesions or their encoded products can also provide biomarkers that can be modulated in certain circumstances. Although acquired genetic lesions would not be excised by a chemopreventive agent, one can assume that some agents will confer a selective advantage for cells not carrying the lesion over those that do. In this instance, the biomarker would be the quantitative reduction in genetically altered cells at the tissue level.
The surrogate end point should have short latency compared with cancer incidence--ideally, months or a few years, compared with the 20- to 40-year period that may be required for cancer development. Intraepithelial neoplasia, even in its early stages, would appear to occur too late for effective chemoprevention studies in asymptomatic subjects. However, as will be discussed below, some cohorts with previous IEN are at high risk for recurrence or new lesions within the time frame of a chemoprevention study.
Some cohorts chosen for chemoprevention trials may be available for too short a period to allow modulation of the lesions most closely related to cancer. For example, we are carrying out trials in patients scheduled for surgery for early-stage prostate cancer and for breast lesions. The biopsy samples in these trials may be most useful for exploring changes in proliferation indices and other early biomarkers rather than actual progression of histologic lesions. Despite the brief duration of the associated trials (1 or 2 months), studies in these cohorts may provide valuable information on modulation of selected biomarkers that are known to be associated with cancer risk.
In early stages of chemopreventive drug evaluation, clinical research efforts focus primarily on short-term phase II trials to measure efficacy and to identify and standardize intermediate biomarker end points. The chemopreventive agents must be able to affect the disease or risk of disease within the relatively short treatment duration of such trials (many span 6 months or less; some may last up to 3 years). There are cohorts at high risk for cancer who are not good candidates for phase II chemoprevention trials, even though they will be targets for chemopreventive intervention. An example is patients at risk because of germ-line mutations (eg, BRCA1) or other family history who do not also have premalignant lesions.
From a practical standpoint, the chemopreventive effect should also be easily measured in the subject population. Tissues that are more accessible and that can be monitored in a relatively noninvasive manner provide better sites for definitive efficacy trials than do less accessible tissues. This does not mean that chemopreventive agents will be ineffective in the more difficult settings, but rather, that initial demonstration of activity may be best carried out in situations where fewer obstacles to measurement exist.
Often, the cohorts in these phase II chemoprevention clinical trials are cancer patients or patients with previous high-risk lesions who have undergone prior treatment. These patients are constantly monitored for possible recurrences and new lesions. It is important that chemoprevention trials work within the constraints of standard treatment so that patients are not at unusual risk. For example, in trials in patients with small colon adenomas (0.5 to 1.0 cm), this may result in frequent monitoring and removal of any adenomas larger than 0.5 to 1.0 cm. As suggested above, in early phase II trials where the primary goal is identification and standardization of biomarkers as end points, normal treatment may also lead to very short-term trials prior to surgery in patients who are scheduled for excision of cancers or high-risk tissue, eg, prostate carcinoma or ductal carcinoma in situ (DCIS) of the breast.
In larger phase II trials and pivotal phase III trials, the eMPHasis is on efficacy evaluation with the standardized and validated biomarkers. These trials are generally of longer duration and are carried out in cohorts that do not initially present IEN severe enough to warrant surgery but are at high risk for progression within a 1- to 5-year time frame. Examples are individuals with atypical breast hyperplasia and multiple biomarker abnormalities, such as estrogen (ER) and epidermal growth factor receptor (EGFR) overexpression, p53 mutations, HER-2/neu, and aneu-ploidy; patients with prostatic intraepithelial neoplasia (PIN); and patients with previous colorectal adenomas or superficial bladder cancers. Cohorts currently being studied in chemoprevention trials that provide the characteristics for biomarker standardization and efficacy evaluation have been discussed elsewhere[24,26,27] and are listed in Table 2.