2004N-0181 - Critical Path Initiative; Establishment of Docket
FDA Comment Number : EC17
Submitter : Miss. Carrie Viccars Date & Time: 08/02/2004 06:08:50
Organization : Elsevier
Category :
Issue Areas/Comments
Comment Issues
1. Hurdle Identification. Please describe the product development issue, the nature of the evaluation tool that is out-of-date or absent, how this problem hinders product development, and how a solution would improve the product development process.
Central to the problems highlighted in the FDA Critical Path report is the in-ability to accurately predict safety pharmacology and toxicological effects at an early stage of drug development and mitigate the risks associated with progressing a potential candidate compound.

Toxicity is typically assessed in preclinical studies but in vivo safety pharmacology and toxicity testing is time-consuming, costly and provides data that often shows little correlation to the human response. Failure of the drug at this stage is costly and the inability to detect idiosyncratic secondary pharmacology-related adverse effects could be catastrophic.

Typically, clinical data is only made publicly available for approved products, thus clinical trial data tends to report trials with positive findings more frequently than those with negative results, resulting in a void of historical data from which to learn.

Although there are a number of parties involved in the generation and collection of chemogenomic data, there is still insufficient data on which to build reliable predictive tools. Good quality data and models would greatly mitigate the risk of downstream failure. This can be illustrated, in the importance of QT prolongation information in order to identify which screens are truly predictive and provide regulatory guidance in the study of QT prolongation, and to enable the creation of models that signal the likelihood of failures at the clinical or post-marketing stage.

This can be solved in an FDA/Elsevier MDL collaboration by I) establishing a pool of quality, reliable data and the models needed to predict toxicology 2) creating a central workspace from which to access these data in solution centric workflows.

6. For each solution identified, please indicate which could be accomplished quickly, in less than 24 months, and which require a long-term approach?
Efforts in describing the solutions above suggest that Elsevier/MDL would be ready to engage negotiations about the creation of the databases with immediate effect. Each part i.e. database of gene expression drug signatures and the central FDA repository could come to a rapid completion and would be available to both industry and the FDA in an estimated 18 months.
7. For each problem identified, what role should FDA play and what role should be played by others?
Elsevier/MDL bring over 25 years of experience in data storage and management, software solution and tool development, and have established a CRADA with the FDA in 2003. Elsevier/MDL are also in the unique position to offer text and data mining tools that utilise the wealth of supporting literature and knowledge, and are competent to design, create and build the databases, and the resulting models.

The FDA would be required to standardise its efforts in collecting preclinical, clinical and post-marketing data by developing the e-submission criteria and the templates used, and we can build the central repository from this.

The creation a working group to undertake the screening for gene expression signatures would require funding from the FDA, and the resulting database, tools, confidentiality algorithms, access and entitlement settings would be designed and built by Elsevier/MDL.

8. What factors should guide FDA in setting priorities among the hurdles and solutions identified?
Here, we have identified solutions to the major hurdles of accurate prediction and mitigation of risk when developing potential pharmaceutical agents and the FDA should be guided by the content of the FDA Critical Path White Paper and the comments submitted to the public docket.

To date there are commonalities in the responses from contributors to the public docket, including central standardisation and aggregation of experimental, pre-clinical and clinical safety pharmacology and toxicity and post-marketing reports. Here Elsevier/MDL outline an approach, utilising data mining, model development and software development to aggregate this information into a workspace environment that would facilitate the use of the data in developing regulatory standards, hypothesis creation, decision making and in silico prediction.

There are obvious opportunities for the FDA to partner with external organisations for collaborative research and development of the tools needed to address the hurdles and the FDA should also be guided by the reality that Elsevier/MDL are uniquely positioned to design, create and build data repositories, prediction tools and supporting software to address the stated bottlenecks and hurdles.

5. Nature of the Solution. For each problem identified, please describe the evaluation tool that would solve the problem and the work necessary to create and implement the tool/solution. For example, would a solution come from scientific research to
The problem of poor data quality is central to the inability to create accurate learning sets and predictive models to enable simulation of therapeutic or toxic effect of candidate drugs.

This can be solved in an FDA/Elsevier MDL collaboration in two ways I) establishing a pool of quality, reliable data and the models needed to predict toxicology 2) creating a central workspace from which to access these data in solution centric workflows.

A central repository would need to aggregate a body of gene expression screening data and would be populated through the establishment of a working group to screen known drugs for gene expression profiles to create a pool of gene signature data that can be used for the 1) Identification of new biomarkers for efficacy and safety, 2) statistically analyze these data in the context of the chemical structure of the drug and create predictive models for targets and safety.

In order to link this screening data to actual outcome it would be necessary to create a central workspace that facilitates the collation and utilization of pre-clinical and clinical safety pharmacology and toxicology data and post-marketing (AERS) data.

Currently the FDA collects clinical trial information in both electronic and paper format but the process would benefit from installation of total e- submission capabilities and development of a common template.

Elsevier/MDL are witness to a number of academic, governmental and commercial initiatives in these areas but can offer a unique position as a neutral intermediary with the ability to guide database structuring and to use experience and expertise to develop tools that interrogate data in a way that protects the source and the intellectual property contained. We are also in the unique position to be able to offer text and data mining capabilities that enable chemical entities and structures to be identified and linked to associated data. All of the data contained within the central repository would be supported by scientific literature, and Elsevier can offer comprehensive resources for data collected at each of the pre-clinical, clinical and post-marketing stages.

Providing for academia, government and industry

Overall this collaboration between the FDA and Elsevier/MDL will result in the creation of a database that presents chemogenomic data alongside proprietary information submitted for regulation within a living `workspace' environment.

This project will also give the FDA a head start in collecting such pharmacogenomic data and through development of a body of data and predictive models the FDA can use the data internally for review and discovery e.g. development of regulations, biomarker discovery but also enable public access to datasets and models, under agreement.

Pharmacogenomic data is costly to create, and although most large companies have efforts in this area, it is out of the reach of many smaller companies. An FDA-Elsevier/MDL collaboration here would also enable this data to be made available to smaller third party organizations for a reduced access charge enabling them to invest R&D money elsewhere and to use the data to screen for anticipated gene expression in order to design and optimize the profile of candidate compounds before advancing them in the drug discovery pipeline.

Academia and industry also benefit from the development of predictive models, drawn from the wealth of human-derived genetic signature data, resulting in the reduced need for the use of animals in safety pharmacology and toxicity testing, quicker screening and optimization of candidate compounds and ultimately the reducing cost and time-lag.