Ying Yuan, PhD

http://www.trialdesign.org

An Integrated Platform for Designing Clinical Trials

 

To facilitate researchers and practitioners to adopt novel adaptive trial designs in practice, we have launched website http://www.trialdesign.org to provide easy-to-use Apps and software for implementing novel phase I, II, I-II clinical trial designs. Our goal is to improve patient care by enhancing the safety, efficiency and success rate of clinical trials through the use of novel adaptive clinical trial designs.

Clinical Trial Design

Recommended Reading

 

Highly recommended reference to learn phase I trial designs !

Zhou, H., Yuan, Y. and Nie, L. (2018) Accuracy, Safety and Reliability of Novel Phase I Trial Designs, Clinical Cancer Research, 24, 4357-4364. (A comprehensive review and comparison of novel phase I trial designs. The results show that the BOIN is simple and has outstanding performance. )

BLAST: Bayesian Latent Subgroup Design for Basket Trials Accounting for Patient Heterogeneity

Although patients enrolled in the basket trial have the same molecular aberration, it is common for the targeted agent to be effective for patients with some types of cancer, but not others. We propose a Bayesian LAtent Subgroup Trial (BLAST) design to accommodate such treatment heterogeneity across cancer types. The innovation of the BLAST design is that it adaptively clusters cancer types within a basket trial into a responsive subgroup and a non-response subgroup, and then uses Bayesian hierarchical model to borrow information within each of the subgroups. Compared to standard Bayesian hierarchical model approach, BLAST yields not only substantially higher power, but also much better controlled type I error rate.

The R code to implement the BLAST design.

Chu Y. and Yuan Y. (2018) BLAST: Bayesian Latent Subgroup Design for Basket Trials Accounting for Patient Heterogeneity, Journal of the Royal Statistical Society: Series C , 67, 723-740.

CBHM: A Bayesian Basket Trial Design Using a Calibrated Bayesian Hierarchical Model

Bayesian hierarchical modeling has been proposed to adaptively borrow information across cancer types to improve the statistical power of basket trials. Although conceptually attractive, research has shown that Bayesian hierarchical models cannot appropriately determine the degree of information borrowing and may lead to substantially inflated type I error rates. We propose a novel calibrated Bayesian hierarchical model approach to evaluate the treatment effect in basket trials.The simulation study shows that our method has substantially better controlled type I error rates than the Bayesian hierarchical model.

The R code to implement the Calibrated Bayesian Hierarchical Model (CBHM) design.

Chu Y. and Yuan Y. (2018) A Bayesian Basket Trial Design Using a Calibrated Bayesian Hierarchical Model, Clinical Trials , 67, 723-740.

Bayesian Optimal Interval (BOIN) Design for Phase I Single-Agent and Drug-Combination Clinical Trials

The Bayesian optimal interval (BOIN) design is a novel phase I clinical trial design for finding the maximum tolerated dose (MTD). It can be used to design both single-agent and drug-combination trials. The BOIN design is motivated by the top priority and concern of clinicians when testing a new drug, which is to effectively treat patients and minimize the chance of exposing them to subtherapeutic or overly toxic doses. The prominent advantage of the BOIN design is that it achieves simplicity and superior performance at the same time. The BOIN design is algorithm based and can be implemented in a simple way similar to the traditional 3+3 design. The BOIN design yields average performance comparable to the continual reassessment method (CRM) in terms of selecting the MTD, but has a lower risk of assigning patients to subtherapeutic or overly toxic doses.

Two types of software are avaialble to implement the BOIN design

1) R package

R package "BOIN" is freely available from CRAN, instruction to install BOIN package in R .

The R manual and the statistical tutorial to use the BOIN design for single-agent and drug-combination trials.

2) Windows desktop program

Windows desktop program with GUI for implementing the BOIN design. This program automatically generates the design protocol.

Templates for protocol preparation

Single-agent trial and drug-combination trial.

Reference:

Zhou, H., Yuan, Y. and Nie, L. (2018) Accuracy, Safety and Reliability of Novel Phase I Trial Designs, Clinical Cancer Research, 24, 74921-4930. (A comprenhensive review and comparison of novel phase I trial designs. The results show that the BOIN is simple and has outstanding performance. Highly Recommended Reference! )

Yuan Y., Hess K.R., Hilsenbeck S.G. and Gilbert M.R. (2016) Bayesian Optimal Interval Design: A Simple and Well-performing Design for Phase I Oncology Trials, Clinical Cancer Research , 22 , 4291-4301.

