Introduction
Pancreatic cystic neoplasms (PCNs) mainly include serous cystic neoplasms (SCNs) and mucinous cystic neoplasms (MCNs), accounting for 10 to 15% of pancreatic cystic lesions (PCLs), and 1 to 2% of pancreatic tumors [
1]. Because of its deep location, slow growth and no clinical symptoms in the early stage, PCNs are easily misdiagnosed [
2]. SCNs and MCNs have different biological behaviors. Relevant studies have reported that only 1 to 3% of SCNs have been transformed into serous cystadenocarcinomas [
3]. To date, only 25–30 serous cystadenocarcinomas have been reported worldwide [
4]. Therefore, SCNs are generally considered benign and can be followed up [
5]. MCNs have malignant potential and are recommended for surgical resection once an adequate diagnosis has been performed [
6,
7]. Therefore, it is crucial to accurately differentiate between SCNs and MCNs for the appropriate treatment.
Computed tomography (CT) and magnetic resonance imaging (MRI) are routine abdominal examinations in China that can effectively screen pancreatic masses. However, due to the limited resolution, these two imaging modalities do not accurately and effectively present the microstructure of PCLs, which increases the difficulty of differential diagnosis of SCNs and MCNs. However, due to its high spatial resolution, endoscopic ultrasound (EUS) can effectively reveal the microstructure of PCLs (such as the septa and thickness of the cystic wall), which greatly improves the diagnostic accuracy of PCLs [
8]. At present, many studies have demonstrated the imaging features of SCNs and MCNs [
9,
10], but these studies do not generate effective diagnostic criteria. The purpose of our study is to provide criteria for the differential diagnosis of SCNs and MCNs by analyzing the imaging features of these two neoplasms by EUS.
Methods
This study was approved by the Ethics Committee of the Chinese People’s Liberation Army General Hospital.
Patients
From April 2015 to December 2017, we prospectively enrolled 88 patients with PCNs who underwent EUS and ultimately received surgery at the Chinese PLA General Hospital. Among these patients, 69 were proven to have MCNs (31 patients) or SCNs (38 patients) by surgical pathology. All patients had neither contraindications to EUS examination nor a history of acute pancreatitis and pancreatic necrosis. All of patients signed informed consent forms.
EUS examination
Before the examination, the patient fasted for at least 8 h. In our study, we used an ultrasonic endoscope (GF-UCT260; Olympus, Tokyo, Japan) in the procedures. To ensure the imaging quality, an echoprobe was routinely covered with a water-filled balloon. During the procedures, the patients were under general anesthesia. The examination was performed by endoscopic physicians with at least 5 years of experience. The EUS findings were recorded in the form of video or picture.
Imaging analysis
Two endoscopic physicians with more than at least 5 years of experience independently completed the analysis of EUS images without knowing the clinical data, other imaging findings and pathological diagnosis of the patients. The final conclusion was drawn after the endoscopic physicians discussed the results and came to an agreement if there was dissent. The observation and recorded items were: size (the longest axis), location (head/neck or body/tail), shape (round, lobulated, and irregular), cystic wall thickness (0–2 mm and > 2 mm, the thickest part of cystic wall was considered thick if it was > 2 mm and thin if it was 2 mm or less), number of septa (0–2 and > 2), and solid components (solid tissues, such as mural nodules, except septa in cystic lesions).
Statistical analysis
SPSS 17 statistical software was used for statistical analysis. The measurement data were presented as the mean ± SD and tested by the t-test. The count data was tested by the chi-square test or continuity correction. Interobserver agreement was assessed by Kappa statistics. The agreement was graded as follows: poor (0.01–0.20), moderate (0.21–0.40), fair (0.41–0.60), good (0.61–0.80), or excellent (0.81–1.00). The sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV) and area under the receiver operating characteristic curves (Az) were used to analyze the efficacy of different imaging features for the differential diagnosis of SCNs and MCNs. The difference was considered statistically significant at P < 0.05.
Discussion
PCLs can be divided into neoplastic cystic lesions and nonneoplastic cystic lesions. Nonneoplastic cystic lesions mainly refer to pseudocysts. Neoplastic cystic lesions mainly include SCNs, MCNs, intraductal papillary mucinous neoplasms (IPMNs) and cystic degeneration of solid tumors. The pancreatic solid-cystic tumors are rare and relatively easy to diagnose. Therefore, it is crucial to differentiate SCNs and MCNs from pseudocysts and IPMNs. Pseudocysts mostly form after inflammation, necrosis or hemorrhage related to pancreatitis or trauma, and are enclosed by a wall with fibrous tissue. Pseudocysts mainly manifest as round cysts, generally without septa [
11]. IPMNs can be classified as main duct IPMNs (MD-IPMNs) and branch duct IPMNs (BD-IPMNs). The MD-IMPNs are characterized by segmental or diffuse dilatation of the main pancreatic duct, which may resemble chronic pancreatitis [
12]. The BD-IPMNs are composed of cysts that communicate with the main pancreatic duct [
12]. Both IPMN and pseudocysts are predominant in men and are prone to pancreatitis [
13,
14]. Our study revealed that SCNs and MCNs occurred more frequently in women, which was consistent with previous studies. It was relatively simple to differentiate pseudocysts and IPMNs from SCNs and MCNs.
