Ectopic Pancreas: CT Findings with Emphasis on Differentiation from Small Gastrointestinal Stromal Tumor and Leiomyoma

Introduction

Ectopic pancreas, also referred to as heterotopic, accessory, or aberrant pancreas, is defined as pancreatic tissue lacking anatomic and vascular continuity with the main body of the gland (^,1–^,3); an ectopic pancreas is usually located in the stomach, duodenum, or jejunum (^,1). Most of the lesions are small and asymptomatic. They are typically discovered incidentally during surgery or autopsy, with an incidence at laparotomy of 0.2% (^,4–^,6). However, some patients with ectopic pancreas develop complications such as pancreatitis, pseudocyst, insulinoma, and pancreatic cancer, which cause clinical symptoms such as abdominal pain, gastrointestinal bleeding, and obstruction (^,4,^,7–^,9).

Ectopic pancreas may manifest as a submucosal mass in the stomach or duodenum and may easily be misinterpreted as another submucosal tumor such as gastrointestinal stromal tumor (GIST) or leiomyoma on imaging studies or at endoscopic examinations, including endoscopic ultrasonography (US). Because GISTs are by far the most common submucosal tumors of the stomach and constitute about 90% of gastric submucosal tumors (^,7), ectopic pancreas can frequently be mistaken for GIST or leiomyoma at endoscopy, US, or computed tomographic (CT) scanning. Furthermore, invasive endoscopic biopsy performed by using standard biopsy forceps also usually cannot enable a definitive diagnosis of submucosal tumor (^,10).

Currently, a number of imaging modalities, including endoscopic US, CT, and barium studies, are used for the detection and characterization of gastric submucosal tumors. Although several studies (^,6,^,10) have demonstrated the diagnostic value of endoscopic US for accurate characterization of gastric lesions, endoscopic US is invasive and dependent on operator skill. On the other hand, CT is the most commonly used noninvasive modality for the preoperative evaluation of gastric tumors. However, to our knowledge, there have been only a few reports (^,11–^,14) that describe the CT findings in ectopic pancreas; in these reports, the authors stated that there were no differential CT findings between ectopic pancreas and other submucosal tumors, except for their location.

The purpose of this study was to describe the CT findings in ectopic pancreas and to identify the features that differentiate this disease from other similarly manifesting gastric submucosal tumors such as GIST and leiomyoma, which are the most common gastrointestinal submucosal tumors.

MATERIALS AND METHODS

Patients

This retrospective study was approved by the institutional review board of Seoul National University Hospital, and the requirement for informed consent was waived. We searched our pathology database from January 2001 to November 2006 by using the search terms “ectopic pancreas and stomach or duodenum,“ “GIST and stomach or duodenum,” and “leiomyoma and stomach or duodenum” and found 32 patients with ectopic pancreas, 287 patients with GIST, and 15 patients with leiomyoma. To determine the precision of our query phrases, we re-searched the database by using various combinations of the following key words: “stomach,” “gastric,” “ectopic pancreas,” “heterotopia,” “GIST,” “leiomyoma,” “duodenum,” “submucosal tumor,” and “mesenchymal tumor.” The additional search revealed the same numbers of patients with ectopic pancreas, GIST, and leiomyoma, as well as a number of patients with other submucosal gastric tumors—16 patients with carcinoid tumor, nine patients with schwannoma, and five patients with lipoma. All lesions were confirmed at histopathologic examination of specimens obtained at surgery or biopsy. To develop a study group of suitable cases in order to compare the CT findings in the three diseases, we used the following inclusion criteria: (a) pathologic diagnosis of GIST, leiomyoma, or ectopic pancreas of the stomach or duodenum; (b) available contrast material–enhanced helical CT images; and (c) lesion smaller than 4 cm in diameter. The criterion of a tumor smaller than 4 cm in diameter was used to avoid potential bias caused by size differences in these tumors, as most previously reported cases of ectopic pancreas and all of our patients with ectopic pancreas had tumors smaller than 4 cm in diameter.