Liu S. and Yuan Y. (2015) Bayesian Optimal Interval Designs for Phase I Clinical Trials, Journal of the Royal Statistical Society: Series C , 64, 507-523.

Lin R. and Yin G. (2017) Bayesian Optimal Interval Designs for Dose Finding in Drug-combination Trials, Statistical Methods in Medical Research, 26(5):2155-2167.

Keyboard: A Novel Bayesian Toxicity Probability Interval Design for Phase I Clinical Trials

The keyboard design provides a useful upgrade to the modified toxicity probability interval (mTPI) design, with a substantially lower risk of overdosing patients and the better precision to identify the maximum tolerated dose (MTD). The keyboard design is an intuitive Bayesian design that conducts dose escalation and de-escalation based on whether the strongest key, defined as the dosing interval that most likely contains the current dose, is below or above the target dosing interval. The keyboard design can be implemented in a simple way, similar to the traditional 3+3 design, but provides more flexibility for choosing the target toxicity rate and cohort size.

The Shiny App to implement the keyboard design.

Reference: Yan, F., Mandrekar, J.S. and Yuan, Y. (2017) Keyboard: A Novel Bayesian Toxicity Probability Interval Design for Phase I Clinical Trials . Clinical Cancer Research, 23 , 3994-4003.

Adaptive Dose Modification (ADM) for Phase I Clinical Trials

Most phase I dose-finding methods in oncology aim to find the maximum-tolerated dose (MTD) from a set of prespecified doses. However, in practice, due to a lack of understanding of the true dose-toxicity relationship, it is likely that none of these prespecified doses is equal or reasonably close to the true MTD. To handle this issue, we propose an adaptive dose modification (ADM) method that can be coupled with any existing dose-finding method to adaptively modify the dose, when it is needed, during the course of dose finding. To reflect clinical practice, we divide the toxicity probability into three regions: underdosing, acceptable and overdosing regions. We adaptively add a new dose whenever the observed data suggest that none of the investigational doses is likely to be located in the acceptable region.

The R code to implement the ADM.

Reference: Chu, Y., Pan, H. and Yuan, Y. (2016) Adaptive Dose Modification for Phase I Clinical Trials. Statistics in Medicine, in press.

Phase I/II Trial Designs to Identify Optimal Biological Dose for Molecularly Targeted Agents

Traditionally, the purpose of a dose-finding design in cancer is to find the maximum tolerated dose (MTD) based solely on toxicity. However, for molecularly targeted agents (MTAs), little toxicity may arise within the therapeutic dose range and the dose-efficacy curves may not be monotonic. We propose three adaptive dose-finding designs to find the optimal biological dose (OBD), which is defined as the lowest dose with the highest rate of efficacy while safe. The first proposed design is parametric and assumes a logistic dose-efficacy curve for dose finding; the second design is nonparametric and uses the isotonic regression to identify the optimal biological dose; and the third design has the spirit of a ``semiparametric" approach by assuming a logistic model only locally around the current dose. We recommend the nonparametric and semiparametirc designs for practical use because of their robust performance.

The R code and the description of the proposed designs

Reference: Zang, Y., Lee, J. and Yuan, Y. (2014) Adaptive Designs for Identifying Optimal Biological Dose for Molecularly Targeted Agents, Clinical Trials, 11, 319-327.

A Bayesian Dose-finding Design for Drug Combination Clinical Trials Based on the Logistic Model

Many trial designs have been proposed to find the maximum tolerated dose for drug combination trials. These designs rely on complicated statisticalmodels that typically are not familiar to clinicians, and are rarely used in practice. The aim of this paper is to propose a Bayesian dose-finding design for drug combination trials based on standard logistic regression, which are relative simple and more familiar to clinicians. Numerical studies show that the proposed design is competitive and outperforms some existing designs.

The R package "dfcomb" and the manual of using the proposed design

Reference: Riviere, M., Yuan, Y., Dubois, F. and Zohar, S. (2014) A Bayesian dose-finding design for drug combination clinical trials based on the logistic model. Pharmaceutical Statistics, 13, 247-257.

A Bayesian Design for Phase II Clinical Trials with Delayed Responses Based on Multiple Imputation

Interim monitoring is routinely conducted in phase II clinical trials to terminate the trial early if the experimental treatment is futile. Interim monitoring requires that patients' responses be ascertained shortly after the initiation of treatment so that the outcomes are known by the time the interim decision must be made. However, in some cases, response outcomes require a long time to be assessed, which causes difficulties for interim monitoring. To address this issue, we propose a Bayesian trial design to allow for continuously monitoring phase II clinical trials in the presence of delayed responses. We treat the delayed responses as missing data and handle them using a multiple imputation approach. Extensive simulations show that the proposed design yields desirable operating characteristics and dramatically reduces the trial duration. The design provides a generalization of the Bayesian phase II design proposed by Thall and Simon (1994, Biometrics). In the case that there is no delayed responses, our design is equivalent to Thall and Simon's design.

