Safety and diagnostic performance of endoscopic ultrasound-guided fine needle biopsy in pancreatic cystic and mixed lesions

Abstract

BACKGROUND

Mixed pancreatic cystic lesions (PCLs) could be a sign of malignant transformation from PCLs. The proper diagnosis of these lesions are crucial for management and avoiding unnecessary treatment. Endoscopic ultrasound (EUS)-guided fine needle biopsy (FNB) may improve diagnostic yield by providing histologic core tissue as well as liquid fraction samples; however, data on its safety and diagnostic performance in cystic and mixed lesions remain limited.

AIM

To evaluate the safety and diagnostic performance of EUS-FNB using a 22G needle in mixed PCLs.

METHODS

This retrospective observational cohort study was conducted at a tertiary referral center between January 2019 and December 2024. Consecutive patients undergoing EUS-FNB (22G) for pancreatic cystic or mixed lesions were included. Solid lesions and pancreatic pseudocysts were excluded. Primary outcome was procedure-related safety. Secondary outcomes included technical success, adequacy, diagnostic yield, pathological spectrum, and concordance with surgical histopathology when available. Follow-up data were collected to assess subsequent management and outcomes.

RESULTS

Among 640 EUS-guided biopsies performed during the study period, 60 patients (mean age 65.0 ± 12.1 years; 36.6% female) underwent EUS-FNB for pancreatic cystic (31.7%) or mixed lesions (68.3%). Technical success was achieved in all cases. Adequate tissue, cytology, or fluid for analysis was obtained in 57 patients (95.0%). No EUS-FNB-related adverse events were observed after review of the medical records. The most frequent diagnosis was pancreatic adenocarcinoma (40.0%), followed by serous cystadenoma (10.0%) and intraductal papillary mucinous neoplasm (10.0%). Overall diagnostic yield was 85.0%. Ten patients underwent surgical resection, with complete concordance between EUS-FNB and surgical histopathology in this selected subgroup. Over a mean follow-up of 17.0 ± 14.3 months, management included chemotherapy, surgery, surveillance, repeat EUS, or drainage according to diagnosis and multidisciplinary assessment.

CONCLUSION

EUS-FNB using a 22G needle could be a feasible approach for tissue acquisition in pancreatic cystic and mixed lesions, with high technical success and specimen adequacy in this retrospective cohort. Higher quality studies are needed to further define its clinical role and safety profile.

Key Words: Pancreatic cystic lesion; Pancreatic mixed lesion; Endoscopic ultrasound; Fine needle biopsy; Cyst fluid

Core Tip: Pancreatic cystic and mixed lesions remain diagnostically challenging because conventional imaging and cyst fluid analysis may not reliably differentiate benign from premalignant or malignant lesions. In this retrospective cohort of 60 patients, Endoscopic ultrasound-fine needle biopsy using a 22G needle was feasible, with 100% technical success, 95% specimen adequacy, 85% diagnostic yield, and no procedure-related adverse events identified after medical-record review. Complete concordance with surgical histopathology was observed in the 10 resected patients, but this selected subgroup is not sufficient to estimate overall diagnostic accuracy. These results suggest that endoscopic ultrasound-fine needle biopsy can provide valuable histologic tissue for lesion characterization, particularly in mixed lesions, while prospective comparative studies remain necessary.

INTRODUCTION

Pancreatic cystic lesions (PCLs) are common; they vary from a simple benign cyst to a highly malignant one[1]. Mixed PCLs (mPCLs), i.e. with a solid part and a cystic part, are rarer. They may reflect a malignant transformation of a cyst (wall nodule, for example)[2-4]. Awareness of these lesions has increased in recent years, especially with the increased incidence of asymptomatic pancreatic cysts in the general population primarily due to improved detection by different advanced imaging modalities[5-7]. Therefore, the proper diagnosis, meticulous differentiation, and staging of these lesions are crucial for proper management and avoiding unnecessary treatment of benign ones and missing early treatment of the malignant/pre-malignant lesions[8-10]. Endoscopic ultrasound (EUS) has become an indispensable tool for diagnosing many pancreatic lesions; it has a benefit for better evaluation of number, location, dimensions, wall thickness, and the content of pancreatic cysts. Also, it is crucial in distinguishing the internal septae and solid areas within the cysts[11].

