Clinical Trials

Background Developments in biotechnology and genomics have increased the focus of biostatisticians on prediction problems. This has led to many exciting developments for predictive modeling where the number of variables is larger than the number of cases. Heterogeneity of human diseases and new technology for characterizing them presents new opportunities and challenges for the design and analysis of clinical trials.

Purpose In oncology, treatment of broad populations with regimens that do not benefit most patients is less economically sustainable with expensive molecularly targeted therapeutics. The established molecular heterogeneity of human diseases requires the development of new paradigms for the design and analysis of randomized clinical trials as a reliable basis for predictive medicine [Simon R. An agenda for clinical trials: clinical trials in the genomic era. Clin Trials 2004; 1:468—70, Simon R. New challenges for 21st century clinical trials. Clin Trials 2007; 4: 167—9.]. Results We have reviewed prospective designs for the development of new therapeutics with candidate predictive biomarkers. We have also outlined a prediction based approach to the analysis of randomized clinical trials that both preserves the type I error and provides a reliable internally validated basis for predicting which patients are most likely or unlikely to benefit from the new regimen. Conclusions Developing new treatments with predictive biomarkers for identifying the patients who are most likely or least likely to benefit makes drug development more complex. But for many new oncology drugs it is the only science based approach and should increase the chance of success. It may also lead to more consistency in results among trials and has obvious benefits for reducing the number of patients who ultimately receive expensive drugs which expose them risks of adverse events but no benefit. This approach also has great potential value for controlling societal expenditures on health care. Development of treatments with predictive biomarkers requires major changes in the standard paradigms for the design and analysis of clinical trials. Some of the key assumptions upon which current methods are based are no longer valid. In addition to reviewing a variety of new clinical trial designs for co-development of treatments and predictive biomarkers, we have outlined a prediction based approach to the analysis of randomized clinical trials. This is a very structured approach whose use requires careful prospective planning. It requires further development but may serve as a basis for a new generation of predictive clinical trials which provide the kinds of reliable individualized information which physicians and patients have long sought, but which have not been available from the past use of post-hoc subset analysis. Clinical Trials 2010; 7: 516—524. http://ctj.sagepub.com

Background The current practice for seeking genomically favorable patients in randomized controlled clinical trials using genomic convenience samples.

Purpose To discuss the extent of imbalance, confounding, bias, design efficiency loss, type I error, and type II error that can occur in the evaluation of the convenience samples, particularly when they are small samples. To articulate statistical considerations for a reasonable sample size to minimize the chance of imbalance, and, to highlight the importance of replicating the subgroup finding in independent studies.

Methods Four case examples reflecting recent regulatory experiences are used to underscore the problems with convenience samples. Probability of imbalance for a pre-specified subgroup is provided to elucidate sample size needed to minimize the chance of imbalance. We use an example drug development to highlight the level of scientific rigor needed, with evidence replicated for a pre-specified subgroup claim.

Results The convenience samples evaluated ranged from 18% to 38% of the intent-to-treat samples with sample size ranging from 100 to 5000 patients per arm. The baseline imbalance can occur with probability higher than 25%. Mild to moderate multiple confounders yielding the same directional bias in favor of the treated group can make treatment group incomparable at baseline and result in a false positive conclusion that there is a treatment difference. Conversely, if the same directional bias favors the placebo group or there is loss in design efficiency, the type II error can increase substantially.

Limitations Pre-specification of a genomic subgroup hypothesis is useful only for some degree of type I error control.