The software package to conduct a phase II trial design using the proposed design. User mannual is included in the package.

Reference: Cai C., Liu S. and Yuan Y. (2014) A Bayesian Design for Phase II Clinical Trials with Delayed Responses Based on Multiple Imputation s, Statistics in Medicine , 33, 4017-4028.

Bayesian Data Augmentation Continual Reassessment Method (DA-CRM) for Phase I Clinical Trials with Delayed Toxicity

A major practical impediment when implementing adaptive dose-finding designs is that the toxicity outcome used by the decision rules may not be observed shortly after the initiation of the treatment. To address this issue, we propose the data augmentation continual reassessment method (DA-CRM) for dose finding. By naturally treating the unobserved toxicities as missing data, we show that such missing data are nonignorable in the sense that the missingness depends on the unobserved outcomes. The Bayesian data augmentation approach is used to sample both the missing data and model parameters from their posterior full conditional distributions. Simulation study shows that the DA-CRM outperformed the existing design (TITE-CRM): treating patients more safely and also selecting the maximum tolerated dose with a higher probability.

The user-friendly professional quality software is available for free download from the Department of Biostatistics at the MD Anderson Cancer Center.

Reference: Liu, S., Yin, G. and Yuan, Y. (2013) Bayesian Data Augmentation Dose Finding With Continual Reassessment Method and Delayed Toxicity. The Annals of Applied Statistics, 7, 2138-2156.

Bridging Continual Reassessment Method (B-CRM) for Phase I Clinical Trials in Different Ethnic Populations

Accumulating evidence shows that the conventional one-size-fits-all dose-finding paradigm is problematic when applied to different ethnic populations. Because of inter-ethnic heterogeneity, the dosage established in a landmark trial for a certain population may not be generalizable to a different ethnic population, and a follow-up bridge trial is often needed to find the maximum tolerated dose for the new population. We propose the bridging continual reassessment method (B-CRM) to facilitate dose finding for such follow-up bridge trials. The B-CRM borrows information from the landmark trial through a novel estimate of the dose-toxicity curve, and accommodates the inter-ethnic heterogeneity using the Bayesian model averaging approach. Extensive simulation studies show that the B-CRM has desirable operating characteristics with a high probability to select the target dose.

The R code to implement the proposed B-CRM design

Reference: Liu, S., Pan, H., Xia, J., Huang, Q and Yuan, Y. (2015) Bridging Continual Reassessment Method for Phase I Clinical Trials in Different Ethnic Populations, Statistics in Medicine, 34, 1681-1694.

Bayesian Model Averaging Continual Reassessment Method (BMA-CRM)

One practical issue of using the continual reassessment method (CRM) is that its operating characteristics may sensitive to the specification of skeleton (i.e., prespecified toxicity probabilities). By using the Bayesian model averaging methodology, this BMA-CRM design solves this issue and substantially improve the robustness of the CRM. The BMA-CRM includes the standard CRM as a special case when only one set of skeleton is specified. This design has been routinely used in the MD Anderson Cancer Center to replace the standard CRM.

The user-friendly professional quality software is available for free download from the Department of Biostatistics at the MD Anderson Cancer Center.

Reference: Yin, G. and Yuan, Y. (2009) Bayesian Model Averaging Continual Reassessment Method in Phase I Clinical Trials, Journal of American Statistical Association, 104, 954-968.

Bayesian Dose Finding for Drug Combination Clinical Trials using Copula Regression

This Bayesian copula-type dose-finding design searches for the MTD (maximum tolerated dose) for clinical trials combining two agents, each with several predefined dose levels. The code is written in C++ and precompiled under Windows.

The executable code and the tutorial.

Reference: Yin, G. and Yuan, Y. (2009). Bayesian Dose Finding in Oncology for Drug Combinations by Copula Regression. Journal of the Royal Statistical Society: Series C, 58, 211-224.

Bayesian Phase I/II Adaptively Randomized Design for Drug Combination Trials

We propose a new integrated phase I/II trial design to identify the most efficacious dose combination that also satisfies certain safety requirements for drug-combination trials. We first take a Bayesian copula-type model for dose finding in phase I. After identifying a set of admissible doses, we immediately move the entire set forward to phase II, in which a novel adaptive randomization scheme is used to favor assigning patients to more efficacious dose combination arms.

