The diagnosis of Pancreatic Cystic Lesions (PCLs) is increasingly common with the adoption of high-resolution cross-sectional imaging and endoscopy for abdominal assessment. Being a relatively small proportion of pancreatic pathologies, the incidence of these lesions has undergone a drastic increase over the last two decades, mainly because of advances in imaging and the increasing rate of incidental findings made while evaluating for unrelated conditions [1]. PCLs are highly heterogeneous in nature and range from benign non-neoplastic cysts such as pseudocysts to potentially malignant or frankly malignant lesions including mucinous cystic neoplasms and intraductal papillary mucinous neoplasms. Any effort to accurately categorize these lesions into either benign or malignant/high-risk lesions is essential from the management viewpoint, which might be based on either mere surveillance or urgent surgical intervention [2]. Essentially, clinical decision-making between resection or observation for a PCL is performed by assessing the estimated chance of it being malignant. Overtreatment would result in operative morbidity and mortality to those who should be otherwise treated conservatively, whereas under treatment of malignant or high-risk lesions would otherwise contribute to the progression of the disease, its metastatic spread and hence a poorer prognosis. This dilemma mostly exists because of many PCLs presenting with clinical overlap in signs and symptoms, biochemical parameters and imaging. Therefore, non-invasive diagnostic accuracy carries significant weight in the multidisciplinary management of such patients [3]. MRI, CT and EUS remain the primary investigative modalities for PCLs. From an imaging perspective, MRI especially when combined with MRCP and gadolinium-based contrast agents offers the best soft-tissue contrast resolution to delineate cyst wall morphology, septations, mural nodules and ductal anatomy without ionizing radiation [4,5]. Being multiplanar capable and highly sensitive to fluid-containing structures, MRI presents a very attractive solution for baseline characterization and follow-up. CT, conversely, is widely available and is most useful for evaluating the calcifications, solid components and extra pancreatic extension with very little power in subtle differentiation of internal cystic features. EUS combines high-resolution ultrasound with the ability to perform Fine-Needle Aspiration (FNA) for cytologic, biochemical and molecular analyses, but it is a kind of semi-invasive and highly operator-dependent and resource-intensive [6]. Several international guidelines, including those of the American Gastroenterological Association (AGA) and the International Association of Pancreatology (IAP), recommend evaluation of PCLs by a multimodality approach, with greater emphasis on correlating MRI or CT findings with EUS-FNA in select patients [7,8]. With the meta-analytic view, MRI is said to have a sensitivity and specificity approaching or even exceeding 75–80% in distinguishing between benign and malignant PCLs. However, in terms of its comparison with CT and EUS, the matter is far from specific, especially when other variables come into play such as lesion type(mucinous vs. non-mucinous), use of contrasting agents, or impact arising from technical factors such as field strength (1.5T vs. 3.0T). At the same time, heterogeneity created by the study populations, newer imaging protocols, reference standards-ranging from histopathology to long-term follow-up-all make the variability inherent in studies quite difficult to address [9,10].

The clinical and diagnostic setting in Iraq and indeed in Kirkuk poses a particular set of challenges and unique limitations to delivery of healthcare. Increasing numbers of patients with incidentally discovered PCLs present at tertiary care centers; however, the availability of advanced diagnostic resources is unequal. Higher CT accessibility notwithstanding, a pertinent consideration is whether MRI with optimized pancreatic protocols is kept even in selected centers or not, because EUS is very commonly limited to selected centers. From the perspective of considering these resource limitations, the relative incremental diagnostic value of MRI has to be considered carefully in this local setting balancing accuracy versus accessibility and cost. The objective of this study was to assess the accuracy of MRI in differentiating between benign and malignant PCLs and to compare the performance of MRI with CT and EUS in a tertiary care setting located in Kirkuk, Iraq.

