Image Data Management and Analysis in Oncology Trials
Medpace Imaging Core Lab (MICL) provides a comprehensive medical image management and analysis services portfolio for oncology trials. The key services and technologies include:
Site Qualification & Training, Ongoing Quality Assurance
Sites are provided with manufacturer (vendor) and model-specific image acquisition parameters to minimize inter-site variability. Key imaging modalities include: CT, MRI, PET, X-ray and bone scans.
The main goal of the initial site qualification phase is to maximize the consistency of image acquisition across sites by controlling various parameters related to scanners, acquisition parameters, contrast injection procedure (if applicable) and anatomical coverage.
Image Data Collection & QC
The image data will be transferred to MICL by the sites using a dedicated web-based image data upload portal. The image data will be electronically uploaded after de-identification and encryption. A study-specific electronic Data Transmittal Form (DTF) will be filled out by the sites during the image submission process. The electronic upload of CT, MRI & PET scans represents a significant gain in terms of time and quality.
The image data will be quality controlled upon receipt at MICL. In case of insufficient image quality or significant deviations with regard to the study’s imaging protocol, repeat scans may be requested by MICL whenever possible.
Tumor Response Assessments
The MICL team has strong scientific and operational expertise in the management of the imaging component of clinical trials in oncology based on a large variety of imaging-based tumor response criteria such as RECIST 1.1, modified RECIST, Cheson, Lugano, irRC, RANO, etc. The imaging department closely cooperates with the Medical Affairs department and its 29 physicians, including 5 oncologists.
Notifications can be sent to the sites for confirmation of progression and/or response. Safety findings will be systematically escalated to Medpace’s clinical team, the Sponsor’s clinical monitors and the site. Such escalations will be expedited by MICL. Whenever needed, the image data and related imaging eCRFs can be accessed online during conference calls for final decision-making.
Conduct of Central Reads and Quality Assurance
Several reading paradigms can be implemented: single reads, two reads plus adjudication, etc. In case of discrepancies amongst the two primary readers for one or more endpoints, a third reader will adjudicate the case. The adjudicator will have access to all the images and evaluation results of the two primary readers and choose one of the available evaluations.
During the course of the study, the adjudication rate is systematically monitored to control the quality of central reads. If the adjudication rate becomes too high, additional training actions may be implemented.
Development of the Image Evaluation System
MICL will develop and validate a web-based Image Evaluation System for central evaluations including study-specific imaging CRFs. The Image Management and Evaluation System will include built-in logical checks to avoid data inconsistencies and omissions.
The imaging data and associated eCRFs will be made available online in a read-only mode for clinical monitoring (Sponsor, CRO) and audit/inspection purposes.
Selection and Training of Central Readers, Quality Assurance of Central Reads
A pool of highly qualified and experienced radiologists will be involved in central reading operations. MICL will suggest names of potential readers, provide their CVs, and the sponsor will review and approve. The sponsor may also suggest other readers.
The readers will be trained by MICL prior to the beginning of central evaluations. MICL will:
Develop an Image Evaluation Manual for central readers
Train the central readers:
Review of key definitions and consensus points (definition of target and non-target lesions, new lesions, etc.)
Presentation of the Image Evaluation System and related eCRFs
Evaluation of test cases (non-study patients)
Review of the overall reading process, deliverables and expectations
Ongoing Site Quality Assurance
In a full-service model, when the EDC and image management and evaluation system are available on the same technology platform (Medpace’s ClinTrak), on-site and central tumor response assessments can be compared in a proactive way. Implementing tumor response criteria which are more recent and less standard than, say, RECIST 1.1, may be challenging with sites and geographical regions which are less experienced. It is therefore crucial to control the quality of on-site evaluations through a systematic comparison of the collected results with central readers’ results. In case of frequent or systematic discrepancies, more training actions can be implemented.
Data Consistency Checks and Exports
Data consistency checks are systematically performed prior to final exports. Built-in logical checks, for example for the calculation of tumor response are executed to ensure the strict application of main definitions and rules, as well as for detecting potential errors in central readers’ results.