Innovative Thinking

For rare disease drug development, one of the major challenges is that there are only limited subjects available for clinical trials. FDA (2019a), however, indicated that the agency does not have an intention to create a statutory standard for rare diseases drug development. In this case, some out-of-the-box

FIGURE 1.2

Demonstrating not-ineffectiveness or effectiveness in active-controlled (top line) and placebo-controlled (bottom line) studies.

innovative thinking designs are necessarily applied for obtaining substantial evidence for the approval of rare disease drug product (Chow, 2020). The out- of-the-box innovative thinking designs include (i) probability monitoring procedure for sample size requirement, (ii) the concept of demonstrating not- ineffectiveness rather than demonstrating effectiveness, (iii) borrowing RWD in support of regulatory approval of rare diseases drug products, and (iv) the use of CID to shorten the process of drug development. Along this line, Chow (2020) and Chow and Huang (2020) proposed an innovative approach for rare diseases drug development by first demonstrating not-ineffectiveness with limited subjects available and then utilizing (borrowing) RWD to rule out the probability of inconclusiveness for the demonstration of effectiveness under a two-stage adaptive seamless trial design. The proposed innovative approach can not only overcome the problem of small patient population for rare diseases, but also achieve the same standard for the evaluation of drug products with common conditions.

Probability Monitoring Procedure for Sample Size - For rare disease clinical development, it is recognized that a pre-study power analysis for sample size calculation is not feasible due to the fact that there are only limited number of subjects available for the intended trial, especially when the anticipated treatment effect is relatively small and/or the variability is relatively large. In this case, alternative methods such as precision analysis (or confidence interval approach), reproducibility analysis, and probability monitoring approach may be considered for providing substantial evidence with certain statistical assurance (Chow et al., 2017). It, however, should be noted that the resultant sample sizes from these different analyses could be very different with different levels of statistical assurance achieved. Thus, for rare disease clinical trials, it is suggested that an appropriate sample size should be selected for achieving certain statistical assurance under a valid trial design. To overcome the problem, Huang and Chow (2019) proposed a probability monitoring procedure for sample size calculation/justification, which can substantially reduce the required sample size for achieving certain statistical assurance.

As an example, an appropriate sample size may be selected based on a probability monitoring approach such that the probability of crossing safety boundary is controlled at a pre-specified level of significance. Suppose an investigator plans to monitor the safety of a rare disease clinical trial sequentially at several times, f,, i = 1,..., K. Let n, and P, be the sample size and the probability of observing an event at time f,. Thus, an appropriate sample size can be selected such that the following probability of crossing safety stopping boundary is less than a pre-specified level of significance

Note that the concept of the probability monitoring approach should not be mixed up the concepts with those based on power analysis, precision analysis, and reproducibility analysis. Statistical methods for data analysis should reflect the desired statistical assurance under the trial design.

Demonstrating Not-Ineffectiveness Versus Demonstrating Effectiveness - For the approval of a new drug product, the sponsor is required to provide substantial evidence regarding safety and efficacy of the drug product under investigation. In practice, a typical approach is to conduct adequate and well-controlled (placebo-controlled) clinical studies, and test the following point hypotheses:

The rejection of the null hypothesis of ineffectiveness is in favor of the alternative hypothesis of effectiveness. Most researchers interpret that the rejection of the null hypothesis is the demonstration of the alternative hypothesis of effectiveness. It, however, should be noted that "in favor of effectiveness" does not imply "the demonstration of effectiveness." In practice, hypotheses (1.2) should be