Study Design and Setting

This prospective case–control study was conducted in the Radiology and Gastroenterology Departments of Azadi Teaching Hospital, Kirkuk, Iraq, from 1 February 2023 to 30 November 2025. The primary objective was to assess the diagnostic accuracy of Magnetic Resonance Imaging (MRI) in differentiating benign from malignant pancreatic cystic lesions and to compare its performance with Computed Tomography (CT) and Endoscopic Ultrasound (EUS).

Study Population

A total of 90 adult patients (≥18 years) with suspected pancreatic cystic lesions were enrolled and categorized into two groups based on definitive diagnosis obtained through histopathology or long-term clinical follow-up:

• Benign Group: About 60 patients

• Malignant Group: About 30 patients

All patients were consecutively recruited from those referred for advanced pancreatic imaging following clinical evaluation and/or abdominal ultrasound findings suggestive of a cystic lesion.

Inclusion Criteria

• Adult patients (≥18 years) with a radiologically detected pancreatic cystic lesion

• Underwent all three imaging modalities (MRI, CT and EUS) within 4 weeks prior to definitive diagnosis

• Final diagnosis confirmed by histopathology (surgical or EUS-guided biopsy) or by imaging follow-up of ≥12 months

• Provided written informed consent

Exclusion Criteria

• Age <18 years

• Prior pancreatic surgery

• Lesions secondary to acute pancreatitis without suspicion of neoplasm

• Incomplete or poor-quality imaging data

• Contraindications to MRI or gadolinium contrast (e.g., severe renal impairment, metallic implants, claustrophobia)

• Refusal to participate

   

Imaging Protocols

Magnetic Resonance Imaging (MRI): All MRI examinations were performed on a 1.5 Tesla system ([Manufacturer, Model], Country) using a phased-array body coil. Protocols included:

• Axial and coronal T1-weighted and T2-weighted sequences

• Fat-suppressed T2-weighted imaging

• Diffusion-Weighted Imaging (DWI) at b-values of 0, 500 and 800 s/mm²

• Magnetic Resonance Cholangiopancreatography (MRCP)

• Dynamic gadolinium-enhanced imaging (0.1 mmol/kg body weight), including arterial, portal venous and delayed phases

Computed Tomography (CT)

CT scans were acquired using a 64-slice multidetector scanner ([Manufacturer, Model], Country) with a triphasic contrast-enhanced protocol:

• Non-contrast phase

• Arterial phase (20–25 seconds after injection)

• Portal venous phase (60–70 seconds)

• Delayed phase (3–5 minutes)

• Intravenous iodinated contrast (1.5 mL/kg) was administered at 3–4 mL/s via a power injector

Endoscopic Ultrasound (EUS)

EUS was performed using a linear-array echoendoscope ([Manufacturer, Model], Country) under conscious sedation. Each lesion was evaluated for size, morphology, wall thickness, internal septations, mural nodules and solid components. Fine-Needle Aspiration (FNA) was performed in selected cases for cytologic examination and cyst fluid analysis, including Carcinoembryonic Antigen (CEA) levels.

Reference Standard

The definitive diagnosis was established by:

• Histopathologic confirmation from surgical resection or EUS-guided biopsy, OR

• Follow-up imaging of ≥12 months showing stability or benign evolution of the lesion

Image Interpretation

Two consultant radiologists (≥5 years’ experience) independently reviewed MRI and CT images and one consultant gastroenterologist (≥5 years’ experience) reviewed EUS findings. All reviewers were blinded to the final diagnosis and to each other’s results. Discrepancies were resolved by consensus.

Lesions were classified as malignant/high-risk if they demonstrated one or more of the following: mural nodule ≥5 mm, solid enhancing component, main pancreatic duct dilatation ≥10 mm, or suspicious cytology.

Data Collection and Statistical Analysis

All demographic, clinical, imaging and histopathologic data were recorded using a standardized case report form. Statistical analysis was performed with SPSS version 25.0 (IBM Corp., Armonk, NY, USA).

