Phase 0 Trials: Expediting the Development of Chemoprevention Agents

Perspective on Reid, et al., p. 347

Chemoprevention has been defined as “the prevention of cancer or treatment of identifiable precancers (intraepithelial neoplasia, IEN)” (1). The potential to intervene prior to the development of cancer is very attractive and confers obvious advantages to both patients and clinicians. Unfortunately, drug development for cancer prevention imposes distinct challenges beyond those associated with drug development for cancer therapy. The scale of chemoprevention trials is often much greater than that of standard treatment trials because of the large number of subjects and long duration of their participation necessary to obtain statistically useful results. For example, the Study of Tamoxifen and Raloxifene (STAR), a randomized, double-blind trial in postmenopausal women at a high risk for developing breast cancer, entered over 19,000 participants over 5 or more years (2, 3). Enrolling healthy individuals, even though at a high risk for developing cancer, raises questions about the long-term safety of the intervention and compliance. Therefore, given the cost and logistics of successfully completing a chemoprevention trial, strategies for the early identification of promising compounds that will expedite getting them into definitive clinical trials could represent a significant advance.

Potential cancer chemopreventive agents must have a high therapeutic index. This requirement enhances these drug's suitability for evaluation in trials conducted under the U.S. Food and Drug Administration (FDA) Exploratory IND (investigational new drug) Guidance, conceived as part of the FDA's “Critical Path” initiative (4). Phase 0 trials are first-in-human clinical studies performed under this guidance. Unlike traditional phase I trials, these studies have no therapeutic or diagnostic intent but instead aim to inform, to enhance the efficiency of, and to increase the chance of success of the subsequent development of the agent (5). Phase 0 objectives are to develop human pharmacodynamic (PD) and/or pharmacokinetic (PK) data (including biodistribution), information that may form the basis for rational drug development decisions (Fig. 1). Only drugs showing sufficient promise are to be evaluated for safety and tolerability in traditional phase I trials. For phase 0 trials, a single dose or a short course (typically fewer than seven days) of low, nontherapeutic, nontoxic doses is administered to a few patients, with tissue and/or blood sampling to evaluate target modulation and the PK profile of the agent. It is therefore essential that the drugs being considered for a phase 0 trial have a high therapeutic ratio in preclinical toxicity models in vivo so that the desired PK or PD effect may be observed without substantial toxicity.

The Exploratory IND Guidance includes examples of 3 types of phase 0 trials that allow, respectively, the determination of biodistribution (via imaging technologies), determination of PKs and bioavailability, and evaluation of the mechanism(s) of drug action (Box 1). These trials provide an opportunity to examine a new agent in humans earlier than traditional dose-finding, toxicity-driven phase I trials (Fig. 2). The amount of preclinical toxicology needed to support a phase 0 trial is determined by the dose and schedule intended to be administered to humans. Therefore, because a limited number of subtherapeutic doses (microdoses or nonmicrodoses) are administered in the phase 0 setting, the preclinical toxicology can also be limited, saving precious time and resources. Phase 0 trials permit identification of potential therapeutic failures earlier in the drug-development process, allowing resources to be reallocated and permitting expeditious evaluation of only the most promising agents.

A lack of predictive (drug sensitivity/resistance) models has hampered drug development across the field of oncology, including chemoprevention. This lack has contributed to the high failure rate for agents in clinical drug development trials, with regulatory approval rates of only 5% to 10% for cancer therapeutics (6). Overall, unfavorable PK, including low bioavailability, accounts for only approximately 10% of oncologic drug-development failures, with the vast majority of the remaining failures due to poor response or sensitivity to drugs with adequate PK. In the process of development, multiple analogues are routinely produced; and the lead compound is determined based on animal data that may not always predict how a molecule will behave in humans. Full-scale toxicology evaluations, manufacturing, and clinical development are often implemented only for the lead compound. Thus, potentially promising compounds may be overlooked because resource constraints limit the ability to evaluate back-up molecules if the lead compound is not initially successful in the clinic. The major advantage of phase 0 trials is their feasibility for assessing in a single trial up to 5 chemical entities or formulations (the maximum number allowed by the FDA) sharing a common biologic target. The specific aim of such studies is to identify the lead compound based on effects in an early human trial rather than in animal models (which may not predict clinical effects), thus helping to combat the lack of predictive models. Under the Exploratory IND Guidance, this identification is permissible within the purview of a single trial in a few patients, thus bypassing the need for full-scale toxicology and manufacturing for all the analogues being evaluated.

Applying the Exploratory IND Guidance in identifying the most bioavailable analogue is highlighted by the report of Reid and colleagues in this issue of the journal (7). These investigators tested multiple formulations of SR 13668 (an indole-3-carbinol analogue that inhibits the Akt pathway), assessing plasma exposures following oral administration of a single 38-mg dose, which is an easily measured, nonmicrodose, in 20 healthy volunteers. This first-in-human trial evaluated 5 different formulations, as well as the effect of food on 1 formulation, using the area under the plasma concentration–time curve (indicating systemic exposure) as the primary endpoint. In light of the small number of patients (three) in each treatment arm, the highest systemic exposure per se, rather than degree of statistical significance versus the control formulation, was prospectively defined as the major criterion for selecting a lead formulation for further development.