Conclusion Complete ascertainment of genomic samples in a randomized controlled trial should be the first step to explore if a favorable genomic patient subgroup suggests a treatment effect when there is no clear prior knowledge and understanding about how the mechanism of a drug target affects the clinical outcome of interest. When stratified randomization based on genomic biomarker status cannot be implemented in designing a pharmacogenomics confirmatory clinical trial, if there is one genomic biomarker prognostic for clinical response, as a general rule of thumb, a sample size of at least 100 patients may be needed to be considered for the lower prevalence genomic subgroup to minimize the chance of an imbalance of 20% or more difference in the prevalence of the genomic marker. The sample size may need to be at least 150, 350, and 1350, respectively, if an imbalance of 15%, 10% and 5% difference is of concern. Clinical Trials 2010; 7: 525—536. http://ctj.sagepub.com

Background Large comparative clinical trials usually target a wide-range of patients population in which subgroups exist according to certain patients’ characteristics. Often, scientific knowledge or existing empirical data support the assumption that patients’ improvement is larger among certain subgroups than others. Such information can be used to design a more cost-effective clinical trial.

Purpose The goal of the article is to use such information to design a more cost-effective clinical trial.

Methods A two-stage sample-enrichment design strategy is proposed that begins with enrollment from certain subgroup of patients and allows the trial to be terminated for futility in that subgroup.

Results Simulation studies show that the two-stage sample-enrichment strategy is cost-effective if indeed the null hypothesis of no treatment improvement is true, as also so illustrated with data from a completed trial of calcium to prevent preeclampsia.

Limitations Feasibility of the proposed enrichment design relies on the knowledge prior to the start of the trial that certain patients can benefit more than others from the treatment. Prolonged accrual and longer-waited outcomes may hinder utilization of the proposed design.

Conclusions The two-stage sample-enrichment approach borrows strength from treatment heterogeneity among target patients in a large-scale comparative clinical trial, and is more cost-effective if the treatment arms are indeed of no difference. Clinical Trials 2010; 7: 537—545. http://ctj.sagepub.com

Background Targeted therapies are becoming increasingly important for the treatment of various diseases. Biomarkers are a critical component of a targeted therapy as they can be used to identify patients who are more likely to benefit from a treatment. Targeted therapies, however, have created major challenges in the design, conduct, and analysis of clinical trials. In traditional clinical trials, treatment effects for various biomarkers are typically evaluated in an exploratory fashion and only limited information about the predictive values of biomarkers obtained.

Purpose New study designs are required, which effectively evaluate both the diagnostic and the therapeutic implication of biomarkers.

Methods The Bayesian approach provides a useful framework for optimizing the clinical trial design by directly integrating information about biomarkers and clinical outcomes as they become available. We propose a Bayesian covariate-adjusted response-adaptive randomization design, which utilizes individual biomarker profiles and patient’s clinical outcomes as they become available during the course of the trial, to assign the most efficacious treatment to individual patients. Predictive biomarker subgroups are determined adaptively using a partial least squares regression approach.

Results A series of simulation studies were conducted to examine the operating characteristics of the proposed study design. The simulation studies show that the proposed design efficiently identifies patients who benefit most from a targeted therapy and that there are substantial savings in the sample size requirements when compared to alternative designs.

Limitations The design does not control for the type I error in the traditional sense and a positive result should be confirmed by conducting an independent phase III study focusing on the selected biomarker profile groups.

Conclusions We conclude that the proposed design may serve a useful role in the early efficacy phase of targeted therapy development. Clinical Trials 2010; 7: 546—556. http://ctj.sagepub.com

Background The rapid advance of molecular genetic technology and of molecular diagnostics companies have set the stage for a new era in personalized treatments. Biomarkers such as gene expressions may be integrated into the anatomically based tumor-node-metastasis staging system to provide information for risk stratification and treatment selection. With the assumption that preliminary results show evidence that a biomarker has predictive value, the marker-based designs are geared to assess the purported predictive value in a clinical trial.

Purpose In this article, we compared the efficiency of the traditional design, which does not involve a biomarker, to several alternative designs in terms of the sample size required in each trial.

Methods We first derived the variance formulas for the two-sample t-tests under the various designs when the biomarker assay is imperfect, and then conducted numerical and simulation studies to evaluate the performance of the various designs. Results Based on numerical and simulation studies, we conclude that the marker-based strategy designs are less efficient than the traditional design in general. Since the biomarker assay is imperfect in a realistic setting, the estimated sample size for each alternative design is influenced by the sensitivity and specificity of the assay and the prevalence of the biomarker in the population of interest as well as the parameters involved in a standard sample size calculation.

