Morphological imaging and CT histogram analysis to differentiate pancreatic neuroendocrine tumor grade 3 from neuroendocrine carcinoma

A Azoulay; J Cros; M-P Vullierme; L de Mestier; A Couvelard; O Hentic; P Ruszniewski; A Sauvanet; V Vilgrain; M Ronot

doi:10.1016/j.diii.2020.06.006

Morphological imaging and CT histogram analysis to differentiate pancreatic neuroendocrine tumor grade 3 from neuroendocrine carcinoma

Diagn Interv Imaging. 2020 Dec;101(12):821-830. doi: 10.1016/j.diii.2020.06.006. Epub 2020 Jul 21.

Authors

A Azoulay¹, J Cros², M-P Vullierme¹, L de Mestier³, A Couvelard², O Hentic⁴, P Ruszniewski³, A Sauvanet⁵, V Vilgrain⁶, M Ronot⁷

Affiliations

¹ Department of Radiology, University Hospitals Paris Nord Val de Seine, Beaujon, Assistance Publique-Hôpitaux de Paris, 92118 Clichy, France.
² Department of Pathology, University Hospitals Paris Nord Val de Seine, Beaujon Assistance Publique-Hôpitaux de Paris, 92118 Clichy, France; Université de Paris, Diderot Paris 7, 75010 Paris, France; INSERM U1149, CRI, Paris, France.
³ Université de Paris, Diderot Paris 7, 75010 Paris, France; Department of Pancreatology, University Hospitals Paris Nord Val de Seine, Beaujon Assistance Publique-Hôpitaux de Paris, 92118 Clichy, France; INSERM U1149, CRI, Paris, France.
⁴ Department of Pancreatology, University Hospitals Paris Nord Val de Seine, Beaujon Assistance Publique-Hôpitaux de Paris, 92118 Clichy, France.
⁵ Department of HPB Surgery, University Hospitals Paris Nord Val de Seine, Beaujon Assistance Publique-Hôpitaux de Paris, 92118 Clichy, France.
⁶ Department of Radiology, University Hospitals Paris Nord Val de Seine, Beaujon, Assistance Publique-Hôpitaux de Paris, 92118 Clichy, France; Université de Paris, Diderot Paris 7, 75010 Paris, France; INSERM U1149, CRI, Paris, France.
⁷ Department of Radiology, University Hospitals Paris Nord Val de Seine, Beaujon, Assistance Publique-Hôpitaux de Paris, 92118 Clichy, France; Université de Paris, Diderot Paris 7, 75010 Paris, France; INSERM U1149, CRI, Paris, France. Electronic address: maxime.ronot@aphp.fr.

PMID: 32709455
DOI: 10.1016/j.diii.2020.06.006

Abstract

Purpose: To compare morphological imaging features and CT texture histogram parameters between grade 3 pancreatic neuroendocrine tumors (G3-NET) and neuroendocrine carcinomas (NEC).

Materials and methods: Patients with pathologically proven G3-NET and NEC, according to the 2017 World Health Organization classification who had CT and MRI examinations between 2006-2017 were retrospectively included. CT and MRI examinations were reviewed by two radiologists in consensus and analyzed with respect to tumor size, enhancement patterns, hemorrhagic content, liver metastases and lymphadenopathies. Texture histogram analysis of tumors was performed on arterial and portal phase CT images. images. Morphological imaging features and CT texture histogram parameters of G3-NETs and NECs were compared.

Results: Thirty-seven patients (21 men, 16 women; mean age, 56±13 [SD] years [range: 28-82 years]) with 37 tumors (mean diameter, 60±46 [SD] mm) were included (CT available for all, MRI for 16/37, 43%). Twenty-three patients (23/37; 62%) had NEC and 14 patients (14/37; 38%) had G3-NET. NECs were larger than G3-NETs (mean, 70±51 [SD] mm [range: 18 - 196mm] vs. 42±24 [SD] mm [range: 8 - 94mm], respectively; P=0.039), with more tumor necrosis (75% vs. 33%, respectively; P=0.030) and lower attenuation on precontrast (30±4 [SD] HU [range: 25-39 HU] vs. 37±6 [SD] [range: 25-45 HU], respectively; P=0.002) and on portal venous phase CT images (75±18 [SD] HU [range: 43 - 108 HU] vs. 92±19 [SD] HU [range: 46 - 117 HU], respectively; P=0.014). Hemorrhagic content on MRI was only observed in NEC (P=0.007). The mean ADC value was lower in NEC ([1.1±0.1 (SD)]×10^-3 mm²/s [range: (0.91 - 1.3)×10^-3 mm²/s] vs. [1.4±0.2 (SD)]×10^-3 mm²/s [range: (1.1 - 1.6)×10^-3 mm²/s]; P=0.005). CT histogram analysis showed that NEC were more heterogeneous on portal venous phase images (Entropy-0: 4.7±0.2 [SD] [range: 4.2-5.1] vs. 4.5±0.4 [SD] [range: 3.7-4.9]; P=0.023).

Conclusion: Pancreatic NECs are larger, more frequently hypoattenuating and more heterogeneous with hemorrhagic content than G3-NET on CT and MRI.

Keywords: Magnetic resonance imaging; Neuroendocrine tumors; OMS 2017; Texture analysis; Tomography, X-ray computed.

MeSH terms

Carcinoma, Neuroendocrine* / diagnostic imaging
Carcinoma, Neuroendocrine* / pathology
Female
Humans
Image Processing, Computer-Assisted
Male
Middle Aged
Neoplasm Grading
Pancreatic Neoplasms* / diagnostic imaging
Pancreatic Neoplasms* / pathology
Retrospective Studies
Tomography, X-Ray Computed