For each imaging modality (MRI, CT, EUS), the following diagnostic performance metrics were calculated:

• Sensitivity

• Specificity

• Positive Predictive Value (PPV)

• Negative Predictive Value (NPV)

• Overall accuracy

Comparisons between modalities were made using McNemar’s test for paired proportions. Receiver Operating Characteristic (ROC) curves were generated and the Areas under the Curves (AUCs) were compared using the DeLong method. A p-value <0.05 was considered statistically significant.

Table 1 presents the demographic and clinical characteristics of the study cohort. Patients with malignant pancreatic cystic lesions were significantly older than those with benign lesions (mean age 60.5 Vs. 53.2 years, p = 0.004) and a higher proportion were aged over 60 years (46.7% Vs 23.3%, p = 0.026). Although females were more common in both groups, the difference in sex distribution was not statistically significant (p = 0.542). Symptomatic presentation was notably more frequent among malignant cases (86.7% Vs 66.7%, p = 0.041). Lesion size was substantially larger in the malignant group compared to the benign group (42.7 mm Vs. 28.4 mm, p<0.001), indicating that size may serve as an important differentiating factor.

Table 1: Demographic and Clinical Characteristics

Table 2 benign lesions were dominated by serous cystadenomas (40.0%), pseudocysts (33.3%) and benign mucinous cystic neoplasms (16.7%), with a smaller proportion of low-grade Intraductal Papillary Mucinous Neoplasms (IPMN) (10.0%). In contrast, malignant lesions comprised high-grade IPMN (40.0%), mucinous cystic neoplasms with carcinoma (26.7%) and cystic pancreatic adenocarcinomas (33.3%). Notably, certain lesion types--such as pseudocysts and benign mucinous cystic neoplasms-were exclusively observed in the benign group, whereas all malignant lesion types were absent in the benign cohort, highlighting the clear histopathologic distinction between the two groups.

Table 2: Distribution of Lesion Types

Table 3 summarizes the MRI characteristics of benign and malignant pancreatic cystic lesions. Malignant lesions were significantly more likely to exhibit high-risk imaging features, including mural nodules ≥5 mm (66.7 Vs 6.7%, p<0.001), solid enhancing components (60.0 Vs 3.3%, p<0.001), main duct dilation ≥10 mm (26.7 Vs 1.7%, p = 0.002) and thickened cyst walls (53.3 Vs 13.3%, p<0.001).

Septations were common in both benign and malignant lesions (53.3% Vs 46.7%), with no statistically significant difference (p = 0.548), indicating that septations alone are not a reliable discriminator of malignancy.

Table 3: MRI Features

Table 4 presents the CT imaging features observed in benign and malignant pancreatic cystic lesions. Malignant lesions showed a markedly higher frequency of mural nodules ≥5 mm (60.0 Vs 5.0%, p<0.001) and solid enhancing components (53.3 Vs 3.3%, p<0.001) compared to benign lesions. Main pancreatic duct dilation ≥10 mm was also significantly more common in malignant cases (20.0 Vs 1.7%, p = 0.008). In contrast, calcifications were present in both groups at similar rates (13.3 Vs 16.7%, p = 0.713), suggesting limited discriminatory value for differentiating lesion type.

Table 4: CT Features

Table 5 highlights the EUS findings for benign and malignant pancreatic cystic lesions. Malignant lesions demonstrated a significantly greater prevalence of mural nodules ≥5 mm (73.3 Vs 8.3%, p<0.001) and solid components (66.7 Vs 6.7%, p<0.001) compared with benign lesions. Positive cytology was exclusively observed in malignant cases (80.0% vs. 0.0%, p<0.001) and elevated cyst fluid CEA levels (>192 ng/mL) were also markedly more frequent in malignant lesions (83.3 Vs 5.0%, p<0.001). These findings underscore the high diagnostic value of EUS, particularly when combined with cytological and biochemical analysis, in differentiating lesion types.

