Danni Yu Eli Lilly and Company (Tue)

Slides:



Advertisements
Similar presentations
Patient Selection Markers in Drug Development Programs
Advertisements

Matthew M. Riggs, Ph.D. metrum research group LLC
Katherine Jenny Thompson
Structural Equation Modeling. What is SEM Swiss Army Knife of Statistics Can replicate virtually any model from “canned” stats packages (some limitations.
Structural Equation Modeling Using Mplus Chongming Yang Research Support Center FHSS College.
Structural Equation Modeling (SEM) Niina Kotamäki.
Federal Institute for Drugs and Medical Devices | The Farm is a Federal Institute within the portfolio of the Federal Ministry of Health (Germany) How.
Statistical Analysis for Two-stage Seamless Design with Different Study Endpoints Shein-Chung Chow, Duke U, Durham, NC, USA Qingshu Lu, U of Science and.
Bayesian posterior predictive probability - what do interim analyses mean for decision making? Oscar Della Pasqua & Gijs Santen Clinical Pharmacology Modelling.
Statistical Issues in Incorporating and Testing Biomarkers in Phase III Clinical Trials FDA/Industry Workshop; September 29, 2006 Daniel Sargent, PhD Sumithra.
HSRP 734: Advanced Statistical Methods July 24, 2008.
Estimation and Reporting of Heterogeneity of Treatment Effects in Observational Comparative Effectiveness Research Prepared for: Agency for Healthcare.
Recursive Partitioning Method on Survival Outcomes for Personalized Medicine 2nd International Conference on Predictive, Preventive and Personalized Medicine.
Multivariate Data Analysis Chapter 4 – Multiple Regression.
“Ghost Chasing”: Demystifying Latent Variables and SEM
Re-Examination of the Design of Early Clinical Trials for Molecularly Targeted Drugs Richard Simon, D.Sc. National Cancer Institute linus.nci.nih.gov/brb.
Structural Equation Modeling Intro to SEM Psy 524 Ainsworth.
Biomedical research methods. What are biomedical research methods? An integrated approach using chemical, mathematical and computer simulations, in vitro.
Thoughts on Biomarker Discovery and Validation Karla Ballman, Ph.D. Division of Biostatistics October 29, 2007.
Assessing Survival: Cox Proportional Hazards Model Peter T. Donnan Professor of Epidemiology and Biostatistics Statistics for Health Research.
Studying treatment of suicidal ideation & attempts: Designs, Statistical Analysis, and Methodological Considerations Jill M. Harkavy-Friedman, Ph.D.
(a.k.a: The statistical bare minimum I should take along from STAT 101)
Hydrologic Modeling: Verification, Validation, Calibration, and Sensitivity Analysis Fritz R. Fiedler, P.E., Ph.D.
DSS Modeling Current trends – Multidimensional analysis (modeling) A modeling method that involves data analysis in several dimensions – Influence diagram.
Background to Adaptive Design Nigel Stallard Professor of Medical Statistics Director of Health Sciences Research Institute Warwick Medical School
Assessing Survival: Cox Proportional Hazards Model
Biostatistics Case Studies 2007 Peter D. Christenson Biostatistician Session 3: Incomplete Data in Longitudinal Studies.
Regression. Population Covariance and Correlation.
1 1 Using Marginal Structural Model to Estimate and Adjust for Causal Effect of Post- discontinuation Chemotherapy on Survival in Cancer Trials Y. Wang,
Lesson Multiple Regression Models. Objectives Obtain the correlation matrix Use technology to find a multiple regression equation Interpret the.
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.
Proposal for End-of-Phase 2A (EOP2A) Meetings Advisory Committee for Pharmaceutical Sciences Clinical Pharmacology Subcommittee November 17-18, 2003 Lawrence.
Measurement Models: Exploratory and Confirmatory Factor Analysis James G. Anderson, Ph.D. Purdue University.
Measurement Models: Identification and Estimation James G. Anderson, Ph.D. Purdue University.
Environmental Modeling Basic Testing Methods - Statistics III.
Innovative Paths to Better Medicines Design Considerations in Molecular Biomarker Discovery Studies Doris Damian and Robert McBurney June 6, 2007.
Biostatistics Case Studies Peter D. Christenson Biostatistician Session 3: Missing Data in Longitudinal Studies.
Clinical Trials - PHASE II. Introduction  Important part of drug discovery process  Why important??  Therapeutic exploratory trial  First time in.
Biomarkers and Surrogates: Underpinnings and Clinical Trial Applications ASENT Annual Meeting March 2009 Marc K. Walton, M.D., Ph.D. Associate Director.
Chapter 17 STRUCTURAL EQUATION MODELING. Structural Equation Modeling (SEM)  Relatively new statistical technique used to test theoretical or causal.
Strategies for Metabolomic Data Analysis Dmitry Grapov, PhD.
Confidential and Proprietary Business Information. For Internal Use Only. Statistical modeling of tumor regrowth experiment in xenograft studies May 18.
Research and Evaluation Methodology Program College of Education A comparison of methods for imputation of missing covariate data prior to propensity score.
Simulation setup Model parameters for simulations were tuned using repeated measurement data from multiple in-house completed studies and baseline data.
Covariance/ Correlation
Sample Size Determination
Project 5 Data Mining & Structural Equation Modeling
Cognitive Biomarker of MS
CJT 765: Structural Equation Modeling
CH 5: Multivariate Methods
5. BRIDGING THE GAP between Pharmacometricians and Biostatisticians
Covariance/ Correlation
A practical trial design for optimising treatment duration
Dose-finding designs incorporating toxicity data from multiple treatment cycles and continuous efficacy outcome Sumithra J. Mandrekar Mayo Clinic Invited.
Multiple Regression Models
Longitudinal Analysis Beyond effect size
PREDICT.
Covariance/ Correlation
Non response and missing data in longitudinal surveys
Handling Missing Not at Random Data for Safety Endpoint in the Multiple Dose Titration Clinical Pharmacology Trial Li Fan*, Tian Zhao, Patrick Larson Merck.
Clinical prediction models
Coiffier B et al. Proc ASH 2011;Abstract 265.
Enhancing Causal Inference in Observational Studies
Biomarkers as Endpoints
Enhancing Causal Inference in Observational Studies
Björn Bornkamp, Georgina Bermann
Structural Equation Modeling
Detecting Treatment by Biomarker Interaction with Binary Endpoints
Yu Du, PhD Research Scientist Eli Lilly and Company
Hong Zhang, Judong Shen & Devan V. Mehrotra
Presentation transcript:

