Comparison of local versus central tumor measurements in a multicenter cancer trial.

Four hundred nineteen solid tumors from 92 patients were measured by a local team consisting of a physician and nurse oncologist or physician's assistant, and centrally by one radiologist. The central radiologist also remeasured 137 tumors on 30 patients to assess intraexaminer agreement. Tumor measurements at specific points in time as well as changes in tumor measurements over time were evaluated. When signed differences were calculated, there was little overall difference between local and central examiners, although three of the eight centers did show a significant difference. However, when absolute differences were calculated, the relative errors of the width times length products ranged from 35 to 55%. Although local and central examiners agreed 75% of the time on change in disease, there was only 42% agreement on the subset of patients who had a remission. In general, the intraexaminer agreement was slightly better than the interexaminer agreement. This study suggests that solid tumor measurements are not particularly reliable, and that survival time remains the most satisfactory endpoint in a cancer clinical trial.