Development of new cancer chemoprevention agents: role of pharmacokinetic/pharmacodynamic and intermediate endpoint biomarker monitoring

^a Author for correspondence. Fax 301-402-0553; e-mail Liebermr{at}dcpcepn.nci.nih.gov.

Abstract

Recently, several promising strategies have been advanced for improving the efficiency of new agent development. These include pharmacokinetic/pharmacodynamic (PK/PD) and intermediate endpoint bio-marker (IEB) monitoring. Here, we review their essential role as practical tools for guiding the evaluation of agents for cancer chemoprevention (CP) and provide examples of CP agents that utilize these approaches. Several important categories of IEBs are delineated, including histologically based (intraepithelial neoplasias and nuclear morphometry). The use of select IEBs combined with a Bayesian method for clinical trial monitoring for rapid identification of ineffective or promising agents is discussed. The similarities between IEB and TDM are described. Finally, we present future tools for enhanced monitoring of CP agents that will impact on laboratory medicine and are also applicable to many other drug classes, e.g., laser capture microdissection and cDNA chip microarrays that assess gene expression patterns of precancerous and cancerous lesions.

Traditionally, agent development has involved evaluation of dose–response relationships and clinically objective endpoints (e.g., survival, cancer incidence). Lieberman and Nelson (1) showed that dose–response can be defined during agent development for a wide variety of drugs and biological agents. In recent years, several innovative strategies have been advanced for improving the efficiency of agent evaluation, including pharmacokinetic/pharmacodynamic (PK/PD)¹ and intermediate endpoint biomarker (IEB)-guided development. The PK/PD approach, which includes therapeutic drug monitoring (TDM), is well suited for those agents with a narrow therapeutic index and substantial PK variability (e.g., digoxin, phenytoin, cyclosporine). The useful role of IEBs is widely recognized for drug classes such as antihypertensives (to treat high blood pressure), lipid-lowering agents (to treat high concentrations of cholesterol), and antiretroviral agents (to treat viral RNA). Rational drug development represents another promising approach that targets key pathways in the pathophysiology of disease.

One therapeutic area that incorporates aspects of all of these tools is chemoprevention (CP) of cancer. Cancer CP can be defined as application of interventions (nutrients, drugs, biologics) that inhibit or reverse the multistage carcinogenic process. Major classes of CP agents include the antimutagens (oltipraz), antiproliferative agents (retinoids), antioxidants (vitamin E), and antiinflammatory agents (sulindac). Recognition of the long latency period for invasive cancer development (e.g., 20–40 years for many epithelial cancers) has led to a strategy in cancer CP agent evaluation that is guided by biomarkers as primary response variables for assessing activity and by histologically based IEBs (e.g., intraepithelial neoplasia) that can serve as valid surrogates for cancer incidence (2). The role of IEBs in cancer CP is distinct from the current application of classic tumor markers in the diagnosis and management of cancer recurrence and remission (see Table 1 ).

Utilization of IEBs provides several important benefits for cancer CP. These include (a) selection of high-risk cohorts for clinical trials; (b) rapid assessment/monitoring of biological activity; (c) reduction in sample size; (d) shorter study duration for achieving important clinical endpoints; and (e) improved estimates of the treatment effect. In this review, we provide an overview and specific examples for the role of PK/PD and the emerging role of IEB monitoring as an efficient framework for guiding clinical trial design and development of CP agents. In this context, CP agents can be viewed as a prototype for other drug classes to which the principles of PK/PD, TDM, and IEBs can be readily extended.

PK/PD-Guided Agent Development

Promising CP agents under development are presented in Table 2 . Regardless of the class and mode of action, it is necessary to characterize the clinical PK, e.g., absorption, distribution, metabolism, and elimination profiles, and PD of these agents in the target population. The relative importance of PK/PD and impact on the therapeutic index for several classes of CP agents is shown in Table 3 . This is exemplified by the new selective estrogen receptor modulators such as raloxifene, which has a better safety profile than tamoxifen.