The morphological features of mPCLs are not independent factors in differentiating malignant from nonmalignant lesions. The combination of both EUS-fine needle aspiration (FNA) findings with cystic fluid tumor marker analysis, along with clinical, radiologic, histologic, genetic, and molecular characteristics, enhances the diagnostic accuracy for PCLs and helps to construct a novel model in the era of PCL diagnosis[8,12]. EUS-guided fine needle biopsy (EUS-FNB) may overcome these limitations of EUS-FNA through acquisition of core tissue samples that provide the ability to assess cyst wall architecture via histology. At this point, ESGE does not clearly recommend use of EUS-FNB for diagnosis of cystic lesions[12].

Although EUS-FNB has increasingly been adopted for pancreatic tissue acquisition, data specifically evaluating its performance in pancreatic cystic and mixed lesions remain limited, particularly in real-world cohorts including lesions with both cystic and solid components. Moreover, few studies have assessed the performance of a standardized 22G EUS-FNB approach in routine clinical practice with longitudinal follow-up and surgical correlation. Therefore, this study aimed to evaluate the safety, feasibility, and diagnostic performance of EUS-FNB using a 22G needle in pancreatic cystic and mixed lesions in a tertiary referral center.

MATERIALS AND METHODS

Study design

This retrospective observational cohort study was conducted at the Gastroenterology Endoscopy Unit, Kremlin-Bicêtre Hospital, Paris, France. We screened 640 consecutive EUS-FNB 22G procedures performed for different indications between January 2019 and December 2024 at our center. Eligibility was assessed by reviewing procedure reports and pre-procedural computed tomography (CT) and/or magnetic resonance cholangiopancreatography (MRCP) imaging. Patients were eligible when EUS-FNB was performed for a pancreatic cystic or mixed lesion. Mixed lesions were defined as lesions with both cystic and solid components on EUS, including mural nodules or a solid component adjacent to or within the cystic lesion. For mixed lesions, the needle should either puncture the liquid part or pass through the liquid part in order to collect the solid part. Patients were retrospectively identified, and follow-up data were retrospectively retrieved from electronic medical records. Exclusion criteria were purely solid pancreatic lesions, typical pancreatic pseudocysts as defined by the revised classification of Atlanta of 2012 (round, non-septated lesions with no internal debris and a recent history of acute pancreatitis or radiologic features of chronic pancreatitis)[13], lesions in which interposed vessels or the main pancreatic duct prevented safe puncture, and severe hemorrhagic risk (platelet count < 50000/µL or prothrombin activity < 50%).

Outcomes

The primary outcome was the safety of EUS-FNB 22G in the diagnosis of pancreatic cystic and mixed lesions. Secondary outcomes included technical success, specimen adequacy, diagnostic yield, pathological spectrum, subsequent management, and concordance with surgical histopathology when available.

Definitions

Safety was defined as the occurrence of any specific EUS-FNB-related adverse events, including pancreatitis, perforation, hemorrhage, or infection. Adverse events were ascertained by reviewing procedure reports, post-anesthesia and endoscopy recovery notes, inpatient and outpatient electronic medical records, emergency readmissions when present, and follow-up records.

Technical success was defined as visible tissue present after biopsy and/or fluid. For visible tissue, a Macroscopic On Site Examination was systematically performed[14]. Macroscopic On Site Examination was deemed positive when a visible white core biopsy sample was obtained in a formalin container. Adequacy was defined as the presence of a specimen histologic or cytologic sufficient to allow diagnosis for tissue and biochemical analysis for fluid samples.

Diagnostic yield was defined as the proportion of procedures resulting in a definitive pathological diagnosis permitting lesion characterization and guiding clinical management. Cases were classified as non-diagnostic when histologic and cytologic analyses remained inconclusive or insufficient for definitive lesion classification.

Benign inflammatory conditions such as chronic pancreatitis were considered diagnostic when pathological findings were concordant with clinical presentation, imaging findings, and follow-up evaluation.

PCLs were classified as mPCLs when EUS showed a cystic lesion with an identifiable solid component, mural nodule, or solid tissue; lesions without a solid component were classified as purely cystic. Classification was based primarily on EUS findings and was cross-checked with CT or MRCP when available.

