Simulation setup Model parameters for simulations were tuned using repeated measurement data from multiple in-house completed studies and baseline data.

Slides:



Advertisements
Similar presentations
Industry Issues: Dataset Preparation for Time to Event Analysis Davis Gates Schering Plough Research Institute.
Advertisements

Phase II/III Design: Case Study
SAMPLE DESIGN: HOW MANY WILL BE IN THE SAMPLE—DESCRIPTIVE STUDIES ?
[1] MA4104 Business Statistics Spring 2008, Lecture 06 Process Monitoring Using Statistical Control Charts [ Examples Class ]
Bayesian posterior predictive probability - what do interim analyses mean for decision making? Oscar Della Pasqua & Gijs Santen Clinical Pharmacology Modelling.
Regression Analysis Once a linear relationship is defined, the independent variable can be used to forecast the dependent variable. Y ^ = bo + bX bo is.
1 Frank Miller, AstraZeneca, Södertälje Estimating the interesting part of a dose-effect curve: When is a Bayesian adaptive design useful? Frank Miller.
Topic 6: Introduction to Hypothesis Testing
Confidence Intervals © Scott Evans, Ph.D..
Comparison of Repeated Measures and Covariance Analysis for Pretest-Posttest Data -By Chunmei Zhou.
Introduction to Inference Estimating with Confidence Chapter 6.1.
Point and Confidence Interval Estimation of a Population Proportion, p
Stat 112: Lecture 9 Notes Homework 3: Due next Thursday
Sample size calculations
Sample Size Annie Herbert Medical Statistician Research & Development Support Unit Salford Royal Hospitals NHS Foundation Trust
The Mimix Command Reference Based Multiple Imputation For Sensitivity Analysis of Longitudinal Trials with Protocol Deviation Suzie Cro EMERGE.
Background to Adaptive Design Nigel Stallard Professor of Medical Statistics Director of Health Sciences Research Institute Warwick Medical School
ANOVA (Analysis of Variance) by Aziza Munir
Biostatistics Case Studies 2007 Peter D. Christenson Biostatistician Session 3: Incomplete Data in Longitudinal Studies.
Biostatistics Case Studies 2008 Peter D. Christenson Biostatistician Session 5: Choices for Longitudinal Data Analysis.
Statistics 1: tests and linear models. How to get started? Exploring data graphically: Scatterplot HistogramBoxplot.
1 An Interim Monitoring Approach for a Small Sample Size Incidence Density Problem By: Shane Rosanbalm Co-author: Dennis Wallace.
1 Statistics in Drug Development Mark Rothmann, Ph. D.* Division of Biometrics I Food and Drug Administration * The views expressed here are those of the.
Biostatistics in Practice Peter D. Christenson Biostatistician Session 6: Case Study.
1 Updates on Regulatory Requirements for Missing Data Ferran Torres, MD, PhD Hospital Clinic Barcelona Universitat Autònoma de Barcelona.
Biostatistics Case Studies 2006 Peter D. Christenson Biostatistician Session 4: An Alternative to Last-Observation-Carried-Forward:
Statistical planning and Sample size determination.
APA Results Section Results.
Relational Discord at Conclusion of Treatment Predicts Future Substance Use for Partnered Patients Wayne H. Denton, MD, PhD; Paul A. Nakonezny, PhD; Bryon.
1 Mohamed Alosh, Ph.D. Kathleen Fritsch, Ph.D. Shiowjen Lee, Ph.D. DBIII, OB, CDER, FDA Efficacy Evaluation in Acne Clinical Trials.
1 STA 617 – Chp10 Models for matched pairs Summary  Describing categorical random variable – chapter 1  Poisson for count data  Binomial for binary.
Irwin/McGraw-Hill © Andrew F. Siegel, 1997 and Methods and Applications CHAPTER 15 ANOVA : Testing for Differences among Many Samples, and Much.