In other words, the rejection of H0 would lead to the conclusion of "not H0," which is Ha, i.e., "not-ineffectiveness" as given in (1.3). As can be seen from Ha in (1.2) and (1.3), the concept of effectiveness (1.2) and the concept of not-ineffectiveness (1.3) are not the same. Not-ineffectiveness does not imply effectiveness in general. Thus, the traditional approach for clinical evaluation of the drug product under investigation can only demonstrate "not-ineffectiveness"

but not "effectivenessThe relationship between demonstrating "effectiveness" (1.2) and demonstrating "not-ineffectiveness" (1.3) is illustrated in Figure 1.2. As it can be seen from Figure 1.2, in a placebo-controlled study, "not-ineffectiveness" consists of two parts, namely, the portion of "inconclusiveness" and the portion of "effectiveness." As a result, the rejection of the null hypothesis of ineffectiveness cannot directly imply that the drug product is effective unless the probability of inconclusiveness, denoted by p,c is negligible, i.e.,

where £ is a pre-specific number which is agreed upon between clinician and regulatory reviewer (Chow and Huang, 2019a).

Note that in active-controlled studies, the concept of demonstrating "not-ineffectiveness" is similar to that of establishing non-inferiority of the test treatment as compared to the active control agent. One can test for superiority (i.e., effectiveness) once the non-inferiority has been established without paying any statistical penalties. More details regarding demonstrating not-ineffectiveness for rare diseases drug development are provided in Chapter 6.

The Use of RWD/RWE - The 21st-century Cures Act passed by the United States Congress in December 2016 requires that the FDA shall establish a program to evaluate the potential use of RWE, which is derived from RWD to (i) support the approval of new indication for a drug approved under section 505 (c) and (ii) satisfy post-approval study requirements. RWD refers to data relating to patient's health status and/or the delivery of health care routinely collected from a variety of sources. RWD sources include, but are not limited to, electronic health record (EHR), administrative claims and enrolment, personal digital health applications, public health databases, and emerging sources. In practice, RWE offers the opportunities to develop robust evidence using high-quality data and sophisticated methods for producing causal-effect estimates regardless randomization method/model is used. In this chapter, we have demonstrated that the assessment of treatment effect (RWE) based on RWD could be biased due to potential selection and information biases of RWD. Although the fit-for-purpose RWE may meet regulatory standards under certain assumptions, it is not the same as substantial evidence (current regulatory standard). In practice, it is then suggested that when there are gaps between the fit-for-purpose RWE and substantial evidence, we should make efforts to fill the gaps for an accurate and reliable assessment of the treatment effect.

In order to map RWE to substantial evidence (current regulatory standard), we need to have good understanding of the RWD in terms of data relevancy/quality and its relationship with substantial evidence so that a fit- for-regulatory purpose RWE can be derived to map to regulatory standard.

As indicated by Corrigan-Curay (2018), there is a value of using RWE to support regulatory decisions in drug review and approval process. However, incorporating RWE into evidence generation, many factors must be considered at the same time before we can map RWE to substantial evidence (current regulatory standard) for regulatory review and approval. These factors include, but are not limited to, (i) efficacy or safety; (ii) relationship to available evidence; (iii) clinical context, e.g., rare, severe, life- threatening, or unmet medical need; and (iv) natural of endpoint/concerns about bias. In addition, leveraging RWE to support new indications and label revisions can help accelerate high-quality RWE earlier in the product lifecycle, providing more relevant evidence to support higher-quality and higher-value care for patients. Incorporating RWE into product labeling can lead to better-informed patient and provider decisions with more relevant information. For this purpose, it is suggested characterizing RWD quality and relevancy for regulatory purposes. Ultimate regulatory acceptability, however, will depend upon how robust these studies can be. That is, how well they minimize the potential for bias and confounding.

More details regarding the use of RWD/RWE for rare diseases drug development are provided in Chapter 8.

Innovative Trial Design - As indicated earlier, small patient population is a challenge to rare disease clinical trials. Thus, there is a need for innovative trial designs in order to obtain substantial evidence with a small number of subjects available for achieving the same standard for regulatory approval. In this sub-section, several innovative trial designs, including и-of-l trial design, an adaptive trial design, master protocols, and a Bayesian design, are discussed.

One of the major dilemmas for rare diseases clinical trials is the unavailability of patients with the rare diseases under study. In addition, it is unethical to consider a placebo control in the intended clinical trial. Thus, it is suggested an п-of-l crossover design be considered. An n-of-1 trial design is to apply n treatments (including placebo) in an individual at different dosing periods with sufficient washout in between dosing periods. A complete n-of-1 trial design is a crossover design consisting of all possible combinations of treatment assignment at different dosing periods.

