Size Measurement and T-staging of Lung Adenocarcinomas Manifesting as Solid Nodules ≤30 mm on CT: Radiology-Pathology Correlation

Abstract

This study aimed to compare long-axis diameter to average computed tomography (CT) diameter measurements of lung adenocarcinomas manifesting as solid lung nodules ≤30 mm on CT, as referenced to pathologic measurements, and to determine the impact of the two CT measurement approaches on tumor (T)-staging of nodules. This institutional review board-approved study included all 274 radiologic solid adenocarcinomas resected at our institution over 10 years. Two observers measured long- and short-axis diameters on pre-resection chest CT in lung and mediastinal windows. T-stages were determined. CT measurements and T-stages were compared to pathology measurements and T-stages using Wilcoxon signed rank test and McNemar test. Inter- and intraobserver variability was determined with intraclass correlation coefficients (ICC) and Bland-Altman plots. For lung and mediastinal windows, nodule size was significantly larger using long-axis diameter rather than average diameter (16.93 vs. 14.92 mm, P <.001; and 14.02 vs. 12.17 mm, P <.001, respectively). The correlation of CT with pathologic measurements was stronger with long-axis than with average diameter (ICC 0.808 vs. 0.730; and 0.731 vs. 0.621, respectively). Lung window measurements correlated stronger with pathology than mediastinal window measurements. CT T-stages differed from pathology T-stages in more than 20% of nodules (P <.001). Inter- and intraobserver variability was small with long-axis and average diameter (ICC range 0.96-0.991, and 0.970-0.993, respectively), but long-axis diameter showed wider scatter on Bland-Altman plots. Long-axis CT diameter is preferable for T-staging because it better reflects the pathology T-stage. Average CT diameter might be used for longitudinal nodule follow-up because it shows less measurement variability and is more conservative in size assessment.