Copyright © 2013, 2009, and 2007, Pearson Education, Inc. Chapter 10 Comparing Two Groups Section 10.1 Categorical Response: Comparing Two Proportions.
CHAPTER 27: One-Way Analysis of Variance: Comparing Several Means
Model-based dose selection for next dose- finding trial 1. Introduction Exploratory clinical development trials often include biomarkers or clinical readout.
Sample Size Determination
Chapter 13 Sampling distributions
 An exposure-response (E-R) analysis in oncology aims at describing the relationship between drug exposure and survival and in addition aims at comparing.
Session 6: Other Analysis Issues In this session, we consider various analysis issues that occur in practice: Incomplete Data: –Subjects drop-out, do not.
How does Biostatistics at Roche typically analyze longitudinal data
The Mixed Effects Model - Introduction In many situations, one of the factors of interest will have its levels chosen because they are of specific interest.
Date | Presenter Case Example: Bayesian Adaptive, Dose-Finding, Seamless Phase 2/3 Study of a Long-Acting Glucagon-Like Peptide-1 Analog (Dulaglutide)
1 Probability and Statistics Confidence Intervals.
Biostatistics Case Studies Peter D. Christenson Biostatistician Session 3: Missing Data in Longitudinal Studies.
Box and Whisker Plots Example: Comparing two samples.
1 Statistics 262: Intermediate Biostatistics Mixed models; Modeling change.
R&R Homework Statgraphics “Range Method”. DATA OperatorPartTrialMeasure B B B B B B326.5 B B B C
Copyright © 2008 Merck & Co., Inc., Whitehouse Station, New Jersey, USA All rights Reserved Pharmacokinetic/Pharmacodynamic (PK/PD) Analyses for Raltegravir.
Practical Solutions Analysing Continuous Data. 2 1)To produce the overall histogram you can use the options exactly as given. This results in the following.
Repeated measures: Approaches to Analysis
Data Analysis Patrice Koehl Department of Biological Sciences
Exposure adjustment in Risk-based monitoring in clinical trials with
Sample Size Determination
Aligning Estimands and Estimators – A Case Study Sept 13, 2018 Elena Polverejan Vladimir Dragalin Quantitative Sciences Janssen R&D, Johnson & Johnson.
Predictive Modeling for Patient Recruitment in Multicenter Trials
Background and Objectives
CHAPTER 29: Multiple Regression*
Monitoring rare events during an ongoing clinical trial
Risk ratios 12/6/ : Risk Ratios 12/6/2018 Risk ratios StatPrimer.
Joanna Romaniuk Quanticate, Warsaw, Poland
Longitudinal Analysis Beyond effect size
Frank Miller AstraZeneca, Södertälje, Sweden
An Introductory Tutorial
Predictive Performance of a Myelosuppression Model for Dose Individualization; Impact of Type and Amount of Information Provided Johan E. Wallin, Lena.
Combined predictor Selection for Multiple Clinical Outcomes Using PHREG Grisell Diaz-Ramirez.
What Do We Know About Estimators for the Treatment Policy Estimand
Handling Missing Not at Random Data for Safety Endpoint in the Multiple Dose Titration Clinical Pharmacology Trial Li Fan*, Tian Zhao, Patrick Larson Merck.