Another useful innovative trial design for rare disease clinical trials is an adaptive trial design. In its draft guidance on adaptive clinical trial design, FDA defines an adaptive design as a study that includes a prospectively planned opportunity for the modification of one or more specified aspects of the study design and hypotheses based on the analysis of (usually interim) data from subjects in the study (FDA, 2010, 2019c). The FDA guidance has been served as an official document describing the potential use of adaptive designs in clinical trials since it was published in 2019. It, however, should be noted that the FDA draft guidance on adaptive clinical trial design is currently being revised in order to reflect pharmaceutical practice and FDA's current thinking.

Woodcock and LaVange (2017) introduced the concept of master protocol for studying multiple therapies, multiple diseases, or both in order to answer more questions in a more efficient and timely fashion. Master protocols include the following types of trials: umbrella, basket, and platform. The type of umbrella trial is to study multiple targeted therapies in the context of a single disease, while the type of basket trial is to study a single therapy in the context of multiple diseases or disease subtypes. The platform is to study multiple targeted therapies in the context of a single disease in a perpetual manner, with therapies allowed us to enter or leave the platform on the basis of decision algorithm. As indicated by Woodcock and LaVange (2017), if designed correctly, master protocols offer a number of benefits, including streamlined logistics, improved data quality, collection and sharing, as well as the potential to use innovative statistical approaches to study design and analysis. Master protocols may be a collection of sub-studies or a complex statistical design or platform for rapid learning and decision-making.

Linder the assumption that historical data (e.g., previous studies or experience) are available, Bayesian methods for borrowing information from different data sources may be useful. These data sources could include, but are not limited to, natural history studies and expert's opinion regarding prior distribution about the relationship between endpoints and clinical outcomes. The impact of borrowing on results can be assessed through the conduct of sensitivity analysis. One of the key questions of particular interest to the investigator and regulatory reviewer is that how much to borrow in order to (i) achieve the desired statistical assurance for substantial evidence, and (ii) maintain the quality, validity, and integrity of the study.

Innovative Approach - Combining the out-of-the-box innovative thinking regarding rare disease drug development described in the previous section, Chow and Huang (2019b) and Chow (2020) proposed the following innovative approach utilizing a two-stage adaptive approach in conjunction with the use of RWD/RWE for rare diseases drug development. This innovative approach is briefly summarized below.

Step 1. Select a small sample size щ at stage 1 as deemed appropriate by the PI based on both medical and nonmedical considerations. Note that щ may be selected based on the probability monitoring procedure.

Step 2. Test hypotheses (3) for not-ineffectiveness at the cq level, a prespecified level of significance. If fails to reject the null hypothesis of ineffectiveness, then stop the trial due to futility. Otherwise, proceed to the next stage.

Note that an appropriate value of cq can be determined based on the evaluation of the trade-off with the selection of a2 for controlling the overall type I error rate at the significance level of a. The goal of this step is to establish non-inferiority (i.e., not-ineffectiveness) of the test treatment with a limited number of subjects available at the a,-level of significance based on the concept of probability monitoring procedure for sample size justification and performing a non-inferiority (not-ineffectiveness) test with a significance level of cq.

Step За. Recruit additional n2 subjects at stage 2. Note that n2 may be selected based on the probability monitoring procedure. Once the non-inferiority (not- ineffectiveness) has been established at stage 1, sample size re-estimation may be performed for achieving the desirable statistical assurance (say 80% power) for the establishment of effectiveness of the test treatment under investigation at the second stage (say JV2, sample size required at stage 2).

Step 3b. Obtain (borrow) N2 - n2 data from previous studies (RWD) if the sample size of n2 subjects is not enough for achieving the desirable statistical assurance (say 80% power) at stage 2. Note that data obtained from the n2 subjects are from RCT, while data obtained from the other N2 - n2 are from RWD.

