This global Phase 2/3 obesity study spanned more than 100 sites. As obesity trials rely on imaging to quantify body composition, DXA scans were selected for their speed, comfort and ability to estimate visceral fat. However, unlike specialised MRI, DXA machines are not designed to accommodate larger body sizes. A notable proportion of participants would not be able to be fully imaged using standard DXA protocols, risking the loss of visceral fat data which is increasingly critical to obesity indication therapies. ICON delivered a scalable imaging solution that would work across varied equipment and site capabilities, ensuring consistent and complete data capture for all enrolled participants.

This project faced a series of challenges over the course of the program of studies including:

DXA, it’s not like MRI. They don’t have specially manufactured machines that accommodate larger individuals. As such, many participants in obesity studies do not fit within the imaging field-of-view. To address this issue, the DXA manufacturers allow the patient to be offset to one side, such that one of their arms and legs are completely within the field of view and the others are not. The software will then mirror the values from the limbs that are within the imaging field into the ones that are not. The software can also calculate the amount of visceral fat within the abdomen, but only if the entire abdomen is within the imaging field. For as many as 10-15% of participants, their torsos could not fully be captured within this mirroring technique, leading to missing visceral fat calculations.

ICON had to implement an imaging solution that would work across the machine variations over 100 sites to capture the critical visceral fat calculations for all participants.

There are two primary imaging techniques that can allow for full DXA imaging and calculations in the event the participant cannot be fully imaged using the traditional or simple offset methods:

1. Modified hand positioning: An alternative hand position can be used where the hand is placed thumb up rather than flat on the table, allowing the patient to be offset and gain several inches of space depending on the participant.
2. Dual-scan method: Acquire and combine two scans; one offset to acquire all single-side limb data for mirroring and one with the participant centred on the table to collect a full abdomen scan.

Quality control

ICON Medical Imaging and Cardiac Safety (IMC) provided detailed site imaging manuals and developed a training module that ensured consistency in scans across all +100 sites. After acquiring accurate and consistent images, the data was centrally read using the manufacturers' software where imaging specialists then combined the results from the doubly scanned participants to identify the correct imaging data to represent a holistic dataset.
 

Communicating options

DXA scans represent a relatively low site and participant burden—they are only 5-10 minutes long and the open table design is more comfortable than other imaging machines may be. However, in opting for the two-scan solution to provide the visceral fat data capture, sponsors must pay for the second scan in about 10% of enrolled participants. With increased competition for obesity trial target populations, avoiding the loss of data from 10% of enrolled participants may be worth the cost of additional scans if appropriately accounted for at the start.

ICON’s imaging strategy enabled full data capture from 100% of participants, including those who would otherwise have been excluded from visceral fat analysis. By combining tailored scanning techniques with robust site training and centralised quality control and analysis, ICON preserved the integrity of the dataset while maintaining operational efficiency. The sponsor avoided data loss from an estimated 10–15% of participants and gained confidence in the reliability of imaging outcomes across a diverse global site network. This approach not only supported the study’s scientific objectives but also demonstrated the value of proactive imaging design in complex obesity trials.