Yu Du, PhD Research Scientist Eli Lilly and Company
Daniel Li and L.J. Wei Harvard University
How Should We Select and Define Trial Estimands
Presentation transcript:

Simulation setup Model parameters for simulations were tuned using repeated measurement data from multiple in-house completed studies and baseline data from the motivating study to match the treatment effect and STD assumed in sample size calculation of the motivating study. Methods Introduction Longitudinal dose-response modelling as primary analysis of a clinical study Karin Nelander, Bengt Hamrén, Susanne Johansson, Magnus Åstrand Quantitative Clinical Pharmacology, Innovative Medicines and Early Development Biotech Unit, AstraZeneca, Gothenburg, Sweden ANOVA or analysis of covariance (ANCOVA) commonly used as primary analysis for clinical studies For studies with repeated measurements Mixed-Effect Model Repeated Measure (MMRM) often used as either primary or secondary analysis Drawback for both methods: No predictions for doses not tested in study No or low information gained from data at intermediate study visits Objective: To evaluate longitudinal dose-response modelling (here LDRM) of all measured data with aim of estimating treatment effect at end of study, both in terms of power and type 1 error for 1.A typical HbA1c study 2.A study with an end-point having higher within patient variability Clinical trial simulations were performed of a placebo controlled multi dose study with end of study HbA1c treatment effect as primary endpoint and for a similar study with an endpoint with higher within patients variability Simulations were performed in R (version 3.2.2, the R Foundation for Statistical Computing) Analyses were performed in SAS (version 9.3, SAS Institute Inc., Cary, NC, USA) For ANCOVA and MMRM proc MIXED was used For LDRM proc NLMIXED was used, with CI computed using the delta method (NLMIXED default) Last value carried forward was used for ANCOVA at interim, and in all ANCOVA analyses baseline measurements of the endpoint was used as covariate Unstructured correlation was used for MMRM 1000 simulated studies were used to evaluate power (portion of CI excluding 0) and precision (average length of CI) 10,000 simulated studies with treatment parameters set to 0 were used to evaluate type 1 error (proportion of CI excluding 0) Figure 3 Typical simulated studies at interim, red line is average per time point Conclusions Longitudinal dose-response modelling can contribute more informed decision making than using ANCOVA or MMRM, providing increased power and precision and sufficient control of type 1 error. In this setting ANCOVA appears to perform better than MMRM. The level of increased power and precision for LDRM is higher for endpoints with higher within individual variability. Supported by Figure 2 Hypothetical HbA1c over time during treatment Results Table 1 Simulation results using original parameters: The first row for each method represents analysis result after all patients have completed the study, and the second row represents analysis results at interim. Low doseHigh dose Power (%) Type 1 err. (%) Length CI (%) Power (%) Type 1 err. (%) Length CI (%) ANCOVA MMRM LDRM Table 2 Simulation results using inflated within individual error: The first row for each method represents analysis result after all patients have completed the study, and the second row represents analysis results at interim. Presented at The PAGE 2016 meeting, Lisbon Portugal, 7-10 th of June 2016 Low doseHigh dose Power (%) Type 1 err. (%) Length CI (%) Power (%) Type 1 err. (%) Length CI (%) ANCOVA MMRM LDRM Background on study motivating evaluation A 28 weeks HbA1c study designed to compare two doses to placebo was ongoing, however, with slow recruitment. The aim was to recruit 195 patients and with 105 recruited patients the study team wanted to explore the option of prematurely terminating the study: Would there be enough power to detect a treatment effect? Could alternative analysis methods increase power? Repeated measured data was simulated according to study protocol with HbA1c at baseline (week 0) and weeks 4, 12, 20, 28. Data for interim included 105 patients, 66 with full data and 39 with partial data (baseline and 1-3 on treatment visits). Final analysis included 195 patients all with full data. Figure 4 Box plot of estimated treatment effect in the case of HbA1c variability (the same patterns is seen with higher variability). Whiskers go out to the most extreme point within 1.5 times the inter quartile range. Observations outside of that range are indicated with a dot Simulate indirect response model for HbA1c Formation of HbA1c represented by a 0-order process governed by k in Elimination represented by a 1-order process controlled by k out Drug effect as inhibitory effect on k in Figure 5 Box plot of length of confidence interval for treatment effect in the case of HbA1c variability (A) and higher within variability (B). Whiskers and dots as for figure 4 The simulations show that the LDRM performs as good as or better than ANCOVA and MMRM (see tables 1 and 2, figures 4 and 5). LDRM provided somewhat better power for the full analysis of the typical HbA1c study. For the interim and the case of higher within patient variability the improvement was higher. The type 1 error was overall similar for the methods and all cases studied. HbA1c k in k out Drug effect as inhibitory on the input rate of HbA1c A B