Data collection

Clinical, imaging, procedural, pathological, and follow-up data were gathered retrospectively from electronic records. Variables included demographics, medical history, symptoms, and initial imaging (CT or MRCP). Lesion details such as size, location, duct diameter, ductal dilatation, mural nodules, and diagnostic assumptions were recorded. EUS-specific data covered lesion morphology, wall features, solid components, duct communication, lymph node status, needle passes, and antibiotic use. Procedural outcomes included technical success, on-site examination, specimen adequacy, and adverse events. Pathology data included histologic/cytologic results and cyst fluid analysis. Follow-up tracked subsequent management, repeat EUS when performed, surgical pathology, survival, cause of death, and duration. For patients with non-diagnostic EUS-FNB despite adequate material, post-procedure management was reviewed separately from the medical record.

EUS-FNB technique and needle

All procedures were performed by experienced endosonographers in a tertiary referral center using the same processor (Eluxeo VP-7000, Fujifilm, Tokyo, Japan), under general anesthesia. All samples were taken with a 22G EUS-FNB needle (Acquire^®, Boston Scientific, Natick, MA, United States). After the needle sheath was inserted into the operating channel of a linear echoendoscope, the needle tip was positioned for puncture. When the needle was within the target lesion, the stylet was removed (slowly in tissue-rich lesions for the slow-pull technique) and a vacuum syringe was attached. Fluid was collected directly into the syringe and transferred into a dry tube for biochemical analysis (carcinoembryonic antigen, carbohydrate antigen 19-9, glucose, lipase, and amylase). Biopsies were directly transferred into a formalin container for histopathological analysis. Needles were systematically rinsed with saline for cytological analysis.

Statistical analysis

All statistical analyses were performed using the Statistical Package for the Social Sciences version 26 (SPSS, Inc., Chicago, IL, United States) and R software version 4.3.1 (R Foundation for Statistical Computing, Vienna, Austria). Categorical variables were expressed as percentages and compared using the χ² test with Fisher’s correction. Continuous variables were expressed as mean ± SD or median with interquartile range (IQR) when appropriate and compared using the t-test or Mann-Whitney U test. Univariate analyses were performed using the Mann-Whitney U test for continuous variables and Fisher’s exact test for categorical variables. Given the limited number of non-diagnostic events and inadequate specimens, multivariable analysis was performed only as an exploratory, hypothesis-generating analysis. Variables with P < 0.10 in univariate analysis and those considered clinically relevant were selectively included to limit model overfitting. A penalized logistic regression model (Firth method) was used to reduce small-sample bias and avoid separation issues[15]. No multivariable result was used to support confirmatory inference.

RESULTS

Baseline patient and imaging characteristics

Between January 2019 and December 2024, a total of 640 consecutive EUS-guided biopsies were performed for different indications at our center. Among these, 60 procedures (9.4%) met the eligibility criteria for pancreatic cystic or mixed lesions and were included in the final analysis (Figure 1). The remaining 580 procedures were not included because they did not correspond to eligible pancreatic cystic or mixed lesions and fulfilled one of the predefined exclusion criteria.

Figure 1 Study flow diagram. Of 640 consecutive endoscopic ultrasound-guided fine needle biopsy 22G procedures screened during 2019-2024, 60 eligible pancreatic cystic or mixed lesions were included. The 580 remaining procedures were non-eligible because they involved other indications or met predefined exclusion criteria. EUS-FNB: Endoscopic ultrasound-guided fine needle biopsy.

The mean age at the time of the procedure was 65.0 ± 12.1 years, and 22 patients (36.6%) were female. A history of diabetes mellitus was present in 20 patients (33.3%). Forty patients (66.7%) were symptomatic at presentation, most commonly with abdominal pain (50.0%), whereas 20 lesions (33.3%) were incidentally discovered. Jaundice was reported in 10.0% of patients, weight loss in 5.0%, and ascites in 1.7% (Table 1).

Table 1 Patient and lesion characteristics in imaging, n (%)/mean ± SD.

