Ph3 lung cancer trial with high adjudication rate among central reviewers

The Issue

1/3 of all cases independently reviewed and adjudicated by pool of 6 radiologists with adjudication rate (intra-reader variability) of 75%.

Readers selected by sponsor, experienced clinical radiologists but no prior blinded read experience, minimal training on RECIST criteria.

The Resolution

Obtain regulatory approval to start read with new reader group; 2 experienced blinded readers and 1 adjudicator (not a pool of readers) used to control performance; extensive training on “grey areas” involving criteria and development of Reader Agreements to manage areas of potential variability created. Performance evaluated at each 10 subjects to “fine tune” reader performance. Final adjudication rate in study was 23%; drug approved based on primary imaging efficacy endpoint.