The study demonstrated that taking SR 13668 with food (the fed state) versus without food (the fasted state) led to optimal oral bioavailability (albeit requiring 8 capsules to deliver the protocol drug dose of 38 mg); furthermore, a lead compound for future development as a chemopreventive agent was clearly identified. Of particular note, the entire study was completed in 5 months. Although no PD endpoints were examined in this trial, its rapid completion and definitive identification of an agent for further study provide support for applying the Exploratory IND in developing new agents for chemoprevention.

Unlike the approach taken by Reid and colleagues, which utilized a dose of SR13688 that approaches the range of doses that would be therapeutic, microdose studies administer 1/100th of the pharmacologically active dose, or a maximum of 100 μg of study drug, and are allowed by both the U.S. FDA and European regulatory agency (European Medicines Agency). By definition, however, drug levels and PK parameters obtained in such studies must be determined using advanced analytic techniques that are not routinely available in academic medical centers. One of the controversies surrounding the nonpharmacologic microdosing approach to the assessment of PK is whether it is possible to extrapolate the PK data obtained from a microdose to the PK of pharmacologically active doses. This concern has been raised specifically (and especially) for agents with a nonlinear PK profile. The literature suggests that a concordance between PK profiles following a microdose versus a therapeutic dose occurs for approximately 70% to 80% of the drugs that have been evaluated in microdose studies (8–10).

Although microdosing is an option for the determination of PK under the FDA's Exploratory IND, the guidance also allows pharmacologically active doses (as used by Reid and colleagues) to be administered for assessing PD parameters, which are not assessed in microdose studies, in addition to PK parameters. The starting dose for these phase 0 studies is defined as 1/50th of the “no observed adverse effect level” (NOAEL) dose in a rodent 2-week toxicology study. Dose escalation in a phase 0 trial is also permitted to determine the dose range for target modulation and/or dose–plasma exposure relationships from PK studies using active drug concentrations. Therefore, real-time results of PK and PD analyses need to be available to allow decisions regarding dose and sampling during the conduct of the trial. This methodology requires close communication between the laboratory scientists and clinical investigators and the formation of a multidisciplinary drug-development infrastructure dedicated to the development and conduct of the preclinical and early clinical phases of the drug-development process.

In 2006, an American Association for Cancer Research Task Force highlighted the concept of prevention or regression of molecular IEN (determined by assessing molecular alterations in the histopathology of the IEN; ref. 1). Stimulated by the feasibility Reid and colleagues demonstrated for the use of pharmacologically active doses in a chemopreventive phase 0 trial, future studies could be directed toward demonstrating target modulation in IEN very early in the development process, thus identifying promising agents at a molecular level while establishing potential endpoints for subsequent trials. The prerequisites for conducting a PD-driven phase 0 chemoprevention trial would be that (i) chemopreventive effects are, in fact, based on the ability of the drug to modulate a specific target, (ii) the agent has a wide therapeutic index with target modulation occurring at nontoxic doses and relatively short exposures, and (iii) the PD effect is robust enough to be reliably measured in small numbers of patients.

Designing phase 0 trials to measure statistically significant target modulation following administration of low doses of a drug in very few patients is challenging (11–13). It requires that rigorously qualified, robust assays for determining the target drug effect be developed early in the drug development process (12); these assays must be capable of use with small tissue samples that contain limited numbers of cells of interest (e.g., premalignant cells), which are typical of chemoprevention. Optimal time points for obtaining tissue or blood samples and standard operating procedures for sample handling and processing that mirror clinical procedures all need to be developed prior to initiating the clinical trial. Questions such as “What defines a PD response in a given patient?” and “What will be considered a promising PD response rate for a dose level?” need to be prespecified and addressed up front in these trials. The tissue of interest also needs to be carefully defined, as there is no assurance that effects of an agent in surrogate tissues such as easily accessed normal skin or peripheral blood lymphocytes will, for example, have any relationship to effects in a premalignant lesion at less accessible sites.

The website Clinicaltrials.gov lists multiple ongoing chemoprevention trials of molecularly targeted agents given for relatively short periods of time (in the range of 3–6 months) to assess effects on premalignant lesions. These studies are being done with agents that have already been developed for other indications, including cancer therapy. Phase 0 trials offer the opportunity to evaluate potentially preventive agents, including molecularly targeted and other agents, much earlier in development, addressing questions related to the drug's effect on IEN and to its pharmacologic properties. Given their early stage and limited preclinical toxicology requirements, phase 0 trials could potentially expand the number of agents that are evaluated for chemoprevention while compressing the drug development timeline.

Neither S. Kummar nor J.H.. Doroshow have any conflicts of interest to report.

This work was supported by federal funds from the National Cancer Institute, NIH.

The content of this publication does not necessarily reflect the views or the policies of the U.S. Department of Health and Human Services nor does the mention of trade names, commercial products, or organizations imply endorsement by the US government.