Variable	Total patients (n = 60)
Sex (female)	22 (36.6)
Mean age at time of procedure (years)	65.0 ± 12.1
History of diabetes mellitus	20 (33.3)
Symptoms
Abdominal pain	30 (50.0)
Jaundice	6 (10.0)
Asymptomatic (accidentally discovered)	20 (33.3)
Ascites	1 (1.7)
Weight loss	3 (5.0)
Initial imaging modality
CT	24 (40.0)
MRCP	34 (56.7)
Missing data	2 (3.3)
Mean lesion diameter (mm)	36.2 ± 29.3
Location
Head	23 (38.3)
Body	13 (21.7)
Tail	13 (21.7)
Isthmus	7 (11.7)
Uncinate	2 (3.3)
Missing data	2 (3.3)
Pancreatic duct dilatation > 5 mm	19 (31.7)
Mural nodule	3 (5.0)
Diagnostic assumption
Cystic lesion	17 (28.4)
Mixed lesion	36 (60.0)
Nodule (solid lesion)	5 (8.3)
Missing data	2 (3.3)

CT: Computed tomography; MRCP: Magnetic resonance cholangiopancreatography.

Initial detection was achieved by MRCP in 56.7% of cases and CT scan in 40.0%. On cross-sectional imaging, the mean lesion diameter was 36.2 ± 29.3 mm. Lesions were most frequently located in the pancreatic head (38.3%), followed by the body (21.7%), tail (21.7%), isthmus (11.7%), and uncinate process (3.3%). Pancreatic duct dilatation > 5 mm was observed in 19 patients (31.7%). A mural nodule was reported on imaging in 5.0% of cases. Based on pre-EUS imaging, lesions were considered mixed in 60.0%, purely cystic in 28.3%, solid nodules in 8.3% (Table 1).

EUS characteristics and procedural details

Table 2 shows EUS findings and procedures. The average lesion diameter was 32.4 ± 16.6 mm. Lesions were found in the pancreatic head (40.0%), body (25.0%), tail (18.3%), isthmus (15.0%), and uncinate process (1.7%). Morphology: (1) 38.3% unilocular; (2) 40.0% multilocular; and (3) 21.7% microcystic (Figures 2 and 3). Mural nodules appeared in 11.7% of cases. Cyst walls were thin (66.7%), thickened (25.0%), or undefined (8.3%) (Table 2).

Figure 2  Mixed unilocular lesion with mural nodule (1 and 2: Diameter of the nodule).

Figure 3  Pure multilocular cystic lesion (1-4: Diameters of microcysts).

Table 2 Lesion characteristics in endoscopic ultrasound and endoscopic ultrasound-guided fine needle biopsy results, n (%)/mean ± SD/median (interquartile range).

Variable	Value
Mean lesion diameter (mm)	32.4 ± 16.6
Location
Head	24 (40.0)
Body	15 (25.0)
Tail	11 (18.3)
Isthmus	9 (15.0)
Uncinate	1 (1.7)
EUS aspect
Unilocular	23 (38.3)
Multilocular	24 (40.0)
Microcystic	13 (21.7)
Mural nodule	7 (11.7)
Wall
Thin	40 (66.7)
Thick	15 (25.0)
Undefined	5 (8.3)
Content
Clear	17 (28.3)
Turbid	4 (6.7)
Non contributive	39 (65)
Calcifications	7 (11.7)
Presence of solid component	41 (68.3)
Lesion type
Mixed cyst	41 (68.3)
Cystic	19 (31.7)
Pancreatic duct dilatation (5-9.9 mm)	19 (31.7)
Pancreatic duct dilatation (≥ 10 mm)	2 (3.3)
Mean dilatation (mm)	2.5 ± 3.6
Communication with pancreatic duct	5 (8.3)
Number of needle passes
1	12 (20.0)
2	39 (65.0)
3	9 (15.0)
Mean number of needle passes	1.9 ± 0.6
Lymph nodes	12 (20.0)
Complications	0 (0.0)
Adequate cells, tissue or fluid for analysis	57 (95.0)
Prophylactic antibiotic	47 (78.3)
Pathological diagnosis
Diagnostic pathology	51 (85)
Pancreatic adenocarcinoma	13 (21.7)
Well differentiated	24 (40.0)
Moderately differentiated	6 (10.0)
Poor differentiated	5 (8.3)
Serous cystadenoma	6 (10.0)
IPMN	6 (10.0)
LGD	2 (3.3)
HGD	4 (6.7)
Mucinous cystadenoma	3 (5.0)
Neuroendocrine tumor	3 (5.0)
Pseudo papillary tumor	1 (1.7)
Pseudocyst	2 (3.3)
Chronic pancreatitis	2 (3.3)
Duplication duodenal cyst	1 (1.7)
Epidermoid carcinoma	1 (1.7)
Fibrosis	1 (1.7)
Normal	1 (1.7)
Non diagnostic pathology	9 (15.0)
Follow-up diagnosis (n = 9)
Pancreatic adenocarcinoma	2 (22.2)
Serous cystadenoma	3 (33.3)
IPMN	1 (11.1)
Granuloma	1 (11.1)
Indeterminate cyst	2 (22.2)
Cyst fluid biochemical analysis
CEA, n = 12	33.4 (4.7-6059.0)
CA19-9, n = 12	1087 (77.9-116082.0)
Glucose, n = 15	1.6 (0.35-4.3)
Lipase, n = 14	628 (126.3-36972.8)
Amylase, n = 3	227 (154.0-19448)