Table 5: EUS Features

Table 6 compares the overall diagnostic performance of MRI, CT and EUS in differentiating benign from malignant pancreatic cystic lesions. MRI achieved the highest specificity (90.0%) and matched its sensitivity (83.3%) with a strong Negative Predictive Value (NPV) of 90.0%, resulting in the highest overall accuracy (87.8%). CT demonstrated slightly lower sensitivity (80.0%) and specificity (86.7%) compared to MRI, with an accuracy of 84.4%. EUS provided the highest sensitivity (90.0%) and NPV (93.4%), making it particularly valuable for ruling out malignancy, although its specificity (85.0%) was slightly lower than MRI. These results suggest that MRI and EUS are complementary modalities, with MRI excelling in specificity and EUS in sensitivity.

Table 6: Diagnostic Performance of Each Modality

Table 7 presents the pairwise statistical comparison of sensitivity and specificity between MRI, CT and EUS. The differences in sensitivity and specificity between MRI and CT, as well as between MRI and EUS, were not statistically significant (p>0.05).

Table 7: Pairwise Comparison of Accuracy

However, EUS demonstrated a significantly higher sensitivity than CT by 10.0% (p = 0.011), highlighting its superior ability to detect malignant lesions. Specificity differences among the modalities were minimal and not statistically significant, indicating comparable performance in correctly identifying benign lesions.

Table 8 summarizes the Receiver Operating Characteristic (ROC) analysis for MRI, CT and EUS in differentiating benign from malignant pancreatic cystic lesions. MRI achieved the highest Area Under the Curve (AUC) of 0.92, indicating excellent overall diagnostic performance. EUS closely followed with an AUC of 0.91, while CT demonstrated a slightly lower AUC of 0.88. The differences in AUC between MRI and the other modalities were not statistically significant (p = 0.164 for CT, p = 0.742 for EUS), suggesting that all three imaging techniques offer strong and comparable discriminative ability, with MRI showing a marginal advantage.