Nonsteroidal antiinflammatory drugs (NSAIDs) and CP.
NSAIDS are prime examples of where PK/PD has had an impact in improving the therapeutic index. Inhibition of cyclooxygenase 2 (Cox-2) and enhancement of apoptosis by NSAIDS (sulindac) make a promising approach for the prevention of sporadic and hereditary colon cancers (3). Furthermore, the development of new selective Cox-2 inhibitors, a novel drug class without the gastrointestinal toxicities of common NSAIDS, e.g., flurbiprofen, is a rational CP strategy for colon cancer. In addition, the R-isomer of flurbiprofen has been shown in animal models to be a potent antiproliferative for normal colonocytes, colon polyps, and adenocarcinomas without the gastrointestinal toxicity of the S-isomer or the racemate (4). R-Flurbiprofen has a novel mechanism of action (induction of apoptosis) independent of Cox-1 or Cox-2-related pathways. Currently, several NSAIDs (sulindac and its sulfone) and Cox-2 inhibitors are being evaluated in CP trials in patients with sporadic and hereditary colon cancer.

Retinoids and CP.
Retinoids are probably the best-studied class of agents in cancer CP. They include >600 natural and synthetic derivatives of vitamin A and retinoic acid (RA). The main factor limiting their clinical use is multisystem toxicity (involving skin and mucosa, liver, bone, the central nervous system, and serum lipids). Fenretinide (4-HPR), a synthetic retinoid and lead agent is less toxic and teratogenic than other retinoids. It is associated with a dose/concentration-related impairment of dark adaptation, attributable to a reduction in tissue/serum retinol concentrations (5). This toxic effect is reversible upon drug interruption, which allows tissue retinol concentrations to recover. In a large phase 3, randomized, placebo-controlled CP trial, daily doses of 200 mg of 4-HPR combined with a short monthly drug holiday (3 days) resulted in a reduction of contralateral breast cancer in premenopausal patients and reduced incidences of impaired dark adaptation (6). Concentrations of 4-HPR and its major metabolite have been observed to be 10–20-fold greater in the breast than in plasma, which may help explain the observed beneficial effect.

Rational combinations and CP.
Combination therapy is another approach for improving the therapeutic index of retinoids and some other CP agents that have serious toxicities. This strategy involves combining agents with different mechanisms of action and nonoverlapping toxicities and avoiding chemoresistance. The combination of 4-HPR and tamoxifen is a case in point. Each agent has marked CP activity in a different cohort of breast cancer patients: 4-HPR in younger subjects (under 45) and tamoxifen in older subjects (over 55). Several groups have initiated combination trials with these agents in patients with breast cancer (Singletary and Lippman, personal communication).

Combinations under development include the versatile antiproliferative agent 2-difluoromethylornithine combined with either tamoxifen, oltipraz, piroxicam, or 4-HPR (7). Other combinations applicable for prostate neoplasia include total androgen blockade (with flutamide and luprolide) as neohormonal adjuvant therapy before prostate cancer surgery, an anti-androgen combined with an anti-estrogen to reduce gynecomastia in high-grade precancerous lesions, or an anti-androgen combined with a 5-reductase inhibitor for low-Gleason-grade prostate cancer.

IEB-Guided Agent Development

IEBs may be subdivided into several categories representing a wide spectrum of biological processes, as shown in Table 4 . The origin of these IEBs closely parallels the progression of genetic/epigenetic changes involved in the multipath cascade that culminates in an invasive cancer phenotype.

Histology-based biomarkers are on the causal pathway to cancer and include preinvasive intraepithelial neoplasias such as colon polyps and carcinoma in situ of the breast, cervix, and prostate. These lesions may be valid as surrogate endpoint biomarkers (SEBs) for cancer incidence (8). Histologic parameters defined by computer-assisted nuclear morphometry represent an extension of the pathologist in quantitating the nuclear morphologic characteristics of the cancer phenotype. A model workstation—microscope, camera, and personal computer for quantitative image analysis—is shown in Fig. 1 .

View larger version (36K): [in this window] [in a new window]   Figure 1. Schematic representation of a computerized quantitative image analysis workstation adapted from the CasTM 200 system.

Cellular and molecular biomarkers are presumed to have biological relevance to carcinogenesis, including measures of proliferation, apoptosis, differentiation, and growth factor-mediated signal transduction. Some of these are proving to be closely correlated with changes in preinvasive lesions (e.g., telomerase activity) and thus could serve as potential SEBs for cancer.

Drug-effect biomarkers reflect the pharmacologic mechanism of action of an agent. Modulation of polyamine biosynthesis through inhibition of ornithine decarboxylase may or may not have a direct association to the carcinogenic process. Recent evidence suggests, however, that under certain circumstances, overexpression of this enzyme can function as an oncogene and contribute to the invasive potential of epithelial cancers (9).