CA19-9: Carbohydrate antigen 19-9; CEA: Carcinoembryonic antigen; EUS: Endoscopic ultrasound; HGD: High grade dysplasia; IPMN: Intraductal papillary mucinous neoplasm; LGD: Low grade dysplasia.

A solid component was seen in 68.3% of lesions, forming the mixed subgroup. Calcifications appeared in 11.7%. Communication with the main pancreatic duct was observed in 8.3%. Pancreatic duct dilation measured 5-9.9 mm in 31.7% and ≥ 10 mm in 3.3%, with an average diameter of 2.5 ± 3.6 mm (Table 2). Upon evaluation of clinical, imaging, and EUS characteristics between pure cystic and mixed lesions, it was found that mixed lesions were significantly larger than cystic lesions on both initial imaging (P = 0.002) and EUS (P = 0.002). No statistically significant differences were identified for other outcomes (Table 3).

Table 3 Comparison of clinical, imaging and endoscopic ultrasound characteristics between pancreatic cystic lesions and mixed pancreatic cystic lesions, n (%)/mean ± SD.

Variable	PCLs (n = 19)	mPCLs (n = 41)	P value
Sex (female)	13 (68.4)	22 (53.7)	0.425
Mean age at time of procedure (years)	61.8 ± 13.4	66.9 ± 11.2	0.167
History of diabetes mellitus	3 (15.8)	17 (41.5)	0.077
Symptoms
Abdominal pain	12 (63.2)	18 (43.9)	0.267
Jaundice	0 (0.0)	6 (14.6)	0.163
Asymptomatic (accidentally discovered)	7 (36.8)	13 (31.7)	0.922
Ascites	0 (0.0)	1 (2.4)	1.000
Weight loss	0 (0.0)	3 (7.3)	0.545
Initial imaging modality
CT	6 (31.6)	18 (43.9)	0.350
MRCP	13 (68.4)	21 (51.2)
Missing data	0 (0.0)	2 (4.9)
Mean lesion diameter on initial imaging (mm)	24.3 ± 8.3	37.1 ± 17.4	0.002
Location on initial imaging
Head/isthmus/uncinate	11 (57.9)	21 (51.2)	0.838
Body	2 (10.5)	11 (26.8)	0.194
Tail	6 (31.6)	7 (17.1)	0.351
Missing data	0 (0.0)	2 (4.9)	1.000
Pancreatic duct dilatation > 5 mm	5 (26.3)	14 (34.1)	0.666
Mural nodule on initial imaging	3 (15.8)	0 (0.0)	0.055
Diagnostic assumption			< 0.0001
Cystic lesion	15 (78.9)	2 (4.9)
Mixed or solid lesion	4 (21.1)	37 (90.2)
Missing	0 (0.0)	2 (4.9)
Mean large lesion diameter on EUS (mm)	24.3 ± 8.3	37.1 ± 17.4	0.002
Location on EUS
Head/isthmus/uncinate	12 (63.2)	22 (53.7)	0.681
Body	2 (10.5)	13 (31.7)	0.112
Tail	5 (26.3)	6 (14.6)	0.466
Mural nodule	4 (21.1)	3 (7.3)	0.193
Pancreatic duct dilatation (≥ 5 mm)	5 (26.3)	16 (39.0)	0.624
Mean dilatation (mm)	2.2 ± 3.9	2.7 ± 3.5	0.608
Number of needle passages	1.7 ± 0.7	2.0 ± 0.5	0.056
Lymph nodes	2 (10.5)	10 (24.4)	0.306
Adequate cells, tissue or fluid for analysis	18 (94.7)	39 (95.1)	1.000
Prophylactic antibiotic	15 (78.9)	32 (78.0)	1.000

CT: Computed tomography; EUS: Endoscopic ultrasound; MRCP: Magnetic resonance cholangiopancreatography; PCLs: Pancreatic cystic lesions; mPCLs: Mixed pancreatic cystic lesions.

