Assembling Analysis Data For NDA and BLA Submissions To The FDA

Photo by National Cancer Institute on Unsplash

INTRODUCTION

The submission of New Drug Applications (NDAs) and most Biologics License Applications (BLAs) to the Food and Drug Administration’s (FDA) Center for Drug Evaluation and Research (CDER) and Center for Biologics Evaluation and Research (CBER) requires a meticulous assembly of analysis data and related files. This paper offers an in-depth look into the process of compiling these crucial elements, adhering to the guidance provided by the FDA and other regulatory bodies. The key deliverables discussed here include analysis datasets, related files, analysis programs, data definition files (define.xml), and the Analysis Data Reviewers Guide (ADRG).

ANALYSIS DATA AND OTHER RELATED DATA

Before delving into the assembly process, it’s crucial to comprehend the meaning of “analysis data” and other related data.

ANALYSIS DATASET DEFINITIONS

The Analysis Data Model Implementation Guide (ADaMIG) v1.1 describes three distinct types of datasets: analysis datasets, ADaM datasets, and non-ADaM analysis datasets:

1. Analysis dataset — This is a dataset employed for analysis and reporting.
2. ADaM dataset — This is a specific type of analysis dataset that either:
3. a. aligns with one of the ADaM defined structures and follows the ADaM fundamental principles, or
4. b. adheres to the ADaM fundamental principles outlined in the ADaM model document and stays as closely as possible to the ADaMIG variable naming and other conventions.
5. Non-ADaM analysis dataset — This is an analysis dataset that doesn’t fall under the ADaM dataset category. Examples include:
6. a. an analysis dataset created following a legacy company standard
7. b. an analysis dataset that doesn’t adhere to the ADaM fundamental principles.

A figure in the ADaMIG v1.1 document elucidates the relationships among these types of datasets, illustrating that an analysis dataset is either an ADaM dataset or a non-ADaM analysis dataset.

ADaM datasets are classified into three standard structural classes:

1. ADSL (Subject-Level Analysis Dataset)
2. BDS (Basic Data Structure)
3. OCCDS (Occurrence Data Structure) if using ADaMIG v1.1; or ADAE (Adverse Event Analysis Dataset) if using ADaMIG v1.0.

There could be instances where none of the standard structures can address a particular analysis need. For instance, none of the standard structures can generate a correlation matrix of time-varying dependent variables. In such a scenario, the unmet analysis need can be catered to by designing a dataset with a non-standard structure

Such a dataset qualifies as an ADaM dataset only if it adheres to all of the ADaM fundamental principles and other ADaM conventions. These ADaM datasets, which cannot follow a standard ADaM structure, are considered part of the ADaM Other class of ADaM datasets.

A non-ADaM analysis dataset refers to any analysis dataset that is not compliant with ADaM. Unlike ADaM datasets, non-ADaM analysis datasets do not have designated structures or classes.

ASSEMBLING THE ANALYSIS DATA

The process of assembling analysis data for NDA and BLA submissions necessitates the preparation of datasets, be they ADaM or non-ADaM, that will be used for analysis and reporting. This involves the following steps:

1. Identifying the Type of Analysis Dataset: The first step is to identify the type of analysis dataset needed for the submission. This could be an ADaM dataset or a non-ADaM dataset, depending on the type of analysis and reporting required.
2. Preparing the Dataset: The dataset is prepared based on the identified type. If it’s an ADaM dataset, it should comply with one of the ADaM defined structures or adhere as closely as possible to the ADaMIG variable naming and other conventions. If it’s a non-ADaM dataset, it should be created according to a legacy company standard or another method that does not follow the ADaM fundamental principles.
3. Verifying Compliance: The prepared dataset should be checked for compliance with the necessary principles or standards, whether ADaM or non-ADaM.
4. Documenting the Dataset: Adequate documentation of the dataset should be prepared, detailing the process and methodology used in its creation. This documentation should be submitted alongside the dataset in the NDA or BLA submission.

ASSEMBLING RELATED FILES

Besides the analysis datasets, other files related to the analysis datasets also need to be compiled for the submission. This includes analysis programs, data definition files (define.xml), and the Analysis Data Reviewers Guide (ADRG).

1. Analysis Programs: Analysis programs refer to the scripts or software used to analyze the dataset. They should be documented and submitted with the dataset.
2. Data Definition Files: The data definition files (define.xml) provide metadata about the dataset, including information about the structure of the dataset, the variables included, and the relationship between the variables. They are crucial for understanding and interpreting the dataset.
3. Analysis Data Reviewers Guide (ADRG): The ADRG is a document that provides an overview of the analysis datasets and related files submitted as part of the NDA or BLA. It includes a description of the datasets and files, the methodology used to create them, and any issues or deviations from the standard procedures. The ADRG is crucial for the reviewers to understand and evaluate the submitted analysis data.

CONCLUSION

Assembling analysis data for NDA and BLA submissions to the FDA is a complex process that requires a thorough understanding of the different types of datasets and the standards or principles they should comply with. It also involves the preparation and documentation of related files, including analysis programs, data definition files, and the Analysis Data Reviewers Guide. By following the guidance provided in the relevant FDA documents and adhering to the principles outlined by the ADaMIG, companies can ensure that their submissions are complete, compliant, and ready for review by the FDA.