Missed Incidental Lung Cancer on Neck CT: Determinants and Impact on Diagnostic Delay and Stage Shift.
Authors
Affiliations (3)
Affiliations (3)
- Department of Radiology, Seoul St. Mary's Hospital, College of Medicine, The Catholic University of Korea, Seoul, Republic of Korea.
- Department of Radiology, Research Institute of Radiological Science and Center for Clinical Imaging Data Science, Severance Hospital, Yonsei University College of Medicine, Seoul, Republic of Korea.
- Department of Radiology, Seoul St. Mary's Hospital, College of Medicine, The Catholic University of Korea, Seoul, Republic of Korea. [email protected].
Abstract
To assess the determinants and clinical impact of missed incidental lung cancers on neck CT and explore the potential role of artificial intelligence-based computer-aided detection (AI-CAD). This study retrospectively screened adults who underwent neck CT at a tertiary care hospital between 2008 and 2024. Patients with prior lung cancer, prior or concurrent chest CT, absence of visible lung lesions on neck CT, lack of histological confirmation, or indeterminate staging or stage shift were excluded. Determinants of missed detection and their impact on diagnostic intervals and stage shift (progression in any T, N, or M category) were evaluated. AI-CAD was retrospectively applied to the missed cases. Of 81,794 patients screened, 123 with lung cancer visible on neck CT were identified (mean age, 65.1 ± 10.5 years; 63 male), of whom 80 (65.0%) were not described in the original reports. Missed detection was more common with CT examinations including CT angiography (odds ratio [OR], 3.40 [95% confidence interval {CI}: 1.13, 11.69]; <i>P</i> = 0.037) and squamous cell carcinoma (OR, 6.84 [95% CI: 1.32, 55.16]; <i>P</i> = 0.037) and was less common with double reading (OR, 0.16 [95% CI: 0.03, 0.81]; <i>P</i> = 0.031), lymphadenopathy (OR, 0.11 [95% CI: 0.04, 0.32]; <i>P</i> < 0.001), and lobulated margins (OR, 0.29 [95% CI: 0.11, 0.79]; <i>P</i> = 0.016). Missed detection was associated with a longer diagnostic interval to pathologic confirmation (median, 27.5 vs. 0 months; <i>P</i> < 0.001) and more frequent stage shift (47.5% vs. 11.6%, <i>P</i> < 0.001), including numerically more frequent progression from stage I-II to III-IV (11.3% vs. 2.3%, <i>P</i> = 0.146). AI-CAD detected 41 of 80 missed lesions (51.3%), without significant variations across imaging or lesion characteristics. Incidental lung cancers on neck CT are rare but are frequently overlooked. Missed detection is associated with substantially longer diagnostic intervals and more frequent stage shifts. AI-CAD has the potential to reduce diagnostic oversight, warranting further validation.