Safety outcomes

No procedure-related adverse events were recorded after review of procedure reports, post-procedure recovery notes, inpatient and outpatient electronic records, and follow-up data. Specifically, there were no cases of post-EUS pancreatitis, hemorrhage, perforation, or infection. However, given the relatively limited sample size and tertiary-center setting, the absence of complications cannot be extrapolated. Prophylactic antibiotics were administered in 78.3% (n = 47/60) of cases (Table 2).

Technical success and adequacy

EUS-FNB 22G achieved 100% technical success, with visible tissue and/or fluid obtained in all cases. One pass was made in 20.0% of patients, two passes in 65.0%, and three passes in 15.0%, with a mean of 1.9 ± 0.6 passes. Adequate material for histologic, cytologic, or biochemical evaluation was obtained in 57 of 60 patients, corresponding to an adequacy rate of 95.0%. Three patients had insufficient material for definitive interpretation. Despite adequate sampling, a definitive pathological diagnosis could not be established in 9 patients (15.0%), who were therefore classified as non-diagnostic. Non-diagnostic cases included lesions with inconclusive histologic and cytologic findings despite sufficient material acquisition. Inflammatory diagnoses such as chronic pancreatitis were considered diagnostic only when supported by clinical and radiological correlation and follow-up data (Table 2).

Pathological diagnosis and diagnostic yield

The histopathological diagnoses obtained via EUS-FNB are summarized in Table 2. Pancreatic adenocarcinoma was identified in 24 patients (40.0%), with subtypes comprising well-differentiated (21.7%), moderately differentiated (10.0%), and poorly differentiated tumors (8.3%). Additional diagnoses included serous cystadenoma in 6 patients (10.0%), intraductal papillary mucinous neoplasm (IPMN) in 6 patients (10.0%; low-grade dysplasia, 3.3% and high-grade dysplasia, 6.7%), while 3 patients (5.0%) had mucinous cystadenoma, 3 patients (5.0%) had neuroendocrine tumor and solid pseudopapillary tumor was found in 1 patient (1.7%). Pseudocyst is diagnosed in 2 patients (3.3%), chronic pancreatitis in 2 patients (3.3%), as well as several rare entities. Overall, a definitive pathological diagnosis permitting lesion classification was achieved in 51 of 60 patients, corresponding to a diagnostic yield of 85.0%. Nine patients (15.0%) were categorized as non-diagnostic because histologic and cytologic findings remained inconclusive.

Cyst fluid biochemical analysis was available in a subset of patients and was summarized using median and IQR when available because of marked variability. Median cyst fluid carcinoembryonic antigen was 33.4 µg/L (IQR: 4.7-6059.0; n = 12), median carbohydrate antigen 19-9 was 1087 KU/L (IQR: 77.9-116082.0; n = 12), median glucose was 1.6 mmol/L (IQR: 0.35-4.3; n = 15), median lipase was 628 U/L (IQR: 126.3-36972.8; n = 14), and median amylase was 227 U/L (IQR: 154.0-19448; n = 3) (Table 2).

In univariate analysis, younger age (odds ratio = 0.91 per year, 95%CI: 0.84-0.98, P = 0.014) and pure cystic lesion type (odds ratio = 5.85, 95%CI: 1.28-26.79, P = 0.023) were associated with non-diagnostic sampling. No independent predictive factors were identified in exploratory Firth multivariable logistic regression analysis.

Follow-up and outcomes

Over a mean follow-up period of 17.0 ± 14.3 months, management strategies comprised chemotherapy for 38.3% of patients, surgical resection for 16.7%, imaging surveillance alone for 21.7%, repeat EUS for 10.0%, and drainage for 1.7%. Surgical resection was performed in 10 patients, with complete concordance (100%) observed between EUS-FNB diagnoses and subsequent surgical histopathology in this selected surgical subgroup. The nine cases classified as non-diagnostic despite adequate sampling were managed according to multidisciplinary clinical and radiological assessment, with repeat EUS and/or imaging surveillance. These cases were not counted as diagnostic successes unless a definitive pathological or multidisciplinary diagnosis was established. Among these nine patients, two were subsequently diagnosed during follow-up with pancreatic adenocarcinoma, three with serous cystadenoma, one with IPMN, one with a granuloma, and two with indeterminate pancreatic cysts without worrisome features.

