CT differentiation of gastric ectopic pancreas from gastric stromal tumors

Background Gastric ectopic pancreas (GEPs) is a rare developmental anomaly which is difficult to differentiate it from submucosal tumor such as gastrointestinal stromal tumor (GIST) by imaging methods. So we retrospectively investigated the CT features of them to help us make the correct diagnosis. and Methods This study enrolled 17 GEPs and 119 GSTs, which were proven pathologically. We assessed clinical and CT features to identify significant differential features of GEPs from GSTs using univariate and multivariate analyses. In univariate analysis, among all clinicoradiologic features, features of age, symptom, tumor marker, location, contour, blurred serosa or fat-line of peritumor, necrosis, calcification, CT attenuation value of unenhancement phase/arterial phase/portal venous phase (CTu/CTa/CTp), the CT attenuation value of arterial phase/portal venous phase minus that of unenhanced phase (DEAP/DEPP), long diameter (LD), short diameter (SD) were considered statistically significant for the differentiation of them. And the multivariate analysis revealed that location, blurred serosa or fat-line of peritumor, necrosis and DEPP were independent factors affecting the identification of them.What's more, ROC analysis showed that the test efficiency of CTp was perfect(AUC= 0.900). Location, blurred serosa or fat-line of peritumor, necrosis and DEPP are useful CT differentiators of GEPs from GSTs. In addition, the test efficiency of CTp in differentiating them was perfect (AUC=0.900).


Introduction
Ectopic pancreas, which was first described by Klob in 1859, is a rare developmental anomaly that is defined as pancreatic tissue lacking anatomic or vascular continuity with the main body of the gland [1][2][3]. It can occur anywhere along distal end of the esophagus to the colon, and of the gastrointestinal lesions, stomach (30%) is the commonest area, followed by duodenum (25%) and jejunum (15%) [3]. Ectopic pancreas is usually discovered incidentally during surgery or autopsy and is generally asymptomatic. However some patients with ectopic pancreas may have clinical symptoms such as abdominal pain, gastrointestinal bleeding, and obstruction when complicated with pancreatitis, pseudocyst, insulinoma, and pancreatic cancer [2,[4][5][6][7].
Although there were many diagnostic imaging methods such as computed tomography (CT), ultrasonography and endoscopic ultrasonography, the diagnosis of ectopic pancreas was difficult due to its non-specific imaging characteristics. Ectopic pancreas may commonly perform as a submucosal mass in gastrointestinal tract [8]and were easily misdiagnosed with other submucosal tumor such as gastrointestinal stromal tumor (GIST) and leiomyoma. As we know, GISTs are the most common subepithelial lesions and account for 90% of gastric submucosal tumors [8,9]. What's more, the standard treatment for a GIST without metastasis is surgical resection and a GIST with metastasis is usually treated by tyrosine inhibitors such as imatinib. However, an ectopic pancreas would not need clinical treatment unless it is symptomatic, in addition, whether to remove suspected ectopic pancreas that is found incidentally is still a controversial issue [10]. So it is of great clinical significance to identify ectopic pancreas and gastrointestinal stromal tumors on image.
As we know, CT is the most commonly used noninvasive modality for preoperative evaluation of gastric tumors due to its well-standardized protocal and easy accessibility. However, CT features of gastric ectopic pancreas (GEPs) as well as differential imaging features of GEPs from GSTs have not yet been investigated completely. Therefore, our study retrospectively investigates the differential CT features of GEPs from GSTs.

Patients
Our retrospective study was approved by the ethics committee of XX hospital and XX hospital and did not require informed consent. All procedures performed in studies involving human participants were in compliance with the 1964 Helsinki Declaration and its later amendments.
We queried pathology database of our institute to derive all histologically proved cases of GEPs from January 2007 to June 2019 and GSTs from January 2016 to June 2019. All patients were histologically proven by surgical resection. Finally total 19 cases of GEPs and 147 cases of GSTs were identified in this query. We included patients who satisfied the following inclusion criteria: (1) patients who had preoperative CT images with optimal gastric distension, (2) patients who had integrated clinical date, (3) the lesion was solitary, (4) the maximum diameter of the tumor was no less than 10 mm. Finally, 17 patients with histopathologically proven GEPs and 119 patients with GSTs comprised our study population (Fig 1).

Clinical features
The clinical features of all patients were analyzed by one author (X.X.) using the medical records of hospital. For each patient, the gender, age, the presence of clinical manifestations and the level of tumor marker (ferrithin) were analysed.

