OBJECTIVES: Clinical outcome assessment criteria, such as RECIST v1.1 and Lugano 2014, provide a standardized approach to assess treatment effectiveness in clinical trials. Similar standardized methodologies for outcome classification may be applied to real-world clinical data. We present the results of a study to explore the feasibility of performing a blinded independent central review (BICR) of positron emission tomography–computed tomography (PET-CT) scans in a real-world data (RWD) cohort.

METHODS: We conducted a retrospective multisite RWD study using Cardinal Health’s Practice Research Network (PRN) to identify adult patients with diffuse large B-cell lymphoma (DLBCL) treated with first line chemoimmunotherapy in US clinical practice between 01JAN2015 and 31DEC2022. Participating PRN sites abstracted clinical data derived from medical records, PET-CT reports, and digital images at treatment initiation and first response assessment. Deidentified images were uploaded for direct digital image evaluation via BICR adjudicated by two lymphoma radiologists using Lugano 2014 criteria to classify treatment response.

RESULTS: We identified 185 eligible patients with DLBCL across 7 PRN sites. PET-CT reports were readily available in every patient’s medical record (n=185, 100%); however, obtaining digital PET-CT images posed significant logistical challenges including submitting a request, frequently on a paper form, and waiting 1-2 weeks for scans to be sent. Then, scans were uploaded to data-capture software, de-identified, and transmitted for BICR, which entailed identifying, contracting, training, and coordinating meetings with expert radiologists. We obtained baseline and follow-up scans for 174 (94%) patients [male n=103 (59%); female n=71 (41%); mean age 66 years] across 6 PRN sites. Inter-rater BICR reached concordance for all patients, observing 82% complete response rate.

CONCLUSIONS: This study demonstrated that pairing manual chart abstraction with deidentified digital image transfer for BICR is challenging, yet feasible. Such approaches may reduce outcome misclassification and increase comparability between RWD and clinical trial data.