The executable code and the readme

Reference: Yuan, Y. and Yin, G. (2011) Bayesian Phase I/II Adaptively Randomized Oncology Trials with Combined Drugs. Annals of Applied Statistics, 5, 924-942.

Bayesian Hybrid Dose-Finding Design for Phase I Oncology Clinical Trials

We propose a Bayesian hybrid dose-finding method that inherits the robustness of model-free methods and the efficiency of model-based methods. In the Bayesian hypothesis testing framework, we adaptively switch between the model-based and model-free methods based on the toxicity information contained at the current dose as measured by the Bayes factor. The design is more robust than parametric model-based methods and more efficient than nonparametric model-free methods.

The simulation code written in R

Reference: Yuan, Y. and Yin, G. (2011) Bayesian Hybrid Dose-Finding Design in Phase I Oncology Clinical Trials, Statistics in Medicine, 30, 2098-2108.

Dose-response Curve Estimation: A Semiparametric Mixture Approach

In the estimation of a dose–response curve, parametric models are straightforward and efficient but subject to model misspecifications; nonparametric methods are robust but less efficient. As a compromise, we propose a semiparametric approach that combines the advantages of parametric and nonparametric curve estimates.

The simulation code written in R and the phase II trial data

Reference: Yuan, Y. and Yin, G. (2011) Dose-response Curve Estimation: A Semiparametric Mixture Approach. Biometrics, 67, 1543-1554.

Mediation Analysis

Bayesian Mediation Analysis

We propose Bayesian analysis of mediation effects. Compared with conventional frequentist mediation analysis, the Bayesian approach has several advantages. First, it allows researchers to incorporate prior information into the mediation analysis, thus potentially improving the efficiency of estimates. Second, under the Bayesian mediation analysis, inference is straightforward and exact, which makes it appealing for studies with small samples. Third, the Bayesian approach is conceptually simpler for multilevel mediation analysis.

Code written in R and readme

BUGS code: simplemed for single-level mediation model, and mlmed for multilevel mediation model

Data: firefighter for single-level mediation, and kennydata for multilevel mediation

Reference: Yuan, Y. and MacKinnon, D.P. (2009) Bayesian mediation analysis. Psychological Methods, 14, 301-322.

Robust Mediation Analysis

The standard mediation analysis methods typically assume that the error distribution is normal and homoscedastic. However, this assumption may be rarely met in practice, which can affect the validity of the mediation analysis. To address this problem, we propose robust mediation analysis based on median regression. Our approach is robust to various departures from the assumption of homoscedasticity and normality, including heavy-tailed, skewed, contaminated, and heteroscedastic distributions.

SAS Macro and R function for robust mediation analysis based on median regression

A simulated heavy-tailed dataset for illustration: heavytails.txt

Reference: Yuan, Y. and MacKinnon, D.P. (2014) Robust mediation analysis based on median regression. Psychological Methods, 19, 1-20.

Bayesian Dynamic Mediation Analysis

Most existing methods for mediation analysis assume that mediation is a stationary, time-invariant process, which overlooks the inherently dynamic nature of many human psychological processes and behavioral activities. We develope a Bayesian dynamic mediation model to describe and estimate the dynamic mediation effect, defined as a mediatoin effect that varies across time or another measurement domain. By taking the nonparametric approach, the proposed method is flexible and able to accommodate any shape of the relationship between time and mediation effects. By modeling mediation effect nonparametrically as a continuous function of time, our method provides a valuable tool to help researchers obtain a more complete understanding of the dynamic nature of the mediation process underlying psychological and behavioral phenomena.

R function for performing Bayesian dynamic mediation analysis, readme file for using the function.

A example dataset for illustration: exampledata.txt

Reference: Huang, J. and Yuan, Y. (2017) Bayesian Dynamic Mediation Analysis. Psychological Methods, 22, 667-686.

Miscellaneous

A Shrinkage Method for Testing the Hardy-Weinberg Equilibrium in Case-Control Studies

we propose a novel shrinkage test for assessing the Hardy-Weinberg equilibrium (HWE). The proposed shrinkage test yields higher statistical power than the likelihood ratio test (LRT) when the marker is independent of or weakly associated with the disease, and converges to the LRT when the marker is strongly associated with the disease. In addition, the proposed shrinkage test has a closed form and can be easily used to test the HWE for large datasets that result from genome-wide association studies.

Simulation code in R