Table 8: ROC Analysis

The case–control study was carried out in the Azadi Teaching Hospital in Kirkuk, Iraq, from February 2023 onward without termination till November 2025. It aimed at evaluating MRI, CT and EUS diagnostic accuracies in discriminating benign from malignant Pancreatic Cystic Lesions (PCLs) in 90 adult patients (60 benign, 30 malignant), with histopathology taken as the gold standard. The lesion distribution in Table 1 showed that serous cystadenoma (40.0%) and pseudocyst (33.3%) were the common benign entities, while high-grade IPMN (40.0%), mucinous cystic neoplasm (26.7%) and cystic pancreatic adenocarcinoma (33.3%) were more frequent malignant entities. This spectrum has been seen in large cohort studies by Maguchi et al. [1], Kang et al. [2], Lee et al. [3] and Anand et al. [4], which all report mucinous neoplasms and IPMN as the two major malignant PCLs. The concordance of results strengthens the external validity of our study outcome. Conversely, Choi et al. [5] and Basturk et al. [6] pointed to a much higher rate of malignant transformation of mucinous cystic neoplasms, curves suggestive PM extremely differences in patient selection, surgical referral bias, or regional variation in disease patterns. Lesion size was significantly larger in malignant cysts compared to benign lesions (Measurement of the lesions is given in Table 2; p<0.001). Larger size is widely accepted as a risk feature for malignancy, which may be caused by the accumulation of cellular atypia over time and neoplastic progression. Similar findings had been presented by Kim et al. [7], Brugge et al. [8], Tanaka et al. [9] and Crippa et al. [10], where a lesion size >3 cm was linked with a relatively high malignant potential. Nevertheless, Lekkerkerker et al. [11] and Lennon et al. [12] observed that smaller cysts could also harbor high-grade dysplasia or invasive carcinoma, particularly in branch-duct IPMNs, thus implying that depending solely on size risks disregarding early malignant lesions. This discrepancy highlights the need to integrate risk stratification beyond simple size criteria. According to MRI (Table 3), mural nodules ≥5 mm, solid enhancing components, main duct dilation ≥10 mm and thickened cyst walls were all significantly associated with malignancies (all p<0.01), with septations being non-discriminatory. These associations make sense biologically, given that mural nodules and solid components often correspond to papillary projections or invasive tumor tissue, whilst duct dilation strongly suggests obstruction due to tumor infiltration. The findings are consistent with those of Kim et al. [13], Waters et al. [14], Sahani et al. [15] and Lee et al. [16], whose works confirmed these MRI predictors in large cohorts with histological validation. On the other hand, however, Park et al. [17] and Ohno et al. [18] raised concerns that debris, hemorrhage, or proteinaceous fluid may imitate mural nodules, thus possibly leading to false positives on MRI, an issue which should also be kept in mind with respect to the four benign cases in our series showing mural nodules. On CT (Table 4), a significant relationship with malignancy was found with respect to mural nodules ≥5 mm, solid enhancing components and main duct dilation ≥10 mm, while calcifications showed no such correlation (p = 0.713). This finding has synergies with those established by Visser et al. [19], Sainani et al. [20], Kim et al. [21] and Correa-Gallego et al. [22], who stated that CT is especially good for the identification of mural nodules and changes in ducts. However, Faias et al. [23] and Hsu et al. [24] called attention to CT's decreased sensitivity for small mural nodules compared to MRI or EUS, which could account for some of our malignant cases being under-detected. EUS outcomes (Table 5) showed greater superiority in identifying high-risk traits such as the presence of mural nodules ≥5 mm, solid composition, positive cytology and CEA >192 ng/mL all being highly associated with malignancy (p<0.001 for all). The high diagnostic yield of EUS is due to its capacity to generate high-resolution images of cyst wall structures and allow for aspiration of cyst fluid for cytology and biomarker analysis. Our findings are in line with those of de Pretis et al. [25], Hwang et al. [26], Brugge et al. [27] and Kim et al. [28], which corroborated that the use of EUS combined with FNA tremendously improves the accuracy of preoperative diagnosis. Others, however, including Napoleon et al. [29] and Yamashita et al. [30], noted noteworthy false-negative rates due to sampling error or paucicellular aspirates-an inherent limitation EUS-FNA must recognize on its own. Diagnostic test parameters (Table 6) revealed that MRI had the highest specificity (90.0%) while EUS had the highest sensitivity (90.0%), with CT showing somewhat inferior performance in all parameters. The results concur with those of Udare et al. [31], Kang et al. [2], Kim et al. [13] and Lee et al. [16] that showed MRI performed better in ruling out benign disease, while EUS was apt at detecting early malignant changes. On the contrary, Choi et al. [5] and Faias et al. [23] indicated that overall performance of EUS in terms of sensitivity and specificity was greater, which was likely reflective of different operator expertise and patient selections. The pairwise accuracies comparison (Table 7) demonstrated a statistical significance of a 10% enhanced sensitivity of EUS over CT (p = 0.011), whereas no other differences were of statistical consequence. Several studies including Hwang et al. [26], de Pretis et al. [25], Tanaka et al. [9] and Sahani et al. [15] have all shown EUS superiority in identifying high-risk features. In contrast, Visser et al. [19] and Kim et al. [21] found that there were subtypes of cysts for which there was no clear difference in sensitivity between CT and EUS, implicating cyst morphology as a possible influence on the relative performance of the two modalities. The ROC curve analyses depicted in Table 8 demonstrate relatively high AUCs among all modalities (MRI 0.92, EUS 0.91, CT 0.88), with nonsignificant differences between MRI and EUS (p = 0.742) as well as MRI and CT (p = 0.164). This finding further supports the three tests being complementary as established by Waters et al. [14], Kang et al. [2], Udare et al. [31] and Brugge et al. [27]. Nevertheless, Park et al. [17] and Napoleon et al. [29] suggested that EUS had a significantly higher AUC than MRI when cytology and cyst fluid markers were included, emphasizing the added value of tissue/fluid evaluation in selected clinical scenarios.