From among the 334 patients, we excluded 236 patients with GISTs and four patients with leiomyomas greater than 4 cm in diameter at pathologic measurement. Furthermore, 18 ectopic pancreases, 18 GISTs, and four leiomyomas were excluded because there were no available CT images. Finally, 33 patients with GISTs (mean age, 56.6 years ± 10.8 [standard deviation]; range, 27–84 years; 16 men [mean age, 56.6 years ± 6.31; range, 41–65 years] and 17 women [mean age, 56.6 years ± 13.95; range, 27–84 years]), 14 patients with ectopic pancreas (mean age, 42.9 years ± 10.1; range, 31–64 years; six men [mean age, 49.5 years ± 11.9; range, 31–62 years] and eight women [mean age, 46.8 years ± 9.13; range, 36–64 years]), and seven patients with leiomyoma (mean age, 49 years ± 5.7; range, 41–59 years; three men [mean age, 52.7 years ± 6.0; range, 47–59 years] and four women [mean age, 52.5 years ± 12.6; range, 41–51 years]) were included in our study (^,Fig 1) (^,15).

Among the 33 patients with GIST, five had abdominal discomfort or pain, two had melena, and one had nausea. The remaining 25 patients with GIST were asymptomatic. Of the 14 patients with ectopic pancreas, four had abdominal discomfort, and one had melena. The remaining nine patients had no symptoms. Only one patient with a leiomyoma had abdominal discomfort, while the remaining six patients with leiomyoma had no symptoms.

Among the 14 instances of ectopic pancreas, 10 were located in the stomach and four were located in the duodenum. However, 32 (97%) of the 33 GISTs and all of the leiomyomas were located in the stomach.

Acquisition of CT Images

CT images were obtained by using a single–detector row helical CT scanner (Somatom Plus 4, Siemens, Erlangen, Germany; or HiSpeed, GE Medical Systems, Milwaukee, Wis) or a multi–detector row CT scanner (Mx 8000, Philips Medical Systems, Cleveland, Ohio; LightSpeed Ultra, GE Medical Systems; or Sensation 16, Siemens Medical Systems, Forchheim, Germany). Each patient received 120 mL of nonionic contrast material (iopromide, Ultravist 370; Schering, Berlin, Germany) through an 18-gauge angiographic catheter inserted into a forearm vein; a mechanical injector was used at a rate of 3.0 mL/sec. For the single–detector row scanners, the CT parameters were as follows: 5–7-mm collimation, 1:1 table pitch, and 5–7-mm reconstruction intervals. For the multi–detector row CT scanners, a 1.5–2.5-mm detector collimation, 12.5–27-mm table feed per gantry rotation, 3–5-mm section thickness, and 3–5-mm reconstruction interval were used. CT images were obtained during the portal venous phase, and the delay range between contrast material administration and scanning was 60–70 seconds.

Image Analysis

Qualitative analysis.—Three radiologists—two attending physicians (J.M.L. and J.Y.C., with 17 and 10 years of experience, respectively) and a clinical fellow (H.S.P., with 8 years of experience)—retrospectively reviewed the CT findings in consensus. They were blinded to the clinical findings and to the histopathologic results. The images were presented to the readers in a random sequence. The images were reviewed on a picture archiving and communications system (Marotech, Seoul, Korea) by using a stack mode. One of the authors (J.Y.K.) who was not involved in the image analysis reviewed the medical records of the patients, including clinical features, surgical findings, and the results of pathologic examination.

To enable us to compare the CT findings in the three disease entities of the stomach and duodenum, the following CT findings were analyzed: (a) location, (b) contour, (c) growth pattern (^,Fig 2^,,), (d) lesion border, (e) enhancement pattern, (f) enhancement grade of the lesion, (g) the presence of surface dimple, (h) prominent enhancement of the overlying mucosa, and (i) intralesional low-attenuation area. Lesion contours were classified as round, ovoid, dumbbell, or flat. Growth patterns were classified as endoluminal, exoluminal, or mixed (^,16). Endoluminal growth was defined as when a tumor was attached to the bowel wall and was completely confined to the bowel lumen without bulging into the extraluminal space. Conversely, exoluminal growth was defined as when a mass was confined to the extraluminal space without bulging into the bowel lumen. A mixed growth pattern was noted when tumors did not show either of the two previously described patterns. Lesion borders were classified as ill defined versus well defined. Enhancement patterns (homogeneous vs heterogeneous) were assessed, and the relative enhancement grade (high, intermediate, or low) of the lesions was judged in comparison to the enhancement of back muscles. The presence of surface dimple, prominent enhancement of overlying mucosa, and low intralesional attenuation (an area with a CT attenuation value < 20 HU) was also analyzed (^,5,^,8,^,11,^,17).

