Gastric heterotopic pancreas and stromal tumors smaller than 3 cm in diameter: clinical and computed tomography findings

Heterotopic pancreatic (HP) masses and stromal tumors (STs) are common gastric submucosa tumors. HP masses are typically found in autopsy or surgery, during which the frequency is approximately 0.2 to 0.25% [1, 2]. Approximately 10–15 per million people worldwide are diagnosed with gastrointestinal STs each year, with most of these tumors located in the stomach [3]. Both the management and prognosis of these two tumors are different [4‐6]. STs are aggressive tumors with a potential tendency for malignancy, and the risk increases as the tumor increases in size. They require resection once detected, and occasionally, chemotherapy is required in cases of metastasis [7]. HP is a congenital anomaly, similar to hamartoma. Most patients are recommended to undergo surveillance because HP is generally asymptomatic, and only a few patients need to be treated because of complications [8]. In view of the above, an accurate preoperative diagnosis of submucosal tumors is critical.

Identifying HP masses and STs only by clinical features is difficult because both can manifest as abdominal pain, abdominal distension, and other symptoms. Although gastroscopic biopsy is regarded as the gold standard for the diagnosis of tumors, limitations include its invasive nature, sampling errors, diagnostic errors, long waiting times for immunohistochemistry results [9], and powerless assessing situations outside the tumor. Computed tomography (CT), as a common imaging examination method, has been used in the preoperative evaluation of submucosal tumors, and can be used as a supplemental tool in differentiating HP masses from STs [10].

Although many studies have been conducted on the imaging features of HP masses and gastric stromal tumors (STs), most have employed endoscopic ultrasonography [11, 12], with few adopting CT characteristics, and most have been case reports [10, 13]. To our knowledge, no studies have reported CT findings for differentiating gastric submucosal tumors less than 3 cm in diameter. Kim [14] reviewed CT findings of HP masses and other gastric submucosal tumors smaller than 4 cm, but size differences between the two groups inevitably led to errors in the results. Therefore, the purpose of this study was to analyze the clinical characteristics and CT findings of HP masses and STs less than 3 cm in diameter, identified in our hospital within a 5-year period, and to identify the features that differentiate one from the other.

This retrospective study was approved by the Institutional Review Board of the First Affiliated Hospital of Zhengzhou University, and the requirement for informed consent was waived.

This retrospective study included gastric lesions less than 3 cm in diameter; therefore, thin-layer images were required to analyze the CT features. Variables with significant differences in primary cohort were compared in validation cohort and there were significant differences in all above variables between two groups in validation cohort.

This study revealed significant differences in tumor location and border between the two groups, which were consistent with a previous report [20]. STs were more often located in the upper part (78%, 39/50), followed by the middle part (16%, 8/50), whereas most HP masses were located in the lower part (66%, 33/50), followed by the middle part (30%, 15/50). This trend of HP masses has been confirmed by many experts [21‐23] and the hypothesis that HP masses are fragments separated from the main pancreas during the embryonic rotation process may explain this distribution [8, 24]. Most STs (90%, 45/50) exhibited well-defined margins, whereas most HP masses (56%, 28/50) exhibited ill-defined margins. These ill-defined margins were related to the histological structure of the HP masses. HP masses consisted of pancreatic acini, ductal components, and islets at different proportions. Approximately 27 to 76.6% of the HP masses exhibited prominent acinar features [14, 21] with lobular architecture. When this component was located in the peripheral part, the margin of the tumor was ill-defined [25]. However, the ill-defined margin was considered to be an adverse factor for the risk grading of STs in a previous study [26].

Significant differences were found in tumor contour and LD/SD ratio. In our study, 42% (21/50) of STs were round, but only 8% of (4/50) HP masses were round. Jang [27] also reported that other submucosal tumors are more likely to be round than HP masses (46.2% vs 6.7%). An HP mass is defined as ectopic flat glandular tissue with pancreatic acinar formation, and therefore commonly resembles the slender appearance of a normal pancreas and often exhibits a broad base on endoscopy [28]. HP masses of the mesentery are more elongated in shape than gastric HP masses, according to Seo [29]. In contrast, STs are real neoplasms, composed of spindle cells, epithelioid cells, or a mixture, with an obvious vertical growth trend. Our results showed that the mean SD for HP masses was significantly smaller than that for STs, but the LD did not differ significantly between the two groups. The greater LD/SD ratio also indicated a wall growth pattern for HP masses. In our study, the diagnostic value of the LD/SD ratio was medium, with an AUC of 0.707. An LD/SD ratio greater than 1.29 was found to be one of the critical imaging features of HP masses for differentiating it from STs. The LD/SD ratio trend was consistent with that mentioned in a previous study [14], but with a lower cut-off value. We concluded that the greater the diameter of an ectopic pancreatic mass, the more obvious the LD/SD ratio trend will be.