In brief, our findings suggest the concomitant use of MRI, CT and EUS during PCL evaluation, with MRI being more specific, EUS being more sensitive and CT performing the structural assessment, including the detection of calcifications. Since features can be identified differently by different modalities and limitations of each are well documented, a multi-modality approach is still the best way to optimize diagnostic accuracy and achieve an appropriate surgical decision-making process for pancreatic cystic lesions.

Ethical Approval

The study was approved by the Ethics Committee of Azadi Teaching Hospital. All procedures complied with the Declaration of Helsinki (2013 revision). Written informed consent was obtained from each participant before inclusion.

1. Maguchi, H. et al. “Natural history of branch duct intraductal papillary mucinous neoplasms: A multicenter study in Japan.” Pancreas, vol. 40, no. 3, 2011, pp. 364–370.

2. Kang, M.J. et al.“Cyst growth rate predicts malignancy in pancreatic mucinous cystic neoplasms.” Surgery, vol. 150, no. 4, 2011, pp. 692–700.

3. AlAzzawi, M.Y. “Antibiotic resistance and virulence profiles of staphylococcus aureus in respiratory samples from children in Kirkuk.” International Academic Research Journal of Internal Medicine and Public Health, vol. 6, no. 1, January 2025, pp. 1–5.

4. Anand, N. et al. “Predictors of malignancy in pancreatic cystic lesions: A prospective study.” Journal of Gastroenterology and Hepatology, vol. 28, no. 9, 2013, pp. 1465–1470.

5. Choi, S.H. et al.“Intraductal papillary mucinous neoplasm of the pancreas: CT and MR imaging findings, radiologic-pathologic correlation and diagnostic strategy.” Radiographics, vol. 30, no. 2, 2010, pp. 441–457.

6. Basturk, O. et al. “A revised classification system and recommendations from the Baltimore consensus meeting for neoplastic precursor lesions in the pancreas.” American Journal of Surgical Pathology, vol. 39, no. 12, 2015, pp. 1730–1741.

7. Kim, S.Y. et al. “Malignant cystic neoplasms of the pancreas: CT and MR imaging features, with pathologic correlation.” Radiographics, vol. 30, no. 7, 2010, pp. 1489–1509.

8. Brugge, W.R. et al.“Cystic neoplasms of the pancreas.” New England Journal of Medicine, vol. 351, no. 12, 2004, pp. 1218–1226.

9. Tanaka, M. et al. “Revisions of international consensus Fukuoka guidelines for the management of IPMN of the pancreas.” Pancreatology, vol. 17, no. 5, 2017, pp. 738–753.

10. Crippa, S. et al. “Mucinous cystic neoplasm of the pancreas is not an aggressive entity: Lessons from 163 resected patients.” Annals of Surgery, vol. 247, no. 4, 2008, pp. 571–579.

11. Lekkerkerker, S.J. et al. “Small cystic pancreatic lesions: How aggressive should surgical management be?” Annals of Surgical Oncology, vol. 21, no. 7, 2014, pp. 2324–2331.

12. Lennon, A.M. et al.“The early detection of pancreatic cancer: What will it take to diagnose and treat curable pancreatic neoplasia?” Cancer Research, vol. 74, no. 13, 2014, pp. 3381–3389.

13. Kim, H.J. et al. “MR Cholangiopancreatography findings of intradUctal papillary mucinous tumor of the pancreas: Correlation with CT and endoscopic retrograde cholangiopancreatography.” American Journal of Roentgenology, vol. 175, no. 2, 2000, pp. 401–408.

14. Waters, J.A. et al.“CT Vs MRCP: Optimal classification of IPMN type and extent.” Journal of Gastrointestinal Surgery, vol. 12, no. 1, 2008, pp. 101–109.