Objective analysis.—To assess the degree of enhancement of the three kinds of tumor, CT attenuation values of the tumors were measured in Hounsfield units by one of the authors (J.Y.K., a 3rd year resident), who was not involved in the qualitative image analysis, by using 16–18-mm² circular regions of interest (ROIs). The ROI cursors were carefully placed to encompass as much of the most strongly enhanced portion of the tumors as possible and to avoid adjacent structures. At least three ROIs were placed on the most strongly enhanced portion of the tumors, and the ROI values were averaged as the mean tumor attenuation. The investigator also measured the long diameter (LD) and the short diameter (SD) of each lesion, and the LD/SD ratios of the lesions were calculated.

Radiologic-Pathologic Correlation

Pathology slides of the ectopic pancreases were reviewed, with focus mainly on the composition of the lesion and the condition of the overlying gastric or duodenal mucosa, by an expert (M.A.K., an attending abdominal pathologist with 11 years of experience) who was blinded to the CT findings. The CT findings in the cases of ectopic pancreas were then compared with the histopathologic findings. For 11 of the 14 patients with ectopic pancreas, pathology slides of the surgical specimens were available for radiologic-pathologic correlation. For the three remaining patients, ectopic pancreas was confirmed only with examination of an endoscopic biopsy specimen in two patients; in the last patient, who underwent surgery, the entire lesion was not fully covered on the pathology slides. Therefore, although it was possible to establish the diagnosis of ectopic pancreas in these three cases of ectopic pancreas, they could not be included in the radiologic-pathologic correlation, which required the whole lesion for evaluation.

Statistical Analysis

To evaluate the differences in demographic data, including age, sex, and symptoms, between the three submucosal tumors, the analysis of variance and χ² tests were used. The χ² test was used to compare the categoric variables of interest in the ectopic pancreas and the other two submucosal tumors (GIST and leiomyoma); these variables included location, contour, growth pattern, border, enhancement pattern, enhancement grade, surface dimple, prominent enhancement of overlying mucosa, and low intralesional attenuation. We used the t test to compare the LD/SD ratios and CT attenuation values between the ectopic pancreases and the other two submucosal tumors. A P value of less than .05 was considered to indicate a significant difference, and a 95% confidence interval was reported for each variable.

Sensitivity and specificity analyses were performed for the LD/SD ratio of each lesion in order to generate a receiver operating characteristic curve and to determine the optimal LD/SD ratio cutoff for the differentiation of the three submucosal lesions. The optimal cutoff point was defined as the value at which the sum of the sensitivity and the specificity was maximized. Odds ratios (with 95% confidence intervals) for differentiating ectopic pancreases from the other two submucosal tumors were calculated for each CT criterion. The sensitivity and specificity of each significant CT criterion and a combination of significant CT criteria were also calculated. Numbers used in this statistical analysis were the numbers of lesions. For statistical analysis, we used software (MedCalc for Windows, version 8.0.0.1; MedCalc Software, Mariakerke, Belgium). Because we wanted to identify only features capable of differentiating ectopic pancreas from other gastric submucosal tumors, statistical analysis was not performed to compare GISTs and leiomyomas.

RESULTS

There was no significant difference in sex distribution between the three submucosal tumors. Furthermore, there were no significant differences in the patients' symptoms or disease states. However, there was a significant difference in age distribution between the patients with GIST and those with ectopic pancreas (mean age, 56.6 years ± 10.8; range, 27–84 years vs mean age, 42.9 years ± 10.1; range, 31–64 years; P < .05).