During the follow-up, 14 patients (23.3%) died. The causes of death were metastatic pancreatic cancer in 78.6% of patients, other malignancies in 7.1%, and non-cancer-related causes in 14.3% (Table 4).

Table 4 Follow-Up and outcomes, n (%)/mean ± SD.

Variable	Total patients (n = 60)
Mean follow-up duration (months) from date of procedure to date of death or last visit	17.0 ± 14.3
Management
Follow up	13 (21.7)
Repeat EUS and diagnosis	6 (10.0)
Chemotherapy	23 (38.3)
Surgical	10 (16.7)
Drainage	1 (1.7)
Missing data	7 (11.6)
Concordance of EUS-FNB compared to surgical pathology for patients who had surgery	10/10 (100)
Death	14 (23.3)
Cause of death
Metastatic pancreatic cancer	11/14 (78.6)
Other death	2/14 (14.3)
Other cancers	1/14 (7.1)

EUS: Endoscopic ultrasound; FNB: Fine needle biopsy.

DISCUSSION

In this single-center retrospective cohort study of 60 patients with pancreatic cystic or mixed pancreatic lesions, EUS-FNB with a 22G core needle was feasible and achieved high technical success and specimen adequacy. Technical success was achieved in all procedures, adequacy was 95.0%, and no procedure-related adverse events were identified after medical-record review. Diagnostic yield was 85.0% and was based on definitive lesion characterization rather than specimen adequacy alone, which explains the difference between adequacy (95.0%) and diagnostic yield (85.0%). Importantly, nearly seventy percent of lesions in our cohort were mixed lesions containing both cystic and solid components, a subgroup that remains underrepresented in the current literature despite its frequent association with malignant transformation and diagnostic uncertainty in daily practice. Surgical histopathology correlated with EUS-FNB diagnoses in all 10 resected cases; however, this selected subgroup is subject to verification bias and should not be interpreted as an estimate of overall diagnostic accuracy.

Historically, EUS-FNA has been widely used for PCLs, but its diagnostic sensitivity for malignancy has been suboptimal. A meta-analysis of 16 studies including 1024 PCL cases reported that pooled sensitivity for malignancy with EUS-FNA cytology was only 51 % (95%CI: 0.45-0.58), despite high specificity (0.94, 95%CI: 0.92-0.96) highlighting the limitations of fluid cytology alone in detecting malignant or high-grade lesions[16].

More recent work has emphasized histologic sampling with biopsy cores. Although van Riet et al[17] did not focus specifically on cystic/mixed lesions, their systematic review of EUS-guided histologic sampling approaches demonstrated that fine-needle biopsy needles generally provide superior diagnostic accuracy and higher core tissue yield compared with aspiration alone (e.g., EUS-FNB diagnostic accuracy up to approximately 87% vs EUS-FNA approximately 80% in solid lesions), reinforcing the trend toward core biopsies in challenging contexts[18].

Comparative data focused on pancreatic cystic disease remain limited, but available cohort studies support the value of histologic sampling. In a recent retrospective series of 100 PCLs undergoing EUS-FNB, histologic diagnostic yield was reported in 60% of cases, with high accuracy for specific cyst subtypes (IPMN 85.7%, MCN 90.5%, serous cystadenoma 95.2%, NET/SPN 95%-100%) compared with surgical histopathology, and appropriate clinical management decisions were significantly more likely when EUS-FNB was diagnostic[19]. These results complement our own findings, where 95% adequacy and 100% concordance in resected cases indicate robust tissue sampling even in mixed lesions.

Direct comparisons between EUS-FNA and EUS-FNB in PCLs is scarce. Apart from studies involving through-the-needle biopsy with microforceps, the literature evaluating EUS-FNB for this indication remains rare[20,21].

