Study group note: Using Real-World Data (RWD) to Support Clinical Trials

01.05.2021

Reading materials

1. https://onlinelibrary.wiley.com/doi/full/10.1002/pds.4932
2. https://www.researchgate.net/publication/330355252_Evaluating_the_Use_of_Nonrandomized_Real_World_Data_Analyses_for_Regulatory_Decision_Making
3. https://www.ncbi.nlm.nih.gov/pmc/articles/PMC7383026/

Meeting note

The definition of real-world data and evidence (RWD, RWE)

is in Table 1 [1]. The collection of diverse data aims to minimise clinical trial design.

Table 1. Definitions of RWD and RWE from some key international regulatory authorities

Different sources of RWD

[3, Figure 1] include electronic health/medical records (EHR/EMR, family history), molecular profiling, mobile healthcare devices/App, registry (disease, adverse events of drugs, etc), literature, social media.

The reasons for collecting RWD:

• Understand patient journey, where are the patient pain points and unmet needs?
• Evaluate drug efficacy and patient adherence, further improve patient experience e.g. dosage, dosage form
• Especially for rare diseases, diseases without a cure e.g. Alzheimer, multiple sclerosis and auto-immune disease (Merck research challenge 2021), to improve patient quality of life

Regulatory authority and regulation:

In US,

• FDA 21st century cures act
• prescription drug user fee act VI

The cancer registry in Germany: (https://www.rki.de/EN/Content/Health_Monitoring/Cancer_Registry/cancer_registry_node.html)

FDA’s proposition on RWD: Framework for FDA’s Real-World Evidence Program https://sbiaevents.com/files2/RWE-Webinar-Mar-2019.pdf

What do we expect for the usage of RWD in drug approval?

RCT + RWD will become more common e.g. drug targeted certain mutation(s) could have broader control groups; also RWD could provide evidence of efficacy and pharmacovigilance.

RCT: Usually 2 pivotal trial III > approval > trial IV

Fast track/accelerating approval: one large II and or one pivotal III trial> conditional approval > rest of the trials as RWD e.g. Pfizer COVID19 vaccine in Israel

Case study

Three case study from [1].

1. BAVENCIO (avelumab) Developed by Merck KGaA in alliance with Pfizer and Eli Lilly
2. BLINCYTO (blinatumomab) Developed by Amgen
3. INVEGA SUSTENNA (paliperidone palmitate) Developed by Janssen

In some scenario, RWD such as history controls from registry and EHR was used in single‐arm, open‐label, Phase II study to acquire conditional or accelerated approvals. Full approval was subsequently granted using confirmatory phase III data or post-marketing efficacy study. On the other hand, RWD was used for the expansion of the label for new indications.

Challenge of using EHR data

• Data quality issue e.g. missing data
• Lack of longitudinal data
• Appropriate data synthesis
• Human factor (bias) during registration e.g. entry of ICD10 code (International Classification of Diseases) in primary and secondary diagnosis
• Stakeholder e.g. hospital’s interest, benefit and billing

How about the strength of Taiwan?

• Well-structured anonymous patient data in National Health Insurance Research Database (https://nhird.nhri.org.tw/en/)
• Taiwan Biobank (https://www.twbiobank.org.tw/new_web_en/)

The eco-system of healthcare data

• Data buyer based on stakeholder’s interest
• Providers such as clinic and insurance company
• Payer e.g. insurance company
• Data clearing houses which buy raw data from the clinical and sell to data analytic company
• Healthcare software company e.g. https://en.wikipedia.org/wiki/Epic_Systems
• Data analytic company analyses data and sell to pharmaceutical companies

Some observations in the industry

• Some conditional drugs were withdrawn by FDA/EMA or the company.
• The shift of data buying focus changed from collecting data of doctor behaviour and judgement e.g. awareness of medicine to collecting patient-related data, connected to the trend of the patient-centric outcome.
• The establishment of HEOR and RWE department in multi-national pharmaceutical companies e.g. Roche, J&J. It could imply that data ownership would transfer from CRO back to pharmaceutical companies.
• More emphasis on data diversity in clinical trials, e.g. social-economic status and ethnicity. These factors are commonly analysed in the meta-analysis of clinical trials.
• Data curation, synthesis, and cleaning could be dealt with by machine learning methods e.g. Flatiron Health, which was acquired by Roche in 2018. (Roche acquired family: Genentech, Foundation Medicine, Flatiron, Spark Therapeutics, GenMark Diagnostics)