Morphologic Analysis

The morphologic CT findings of ectopic pancreas, GIST, and leiomyoma are summarized in ^,Table 1. The distribution of tumor location was significantly different between these three tumors (P < .001). The most common location for each tumor was the antrum (50% [seven of 14]) or duodenum (29% [four of 14]) for ectopic pancreas, the body (39% [13 of 33]) or fundus (27% [nine of 33]) for GISTs, and the cardia (86% [six of seven]) for leiomyoma. With regard to lesion shape, nine (64%) of 14 ectopic pancreases were flat or ovoid, whereas most GISTs (91% [30 of 33]) and leiomyomas (86% [six of seven]) were round or ovoid. Furthermore, ectopic pancreases showed a tendency toward endoluminal growth (86% [12 of 14]), while GISTs and leiomyomas showed an even distribution of endoluminal or exoluminal growth (P = .043). Neither surface dimple nor low intralesional attenuation was a significant factor in differentiating ectopic pancreas from submucosal tumors.

The three types of tumors showed different enhancement patterns, but there was no significant difference—that is, ectopic pancreases showed predominantly heterogeneous enhancement (71% [10 of 14]), while the other tumors did not (^,Figs 3^,, E1 [ http://radiology.rsnajnls.org/cgi/content/full/252/1/92/DC1]). In terms of enhancement grade, ectopic pancreases and GISTs showed higher attenuation than back muscle, whereas leiomyomas showed lower attenuation. The difference in enhancement grade between ectopic pancreases and leiomyomas was significant (P < .05). Representative images are presented in ^,Figures 4^,,–^,5^,, E1, E2 ( http://radiology.rsnajnls.org/cgi/content/full/252/1/92/DC1). Prominent enhancement of the overlying mucosa was detected in only four patients (29%) with ectopic pancreas (P = .004).

Quantitative Image Analysis

Results of the quantitative analysis are presented in ^,Table 2. Ectopic pancreases and GISTs showed higher mean CT attenuation than leiomyomas: 106.07 HU ± 30.37 (standard deviation) for ectopic pancreases versus 98.1 HU ± 42.11 for GISTs and 57.3 HU ± 19.13 for leiomyomas.

The mean LD/SD ratios for the ectopic pancreases and the other submucosal tumors (GISTs and leiomyomas) were 1.5 ± 0.37 and 1.27 ± 0.30, respectively (^,Fig 6). The mean LD/SD ratio for the ectopic pancreases was significantly higher than that for the other two submucosal tumors (P < .05). At receiver operating characteristic curve analysis, we set the cutoff value for the LD/SD ratio of the lesion as 1.4 for the differential diagnosis of ectopic pancreas and the two gastric submucosal tumors (GIST and leiomyoma), which yielded a sensitivity of 64.3%, a specificity of 82.5%, and an area under the receiver operating characteristic curve of 0.72.

Radiologic-Pathologic Correlation

In our study, ectopic pancreases for which pathology slides were available could be classified into three subtypes according to their histopathologic composition—that is, predominantly pancreatic acini (three of 11, 27%), predominantly ducts (one of 11, 9%), and mixed (seven of 11, 64%). All three ectopic pancreases with predominantly pancreatic acini showed greater enhancement and had a higher CT attenuation value in the portal venous phase than did the pancreas. On the other hand, the one ectopic pancreas that was predominantly composed of ducts had lower CT attenuation values than the pancreas and even the back muscles. The other seven ectopic pancreases with mixed composition showed variable CT attenuation values compared with the pancreas (higher attenuation, three [43%]; isoattenuation, one [14%]; lower attenuation, three [43%]). Representative images are shown in ^,Figures 3^,, ^,4^,,, E1 ( http://radiology.rsnajnls.org/cgi/content/full/252/1/92/DC1).

Furthermore, the enhancement pattern (heterogeneous vs homogeneous) of the ectopic pancreases differed according to the microscopic composition of the ectopic pancreases. The three ectopic pancreases with predominantly pancreatic acini showed a homogeneous enhancement pattern. However, all seven lesions with a mixed composition of acini and ducts showed a heterogeneous enhancement pattern.

