Systematic double reading for oncological PET/CT scans: insights from a prospective multicentre study in 678 patients

As PET/CT's role in oncology expands, with increasing demands for staging, restaging, and therapy monitoring across a range of malignancies, and with the rise of theranostics, PET facilities are experiencing higher workloads, raising concerns about the accuracy of reports. This prospective study evaluates the frequency and nature of errors identified through a double-reading protocol.

Patients scheduled for baseline or follow-up PET/CT scans were prospectively included in the study between May and June 2024. PET scans included [18 F]FDG, [18 F]Choline, [18 F]Dopa, and [18 F]PSMA. Each scan was independently reviewed by a second reader before finalizing the report. Significant changes were discussed before validation, and a third reader's input was sought if consensus was needed. Data collection focused on potential factors influencing PET report errors, using a survey completed by second readers evaluating report changes (minor versus impactful), reader confidence, modified diagnostic hypotheses, and whether the initial reader was a permanent doctor or a locum doctor. Information regarding the indication, the number of previous scans for comparison, and the timing (period of the day) of PET readings was also recorded.

Two-thirds of second readings were completed within 2-5 min. Changes were reported in 17% of PET reports (117/678), although impactful changes were rare. In therapy monitoring, only 5 out of 20 changes were significant. In staging and restaging, two major changes were noted. Most changes involved phrasing or formatting (59%), laterality (10%), and target lesions that did not affect TNM staging (29%). Regarding the degree of certainty, 18 changes were observed: 8 cases shifted from doubtful to certain, and 10 from certain to doubtful, necessitating follow-up scans. Factors such as the timing of the first reading, indication for PET, number of previous comparative PET scans, PET request clarity, first reader's experience, type of PET tracer, and patient BMI, as a surrogate of image quality, showed no significant correlation with changes in PET reports.

Systematic double reading effectively identifies and corrects both rare impactful and minor errors in PET reports, though no specific predictive factors for error occurrence were identified, indicating that double reading, if performed, must be systematic.