NASH Placebo Arm Data Overview

E-mail Print

Overview

The Forum for Collaborative Research’s Data & Analysis Center at the University of California Berkeley (UCB) School of Public Health (SPH), a neutral venue for data sharing and analysis is working to initiate the Liver Forum’s Placebo Arm Database (PDB) project, an integrated patient-level database from completed phase 2 and phase 3 NASH studies. The Liver Forum, a project of the Forum for Collaborative Research, is a neutral and independent venue for regulators, industry, academicians and patients to engage in dialogue and deliberation to accelerate drug development in the field of liver diseases, focusing at this time on NAFLD/NASH. Regulatory experts from FDA (CDER: DGIEP, OND, OTM and CDRH) and EMA are involved in Liver Forum proceedings, including all working groups. The Forum has had high track record in proceedings which have informed regulatory guidance in all disease areas the Forum works in.

Mission

To facilitate responsible use of data through collaboration to honor patient contributions to clinical research.

Why?

Data sharing offers opportunities to spur innovation and can generate new knowledge to facilitate drug development . Data sharing can: 1) Develop better tools to stratify patient populations/assess treatment benefits; 2) Promote development of biomarkers, simulation tools to improve clinical trial design; improve likelihood of success; 3) Reduce risk/increase confidence; and 4) Reduce time, size and cost of phase 3 trials through optimization.

Despite recent improvements in the clinical development process, more than one-third of new drugs fail between phase III and launch. Considering the burden on patients and the time and expense of clinical trials, regulators and industry members endorsed the formation of a "Pooled Placebo Arm Database" as a unique contribution complementing other type of observational cohorts, such as TARGET NASH and NASH CRN. The PDB is distinctive in that it includes patients meeting specific inclusion/exclusion criteria, intensive follow-up, paired biopsy data and standard plus investigational biomarkers.

Regulatory problems to address

The need for long-term maintenance of patients in a placebo arm in cases where no biomarkers qualified for surrogate endpoints exist. Other problems with recruitment of patients into long-term clinical trials, including a rapid development space (high early development index), with competition for patient recruitment. The high development pace is challenged by lack of non-invasive diagnostics to confirm disease status and entry/exclusion criteria.

Aim

To combine the power of a PDB and Targeted Machine Learning to advance regulatory science and increase the quality, efficiency and output of clinical trials to accelerate drug development for an unmet medical need of great public health significance: non-alcoholic fatty liver diseases (NAFLD), specifically, non-alcoholic steatohepatitis (NASH) including indeterminate NASH, NASH with and without fibrosis, and cirrhosis.

Objectives

  1. Develop greater understanding of the natural history of NASH, specifically, what determines the progression of NASH vs. the reversal of NASH in untreated patients (e.g., patients in placebo arms). Symptoms of NASH are often silent or non-specific to NASH which makes it difficult to diagnose. As a result, the disease is commonly diagnosed in late stages.
  2. Reduce placebo-arm burden in future phase 3 (phase 4) trials.
  3. Address inclusion/ exclusion criteria for NASH clinical trials.
  4. Identify adverse events.
  5. Establish a framework for increasing specificity of patient selection.

Strategy

By pooling placebo arms from over 30 completed Phase 2 and Phase 3 NASH studies, and applying novel targeted learning analytic methods, we can decrease the number of patients in placebo control arms, reduce the need for long term maintenance of patients in a placebo control arm, and establish a framework for biomarker identification and evaluation.

Results

Need of fewer patients to demonstrate efficacy by increasing the utility and information output of individual and pooled datasets. In economic terms, replacement or reduction in size of a placebo arm carries significant economic benefit. Estimated costs per patient in clinical trials range from $25,000 to $70,000 – in the NASH space, the cost can be considerably higher. Thus, eliminating even 100 patients from a clinical trial may save $2.5 – $7.0 million. Based on these potential savings, the return on investment (ROI) based on a total cost of $500,000 (total annual PDB construction, management, oversight, governance, and housing costs) is estimated to be 500 – 1,400 percent.

Partners

  • Regulatory Agencies: Regulatory experts from FDA (CDER: DHN, DHN, OND, OTM and CDRH) and EMA - regulators will have input at each stage of discussion and project development.
  • Berkeley Center for Targeted and Machine Learning: the center brings the rigor and power of classical statistics together with advances in data mining and machine learning to accelerate the development and dissemination of causal inference methods that bring robust insight and evidence to important public health related questions.
  • Industry: Participating Liver Forum members will share data under the Forum’s safe keeping environment.
  • The PDB will interact extensively with all other relevant groups including the NIMBLE and LITMUS consortia.

