Strategies for phase II cancer chemoprevention trials: cervix, endometrium, and ovary

Well-designed and conducted Phase II clinical trials are very important to cancer chemoprevention drug development. Three critical aspects govern the design and conduct of these trials--well-characterized agents, suitable cohorts, and reliable biomarkers for measuring efficacy that can serve as surrogate endpoints for cancer incidence. Requirements for the agent are experimental or epidemiological data showing chemopreventive efficacy, safety on chronic administration, and a mechanistic rationale for the chemopreventive activity observed. Agents that meet these criteria for chemoprevention of cervical cancer include antiproliferative drugs (e.g., 2-difluoromethylornithine), retinoids, folic acid, antioxidant vitamins and other agents that prevent cellular oxidative damage. Because of the significant cervical cancer risk associated with human papilloma virus (HPV) infection, agents that interfere with the activity of HPV products may also prove to be effective chemopreventives. In endometrium, unopposed estrogen exposure has been associated with cancer incidence. Thus, pure antiestrogens and progestins may be chemopreventive in this tissue. Ovarian cancer risk is correlated to ovulation frequency; therefore, oral contraceptives are potentially chemopreventive in the ovary. Recent clinical observations also suggest that retinoids, particularly all-trans-N-4-hydroxyphenylretinamide, may be chemopreventive in this tissue. The cohort should be suitable for measuring the chemopreventive activity of the agent and the intermediate biomarkers chosen. In the cervix, patients with cervical intraepithelial neoplasia (CIN) and in endometrium, patients with atypical hyperplasia, fit these criteria. Defining a cohort for a Phase II trial in the ovary is more difficult. This tissue is less accessible for biopsy; consequently, the presence of precancerous lesions is more difficult to confirm. The criteria for biomarkers are that they fit expected biological mechanisms (i.e., differential expression in normal and high-risk tissue, on or closely linked to the causal pathway for the cancer, modulated by chemopreventive agents, and short latency compared with cancer), may be assayed reliably and quantitatively, measured easily, and correlate to decrease cancer incidence. They must occur in sufficient incidence to allow their biological and statistical evaluation relevant to cancer. Since carcinogenesis is a multipath process, single biomarkers are difficult to validate as surrogate endpoints, perhaps appearing on only one or a few of the many possible causal pathways. Panels of biomarkers, particularly those representing the range of carcinogenesis pathways, may prove more useful as surrogate endpoints. It is important to avoid solely on biomarkers that do not describe cancer but represent isolated events that may or may not be on the causal pathway or otherwise associated with carcinogenesis. These include markers of normal cellular processes that may be increased or expressed during carcinogenesis. Chemoprevention trials should be designed to evaluate fully the two or three biomarkers that appear to be the best models of the cancer. Additional biomarkers should be considered only if they can be analyzed efficiently and the sample size allows more important biomarkers to be evaluated completely. Two types of biomarkers that stand out regarding their high correlation to cancer and their ability to be quantified are measures of intraepithelial neoplasia and indicators of cellular proliferation. Measurements made by computer-assisted image analysis that are potentially useful as surrogate endpoint biomarkers include nuclear polymorphism comprising nuclear size, shape (roundness), and texture (DNA distribution patterns); nucleolar size and number of nucleoli/nuclei; DNA ploidy, and proliferation biomarkers such as S-phase fraction and PCNA...