When preparing for a presentation at a conference highlighting some of the complexities and challenges associated with trials containing imaging endpoints., I put together the following checklist of best practices for managing and processing imaging data in clinical trials.
1. Understand the Imaging Review Charter
Clinical trials with imaging endpoints require an Imaging Review Charter (IRC). The IRC serves as a roadmap for standardizing and interpreting data coming from trials containing imaging endpoints, providing a comprehensive and detailed description of the clinical trial imaging methodology. It is important to ensure that the Charter and export specifications document match with respect to the primary endpoints. Although additional data or assessments not described in the Charter may be exported, the key assessments have to match the content of the Charter. An understanding of the IRC will help ensure that a data manager is in tune with the overall flow of data for the trial and is up to speed with all imaging data that will fall under their supervision.
2. Understand your Imaging Data
Data managers should be familiar with the imaging endpoint(s) being measured in a given study. Different endpoints provide different data outputs. Some data are quantitative at the time of acquisition (e.g. PET and DXA scans) while other data are derived from image measurements (e.g. lesion area or volume). Another type of data output is scoring systems, which are commonly used in many therapeutic areas (Genant for osteoporosis or Sharp modified for RA) and provide semi-quantitative data. Familiarizing yourself with the measurements feeding into an eCRF is crucial to understanding and validating data and will facilitate the development of appropriate edit checks for a study.
3. Develop the Edit Checks Being Applied to Your Imaging Data Early in the Process
Developing optimum edit checks for each imaging endpoint is important to ensure high-quality data. Different imaging endpoints will require different edit checks due to the inherent variability of different modalities and measurements. Longitudinal studies (e.g. lesion tracking for oncology studies) provide multiple measurements and track differences over time, making it necessary to understand the extent of variability which can be tolerated in a given measurement. Imaging core labs are often tasked with performing edit checks, so it is critical for the data manager to understand these edit checks and the rationale behind choosing them.
4. Understand the Read Design
The read design for a clinical trial ultimately dictates the imaging data that will pass through the hands of the data manager. Different clinical trials utilize different reader paradigms, from a relatively straight-forward single reader to more complex paired reads with adjudication. The choice of reading paradigm is based on a number of factors including the study phase, regulatory compliance, operational efficiency, and cost-benefit. By understanding the selected reader paradigm, data managers can understand the flow of data in a trial and can anticipate the amount of data he/she will be handled throughout the course of the trial.
5. Visit the Core Lab
Although this may sound obvious, I encourage all data managers working on an outsourced clinical trial to establish a relationship with the vendor of your study. For clinical trials in which an imaging core lab is utilized for centralized image analysis, it is important to be involved in communications with the core lab from the start of the trial. Visiting the core lab and participating in conference calls between the sponsor and core lab are good ways to ensure open lines of communication during the course of the trial.
With medical imaging playing a prominent role in today's clinical trials, data managers must be aware of the challenges associated with managing complex imaging data.