Insulin-like growth factor-1 (IGF-1)-guided development of 4-HPR.
IGF-1 underscores the successful use of IEBs in guiding CP agent development. Several lines of evidence support the biological role of the IGF family of ligands/receptors in the proliferation of breast cancer cells (10). The effect of 4-HPR on plasma concentrations of IGF-1 was studied in a consecutive cohort of stage 1 breast cancer patients from the large phase 3 breast cancer prevention trial. The study dealt with 32 women receiving 200 mg of 4-HPR per day and 28 untreated controls. At baseline, there was no difference in IGF-1 concentrations between the two groups. After follow-up, plasma IGF-1 concentrations were unchanged in controls (163.3 ± 7.4 µg/L) but were less in patients treated with 4-HPR (134.6 ± 8.1 µg/L). As Fig. 2 shows, there was a distinct interaction between 4-HPR, IGF-1, and age (menopausal status). The decrease in IGF-1 concentrations induced by 4-HPR was much more pronounced in younger premenopausal patients, whereas an age-related decline in controls was observed over time.

View larger version (13K): [in this window] [in a new window]   Figure 2. Relationship between 4-HPR treatment, change in IGF-1 plasma concentrations, and age (menopausal status) in breast cancer patients. Adapted from Torrisi et al. [11]; used with permission.

Moreover, there was a cross-over effect such that IGF-1 concentrations tended to increase in 4-HPR-treated patients with increasing age (11). Because the observed reduction in contralateral breast cancer correlates with the changes in IGF-1, 4-HPR-induced modulation (decrease) of circulating plasma concentrations of IGF-1 in premenopausal women may be a valid SEB for the reduction in breast cancer incidence.

Retinoic acid receptor (RAR)-guided development of retinoids.
Upregulation of expression of certain RAR subtypes appears to be an important indicator of drug efficacy for 13-cis-RA and possibly 9-cis-RA in head and neck preinvasive lesions as well as in dysplastic lesions of the lung (12). Lotan et al. (13) evaluated biopsy specimens of 52 patients with oral premalignant lesions for expression of RARs before and after 13-cis-RA in comparison with normal controls. Only ~40% of the patients with oral premalignant lesions expressed RAR-ß mRNA at baseline, whereas 90% did after 13-cis-RA treatment.

Monitoring of IEBs in Clinical CP Trials

Development and evaluation of agents for cancer CP is guided by a set of principles promulgated by a joint working group of the National Cancer Institute and the Food and Drug Administration (14). An abbreviated summary of this developmental scheme is presented in Table 5 . The phases of agent development encompass preclinical evaluation (in vitro and in vivo models) and the full range of clinical studies (phases 1–4). The level of evidence required for the evaluation and validation of candidate SEBs is that of the randomized controlled clinical trial. SEBs can be used for accelerated drug approval under the Code of Federal Regulations (CFR-314.510). Criteria for statistical validation of a surrogate endpoint in these trials have been proposed by Prentice (15).

IEBs play a critical role in the design and analysis of phase 1 and 2 clinical trials of CP agents in cohorts of patients with preinvasive neoplasia. The primary use of IEBs is to serve as indicators of biological response and as SEBs for cancer incidence reduction. Furthermore, to improve statistical power, the primary IEB is often modeled as a gaussianly distributed continuous random variable (mean, SD).

Bayesian monitoring and IEBs.
Recently, to improve the efficiency of phase 2 trials, we have begun to evaluate a Bayesian monitoring scheme for interim analyses as an alternative to the standard group sequential methods. The conceptual basis for this approach is shown in Fig. 3 . The Bayesian stopping rule is designed (n = 30 patients) so that it is quicker to stop if the evidence suggests ineffectiveness rather than only if it suggests biological activity; this allows resources to be concentrated on those agents that show the most promise in this early stage of testing.

View larger version (12K): [in this window] [in a new window]   Figure 3. Conceptual basis for Bayesian monitoring and analysis of modulation in intermediate endpoint biomarkers during a clinical intervention (chemoprevention) trial. Adapted from Freedman et al. Stat Med 1994;13:1371–83; used with permission.

In a series of clinical trial simulations and sensitivity analyses based on literature-derived values for a growth factor (transforming growth factor- expression), the Bayesian method compared favorably with standard group sequential methods. Furthermore, the Bayesian method retained simplicity of interpretation and provided flexibility in monitoring as the data accumulated during a trial without the need to adjust for type 1 error at each interim look (15A).