A retrospective series of 90 PCLs showed similar overall diagnostic yield between EUS-FNA and EUS-FNB (94.2% vs 94.7%), with higher adequacy in the EUS-FNB group (81.6% vs 71.2%), although differences did not reach statistical significance[22]. This suggest broadly similar diagnostic yields but a trend toward higher tissue adequacy with EUS-FNB. Importantly, adverse event rates were equally low in both techniques, consistent with the excellent safety profile seen in our cohort.

Current ESGE technical guidance does not clearly recommend EUS-FNB for PCLs because evidence remains limited and comparative data are scarce. Our 85.0% diagnostic yield and complete surgical concordance in 10 resected patients are encouraging, but they are not sufficient to support a change in guideline recommendations because of the retrospective single-center design, absence of a comparator group, and small surgically verified subgroup. These findings should instead be considered hypothesis-generating and support the need for prospective multicenter comparative studies or randomized controlled trials before EUS-FNB can be recommended more broadly for PCLs.

In this study, younger age and pure cystic lesion were associated with a higher rate of non-diagnostic samples in the univariate analysis, but not in the multivariate analysis. Age was not associated with lesion size or the presence of a solid component. This likely reflects a lack of statistical power related to the limited sample size and low number of non-diagnostic events. To date, only one retrospective study has demonstrated an inverse association between age and the diagnostic yield of EUS-guided FNA in PCLs, which contrasts with our results[23]. No independent predictive factors were identified in exploratory Firth multivariable logistic regression analysis. These analyses should be interpreted as hypothesis-generating because of the limited number of non-diagnostic cases and inadequate specimens. Nevertheless, age may influence diagnostic yield in this specific context. Age has been identified as a risk factor for increased size of PCLs and their malignant transformation and may influence the diagnostic yield of EUS-guided puncture[24,25]. This issue warrants further investigation through higher-quality studies. In contrast, the literature clearly shows that the sensitivity of EUS-guided sampling is higher for solid lesions than for cystic lesions. These findings are consistent with our univariate analysis[26,27].

This study has several limitations. The retrospective, single-center design inherently restricts generalizability and may introduce selection bias. Although 640 EUS-FNB procedures were screened, only 60 met the eligibility criteria for pancreatic cystic or mixed lesions, and granular mutually exclusive counts for all excluded procedures could not be reconstructed reliably from the retrospective procedure database. In addition, the absence of adverse events should be interpreted cautiously, as the relatively small cohort size, exclusion of high-risk patients, and procedures performed by experienced operators in a tertiary referral center may have limited the onset of complications. The relatively small sample size precludes detailed subgroup analyses; the small number of non-diagnostic cases and inadequate specimens makes multivariable analysis exploratory only. Surgical concordance was available for only 10 selected resected patients and is therefore subject to verification bias; it should not be interpreted as overall diagnostic accuracy. Finally, the absence of an EUS-FNA comparator group prevents head-to-head evaluation and any conclusion regarding superiority over alternative sampling strategies, although this reflects real-world practice at our center, where EUS-FNB needles have replaced EUS-FNA needles for diagnostic EUS. Cyst fluid biochemical analyses were available in a limited number of patients and were heterogeneous in nature. Molecular and genetic analyses were not conducted on the collected samples; however, certain studies indicate that cyst fluid biomarkers may be valuable in differentiating pancreatic cysts[28].

Despite these limitations, and without demonstrating superiority over EUS-FNA, our study provides real-world evidence supporting the feasibility of a standardized EUS-FNB strategy in a heterogeneous cohort of pancreatic cystic and mixed lesions. The study draws attention to several clinically relevant findings: (1) EUS-FNB achieved good specimen adequacy and diagnostic yield in pancreatic cystic and mixed lesions; (2) No procedure-related adverse events were identified in this selected tertiary-center cohort; and (3) Acquisition of histologic core tissue may facilitate architectural analysis and histologic subtyping, particularly in lesions with mixed solid and cystic components where fluid cytology alone may be limited. To establish the best use of EUS-FNB within diagnostic processes and compare it to other sampling methods, further prospective multicenter studies and randomized controlled trials are necessary.

CONCLUSION

EUS-FNB using a 22G needle could be a feasible approach for tissue acquisition in pancreatic cystic and mixed lesions, with high technical success, high specimen adequacy, and no procedure-related adverse events identified in this single-center retrospective cohort. These findings support its potential clinical utility in selected patients; however, prospective multicenter comparative studies and/or randomized control trials are required to determine its relative performance, safety profile, and place in future diagnostic algorithms.