Objectives: To identify the CT features in differentiating grade 3 neuroendocrine carcinomas from grade 1/2 neuroendocrine tumours.
Methods: This study included 161 patients with surgically confirmed pancreatic neuroendocrine neoplasms. Pathology slides were reviewed to determine the tumour grade. CT image analysis included size, pattern, calcification, margin, pancreatic duct dilatation, bile duct dilatation, vascular invasion, arterial enhancement ratio, and portal enhancement ratio. We used 2 cm, 3 cm, and 4 cm as cutoff values of tumour size and 0.9 and 1.1 of enhancement ratio to determine the sensitivity and specificity.
Results: Pathology analysis identified 167 lesions in 161 patients. 154 lesions (92 %) were grade 1/2 and 13 (8 %) were grade 3. Portal enhancement ratio (< 1.1) showed high sensitivity and specificity 92.3 % and 80.5 %, respectively in differentiating grade 3 from grade 1/2. It showed the highest odds ratio (49.60), followed by poorly defined margin, size (> 3 cm), bile duct dilatation, and vascular invasion. When at least two of these five criteria were used in combination, the sensitivity and specificity for diagnosing grade 3 were 92.3 % (12/13) and 87.7 % (135/154), respectively.
Conclusions: By using specific CT findings, grade 3 can be differentiated from grade 1/2 with a high diagnostic accuracy leading to an appropriate imaging staging.
Key points: • Neuroendocrine carcinomas should be differentiated from neuroendocrine tumours. • Neuroendocrine carcinomas can be differentiated from neuroendocrine tumours on dynamic CT. • Neuroendocrine carcinomas show iso- or hypo-enhancement on portal venous phase CT.