Cohorts for CP trials.
The third component of the triad for a successful CP program (the first component being agent selection and the second being biomarker validation) is the identification of suitable patient cohorts. Using the CP of prostate cancer as a prototype, Table 6 presents general and specific strategies for this identification.

Analogy of IEB monitoring to TDM principles.
IEB monitoring and TDM share several common features, as summarized in Table 7 . Both can serve as surrogate measures for objective clinical responses. Both can have a target range, i.e., therapeutic range of drug concentrations or expression values for guiding dosing requirements. A relative therapeutic index, which reflects the safety margin between toxic and effective doses, can be defined for TDM-guided agents. Similarly, a relative CP index can be defined for IEB-guided agents. In addition, both IEBs and TDM require careful measurements of analytes combined with quality assurance, quality control, and laboratory validation.

Future Direction of IEBs in Cancer CP

One of the most innovative approaches under development is the Cancer Genome Anatomy Project of the National Cancer Institute (16). This approach is a hybrid, bringing together computer science and genetics and creating a new field known as genomics. The aim of the project is to develop a user-friendly computer database containing a complete catalog of all the genes expressed in cancer cells (tumor gene index). The initial focus will be on five big killers—cancers of the prostate, breast, colon, lung, and ovary.

Several critical technologies are being used. The first is laser capture microdissection, used as a tool of the pathologist for acquisition of histopathogically defined pure cell populations from human precancer and cancer (17). Another critical technique involves the cloning/amplification of cellular RNA for the construction of full-length cDNA libraries that are representative of the expressed repertoire of genes from normal, precancer, and cancer lesions. The third new technology involves microchip cDNA probes and oligonucleotide arrays.

Several applications of these new technologies for cancer CP are indicated in Table 8 . Recently published experience with DNA microarrays includes sequence analysis of mutations in heterozygous carriers of BRCA-1 (18) and analysis of changes in gene expression in malignant melanoma cells before and after suppression of the tumorigenic phenotype (19). The first commercial applications of this new chip technology involve mutational analysis of the p53 tumor suppressor gene and of the human immunodeficiency virus. Furthermore, the pattern of differential gene expression among normal cells and preinvasive and invasive cancer cells should be a powerful tool for risk stratification and histologic classification.

In addition, the ability of CP agents to modulate gene expression and thus the genetic profile of a precancerous lesion should lead to the development of novel panels of IEBs (genes/proteins) for assessing biological response and predicting reduction of cancer incidence. These emerging technologies will have important implications for cancer CP and the clinical laboratory as we enter the new era of molecular medicine and genomics.

Applications to Other Drug Classes

The utility of PK/PD, TDM, and IEB monitoring for the efficient development of CP agents for cancer should be readily applicable to other drug classes and therapeutic indications. For example, SEBs have been successfully incorporated into the development of cardiovascular agents (antihypertensives), lipid-altering agents (HMG coreductase inhibitors), and antiretroviral agents (protease inhibitors). In addition, the new molecular tools being developed for cancer CP (see Table 8 above) hold considerable promise for many classes of therapeutic agents.

In summary, the integration of these tools into new drug-evaluation strategies, especially into randomized controlled intervention trials, provides a rigorous scientific framework for the development, validation, quality control, and monitoring of innovative assays of drug efficacy and toxicity.

Footnotes

Chemoprevention Branch, National Cancer Institute, 6130 Executive Blvd., Rm. EPN 201, Rockville, MD 20852.

¹ Nonstandard abbreviations: PK, pharmacokinetic(s); PD, pharmacodynamic(s); IEB, intermediate endpoint biomarker; CP, chemoprevention; NSAID, nonsteroidal antiinflammatory drug; Cox, cyclooxygenase; RA, retinoic acid; 4-HPR, fenretinide [N-(4-hydroxyphenyl)retinamide]; SEB, surrogate endpoint biomarker; IGF, insulin-type growth factor; and RAR, retinoic acid receptor.