SCNs have different biological characteristics than MCNs. SCNs are generally benign, with only 1 to 3% malignant potential [
3], and can be followed up [
5]. Approximately 10% of SCNs are manifested as unilocular without septa, which are easily misdiagnosed as MCNs [
13]. MCNs have malignant potential and are recommended for surgical resection once an adequate diagnosis has been performed [
6,
7]. Therefore, it is of great significance to correctly differentiate between SCNs and MCNs for the appropriate treatment. In 2005, Sahani et al. [
15] proposed a simple imaging-based classification system for guiding the management of PCLs. PCLs were classified into unilocular cysts, microcystic lesions, macrocystic lesions and cysts with a solid component. However, this classification system did not effectively identify SCNs and MCNs. In 2017, Zhang WG et al. [
16] first proposed a new criterion to differentiate between SCNs and MCNs by EUS. This study enrolled only 41 patients diagnosed with SCNs and MCNs. The sample size included was limited. At present, there is no uniform standard for the differential diagnosis of SCNs and MCNs, and it is still difficult to accurately differentiate between SCNs and MCNs.
In our study, SCNs and MCNs were predisposed to occur in middle-aged women (49.16 years vs 46.39 years), which was slightly different from previous studies. Relevant studies have demonstrated that MCNs occurred almost exclusively in women (> 98%) and were generally diagnosed in patients in their 40s and 50s [
17,
18], while SCNs occurred more commonly in women, who typically presented in their 60s [
19]. The difference may be related to the insufficient sample size in our study.
SCNs and MCNs have different histological characteristics. The location distribution of SCNs was different from that of MCNs. MCNs were prone to occur in the body/tail location, and SCNs were more likely to occur in the head/neck location, which was in accordance with previous literature [
20,
21]. SCNs are mainly lobulated, while MCNs are mainly round-like [
22]. Multiple thin septa can be detected in SCNs. According to the number and size of daughter cysts, SCNs can be classified as microcystic, mixed macrocystic and microcystic, macrocystic, and solid types. Microcystic SCNs are composed of multiple cysts of varying sizes, from a few millimeters up to two centimeters. Macrocystic SCNs are characterized by a predominantly or exclusively unilocular pattern [
23], which is also called an “oligo-cystic pattern”. Regarding septa, the “oligo-cystic pattern” refers to the 0–2 septa pattern. Previous studies have demonstrated that microcystic SCNs accounted for approximately 45%, while macrocystic SCNs accounted for approximately 32% [
24], which was similar to our results. MCNs usually present as round macrocystic lesions with no or few septa [
12]. The cystic wall of SCNs is thinner than that of MCNs. Khurana B et al. [
25] revealed that the cystic wall was less than 2 mm thick in four (80%) of the five SCNs.
Head/neck location, lobulated shape, thin wall (0–2 mm) and > 2 septa were the specific imaging features in the diagnosis of SCNs. The Az value of a single imaging feature in the diagnosis of SCNs is not ideal. When any two imaging features were combined, we obtained the highest Az value (0.824) in diagnosing SCNs, as well as the appropriate sensitivity (84.2%), specificity (80.6%), PPV (84.2%), NPV(80.6%) which was almost consistent with study by Sun Y [
14]. The four specific imaging features used in the study by Sun Y [
14] for the diagnosis of SCNs were head/neck location, lobulated shape, thin wall (0–2 mm) and honeycomb pattern, which were slightly different from the imaging features used in our study. We believed that the honeycomb pattern was more specific for the diagnosis of SCNs than the number of septa > 2 but excluded a large portion of atypical SCNs. In our study, > 2 septa were eventually used as one of the four EUS image features, and we achieved good diagnostic efficacy.
Body/tail location, round shape, thick wall and 0–2 septa were the specific imaging features for the diagnosis of MCNs. Compared with any other single imaging signs in the diagnosis of MCNs, round shape had the highest Az value (0.756), with a relatively appropriate sensitivity (77.4%), specificity (73.7%), PPV (70.6%) and NPV (80.0%). When the imaging features were combined, the combined diagnosis of any three features could obtain the highest Az value (0.808), as well as the appropriate sensitivity (77.4%), specificity (84.2%), PPV (80.0%) and NPV (82.1%), which was almost consistent with the results reported by Sun Y [
14]. The combined diagnosis of any three features was more advantageous than the round shape feature in the diagnosis of MCNs.
There are three limitations in our research. First, the sample size is not large enough; second, the criteria are only applicable to the differential diagnosis of SCNs and MCNs, but not to the diagnosis of other PCLs; third, due to the limited sample size, we did not conduct a comparative study of macrocystic SCNs and MCNs.
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.