With regard to the overlying mucosa, among the four ectopic pancreases with prominent enhancement of the overlying mucosa, two showed marked inflammation of the overlying gastric mucosa, but the remaining two, in which the diagnosis was confirmed only with endoscopic biopsy, had no pathology slides appropriate for evaluating the overlying mucosa (^,Fig 4^,,). Among the 10 ectopic pancreases without prominent enhancement of the overlying mucosa, the nine with available pathology slides showed mild gastritis (n = 3) or no marked abnormality (n = 6).

Sensitivity and Specificity Values for CT Diagnosis

Table 3 summarizes the sensitivity and specificity values of each significant CT criterion for differentiating ectopic pancreas from the two other submucosal tumors. Except for prominent enhancement of the overlying mucosa, the highest odds ratio was achieved for the border of the lesion, followed by the location, the LD/SD ratio, and the growth pattern of the lesion.

When at least two of these five criteria were used in combination, the sensitivity and specificity for diagnosing ectopic pancreas were 100% (14 of 14) and 82.5% (33 of 40), respectively. When any four or five of these criteria were used, a specificity of 100% was achieved (^,Table 4).

DISCUSSION

Our study results demonstrate that five CT findings—that is, prominent enhancement of the overlying mucosa, location, LD/SD ratio, growth pattern, and lesion border—are statistically significant predictors in the differentiation of ectopic pancreas from GIST and leiomyoma. In addition, when these CT imaging findings were used in combination as the diagnostic criteria, ectopic pancreas could be differentiated from submucosal tumors with a high degree of diagnostic accuracy. When the results of our study are applied in differentiating ectopic pancreas from other submucosal tumors at CT, they may be helpful in establishing an accurate diagnosis, thereby avoiding unnecessary surgery or invasive procedures such as endoscopic US in patients with asymptomatic ectopic pancreas.

In our study, an LD/SD ratio greater than 1.4 was found to be the one of the important morphologic CT features of ectopic pancreas for differentiating it from other tumors (P < .05). Considering that histologically, ectopic pancreas of the stomach is not a true neoplasm but is a hamartoma of flat glandular tissue with pancreatic acinar formation and duct development, (^,4,^,7,^,9,^,14), it is not surprising that ectopic pancreases may have a flat-ovoid shape (high LD/SD ratio) and an ill-defined border. Only after intraglandular cyst formation, an ectopic pancreas may manifest as a large protruding submucosal mass with a rather oval or round shape (^,14,^,16). Furthermore, interestingly, ectopic pancreas more often tends toward endoluminal growth (86%), but GIST tends toward an exophytic or mixed growth pattern. This finding is in accordance with the results of a previous study (^,12) regarding the CT findings of GISTs that showed that large GISTs tend toward exophytic growth.

In our study, 10 (71%) of 14 cases of ectopic pancreas showed an ill-defined margin at CT. This result is consistent with other reports that describe the endoscopic US features of ectopic pancreas (^,17,^,18). According to these previous studies, this feature of ectopic pancreas is explained by the lobular structures of the acinous tissue. However, in previous studies (^,11) regarding the CT features of ectopic pancreas, the ill-defined margin was not considered to be the characteristic CT finding.

Furthermore, prominent enhancement of overlying mucosa was detected exclusively in patients with ectopic pancreas (four of 14, 29%). This finding was also described in a previous study by Cho et al (^,11). Considering our histopathologic finding that the two ectopic pancreases with available pathology slides and prominent enhancing overlying mucosa showed marked microscopic gastritis, prominent enhancement of the overlying mucosa of the ectopic pancreas at CT may be explained by the repeated inflammatory changes caused by ectopic pancreas.

In our series, seven (50%) of the 14 cases of ectopic pancreas were identified in the gastric antrum. On the contrary, five (15%) of 33 GISTs and one (14%) of seven leiomyomas were located in the gastric antrum. This finding about the location of ectopic pancreases is consistent with the well-known fact that ectopic pancreases in the upper gastrointestinal system are commonly found in the gastric antrum within 6 cm of the pyloric canal (^,4,^,7–^,10).

