Use of Real-World Data in Clinical Drug Development Julie Cong, Susan Wang Boehringer Ingelheim, China JSM 2018

Real-world data (RWD) and Evidence (RWE) - FDA Voice blog by Dr. Scott Gottlieb on July 2018

Regulatory use of RWD/RWE Post-approval benefits and risks monitoring through the FDA’s Sentinel System and the National Evaluation System for health Technology (NEST) Label expansion New $100M medical data enterprise proposal to build a modern system relying on Electronic Health Records (EHR). Innovative statistical methods needed e.g. Bayesian and propensity score methods - FDA Voice blog by Dr. Scott Gottlieb on July 2018

Outline Overview of how real-world data can be used to optimize clinical drug development Hypothesis generation Patient availability Site feasibility Potentially minimize control arm Tools/technologies needed Summary and take-home message

Generate a more feasible and clinical relevant hypothesis Treatment effect size Historical trials, publications RWE can rapidly explore and test hypotheses across diverse datasets. Indication and severity of disease of interest Outcomes Unmet needs Treatment journey

RWE to optimize trial design Information from RWE regarding Real-world treatment length and adherence data Patient retention in RCT Frequency of visits, trial procedures to minimize trial burden Reduce protocol amendment and interaction with IRB

Patient availability and feasibility Assess feasibility based on inclusion and exclusion criteria against available RWD when writing trial protocol Availability of target patients in candidate sites Look for potential sites with more patients

Example of locating target patients by geographical data (Fish where the fish are) Picture taken from Bill Gwinn 2015 DIA

Evidence-driven site feasibility Non-recruiting sites contribute to big financial loss Traditional site survey administration is deemed inefficient. Lasagna’s Law RWE assess patient engagement investigator skill sets PRO data

Evidence-driven site feasibility Picture taken from Bill Gwinn 2015 DIA

Example on how site selection influences trial conduct Table taken from Johnson 2015

Synthetic Control Arm (SCA) RWE serves as a control arm (synthetic control arm, SCA) When placebo is unethical, active comparator is not available, adherence to placebo arm is problematic Terminal illness, rare disease

Synthetic Control Arm (SCA) Conventionally, SCA was built on historical or previous trial information. SCA based on previous trials have the biases of those trials, can be reduced by considering multiple trials, e.g. Berry et al 2017, Medidata archival with >3000 trials

Example Berry et al 2017 Matched 340 patients from 7 Medidata archived trials, not real-world data Particular phase I/II trial This also helps develop early endpoints as predictor of long-term survival in AML patients.

Synthetic Control Arm (SCA) With RWE, real-world practices; potentially large SCA Challenges using RWD to create SCA: Real-world data unorganized not collected in a controlled environment Methods to be considered Propensity score matching to match demographics, baseline characteristics, treatment patterns etc.

Statisticians be well prepared for RWE Innovative statistical methods needed Big data Bayesian framework Propensity score methods Machine learning technologies Neural networks Text mining

Checklist for Regulatory – Grade RWE High quality Complete Transparent Generalizable Timely Scalable --- RA Miksad and AP Abernethy. 2018

Take-home message Appropriate use of RWE can help expedite drug development Limitation still exists RWE with better quality needed Regulatory agencies provide clarity guidelines on using RWE for drug approval Call for statisticians on innovative statistical methods to ensure RWD quality and broader usage of RWE in drug development

References DA Berry et al. 2017. Journal of Clinical Oncology 35, no. 15_suppl (May 20 2017) 7021-7021. R Ahmed et al. Whitepaper on “Using Real-world evidence to otimize clinical trials” Shyft Analytics Bill Gwinn 2015 DIA ‘Data-Driven Study Feasibility Assessment and Impact on Successful Execution of Clinical Trial Protocols Otis Johnson 2015 Clin. Invest. (Lond.) “An evidence-based approach to conducting clinical trial feasibility assessments”. RA Miksad and AP Abernethy. 2018. “Harnessing the Power of Real-World Evidence (RWE): A Checklist t Ensure Regulatory-Grade Data Quality”

RWE What questions do you have for me?