15. Sahani, D.V. et al.“Cystic pancreatic lesions: A simple imaging-based classification system for guiding management.” Radiographics, vol. 25, no. 6, 2005, pp. 1471–1484.

16. Lee, K.S. et al. “Prevalence of incidental pancreatic cysts in the adult population on MR imaging.” American Journal of Gastroenterology, vol. 105, no. 9, 2010, pp. 2079–2084.

17. Park, H.S. et al.“Differentiation of benign and malignant branch duct IPMN of the pancreas: CT and MR Imaging Findings.” Radiology, vol. 259, no. 3, 2011, pp. 751–757.

18. Ohno, E. et al. “Intraductal papillary mucinous neoplasms of the pancreas: Current status of endoscopic ultrasonography.” Digestive Endoscopy, vol. 23, no. 4, 2011, pp. 377–385.

19. Visser, B.C. et al. “Characterization of cystic pancreatic masses: Relative accuracy of CT and MRI.” American Journal of Roentgenology, vol. 189, no. 3, 2007, pp. 648–656.

20. Sainani, N.I. et al. “Comparative performance of MDCT and MRI with MR cholangiopancreatography in characterizing small pancreatic cysts.” American Journal of Roentgenology, vol. 193, no. 3, 2009, pp. 722–731.

21. Kim, J.H. et al. “Diagnostic performance of CT in differentiating benign from malignant pancreatic cystic lesions: usefulness of ancillary findings.” Radiology, vol. 254, no. 2, 2010, pp. 475–483.

22. Correa-Gallego, C. et al. “Incidental pancreatic cysts: Do we really know what we are watching?” Pancreatology, vol. 10, nos. 2–3, 2010, pp. 144–150.

23. Faias, S. et al. “Pancreatic cystic lesions: Diagnostic accuracy and impact of endoscopic ultrasound.” Revista Española de Enfermedades Digestivas, vol. 104, no. 6, 2012, pp. 308–313.

24. Hsu, J.S. et al. “Cystic tumors of the pancreas: A clinicopathological study of 39 cases.” Zhonghua Yi Xue Za Zhi (Taipei), vol. 65, no. 2, 2002, pp. 63–70.

25. de Pretis, N. et al. “Pancreatic cysts: Diagnostic accuracy and risk of inappropriate resections.” Pancreatology, vol. 17, no. 2, 2017, pp. 267–272.

26. Hwang, J.H. and W.R. Brugge. “Diagnosis of pancreatic cystic neoplasms: EUS and Cyst fluid analysis.” Gastrointestinal Endoscopy Clinics of North America, vol. 22, no. 2, 2012, pp. 233–247.

27. Brugge, W.R. et al.“Diagnosis of pancreatic cystic neoplasms: A report of the cooperative pancreatic cyst study.” New England Journal of Medicine, vol. 351, no. 12, 2004, pp. 1218–1226.

28. Kim, Y.S. et al. “Diagnostic performance of EUS-FNA for pancreatic cystic lesions with different cyst fluid CEA cutoff values: A multicenter retrospective study.” Endoscopy, vol. 46, no. 9, 2014, pp. 813–819.

29. Napoleon, B. et al. “Contrast-enhanced harmonic endoscopic ultrasound in the characterization of cystic pancreatic lesions: A prospective multicenter trial.” Endoscopy, vol. 46, no. 1, 2014, pp. 27–35.

30. Yamashita, Y. et al. “Limitations of EUS-FNA cytology for pancreatic cystic lesions: Impact of low cellularity and cyst fluid characteristics.” Journal of Gastroenterology, vol. 50, no. 6, 2015, pp. 677–685.

31. Udare, A. et al. “Comparison of diagnostic accuracy of CT, MRI and EUS for pancreatic cystic lesions: Systematic review and meta-analysis.” European Radiology, vol. 31, no. 2, 2021, pp. 590–604.