Novel Designs of Early Phase Trials for Cancer Therapeutics

Shivaani Kummar, Chris Takimoto

• Publisher: Academic Press
• Year: 2018
• Print ISBN: 9780128125120
• eBook ISBN: 9780128125700
• Topic: Medizin
• Subtopic: Onkologie

Novel Designs of Early Phase Trials for Cancer Therapeutics provides a comprehensive review by leaders in the field of the process of drug development, the integration of molecular profiling, the changes in early phase trial designs, and endpoints to optimally develop a new generation of cancer therapeutics. The book discusses topics such as statistical perspectives on cohort expansions, the role and application of molecular profiling and how to integrate biomarkers in early phase trials. Additionally, it discusses how to incorporate patient reported outcomes in phase one trials.This book is a valuable resource for medical oncologists, basic and translational biomedical scientists, and trainees in oncology and pharmacology who are interested in learning how to improve their research by using early phase trials.- Brings a comprehensive review and recommendations for new clinical trial designs for modern cancer therapeutics- Provides the reader with a better understanding on how to design and implement early phase oncology trials- Presents a better and updated understanding of the process of developing new treatments for cancer, the exciting scientific advances and how they are informing drug development

Chapter 1

Changing Landscape of Early Phase Clinical Trials

Beyond the Horizon

Khanh Do¹, Chris H. Takimoto² and Shivaani Kummar³, ¹Dana-Farber Cancer Institute, Boston, MA, United States, ²Forty Seven, Inc., Menlo Park, CA, United States, ³Stanford University, Palo Alto, CA, United States

Abstract

The field of oncology has evolved dramatically over the past 60 years. The growing complexity of clinical trials presents a number of logistical challenges, driving the escalating costs of drug development. As trial designs evolve to meet the pace of early drug development, financial considerations need to be taken into account in addition to safety and efficacy.

Keywords

Drug approval; trial design; molecular targeted agents; precision medicine

1.1 Historical Perspective

The field of drug development has evolved dramatically over the past 60 years, beginning with the first law put in place to protect against misbranding and adulteration of foods, drinks, and drugs—the 1906 Pure Food and Drugs Act. It would take an additional 30 years before the 1938 Food, Drug and Cosmetic Act was passed, which required premarket proof of safety before a drug could go on the market. This was enacted in response to several deaths that occurred when the elixir sulfanilamide, which contained a solvent analog of antifreeze, was marketed as an antiinfective. More recently, 2012 marked the 50-year anniversary of the 1962 Kefauver-Harris Amendment to the Food, Drug and Cosmetic Act that required drug manufacturers to provide proof of efficacy and safety before a drug application could be approved. This landmark piece of legislation was enacted in response to birth defects arising from the use of thalidomide, which at the time was marketed as a sedative and widely prescribed in Europe and Canada. Shortly after this, the FDA formalized the drug review process, delineating each step and phase in the development of investigational agents. Informed consent would now be required of participants in clinical trials and reporting of adverse drug reactions to the FDA would be mandated. Together with increasing legislative hurdles came longer review processes before a drug could come to market. Currently, in the field of oncology, the likelihood of FDA approval for drugs entering clinical development in Phase 1 studies is estimated at 6.7%, the lowest of all investigational drugs reviewed by the FDA during the period of January 1, 2003 to December 31, 2011 [1]. In response to increasing pressure from patients, patient advocates, and Congress to improve patient access to investigational drugs, the FDA Safety and Innovation Act of 2012 (FDASIA) was passed, which established two modifications to the Food, Drug and Cosmetic Act. First, this Act allowed for a new “breakthrough therapy” designation for investigational drugs. Second, in an effort to expedite the development and review of drugs intended to treat a “serious condition” and have “preliminary clinical evidence indicating that the drug demonstrates substantial improvement over available therapy on a clinically significant endpoint” it allowed for expansion of the statute regarding accelerated approval of investigational agents [2].

