Pancreatic neuroendocrine tumors: Correlation between the contrast-enhanced computed tomography features and the pathological tumor grade

https://doi.org/10.1016/j.ejrad.2015.05.005Get rights and content

Highlights

  • Updated WHO classification divided pancreatic neuroendocrine tumors into G1, G2, and G3.

  • We investigate the relationship between CT findings and pathological tumor grades (G1 and G2).

  • A larger tumor size was associated with G2 pancreatic neuroendocrine tumors (PanNETs).

  • Non-hyperattenuation during the portal venous phase was associated with G2 PanNETs.

Abstract

Objective

To determine whether CT features can predict the pathological tumor grades of pancreatic neuroendocrine tumors (PanNETs) according to the recent WHO classification.

Materials and methods

In all, 28 patients with histologically confirmed PanNETs underwent preoperative contrast CT examinations. Thirteen tumors were classified as G1 and 15 as G2. Two radiologists independently evaluated the CT features (tumor delineation, peripancreatic vascular involvement, upstream pancreatic duct dilatation, N (regional lymph node metastasis) and M (distant metastasis) grades, tumor homogeneity, cystic or necrotic change, and tumor conspicuity). The tumor sizes and Hounsfield unit values of all PanNETs during each phase on CT were measured by one radiologist. We compared the CT features between pathological tumor grades using Fisher's exact test for nominal scales and Mann–Whitney U test for ordinal scales or continuous variables. Additionally, we evaluated the performances of the CT findings and their combinations to diagnose G2 tumors.

Results

G2 tumors showed significantly larger in tumor size than G1 tumors (p = 0.029). All 4 tumors with hepatic metastases were G2. Non-hyperattenuation compared with pancreatic parenchyma during portal venous phase (PVP) was significantly associated with G2 (p = 0.016). The accuracy for G2 diagnosis of tumor size (≥20 mm), M grade (M1), and tumor conspicuity (non-hyperattenuation during PVP) were 71%, 61%, and 71%, respectively, while the accuracy of their combination was 82%.

Conclusion

Contrast-enhanced CT features (tumor size, M grade, and tumor conspicuity during PVP) can predict the pathological tumor grades of PanNETs.

Introduction

Pancreatic neuroendocrine tumors (PanNETs) are rare pancreatic neoplasma with an annual incidence of 2.2 per 1,000,000 individuals [1]. However, PanNETs are clinically important, given their high rate of malignancy. Contrast-enhanced computed tomography (CT) is a widely accepted technique for the detection and staging of pancreatic tumors. This technique is usually performed as a multiphase dynamic study, and the pancreatic parenchymal or portal venous phase can facilitate the detection of PanNETs [2], [3]. Many studies have reported the radiologic findings of PanNETs [4], [5], [6]. Typical PanNETs show hyperattenuation during the pancreatic parenchymal and portal venous phases because of their rich blood capillary networks [5].

In 2000, the World Health Organization (WHO) classification for PanNETs identified lesions exhibiting malignant behaviors on the basis of tumor cell differentiation [7]. Subsequently, in 2010, the updated WHO classification divided neuroendocrine tumors into the categories of NETG1, NETG2, and NEC, according to the pathological tumor grade, which is an independent prognostic factor of survival in patients with PanNETs [8], [9], [10]. Therefore, pretreatment prediction of the PanNET pathological tumor grade according to the recent WHO classification is very important in determining an efficient treatment strategy. Studies on the relationship between contrast-enhanced CT findings and prognostic factors, including pathological tumor grades according to the previous WHO PanNET classification, have been reported [11], [12], [13]. However, there is a paucity in research on the relationship between contrast-enhanced CT findings and pathological tumor grades according to the recent WHO PanNET classification [14]. Therefore, the purpose of this study was to determine whether CT features can predict the pathological tumor grades of PanNETs according to the recent WHO classification.

Section snippets

Patients

Institutional ethics review board approval was obtained and informed consent was waived for this retrospective study. We queried our institutions’ pathology databases to identify all histologically proven PanNET cases during a period from September 2004 to November 2011, and identified 30 patients with PanNETs who were considered for inclusion in this retrospective study. All patients underwent preoperative dynamic-enhanced CT examinations to evaluate pancreatic tumors. We excluded 2 patients

Results

The interobserver agreements ranged from moderate to near-perfect (κ = 0.70 for tumor margin delineation, κ = 1.00 for peripancreatic vascular involvement, κ = 0.91 for upstream pancreatic duct dilatation, κ = 0.70 for the N grade, κ = 0.84 for the M grade, κ = 1.00 for tumor homogeneity during the pancreatic parenchymal phase, κ = 0.85 for tumor homogeneity during the portal venous phase, κ = 0.85 for tumor homogeneity during the delayed phase, κ = 0.87 for cystic or necrotic change, κ = 0.92 for tumor

Discussion

In this study, we sought to determine whether the CT features of PanNETs could facilitate the differentiation of G2 from G1. Gallotti et al. [12] classified PanNETs into 2 groups according to the 2000 WHO classification as follows: benign PanNETs, which were defined as a well-differentiated endocrine tumor with benign behavior; and non-benign PanNETs, which included well-differentiated endocrine tumors with uncertain behavior, well-differentiated endocrine carcinomas, and poorly differentiated

Conflicts of interest

None of the authors have any conflicts of interest associated with this study.

References (27)

  • E. Solcia et al.

    World Health Organization international histological classification of tumours: histological typing of endocrine tumours

    (2000)
  • D.S. Klimstra et al.

    Neuroendocrine neoplasms of the pancreas

  • U.F. Pape et al.

    Prognostic relevance of a novel TNM classification system for upper gastroenteropancreatic neuroendocrine tumors

    Cancer

    (2008)
  • Cited by (81)

    • The diagnostic value of intravoxel incoherent motion diffusion-weighted imaging for distinguishing nonhypervascular pancreatic neuroendocrine tumors from pancreatic ductal adenocarcinomas

      2022, European Journal of Radiology
      Citation Excerpt :

      However, ADC values of nonhypervascular PNETs were significantly lower than PDACs in our study. The enhancement pattern of PNET was related with tumor grade [24], and nonhypervascular PNETs were more commonly high-grade tumors (G2/G3 grades) in our study, as previously reported in the literature [14,15]. As tumor grade increases, the increase of cell density and nuclear-to-cytoplasmic ratios lead to the restricted diffusion of water molecules [25], resulting in a lower ADC value in the nonhypervascular PNETs.

    • Non-gaussian models of 3-Tesla diffusion-weighted MRI for the differentiation of pancreatic ductal adenocarcinomas from neuroendocrine tumors and solid pseudopapillary neoplasms

      2021, Magnetic Resonance Imaging
      Citation Excerpt :

      According to recent reports, up to 42%–49% of PNETs may not present arterial enhancement, and these non-hypervascular PNETs may be difficult to distinguish from PDACs [10,11]. So, the differential diagnosis of solid pancreatic tumors based on both clinical and imaging manifestation is a challenge when the findings are atypical [15]. Therefore, new abdominal imaging techniques are needed to improve the diagnostic performance of differentiating PDACs from SPNs and PNETs.

    View all citing articles on Scopus
    View full text