References

1. Lieberman R, Nelson R. Dose–response and concentration–response relationships: clinical and regulatory perspectives. Ther Drug Monit 1995;15:498-502.
2. Kelloff GJ, Boone CW, Crowell JA, Steele VE, Lubet R, Doody LA. Surrogate endpoint biomarkers for phase 2 cancer chemoprevention trials. J Cell Biochem 1994;19:1-9.
3. Boolbol SK, Dannenberg AJ, Chadburn A, Martucci C, Guo XJ, Ramonetti JT, et al. Cyclooxygenase-2 overexpression and tumor formation are blocked by sulindac in a murine model of familial adenomatous polyposis. Cancer Res 1996;56:2556-2560. [Abstract/Free Full Text]
4. McCracken JD, Jacoby RF, Kantoci D, Murray ED, Wechter W. The effect of R-flurbiprofen on inhibition of polyp formation in the Min mouse model [Abstract]. Gastroenterology 1996;110:A555.[Web of Science]
5. Formelli F, Clerici M, Campa T, Di Mauro MG, Magni A, Mascotti G, et al. Five-year administration of fenretinide:pharmacokinetics and effects on plasma retinol concentrations. J Clin Oncol 1993;11:2036-2042. [Abstract/Free Full Text]
6. Rotmenze N, DePalo G, Formelli F, Costa A, Marubini E, Campa E, et al. Long-term tolerability of fenretinide (4-HPR) in breast cancer patients. Eur J Cancer 1991;27:1127-1131.
7. Kelloff G, Hawk E, Crowell JA, Boone CW, Nayfield S, Perloff M, et al. Strategies for identification and clinical evaluation of promising chemopreventive agents. Oncology 1996;10:1471-1480. [Medline] [Order article via Infotrieve]
8. Kelloff GJ, Hawk ET, Karp JE, Crowell JA, Boone CW, Steele VE, et al. Progress in clinical chemoprevention. Semin Oncol 1997;24:241-252. [Web of Science][Medline] [Order article via Infotrieve]
9. Smith MK, Goral M, Wright JH, Matrisian LM, Morris RJ, Klein-Szanto AJP, et al. Ornithine decarboxylase overexpression leads to increased epithelial tumor invasiveness. Cancer Res 1997;57:2104-2108. [Abstract/Free Full Text]
10. Macaulay VL. Insulin-like growth factors and cancer. Br J Cancer 1992;65:311-320. [Web of Science][Medline] [Order article via Infotrieve]
11. Torrisi R, Pensa F, Orengo MA, Catsafados E, Ponzani P, Boccardo F, et al. The synthetic retinoid fenretinide lowers plasma insulin-like growth factor 1 levels in breast cancer patients. Cancer Res 1993;53:4769-4771. [Abstract/Free Full Text]
12. Khuri FR, Lippman SM, Spitz MR, Lotan R, Hong WK. Molecular epidemiology and retinoid chemoprevention of head and neck cancer. J Natl Cancer Inst 1997;89:199-211. [Abstract/Free Full Text]
13. Lotan R, Xu XC, Lippman SM, Ro JY, Lee JS, Lee JJ, et al. Suppression of retinoic acid receptor-beta in premalignant oral lesions and its upregulation by isotretinoin. N Engl J Med 1995;332:1405-1410. [Abstract/Free Full Text]
14. Kelloff GJ, Johnson JR, Crowell JA, Boone CW, DeGeorge J, Steele VE, et al. Approaches to the development and marketing approval of drugs that prevent cancer. Cancer Epidemiol Biomarkers Prev 1995;4:1-10. [Abstract]
15. Prentice R. Surrogate endpoints in clinical trials: definition and operational criteria. Stat Med 1989;8:431-440. [Web of Science][Medline] [Order article via Infotrieve]
16. Cronin KA, Lieberman R, Freedman LS, Weiss HL, Beenken SW, Kelloff GJ. Bayesian sequential analysis of intermediate end points in cancer prevention trials [Abstract]. Proc Am Assoc Cancer Res 1998;A2065..
16. News Comment. A catalog of cancer genes at the click of a mouse. Science 1997;276:1023-1024. [Free Full Text]
17. Emmert-Buck M, Bonner RF, Smith PD, Chuaqui RF, Zhuang Z, Goldstein SR, et al. Laser capture microdissection. Science 1996;274:998-1001. [Abstract/Free Full Text]
18. Hacia JG, Brody LC, Chee MS, Fodor SP, Collins F. Detection of heterozygous mutations in BRCA1 using high density oligonucleotide arrays and two-colour fluorescence analysis. Nature Genet 1996;14:441-447. [Web of Science][Medline] [Order article via Infotrieve]
19. DeRisi J, Penland L, Brown PO, Bitner ML, Meltzer PS, Ray M, et al. Use of a cDNA microarray to analyze gene expression patterns in human cancer. Nature Genet 1996;14:457-460. [Web of Science][Medline] [Order article via Infotrieve]