Historically, the majority of oncology drugs in development prior to the 1990s were cytotoxic agents. The principles of early drug development initially focused on defining the highest tolerable dose based on observations of direct correlation between dose and cell killing of cytotoxic agents [3]. Phase 1 trials accordingly serve as the cornerstone of drug development, shepherding the transition from the preclinical to clinical stage, testing the safe and maximum tolerable dose (MTD) of a drug in order to define the recommended dose to be carried forward in Phase 2 trials. Rule-based designs have traditionally been the most accepted and widely used of early phase clinical trial designs [4]. In particular, the 3 + 3 design remains the prevailing method used in Phase 1 clinical trial design. The structure of this design assumes that toxicity increases with dose and involves enrollment of three-patient cohorts in escalating preestablished dose levels with the starting dose extrapolated from animal toxicology data. Escalation proceeds until two patients experience dose-limiting toxicities (DLT) in a cohort of three to six patients, whereupon the dose level below this toxic dose level is designated the recommended dose for Phase 2 trials (RP2D). While this design is simple to implement and allows for gathering of data to establish PK-toxicity curves, critics have argued that it requires an excessive number of escalation steps, which prolongs the time required to reach MTD. This also increases the exposure of a disproportionate number of patients to subtherapeutic doses [5–7].

Over the past decade and a half, evolving knowledge of the human genome and molecular pathways has resulted in an exponential growth in the development of molecularly targeted agents (MTAs). Unlike cytotoxic agents, MTAs can demonstrate delayed or cumulative low-grade toxicities that may not be captured within a predefined DLT-assessment window. Additionally, dose may not directly correlate with toxicity or efficacy, depending on the mechanism of action of the agent and the exposures required for target engagement. In response to these challenges, newer strategies for dose escalation have included accelerated titration designs and model-based designs. Accelerated titration designs have the advantage of allowing for rapid dose escalation in single patient cohorts, as well as intrapatient dose escalation, thereby minimizing the proportion of patients potentially treated at subtherapeutic doses [8]. An analysis of 270 published Phase 1 studies from 1997 to 2008 showed that studies using accelerated titration designs resulted in the evaluation of a greater number of dose levels (7 vs 5, P=0.0001) and reduced numbers of patients treated at doses below the RP2D (46% vs 56%, P=0.0001) [9]. In accelerated titration designs, dose escalations occur in increments of either 40% or 100% until a DLT or two moderate toxicities are observed, whereupon the dose escalation and stopping rules revert to the 3 + 3 design. Some of these accelerated design have the drawback of allowing for intrapatient dose escalation with regard to delineating toxicity data. Specifically, a single patient may contribute data for more than one dose level and delayed toxicities may be masked by presumed cumulative toxicities. Additionally, analyses comparing 3 + 3 design and accelerated titration designs have not shown convincing data that accelerated titration designs shorten the overall accrual time nor increase the efficacy of Phase 1 trials [9]. Other rule-based designs have been proposed, including the isotonic regression model [10], improvements on the original up-and-down design [11], accelerated biased coin up-and-down design [12], and the “rolling six” design [13]. Attempts have also been made to propose the use of pharmacologic data to guide dose escalation, however, the logistical practicality of this model, requiring real-time patient pharmacokinetic (PK) data and variability in the PKs between patients and challenges in extrapolation of plasma exposure data from one patient to the next, has limited the widespread acceptance of this model [14].

Alternatively, model-based designs use mathematical modeling based on Bayesian probability statistics to predict a dose level, which would produce a prespecified probability of DLT using real-time cumulative toxicity data from all enrolled patients, thereby producing a dose-toxicity probability curve that allows for computation of the optimal safe dose for the next cohort of patients. The continual reassessment method (CRM) was the first Bayesian model-based method to be adopted in Phase 1 trial designs [15]. In this design, the estimation of probability of encountering a DLT is updated for each new patient entering the study, allowing for multiple dose escalations and de-escalations, until the prespecified probability of DLT at the estimated MTD level is achieved. Multiple modifications of the original CRM design have been developed with the aim of enhancing safety, including restricting dose escalation to one level at a time [16], allowing for treatment of several patients at the same dose level [17], expanding the cohort of patients at the RP2D [18,19], and implementing overdose control in an effort to limit exposing patients to potentially higher toxic doses [20,21]. More recently, additional modifications of the CRM model have been proposed to account for low-grade chronic toxicities often seen with MTAs, e.g., using ordinal toxicity outcomes [22–26] and late-onset or cumulative toxicities using a time-to-event continual reassessment method (TITE-CRM) [27]. A practical challenge with the implementation of model-based approaches is the need for real-time biostatistical support, which may not be readily available at all institutions.