CT image acquisition
All Enhanced CT images were obtained from multidetector CT scanners (SOMATOM, Sensation 16, Siemens, Forchheim, Germany and SOMATOM, Defnition AS+, Siemens, Forchheim, Germany). All patients drank 600-800 ml of water before CT examination. The CT scan parameters were set as follows: for SOMATOM Sensation 16, beam collimation = 1.2 mm × 16, pitch = 1, kVp/effective mA = 120/300, rotation time = 0.5 seconds and reconstruction section thickness=5mm; for SOMATOM, Defnition AS+, beam collimation = 1.2 mm × 32, pitch = 1, kVp/effective mA = 120/160, rotation time = 0.5 seconds, and reconstruction section thickness=5mm. Patients were injected with non-ionic contrast material (Ultravist; 300mg I/ml, Bayer Schering Pharma AG, Berlin, Germany or Iopamidol; 300mg I/ml, Bracco Sine Pharma AG, Shanghai, China) with antecubital venous access at a rate of 3.0 ml/sec and a total of 90-120ml (1.5 mL per kilogram of body weight) was injected by using the CTcompatible power injector during arterial and parenchymal phase. The scanning delay for arterial imaging was determined by using automated scan-triggering software Arterial scanning automatically began 7.0 second after the trigger attenuation threshold (100 HU) was reached at the level of the superior abdominal aorta and parenchymal scanning began at a delay of 45 second after arterial scanning,

Imaging analysis
All CT images were respectively interpreted by two radiologists (C.L. and J.W.) who had 4-yr (C.L.) and 12-yr (J.W.). Both radiologists were aware that the study population had either GEPs or GSTs, but were blinded to their histological subtypes. The following CT features of the primary gastric lesion were assessed: (1) the long diameter (LD) and short diameter (SD), (2)

Statistical analysis
All statistical analyses were performed using commercial software, SPSS 22.0 for Windows (SPSS Inc, Chicago, IL, USA). The prevalence of clinical and CT features were compared using Student's t-test and χ 2 test, and binary logistic regression analyses were performed to achieve the most significant differential CT features. Receiver operating characteristic (ROC) curve analysis was performed to ascertain the optimal cut-off value of significant quantitative CT features, such as LD, SD, CTu, CTa, CTp, DEAP, DEPP and to obtain the value of sensitivity and specificity of qualitative CT features, such as location, contour, the presence of serosal invasion or fat-line of peritumor, necrosis and calcification. A P value< 0.05 was considered statistically significance.

Clinical features
The clinical characteristics of GEP and GST patients were summarized in Table 1. There was no significant difference in gender distribution between the two groups. However, there were significant differences in mean age, symptom and the level of serum tumor marker (P 0.05).

Qualitative and Quantitative Image Analysis
Results of the qualitative and quantitative images analysis were presented in Table 2. The location of tumor was significantly different between the two groups (P 0.05) -Or rather, there was a significant difference in the distribution of tumors in gastric fundus between two groups (p<0.05). The majority of GEPs were located in body (64.7% [11/17] and 34.26HU±15.57for DEPP) (P 0.05). Moreover, the LD and SD were also the significant CT features in differentiating GEPs from GSTs (P0.05). In addition, multivariate analysis revealed that location, leakage or fat-line of peritumor (Fig 2&3), necrosis and DEPP were independent factors affecting the identification of GEPs and GSTs. Table 3 shows the sensitivity and specificity values of each significant CT feature for differentiating GEPs from GSTs. According to the ROC analysis (Fig 4), the largest area under the curve (AUC) was the CTp (0.900), followed by the DEPP (0.832), the CTa (0.821), the DEAP (0.806), the SD (0.757), the CTu (0.727) and the LD (0.726), which were significantly continuous variables differentiated GEPs from GSTs. Using clustered boxplot to study the significant continuous variables-CT attenuation value and diameter for differentiating GEPs from GSTs, we found the CT median attenuation value was universally higher for the GEPs than that of the GSTs, no matter what CTu, CTa, CTp, DEAP or DEPP, while the diameter was universally smaller, no matter LD or SD (Fig 5).

