Analysis of the cause of discordance between two radiologists on the assessment of radiographic response and progression for subjects enrolled in breast cancer clinical trials employing blinded independent central review.

6044 Background: The Food and Drug Administration (FDA) advocates blinded independent central review (BICR) of radiographic exams for oncology registration studies when the primary endpoint is based on tumor measurements. Current FDA guidance recommends multiple BICR readers evaluating each subject. This introduces the potential for discordance between readers on subject outcome resulting in adjudication by a third reader. There are no published metrics regarding the cause of discordance between BICR radiologists. Methods: BICR data was reviewed to identify 459 cases in which 2 readers were discordant in outcome. Cases were reviewed to determine the cause of discordance. Results: In 37% of cases (168/459) discordance resulted from differences inlesion selection. In 137 cases the difference was justifiable, but in 31, the choice of lesions by one reader was not thought to represent the overall extent of disease. In 30% of cases (139/459) discordance resulted from a difference in the perception of new lesions. In 88 cases the difference was justifiable, but in 51, one reader's assessment was deemed incorrect during review. In 13% of cases (58/459) discordance resulted from a difference in the qualitative assessment of nontarget disease progression. In 39 cases the difference was justifiable, but in 19, one reader's assessment was deemed incorrect. In 11% of cases (52/459) image quality issues resulted in a justifiable difference between readers. In 9% of cases (41/459) discordance resulted from a difference in lesion measurements between readers. In 37 cases the difference was justifiable, but in 4, one reader's measurements were deemed incorrect. In one case, discordance resulted from a lack of clinical information during the review. Conclusions: Some factors that cause discordance are process-driven or due to justifiable interpretation differences as was observed in 77% of cases (354/459). Separate from these factors are reader assessment errors such as those identified in 23% of cases (105/459). The BICR process of adjudication by a third reader identifies and mitigates these interpretive errors. Author Disclosure Employment or Leadership Position Consultant or Advisory Role Stock Ownership Honoraria Research Funding Expert Testimony Other Remuneration RadPharm, Inc. RadPharm, Inc. RadPharm, Inc.