There is a growing trend to evaluate predictive biomarkers in enriched patient populations in early phases of clinical trial. Novel biomarker-driven, adaptive clinical trial designs can facilitate rapid evaluation of drugs, help validate multiple predictive biomarkers, minimize exposure of patients to ineffective therapies, and potentially allow accelerated drug approval in molecularly defined populations.
At a recent FDA/DIA forum Bob Temple from the FDA presented on the topic of enrichment (slides available on the FDA website).
Enrichment is defined as the prospective use of any patient characteristic - including demographic, pathophysiologic, historical, and genetic - to select patients for a study population in which detection of a drug effect is more likely than it would be in an unselected population. While the main reason for enrichment is study efficiency (increasing the chance of success, often with a smaller sample size,) it also provides major benefits of individualization, directing treatment where it will do the most good and sparing people who cannot respond to potential harm.
There are three broad categories of enrichment: noise reduction (seeks to eliminate needless variability to improve the likelihood of distinguishing between two treatments); predictive (selects individuals who are likely to respond); and prognostic (selects high-risk patients for risk reduction studies.) The latter two enrichment approaches have an impact on "generalizability" of the results.
For Phase I oncology clinical trials, the primary objectives are to establish the safety of either a new drug or a new drug combination, to determine maximum tolerated dose (MTD), and verify recommended dose for Phase II. Most phase I studies have a two-stage design: 1) dose escalation of 12 to 24 patients for MTD assessment and 2) expansion cohorts at the recommended Phase 2 dose to obtain additional safety data and preliminary efficacy signals in specific patient populations of approximately 10-40 patients.
In designing a Phase I clinical trial, every aspect of the study drug must be considered. This includes mechanism of action, class effect data, preclinical data, and competitive data. In addition, certain thought must be given to defining the patient population. For instance, should therapy be restricted to certain tumor types or tumors with a specific target present? If the answer is "no," then patient inclusion can be kept broader. If "yes," the challenge is which predictive marker of enrichment of more responsive patients will be used.
A biomarker is defined as "a characteristic that is objectively measured and evaluated as an indicator of normal biological processes, pathogenic processes, or pharmacologic responses to a therapeutic intervention." Utilization of appropriate predictive biomarkers based on strong scientific rationales has helped improve clinical trial success and accelerate regulatory approval of anticancer agents by identifying a subset of patients most likely to respond to treatment. In particular, when a small fraction of the total patient population has the predictive marker and is a potential responder, predictive enrichment can be critical. On the other hand, a study in an unselected population would have little chance of success. It also prevents exposure to drugs in patients unlikely to respond, avoiding potential harm.
Putative predictive biomarkers for a targeted anticancer agent are typically identified and characterized in preclinical studies. Once a method is developed and validated for clinical sample analysis, candidate biomarkers can be tested in early phases of clinical studies and confirmed in appropriately designed phase II or phase III clinical trials. It is a growing trend to evaluate predictive biomarkers early in phase I clinical trials, especially in enlarged expansion cohorts at the maximum tolerated dose (MTD) or at the recommended phase II dose (RP2D).
The goal of phase I studies incorporating predictive biomarkers is to identify early signals that link putative biomarkers to objective responses and clinical benefits.
In oncology clinical research, biomarkers can be divided into different categories based on their functions and clinical applications, including pharmacokinetic, pharmacodynamic, prognostic, predictive, pharmacogenomic and outcome biomarkers. The performance characteristics of the selection marker (sensitivity, specificity, and predictive value) are important.
Several factors need to be taken into consideration when using this biomarker-based patient selection. For instance, the presence of the biomarker may not be representative of the disease due to the heterogeneity or its biology. There is also a need for strong biological hypothesis, supportive preclinical data, and tumor models. It's also possible that the predictive test may not be accurate and the subpopulation most likely to benefit may not be reliably identifiable. And finally, the targeted pathway might not be completely blocked by the targeted agent, or a rapid acquisition of resistance could occur due to cross-pathway escape.
Given the limited impact of most targeted therapies as a single agent, and the complexity of cancer biology, combination therapies are frequently needed. Using drugs with different targets in combinations helps fight tumors that are fueled by numerous, often redundant, genetic anomalies. Targeted drug combinations also aim to selectively inhibit mechanisms essential to cancer pathogenesis and limit off-target effects and under-desired toxicities. Possible strategies for choice of targeted drug combination include: maximizing target inhibition of a single target; maximizing pathway inhibition; targeting multiple pathways; and combining different treatment principles.
Traditionally, it was difficult to test two experimental drugs in combination because the regulatory system was geared to assessing a single drug at a time. However, much progress has been made recently regarding the co-development of two (or more) unmarketed investigation drugs for use in combination. The FDA issued guidance in 2010 on the co-development of two unmarketed drugs. The guidance described the criteria for determining when co-development is an appropriate option, made recommendations about nonclinical and clinical development strategies, and addressed certain regulatory process issues.
Clinical experience suggests that molecular prescreening and biomarker enrichment strategies in Phase I trials with targeted therapies will improve the outcomes of cancer patients. Detecting a new drug's clinical benefit in early stages of development helps avoid exposing large number of patients to single NMEs that show modest activity in early clinical testing. In addition, there are multiple options potentially worth testing for combining the targeted drug with either other targeted agents or chemotherapy.
For more information on how Proof-of-Concept data is being used in Phase I oncology trials, the following webinar is available.
Martine Poelman, M.D., is an Executive Medical Director at Covance's Clinical Development Services. She has more than 20 years experience with drug development in oncology. Her primary focus is on early development and hematological malignancies.