Discussion
Gastric ectopic pancreas (GEPs) is a pancreatic issue which was found outside its normal localization and without any anatomical or vascular connection with pancreas [3,11]. Although the majority of patients with GEPs were asymptomatic, a few patients may have clinical manifestations due to stimulation of hormones and enzymes secreted by the ectopic pancreas [11]. As the most common subepithelial lesions, GISTs account for 90% of gastric submucosal tumors [8,9], And it is difficult to differentiate GEPs from GSTs, so we compare the CT features of GEPs and GSTs to help us make the correct diagnosis.
Our study demonstrated that these CT findings like location, contour, the presence of leakage or fatline of peritumor, necrosis, calcification, CTu, CTa, CTp, DEAP, DEPP, LD and SD were statistically significant for the differentiation of GEPs and GSTs. Besides the multivariate analysis revealed that location, blurred serosa or fat-line of peritumor, necrosis and DEPP were independent factors affecting the identification of GEPs and GSTs and CTp performed the perfect test efficiency in differentiate them. Our results may be useful for us to diagnose so as to avoid unnecessary surgery or invasive endoscopic US in patients with asymptomatic ectopic pancreas.
Our study indicated that 0 (0%) of 17 cases of GEPs were located in gastric fundus and the majority of GEPs (64.7%) were located in gastric body, what's more, only 5 cases (29.4%) were located in gastric antrum. In contrast, 41 (34.4%) of 119 GSTs were located in gastric fundus and 56 (47.1%) cases were located in gastric body. However, the location of GEPs is inconsistent with the well-known fact that ectopic pancreases in upper gastrointestinal system are usually located in the gastric antrum [6][7][8][12][13][14], and this may be resulted from mall sample size of GEPs.  [14,15] and consistent with typical endoscopic ultrasonography (EUS) characteristics, that is indistinct borders, lobulated margins, presence of anechoic duct-like structures, a mural growth pattern, and localization within two or more layers [16,17]. The feature of blurred serosa is very closely correlated with the histological feature of a lobular structure of the acinous tissue at the margin [16]. Since most GEPs are exophytic growth pattern (64.7% [11/17]) and GEPs is not a true neoplasms but a hamartoma that flat pancreatic acinar formation and duct development histologically [6][7][8], so it is of high possibility for GEPs to have fat space between the tumor and serosal layer. As for GISTs, the main endoscopic finding of it is a nonspecific smooth bulge covered with normal mucosa, which is common to all subepithelial lesions [9], so the possibility for GISTs to have fat space between the tumor and serosal layer is extremely low. What's more, Mietinenn [18] reported the metastatic risk of GISTs increases according to the tumor size irrespectively of the mitotic count and the probability of malignancy was significantly increased when the tumor was larger than 5cm in diameter. In our series, 86 cases of GISTs (72.3% [86/119]) were less than 5cm in diameter and it is of low probability for GISTs in our study to invaded the serosal layer, which lead to the seldom presence of blurred serosa.
In our series, 0 (0%) of the 17 cases of ectopic pancreas had the presence of calcification, surface ulceration and lymph node, just 1 (5.9%) case of GEPs had necrosis, our results were supported by the fact that GEPs is not a true neoplasm but a hamartoma of flat pancreatic acinar formation and duct development [6][7][8]. Our study also showed that the LD and SD of GEPs were shorter than GSTs, but the previous study [14] regarding the CT features of GEPs didn't regard it as a characteristic CT finding.
Our study demonstrated that the CT attenuation values of CTu, CTa and CTp of GEPs were significantly higher than that of GSTs. Besides, the degree of enhancement was much heavier for the GEPs than that of GSTs, both in the DEAP and DEPP. What's more, ROC analysis showed that the test efficiency of CTp in differentiating them was perfect (AUC=0.9). The majority of GEPs appeared as homogeneously extramucosal masses with similar or higher attenuation to pancreas and this result may be attributable to the histologic similarity of GEPs to normal pancreatic tissue, especially acini.
Microscopically, GEPs consist of pancreatic acini and ducts and rarely contain islet cells. If the GEPs were mainly composed of pancreatic acini, the lesions would show greater enhancement and have a higher CT attenuation value in portal venous phase than the pancreas. On the contrary, if the GEPs consisted of predominantly ducts, the lesions would have lower CT attenuation values than pancreas and even the back muscles [14,19,20]. Our result showed that 16 of 17 GEPs mainly consisted of pancreatic acini and 1 of 17 GEPs contained many ducts and a few acini, and it is similar to the report of Yamagiwa H et al [20]. But for GSTs, as we know, GSTs had malignant potential and fast growth rate so as to tumor cell prone to degeneration, which undoubtedly decreased the degree of enhancement of GSTs [9,21].
Our study has several limitations. Firstly, this was a retrospective study, the selection bias and the use of various CT scanners were inevitable. Secondly, only two types of gastric mesenchymal tumors were compared and other subepithelial lesions were ignored. Besides, in order to avoid the influence of size bias, we excluded larger GSTs because most of larger-sized gastric mesenchymal tumors can be diagnosed as GST eventually [22]. Thirdly, our results were based on clinical data obtained from an eastern country at a single institution, which could not reflect fully the image differences between the different races and regions of two groups.
In conclusion, gastric ectopic pancreas had characteristic CT findings that differ from those of gastric submucosal tumors such as GSTs. When characteristic CT imaging findings are used in combination, ectopic pancreas can be differentiated from with a high degree of diagnostic accuracy.

3.Availability of data and materials:
The data underlying this paper are available upon request to the corresponding author due to ethical restrictions.

4.Competing interests:
The authors declare that they have no competing interests.      Figure 1 Flowchart of study base on recommended standards for differentiating diagnosis accuracy.    Clustered boxplot graph shows that the CT median attenuation value was universally higher for the GEPs than that of the GSTs, no matter what CTu, CTa, CTp, DEAP or DEPP, while the diameter was universally smaller.