Step 4. Combined data from both Step 3a (data obtained from RCT) and Step 3b (data obtained from RWD) at stage 2, perform a statistical test to eliminate the probability of inconclusiveness. That is, perform a statistical test to determine whether the probability of inconclusiveness has become negligible at the a2-level of significance. For example, if the probability of inconclusiveness is less than a pre-specified value (say 5%), we can then conclude that the test treatment is effective.

In summary, for review and approval of rare diseases drug products, Chow and Huang (2020) proposed first to demonstrate not-ineffectiveness with limited information (patients) available at a pre-specified level of significance of «1. Then, after the not-ineffectiveness of the test treatment has been established, collect additional information (RWD) to rule out the probability of inconclusiveness for the demonstration of effectiveness at a pre-specified level of significance of a2 under the two-stage adaptive seamless trial design.

More details regarding innovative approach for the assessment of rare diseases drug development are provided in Chapter 9.

Remarks - In this chapter, some out-of-the-box innovative thinking designs regarding rare disease drug development are described. These innovative thinking designs include (i) probability monitoring procedure for sample size calculation/justification for certain statistical assurance, (ii) the concept of testing non-inferiority (i.e., demonstrating not-ineffectiveness) with a limited number of subjects available, (iii) utilizing (borrowing) RWD from various data sources in support of regulatory approval of rare diseases drug products, and (iv) the use of a two-stage adaptive seamless trial design to shorten the process of drug development. Combining these innovative thinking designs, under a two-stage adaptive seamless trial design, Chow and Huang (2019b) and Chow (2020) proposed an innovative approach for rare diseases drug development by first demonstrating not-ineffectiveness based on limited subjects available and then utilizing (borrowing) the RWD to rule out the probability of inconclusiveness for the demonstration of effectiveness. Chow and Huang's proposed innovative approach can not only overcome the problem of small patient population for rare diseases, but also achieve the same standard for the evaluation of drug products with common conditions.

Aim and Scope of This Book

This book is intended to be the first book entirely devoted to the discussion of innovative design and analysis for rare diseases drug development. The scope of this book will focus on some out-of-the-box innovative thinking and approach for rare diseases drug development. Along this line, an innovative approach based on these out-of-the-box innovative thinking designs is proposed for obtaining substantial evidence that will meet the same standard for the approval of rare drug products (Chow and Huang, 2019b; Chow, 2020; Chow and Huang, 2020).

This book consists of 14 chapters concerning regulatory requirement, innovative design, and analysis for rare diseases drug development. This chapter provides some background regarding rare diseases drug development. Also included in this chapter are regulatory perspectives, including regulatory incentives/guidance and brief introduction of the out-of-the-box innovative thinking for rare diseases drug development. Chapter 2 provides some basic considerations concerning rare diseases drug development. Chapter 3 focuses on hypotheses testing for the clinical evaluation of rare diseases drug products. Chapter 4 discusses the endpoint selection in rare diseases clinical trials. Also included in these chapters is the potential use of biomarker and development of therapeutic index. Chapter 5 provides the clinical strategy for non-inferiority and equivalence margin selection based on risk assessment. Chapter 6 evaluates the probability of inclusiveness after not-ineffectiveness has been established. Chapter 7 discusses the sample size requirement for rare diseases drug development based on a newly proposed probability monitoring procedure. Chapter 8 reviews FDA's recent clinical initiative for potential use of RWD/RWE in support of regulatory approval of new indications and/or labeling change. Chapter 9 gives an innovative approach for rare diseases drug development. Chapters 10-12 cover some useful complex innovative trial designs, including the и-of-l trial design, a two-stage adaptive trial design, and master protocols (the platform trial design) for rare diseases drug development, respectively. Chapters 13-14 provide discussions on case studies for gene therapy for rare diseases and NASH (Non-Alcoholic SteatoHepatitis) study for liver diseases with unmet medical need, respectively.

 
Source
< Prev   CONTENTS   Source   Next >