1.2 Current Trends

With evolving selectivity of each generation of MTAs, the biologic effective dose has emerged as the more relevant endpoint in early phase clinical trials. While toxicity remains an important endpoint, the highly selective nature of MTAs results in widening of the dose-toxicity curves and a RP2D that may be well below the MTD. Increasingly, more trials are now incorporating mandated biopsies to further explore the pharmacodynamic (PD) effects of receptor occupancy and target inhibition in tumors in an effort to characterize the biologically active dose. A greater emphasis is also being placed on characterization of the PK–PD relationship to guide the decision on the declaration of the RP2D. In line with this shift in paradigm in oncology drug development, various novel approaches including the TriCRM method have been proposed, to address the incorporation of toxicity and efficacy data into the estimation of the biologically effective dose [28]. In response to the complexity of incorporating PD endpoints in early clinical trial modeling, the Task Force on Methodology for the Development of Innovative Cancer Therapies was developed to provide guidance on dose escalation methods specific for molecularly targeted compounds [29]. The Task Force acknowledged the importance of establishing the “biologically active” dose range for future development of targeted agent combinations where overlapping toxicities have the potential to limit the tolerability of administering both agents in full doses. With increasing emphasis on PD-driven trials, more attention has been drawn to the challenges of tissue acquisition and assay performance. In an effort to address the call for development of more sensitive and specific biomarkers and enhance the efficiency of investigational drug development, the National Cancer Institute Investigational Drug Steering Committee and Biomarker Task Force was charged with the development of guidelines for the incorporation of biomarker studies in early clinical trials of novel agents and set the standards for assay performance [30]. Although the evaluation of safety remains the primary goal of early clinical trials, assessment of efficacy and PK/PD parameters are emerging as the key objectives in the new era of drug development where advances in next-generation sequencing can provide rapid genomic mutation profiles of tumors.

Phase 1 trials are increasingly being used as a platform to explore predictive biomarkers and to enable early evaluation of antitumor efficacy by enriching subsets of patients selected according to molecular criteria who are expected to most likely respond to a particular MTA, focusing on ever smaller subsets of patients. This paradigm shift in oncology drug development has culminated in the “precision-medicine” based approach where the selection of patients is limited to certain mutations of interest or presence of target, based on the mechanism of action of the agent being studied. The ability to identify and treat specific subsets of patients based on presence of a molecular target has increased the complexity of early phase trials and has necessitated multicenter collaborations.

These “patient enrichment” strategies have been utilized in order to enroll fewer patients, demonstrate larger treatment effects, and expedite the drug development process. However, apart from identifying the mutations of interest and designing agents that effectively target these alterations; this approach presents the logistical challenge of patient selection and timely enrollment. The clinical trial demonstrating high response rate and clinical benefit for crizotinib in patients with nonsmall cell lung cancer carrying the EML4-ALK rearrangement screened approximately 1500 tumor samples to enroll the required 82 patients [31]. The need to obtain archival or fresh tumor tissue for screening, laboratory infrastructure to perform adequately qualified assays in a clinically relevant time frame for patient selection, and treatment of selected patients requires an infrast...

Table of contents

• Cover image
• Title page
• Table of Contents
• Copyright
• Dedication
• List of Contributors
• Preface
• Chapter 1. Changing Landscape of Early Phase Clinical Trials: Beyond the Horizon
• Chapter 2. Requirements for Filing an Investigational New Drug Application
• Chapter 3. The Evolution of Phase I Trials, Past, Present, and Future: A Biostatistical Perspective
• Chapter 4. Evolving Early Phase Trial Designs: A Regulatory Perspective
• Chapter 5. The Challenges of Implementing Multiarmed Early Phase Oncology Clinical Trials
• Chapter 6. Designing Trials for Drug Combinations
• Chapter 7. Dose Selection of Targeted Oncology Drugs in Early Development: Time for Change?
• Chapter 8. Integrating Biomarkers in Early-Phase Trials
• Chapter 9. Can Early Clinical Trials Help Deliver More Precise Cancer Care?
• Chapter 10. Role of Imaging in Early-Phase Trials
• Chapter 11. Clinical Trial Design in Immuno-Oncology
• Chapter 12. Cell-Based Therapies: A New Frontier of Personalized Medicine
• Chapter 13. Incorporating Patient-Reported Outcomes Into Early-Phase Trials
• Index