Danni Yu Eli Lilly and Company 2018-07-31 (Tue) Discovering biomarkers jointly modeled with multiple efficacy variables in early phase clinical trials Danni Yu Eli Lilly and Company 2018-07-31 (Tue) Use a minimum font size of 14 points if page setup is ‘on-screen show (16:9)’ E-Poster Number 28

Background “Biomarker is a characteristic that is objectively measured and evaluated as an indicator of normal biological process, pathogenic processes, or biological responses to a therapeutic intervention.” Drug-related biomarkers include, but not limited to, baseline predictive biomarkers for patient stratification, efficacy and toxicity; longitudinal pharmacodynamics biomarkers for target engagement evidence, dose decision, or early response detection. Biomarkers studies need to be involved in all the clinical trial stages. Preclinical Lab Studies Phase I Human Safety Phase I/II Safety Confirm Phase III Efficacy & Safety Sample size of phase I trials for biomarker identification is typically small, e.g. n=20~50. E-Poster Number 28 1 of 5

Problem and proposed solution Sensitivity can be low due to the potential bias from small group systematic errors. Traditional approach with one efficacy variable in the model makes decision difficult if it is on-the-boundary significance. Jointly considering tumor change and survival time may help. Integrative approach is required. Observe difference in tumor change between two baseline biomarker levels, but marginal significance in KM. How to confirm in the next step? Structural Equation Modeling (SEM) fits networks of constructs to data and can impute relationships between unobserved latent variables from observed variables. It combines exploratory factor analysis and multiple multi-variates regressions (Ullman 2001) through structural equations. E-Poster Number 28 2 of 5

Simulation and analysis method Simulate the biological and clinical association from predictive markers to efficacy variables. Setup structural equations Observed variables Baseline markers: bmk.t1, bmk.fc1, bmk.fn1… ChangeFromBaseline markers: pmk.t1, pmk.f1… Tumor change ‘tc’, survival time ‘tte’ and ‘cnsr’ Latent variables ‘eff =~ tc + tte’, ‘bsl =~ bmk.t1+bmk.fc1+bmk.fn1+…’ and ‘log =~ pmk.t1+pmk.f1+…’ Regressions ‘eff ~ bsl + log’ if only one marker in bsl or log, then the latent variable is removed. Residual covariance ‘tte ~~ tc’ assume the correlation between tumor change and survival time is nonzero. Predictive marker: bmk.t1~Ber(prevalence). Tumor change: tc~Gamma(sp, rt) where {rt, sp}= F[bmk.t1]. Survival time: tte~Weibull(sp, sc) where sp=1 and sc=G[tc, bmk.t1] if death, max(G[tc,bmk.t1]) if censored. ChangeFromBaseline markers: pmk.t~N(mu,sd) where {mu, sd}=Z[tc]. Markers not related with any efficacy variables: bmk.fc~Ber(p) where p~U(0.2,0.8), bmk.fn and pmk.f~N(m,s) where m~N(0,1) and s~Chisq(1). sp: shape parameter; rt: rate parameter; sc: scale parameter E-Poster Number 28 3 of 5

Single Y variable pval=0.064 Results: The proposed method and the BEACH tool providing GUI for users help identify meaningful baseline biomarkers. Single Y variable pval=0.064 Data Simulation in BEACH SEM analysis in BEACH: bmk.t1 is identified. tc tte eff Predictive bmk The predictive biomarker ‘bmk.t1’ was not significant in separate models but meaningful association was identified while structuring ‘tc’ and ‘tte’ in the joint model. n=30 E-Poster Number 28 4 of 5

Conclusion & Future work Reference & Acknowledgement SEM can be used to jointly analyze biomarkers in multiple multi-variates models. It enables us to increase the sensitivity of identifying biomarkers with joint efficacy approach. A user-interface GUI tool coded in R/Shiny BEACH platform (Yu 2018) is created to implement the analyses dynamically and interactively. The proposed simulation method has a potential to estimate the minimum sample size with controlled FDR. Real clinical data with more efficacy variables can be applied to further improve this method with adjusted structural models. Yu, D. and Man, M. (2018). BEACH: an open platform for building interactive and automatic analysis powered by R/Shiny. AD05 PharmaSUG. Ullman, J. B. (2001). Structural equation modeling. In B. G. Tabachnick & L. S. Fidell (Eds.), Using multivariate statistics (4th ed.). Needham Heights, MA: Allyn & Bacon. Yves, R. (2012). lavaan: An R package for structural equation modeling. Journal of statistical software vol. 048 i02. Acknowledge Dr. Yanping Wang and Dr. Michael Man for their advice. E-Poster Number 28 5 of 5