According to the results of radiologic-pathologic correlation in our study, the degree and pattern of enhancement of ectopic pancreases reflected the microscopic composition of the lesions—that is, the predominantly acini-dominant type showed more homogeneous enhancement and a stronger degree of enhancement than the other types. In addition, leiomyomas showed a distinct enhancement grade and pattern compared with the ectopic pancreases and the GISTs. Most leiomyomas showed a homogeneous enhancement pattern (five [71%] of seven), a low or intermediate enhancement grade (six [86%] of seven), and were predominantly located in the cardia (six [86%] of seven).

Our study had several limitations. First, because this was a retrospective study, some degree of selection bias and the use of various CT scanners could not be avoided. Second, we did not analyze the diagnostic performance of helical CT by using receiver operating characteristic analysis with multiple reviewers in both blinded and individual fashion. Third, because we used the inclusion criterion of a tumor smaller than 4 cm in diameter, a large number of GISTs and leiomyomas were excluded. In our study, we excluded submucosal tumors greater than 4 cm in diameter, and we compared ectopic pancreases and other small submucosal tumors. Nevertheless, given the fact that many gastric submucosal tumors are usually detected as small-diameter tumors with the more frequent use of endoscopy and cross-sectional imaging modalities in evaluating abdominal diseases, such tumors may be difficult to differentiate on the basis of the diameter measurement only. Therefore, comparing ectopic pancreases and other small submucosal tumors may be more clinically important. Hence, we attempted only to identify differential findings between ectopic pancreases and other small submucosal tumors.

In conclusion, ectopic pancreas of the stomach and duodenum has characteristic CT findings that differ from those of gastric submucosal tumors such as GIST and leiomyoma. When characteristic CT imaging findings are used in combination, ectopic pancreas can be differentiated from submucosal tumors with a high degree of diagnostic accuracy.

ADVANCE IN KNOWLEDGE

IMPLICATION FOR PATIENT CARE

Figure 1:

Download as PowerPointOpen in Image Viewer

Figure 2a:

Download as PowerPointOpen in Image Viewer

Figure 2b:

Download as PowerPointOpen in Image Viewer

Figure 2c:

Download as PowerPointOpen in Image Viewer

Figure 3a:

Download as PowerPointOpen in Image Viewer

Figure 3b:

Download as PowerPointOpen in Image Viewer

Figure 4a:

Download as PowerPointOpen in Image Viewer

Figure 4b:

Download as PowerPointOpen in Image Viewer

Figure 4c:

Download as PowerPointOpen in Image Viewer

Figure 5a:

Download as PowerPointOpen in Image Viewer

Figure 5b:

Download as PowerPointOpen in Image Viewer

Figure 6:

Download as PowerPointOpen in Image Viewer

Table 1. Categoric Data: CT Findings in Ectopic Pancreas, GIST, and Leiomyoma

Note.—Data are numbers of patients, with percentages in parentheses. Percentages may not add up to 100% because of rounding.

^* Between ectopic pancreas and other tumors (GIST and leiomyoma); calculated with χ² test.

Table 2. Quantitative Image Analysis Results for Three Types of Gastric Submucosal Tumor

Note.—Data are means ± standard deviations.

^* Between ectopic pancreas and other tumors (GIST and leiomyoma); calculated with Student t test.

Table 3. Sensitivity and Specificity of CT Findings in Diagnosis of Ectopic Pancreas among Gastric Submucosal Tumors

Note.—Unless otherwise specified, data in parentheses are numbers of patients.

^* Data in parentheses are 95% confidence intervals. NA = not applicable (no GISTs or leiomyomas showed prominent enhancement of overlying mucosa).

Table 4. Combined CT Findings in Patients with Gastric Submucosal Tumors

Note.—Data are numbers of patients with one or more of the following lesion findings at CT: flat-ovoid shape (LD/SD ratio > 1.4), typical location (antrum, pylorus, or duodenum), endoluminal growth pattern, ill-defined border, and prominent enhancement of overlying mucosa. Numbers in parentheses are percentages.