Highly significant differences were observed between the two groups in both qualitative and objective analyses. Qualitative analysis results indicated that both lesions presented homogeneous enhancement and had a greater enhancement grade in the venous phase than in the arterial phase. Hence, the CT attenuation values in the venous phase were used in subsequent calculations. In objective analysis, we assessed the enhancement grade by using the CT enhancement value and the enhancement ratio, rather than the values in the venous phase. The application of the enhancement value and ratio may reduce the influence of differences in machine and individual variations, which have been a factor in many previous studies [30, 31]. The presence of marked enhancement is a crucial finding with regard to avoiding misdiagnosis of HP masses and distinguishing them from STs. The study by Kim [14] indicated that the enhancement grade of HP masses has a close relationship with histological components. HP masses with predominant acini present greater enhancement than those with predominant ducts [22, 32]. Additionally, the enhancement ratio of tumor to pancreas is also crucial for distinguishing HP masses from STs. The average ratio of HP masses was significantly greater than that of STs and closer to 1, which is consistent with previous magnetic resonance imaging findings [27]. The AUC of the enhancement ratio to pancreas was greater than that of the enhancement value and the enhancement ratio, which indicated the ratio of the CT value of an HP mass to that of the pancreas is more valuable for identification than the variables (the enhancement value, the enhancement ratio) of the lesion itself.

In our study, both tumors predominantly exhibited an endo-luminal growth pattern; 30% of STs exhibited a predominantly exo-luminal or mixed growth pattern, which was lower than the proportion reported in a previous study on tumors less than 4 cm in diameter [14]. In addition, STs with a mean diameter of 10 cm also exhibit an obvious exo-luminal growth pattern [33]. We concluded that the larger a tumor is, the more obvious the exo-luminal growth pattern will be. The smaller a tumor is, the more difficult identification will be.

In our study, some findings, such as the presence of prominent thickness of the overlying mucosa, central umbilication, and multiple lesions, were detected in only a few (10, 6, 2%, respectively) patients with HP masses, and other findings, such as calcification and ulceration were detected in only a few (10, 8%, respectively) patients with STs. We assessed whether these findings constitute specific characteristics. First, central umbilication suggesting a rudimentary duct is present radiographically in only 16 to 25% of HP masses [14, 34], but present endoscopically in 35 to 60% of HP masses [12, 35]. Endoscopic ultrasonography is superior to CT in the assessment of mucosal surfaces. To our knowledge, no studies have reported STs with central umbilication, but ulceration may be confused with central umbilication, which was observed in 8% of STs in our study. Second, recurrent inflammatory changes caused by HP masses may explain the prominent thickness and enhancement of the overlying mucosa indicating microscopic gastritis in HP masses in a previous study [14]. Third, only one patient with HP exhibited multiple lesions; however, a previous study reported the presence of multiple lesions as a typical characteristic of succinate dehydrogenase deficient STs [36]. Therefore, the presence of multiple lesions is a rare and non-specific feature. Finally, HP with calcification has been reported [37], indicating that calcification is also not a specific feature.

Our study had several limitations. First, two CT scanners were used in our retrospective study, resulting in nonconformity of the scanned parameters and volumes of contrast media. However, we believe that morphological features may be unaffected by nonconformity, and the application of the enhancement values, enhancement ratios, and the enhancement ratios to pancreas may greatly reduce the influence of nonconformity. Second, histological analysis was not performed and radiologic-pathologic correlation was not assessed in our study because a complete pathological specimen was not available for all the included patients, and we could not guarantee that the sample and the ROI were at the same level. Third, HPs of mainly cystic composition were excluded. This induces a bias, but considering their particular CT findings the diagnosis of mainly cystic HP does not constitute a significant radiological problem. Finally, logistic regression analysis, texture analysis, and nomography [38] were not performed in this study. Larger prospective investigations are needed to confirm the present findings.

In conclusion, HP has characteristic CT features for differentiating it from STs in the stomach. No significant differences were observed in growth patterns between the two lesions less than 3 cm in diameter. An LD/SD ratio greater than 1.29, an enhancement value greater than 27.50 HU, an enhancement ratio greater than 0.66, and an enhancement ratio to pancreas greater than 0.72 were critical CT features for differentiating HP masses from stromal tumors in our study.