Determining size in an era when size matters: Interobserver comparison of CT lung tumor size measurements and accuracy of radiologic size estimation compared to pathologic size.

[BACKGROUND] Limited data is available on the diagnostic performance of CT tumor size measurements in the setting of updated lung cancer management including sublobar resection for select tumors ≤2 cm, and neoadjuvant chemoimmunotherapy for tumors ≥4 cm. Our aim was to compare radiographic with pathologic tumor size and evaluate inter-radiologist variability in CT tumor size assessment.

[METHODS] This retrospective study included patients who underwent lung cancer resection from 1/2022-12/2023. Maximal tumor diameter on the preoperative CT report and pathology report were recorded. Diagnostic performance of CT for clinically relevant thresholds above 2 or 4 cm was examined using receiver operating characteristic (ROC) analysis. To evaluate interobserver variability, two radiologists independently measured maximal tumor diameter on CT for a subset of 92 cases and measurements were compared with intraclass correlation coefficient (ICC) and Bland-Altman analysis.

[RESULTS] A total of 316 patients were included [median age: 69 years (IQR 64-74); 60.4% female]. Median diameter on CT was 18 mm vs 20 mm on pathology (p < 0.001). For the 2 cm threshold, CT demonstrated an AUC of 0.84, 89% sensitivity, 79% specificity. For the 4 cm threshold, CT measurement demonstrated an AUC of 0.79, 61% sensitivity, 98% specificity. Interobserver agreement between radiologists was high (ICC = 0.91, 95% CI 0.87-0.94).

[CONCLUSIONS] While CT measurements underestimated tumor size compared to pathology, CT demonstrated strong diagnostic performance for correct pathologic size categorization for a 2 cm threshold, suggesting utility for determining lobar versus sublobar resection. For larger tumors with a threshold of 4 cm, CT demonstrated lower sensitivity, potentially leading to underutilization of neoadjuvant therapy.