Limitations Due to limitations of a simulation study, it is not clear whether our results can be generalized to other parameter settings that are different from the ones used in the simulation study.

Conclusions The marker-based strategy designs are less efficient than the traditional design in general. If there is no treatment effect among marker-negative patients, it is still feasible to use the marker-based strategy design I if the assay sensitivity is high. If the treatment effect among marker-negative patients is half of the effect among marker-positive patients, the marker prevalence must be relatively high and the sensitivity of the assay must be very high for the marker-based strategy design I to approximate the efficiency of the traditional design. The efficiency of the marker-based strategy design II relative to the traditional design is low in all scenarios considered under the current study. Clinical Trials 2010; 7: 557—566. http://ctj.sagepub.com

Background With the advent of targeted therapies, biomarkers provide a promising means of individualizing therapy through an integrated approach to prediction using the genetic makeup of the disease and the genotype of the patient. Biomarker validation has therefore become a central topic of discussion in the field of medicine, primarily due to the changing landscape of therapies for treatment of a disease and these therapies purported mechanism(s) of action.

Purpose In this report, we discuss the merits and limitations of some of the clinical trial designs for predictive biomarker validation using examples from ongoing or completed clinical trials.

Methods The designs are broadly classified as retrospective (i.e., using data from previously well-conducted randomized controlled trials (RCT)) versus prospective (enrichment or targeted, unselected or all-comers, hybrid, and adaptive analysis). We discuss some of these designs in the context of real trials.

Results Well-designed retrospective analysis of prospective RCT can bring forward effective treatments to marker defined subgroup of patients in a timely manner. An example is the KRAS gene status in colorectal cancer — the benefit from cetuximab and panitumumab was demonstrated to be restricted to patients with wild type status based on prospectively specified analyses using data from previously conducted RCTs. Prospective enrichment designs are appropriate when compelling preliminary evidence suggests that not all patients will benefit from the study treatment under consideration; however, this may sometimes leave questions unanswered. An example is the established benefit of trastuzumab as adjuvant therapy for breast cancer; a clear definition of HER2-positivity and the assay reproducibility have, however, remained unanswered. An all-comers design is optimal where preliminary evidence regarding treatment benefit and assay reproducibility is uncertain (e.g., EGFR expression and tyrosine kinase inhibitors in lung cancer), or to identify the most effective therapy from a panel of regimens (e.g., chemotherapy options in breast cancer).

Limitations The designs discussed here rest on the assumption that the technical feasibility, assay performance metrics, and the logistics of specimen collection are well established and that initial results demonstrate promise with regard to the predictive ability of the marker(s).

Conclusions The choice of a clinical trial design is driven by a combination of scientific, clinical, statistical, and ethical considerations. There is no one size fits all solution to predictive biomarker validation. Clinical Trials 2010; 7: 567—573. http://ctj.sagepub.com

Background The genomics revolution is still in its infancy, and there is much to learn about how to transform biological knowledge into useful medicines to further public health. At the bedside, we are asking how and why individual patients respond to different drug treatments in different ways. In addition to genetic mechanisms, there are many clinical markers (e.g. medical history, disease severity) as well as social/environmental factors (e.g. smoking habits) that can be used to identify who may or may not respond to treatment.

Purpose This issue has some considerable statistical complexity, and different approaches to the analysis of clinical trials may yield more interesting insights into the problem. Novel applications of statistical methods will be discussed, and examples will be used to demonstrate sub-group identification.

Methods In order to evaluate many potential predictors of response, we use recursive partitioning methods to identify predictor variables and their cut-off values to define sub-groups of patients with differential treatment response. Validation of this variable/model selection approach was done using independent data from other clinical trials.

Results In one example, a classification tree was developed using baseline measures to define important sub-groups of patients that responded much better than the overall mean response in the study. In a second example, a classification tree was built based on measures of response early in treatment to predict longer-term responders and nonresponders.