Leadership

The Liver Forum’s Placebo Arm Database Project is led by an Executive Committee and managed by Forum staff. Neutrality and objectivity constitute the main principles of the PDB Project. This is achieved by having an Executive Committee to underscore the collaborative process with representation and active engagement of scientific experts from all stakeholder groups including: academia, industry, patient organizations, and regulatory agencies. Industry members on the Executive Committee rotate on a bi-annual basis. The Executive Committee is tasked with providing scientific leadership, general feedback on project development, sustainability, new directions, and help identifying area of focus. Once the database is established, the Executive Committee, in addition to ad hoc scientific expert advisees will be in charge of reviewing research proposals and analyses requests from industry and non-for-profit organizations to be performed by the Forum for Collaborative Research’s Data & Analysis Center on their behalf. As well, the Executive Committee will provide guidance and suggestions on analyses to be performed.

Executive Committee:
Bettina Hansen, Senior Biostatistician, Erasmus MC Netherlands
Others are currently under recruitment

Database Security & Infrastructure

Data will be housed in a secure research data and compute (SRDC) platform developed at UC Berkeley for researchers working with highly sensitive (P4 level) data. P4 level is the highest level of data security available to accommodate sensitive information, up to protected health information (PHI) / patient records. Although the PDB will include only deidentified data, because of the amount of data it will house we will go over and above the required campus security for deidentified data.

Process & Outputs

Academia, industry, patient organizations, and governmental agencies can request analyses to be performed on their behalf. For example, project members can request Forum statisticians to perform confidential analyses specific to their drug development program.

Regulatory agencies will be granted access to the database directly for purposes of analyses to inform their regulatory processes. All requesters may make analysis results publicly available through the development of white papers and manuscripts in collaboration with the Forum for publication to expand the knowledge in the field or for their own confidential internal analysis. Research and analysis proposals will be evaluated on the basis of the following criteria: (1) is the proposal in alignment with the mission and overall goals and objectives of the database, (2) is the proposal feasible, (3) is the individual submitting the proposal a scientific expert from an organization committed to and actively engaged in research and development in NAFLD/NASH, PSC, and/or liver fibrosis. Approved requests will be analyzed by Forum statisticians and results will be shared with the approved organization.

Working Groups

Working groups carry on the work of the Placebo Arm Database Project by addressing priority areas identified by both the Executive Committee and project members. Expected outcomes of the working groups include recommendations, position papers, reports, manuscripts for submission to peer-reviewed journals, and presentations at conferences with the goal of disseminating findings. Throughout the year, working groups communicate and collaborate by conference calls, email, in-person meetings, and workshops as needed. There is no time requirement for participation; however, working group members are expected to be active participants and participate in as many calls and activities as possible.
Current working groups include:

  1. The NASH Trials Placebo-Arm Cohort Working Group, co-led by Michael Cooreman and Manal Abdelmalek, to further aid the field in answering key questions such as what determines progression of NASH vs. reversal of NASH in untreated patients.
  2. The Statisticians Working Group will be composed of Forum statisticians and statisticians from the FDA, academia and industry. This group will work together to assess the importance of baseline and time-varying factors in predicting spontaneous resolution vs. NASH progression, identify novel baseline patient stratifications for these outcomes, and develop individualized machine learning predictive algorithms for these outcomes. These analyses will benefit future trial design and analysis and inform inclusion/exclusion criteria for all subsequent studies. Potential analyses include:
    1. Algorithm to predict placebo outcome based on trial design criteria
    2. Sensitivity analyses on surrogate/histological endpoint accuracy
    3. Kappa coefficient calculations for inter- and intra-observer reliability
    4. Confidence levels for placebo estimates
    5. Attempt alternate statistical analysis techniques and/or apply common techniques used in these trials for better comparability
    Additional outputs of this working group will include annual closed door statistical workshops.

SPONSORSHIP OPPORTUNITIES/S

Sponsorships are based on data and funding contributions. Please contact Chris Hoffman, IT and Operational Director at This e-mail address is being protected from spambots. You need JavaScript enabled to view it for information on how to become a sponsor.



SPONSOR/S

 

pfizer

 

References



1) Thompson, A., & Parekh, A. (2021). Value of Data Sharing to Advance Drug Development: A Regulatory Perspective. Therapeutic Innovation & Regulatory Science, 1-3.

2) Galson, S., Austin, C. P., Khandekar, E., Hudson, L. D., DiMasi, J. A., Califf, R., & Wagner, J. A. (2020). The failure to fail smartly. Nature reviews. Drug Discovery.

3) Siddiqui, M.S., Harrison, S.A., Abdelmalek, M.F., Anstee, Q.M., Bedossa, P., Castera, L., Dimick‐Santos, L., Friedman, S.L., Greene, K., Kleiner, D.E., Megnien, S., Neuschwander‐Tetri, B.A., Ratziu, V., Schabel, E., Miller, V., Sanyal, A.J. and (2018), Case definitions for inclusion and analysis of endpoints in clinical trials for nonalcoholic steatohepatitis through the lens of regulatory science. Hepatology, 67: 2001-2012. https://doi.org/10.1002/hep.29607