Malignant pleural mesothelioma: Inter-observer variability in the assessment of tumor response classification

Abstract

7194 Background: Measurement of tumor thickness on the computed tomography (CT) scans of patients with malignant pleural mesothelioma (MPM) is tedious, and measurement consistency across sequential CT scans remains a significant clinical challenge. This study documents inter-observer variability in the tumor response classification of MPM. Methods: Three observers familiar with the clinical measurement protocol analyzed 22 CT scans from different MPM patients. Given previously measured baseline CT scans, observers measured tumor thickness in follow-up scans through a computer interface that allowed for the side-by-side viewing of the two scans. Observers made tumor thickness measurements in the follow-up scans during each of two independent sessions that differed in the manner in which baseline scan measurements were presented. One session visually displayed baseline measurements as line segments superimposed across the measurement sites on the baseline scan, while the other session used a written report of baseline measurements without a visual aid. Tumor response classification was determined from the differences in summed measurements between the two scans of each patient in accordance with RECIST guidelines. Results: 88% of pair-wise comparisons among the three observers exhibited concordant tumor response classifications using the visual aid; a 79% concordance rate was obtained based on the written report only. Statistically significant differences in summed measurement differences were observed for two of the three pair-wise observer comparisons based on the written report (p<0.05), while no statistically significant differences between observers were observed when follow-up measurements were based on the visual aid. Conclusions: This study documents the degree of inter-observer variability that exists in MPM tumor response classification. This variability may be reduced with a computer interface that presents a visual aid of baseline scan measurements. Incorporation of this interface into ongoing clinical trials is anticipated. Author Disclosure Employment or Leadership Consultant or Advisory Stock Ownership Honoraria Research Funding Expert Testimony Other Remuneration R2 Technology, Inc. R2 Technology, Inc.