Limitation Classification algorithms can be prone to over-fitting, and validation of results is an important consideration. Obviously, analyses are limited by the available predictor variables.

Conclusions Using classification trees proved to be very useful in evaluating large numbers of potential predictors to find sub-groups of patients with exceptional response. The method is easy to use, and clinicians can easily interpret and implement results. This approach can be helpful in tailoring treatments to individual patients. Clinical Trials 2010; 7: 574—583. http://ctj.sagepub.com

Background With better understanding of the disease’s etiology and mechanism, many targeted agents are being developed to tackle the root cause of problems, hoping to offer more effective and less toxic therapies. Targeted agents, however, do not work for everyone. Hence, the development of target agents requires the evaluation of prognostic and predictive markers. In addition, upon the identification of each patient’s marker profile, it is desirable to treat patients with best available treatments in the clinical trial accordingly.

Methods Many designs have recently been proposed for the development of targeted agents. These include the simple randomization design, marker stratified design, marker strategy design, efficient targeted design, etc. In contrast to the frequentist designs with equal randomization, we propose novel Bayesian adaptive randomization designs that allow evaluating treatments and markers simultaneously, while providing more patients with effective treatments according to the patients’ marker profiles. Early stopping rules can be implemented to increase the efficiency of the designs.

Results Through simulations, the operating characteristics of different designs are compared and contrasted. By carefully choosing the design parameters, types I and II errors can be controlled for Bayesian designs. By incorporating adaptive randomization and early stopping rules, the proposed designs incorporate rational learning from the interim data to make informed decisions. Bayesian design also provides a formal way to incorporate relevant prior information. Compared with previously published designs, the proposed design can be more efficient, more ethical, and is also more flexible in the study conduct.

Limitations Response adaptive randomization requires the response to be assessed in a relatively short time period. The infrastructure must be set up to allow timely and more frequent monitoring of interim results.

Conclusion Bayesian adaptive randomization designs are distinctively suitable for the development of multiple targeted agents with multiple biomarkers. Clinical Trials 2010; 7: 584—596. http://ctj.sagepub.com

Background The Clarification of Optimal Anticoagulation through Genetics (COAG) trial is a large, multicenter, double-blinded, randomized trial to determine whether use of a genotype-guided dosing algorithm (using clinical and genetic information) to initiate warfarin treatment will improve anticoagulation status when compared to a dosing algorithm using only clinical information.

Purpose This article describes prospective alpha allocation and balanced alpha allocation for the design of the COAG trial.

Methods The trial involves two possibly heterogeneous populations, which can be distinguished by the difference in warfarin dose as predicted by the two algorithms. A statistical approach is detailed, which allows an overall comparison as well as a comparison of the primary endpoint in the subgroup for which sufficiently different doses are predicted by the two algorithms. Methods of allocating alpha for these analyses are given — a prospective alpha allocation and allocating alpha so that the two analyses have equal power, which we call a ‘balanced alpha allocation.’

Results We show how to include an analysis of the primary endpoint in a subgroup as a co-primary analysis. Power can be improved by incorporating the correlation between the overall and subgroup analyses in a prospective alpha allocation approach. Balanced alpha allocation for the full cohort and subgroup tests to achieve the same desired power for both of the primary analyses is discussed in detail.

Limitations In the COAG trial, it is impractical to stratify the randomization on subgroup membership because genetic information may not be available at the time of randomization. If imbalances in the treatment arms in the subgroup are found, they will need to be addressed.

Conclusions The design of the COAG trial assures that the subgroup in which the largest treatment difference is expected is elevated to a co-primary analysis. Incorporating the correlation between the full cohort and the subgroup analyses provides an improvement in power for the subgroup comparison, and further improvement may be achieved via a balanced alpha allocation approach when the parameters involved in the sample size calculation are reasonably well estimated. Clinical Trials 2010; 7: 597—604. http://ctj.sagepub.com