Endoscopic ultrasound-guided fine-needle aspiration (EUS-FNA) is an accurate and safe technique for cytopathologic diagnosis of pancreatic mass lesions. The reported diagnostic accuracy ranges from 65% to 96%, with a median accuracy of 88%.¹ The diagnostic accuracy is very high, especially for pancreatic cancer, and EUS-FNA is currently becoming a standard technique for the histologic diagnosis of pancreatic mass lesions.^1,2 However, various difficulties can be encountered during the procedure, and the location of the mass lesions in the pancreatic head is a technically demanding condition during the EUS-FNA procedure. Inappropriate positioning of the mass lesion during sonographic visualization can be caused by unstable scope positioning. Although the lesion should be visualized in a stable position for the needle puncture, this is not always easy to accomplish. Positioning the echoendoscope in the descending portion of the duodenum (D2) presents technical difficulties in patients with duodenal deformity (short D2 cases). The cascade stomach also hinders the echoendoscope from maintaining a stable watershed area position.

The optimal location of the mass lesion for needle puncture during EUS-FNA is below the probe and slightly left of the center.³ However, pancreatic head lesions are often located to the right of the center below the probe on the EUS image. The needle maneuverability (especially for 19G and 22G needles) with the long scope position at the duodenal bulb can be hampered due to the bending of the scope.^4,5 Additionally, visualization of the lesion may be difficult after needle insertion because the increased rigidity of the scope shaft limits the scope maneuverability. Although no significant differences were noted in the diagnostic performance of EUS-FNA between the pancreatic head and body/tail lesions in previous studies,^4,6 many endosonographers feel that the diagnostic yield of EUS-FNA for the pancreatic head lesions is not as satisfactory as that for the lesions in the pancreatic body and tail in real clinical circumstances.

In the present study, we aimed to explore the clinical and endoscopic factors that determined the diagnostic performance of EUS-FNA for mass lesions in the pancreatic head.

Patients

This is a single-center retrospective cohort study. Medical records and images from picture archiving and communication system (Infinitt Healthcare) were collected from 96 consecutive patients with suspected mass lesions in the pancreatic head detected by contrast-enhanced multidetector-row computed tomography and/or magnetic resonance imaging, who underwent EUS-FNA for cytologic and/or histologic confirmation in our institution between June 2012 and April 2017. The inclusion criteria were patients over 19 years old who would undergo EUS-FNA for tissue confirmation of the mass lesions in the pancreatic head in the given study enrollment period. In contrast, exclusion criteria were 1) patients with pancreatic cystic lesions, 2) those without a final clinical diagnosis, or 3) those who underwent EUS-FNA with both short- and long scope position at the same session (multiple punctures from different positions, because we could not distinguish which position yield true positive cytopathologic results). Final diagnoses were confirmed by pathologic analyses of the specimens obtained from surgical resection or exploratory laparotomy in 20 patients (25.6%), whereas 10 patients (12.8%) had final diagnoses by clinical follow-up data of at least 12 months with negative repeated imaging and a clinical course compatible with benign disease. This study was conducted in accordance with the principles of the Helsinki Declaration. The study protocol was approved by the ethics committee of Kangbuk Samsung Hospital (KBSMC 2017-11-006, approved on 2017-11-07). Informed consent requirement was waived by the ethics committee of Kangbuk Samsung Hospital because only deidentified data were retrospectively accessed.

EUS-FNA procedure

EUS was performed by an experienced endosonographer (HJ Kim). Midazolam was administered to sedate the patient and meperidine was used to provide analgesia, as prescribed by the attending endosonographer. A curvilinear echoendoscope (GF-UCT 260; Olympus Korea Inc.) with corresponding display units (Aloka Prosound SSD Alpha 10; Aloka Korea Co.) was used for EUS-FNA, together with 22G and 25G EchoTip ProCore needles (Cook Medical Korea). The mass lesions in the pancreatic head were sonographically detected at 7.5 MHz with color Doppler imaging. EUS-FNA was subsequently performed. The echoendoscope position (long vs. short scope position; Fig. 1) and needle selection were determined, based on the discretion of the attending endosonographer, who selected the most accurate and appropriate scope position for visualizing the mass lesion for needle puncture (below the probe and slightly left of the center). All targeted lesions were punctured with a stylet in the needle. After withdrawal of the stylet, the needle was moved back and forth within the target lesion using the fanning technique. The needle was subsequently withdrawn. Each needle pass was performed in most of our cases without suction. However, when the target lesion was suspected to be fibrotic and the volume of aspirates was small or the specimens were not suitable, a negative pressure suction with a 10-mL syringe was performed.

Figure 1. Representative sonic views for long- and short scope position. (A) Sonic view of long scope position depicts a hypoechoic tumor neighboring portal vein or superior mesenteric vein, distal common bile duct and head portion of pancreas parenchyma. The mass lesion usually located near the echoendoscope. (B) Sonic view of short scope position depicts a hypoechoic tumor with no distinctively seen neighboring distal common bile duct, portal vein and superior mesenteric vein. The mass lesion usually located far from the echoendoscope.

Processing and analysis of EUS-FNA specimens

EUS-FNA specimens were expressed into the slides using a stylet or 10 mL air-filled syringe, and the specimens were prepared and evaluated with gross examination by an experienced endosonographer before any staining was performed. Specimen adequacy was evaluated according to our previously suggested scoring system for the determination of macroscopic specimen adequacy obtained from EUS-FNA.⁷ Hence our institution does not have resources for rapid on-site evaluation (ROSE) by an attending cytopathologist, the pass number of EUS-FNA is determined by the attending endosonographer to procure the target number (usually 4–5) of adequate tissue cores. Tissue cores of the expressed specimens into the slides were subsequently placed into 10% formalin and processed for paraffin embedding according to the standard routine methods. The 5-μm sections were stained with hematoxylin and eosin for conventional histology and subsequent immunohistochemical staining, if indicated. Other blood clot-like or saline-flushed materials were expressed into the slides, and smears were prepared with the conventional direct smear method by an experienced endosonographer. The slides were fixed by immediate immersion into an alcohol-based solution and subsequently subjected to a modified Papanicolaou stain (Surgipath Medical Industries Inc.) for cytopathologic review. The cytologic and histologic findings were classified as negative for malignancy if no malignancy was present, the diagnosis was indeterminate, or only suspicious malignancy. In contrast, a sample was categorized as malignant if it unequivocally presented malignancy or dysplasia.

Standards of diagnostic confirmation

1. Histopathologic diagnoses of specimens obtained from surgical resection or exploratory laparotomy.

2. Unequivocally positive histologic or cytologic findings obtained from EUS-FNA and follow-up data are compatible with a progressive malignant disease.

3. Clinical follow-up data of at least 12 months with negative repeated imaging and a clinical course compatible with a benign disease.

Statistical analyses

The primary outcome was the overall sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV), and accuracy of EUS-FNA to establish a cytopathologic confirmation in patients with suspected mass lesions in the pancreatic head. Meanwhile, the secondary outcome was to explore the clinical and endoscopic parameters that were independent determinants of diagnostic yield of EUS-FNA in these patients. Patient characteristics are presented as mean ± standard deviation and the other results are expressed as median (range). Categorical variables are expressed numerically as a percentage. The Mann–Whitney U test was used for continuous variables, while the Fisher’s exact test was used for categorical variables. The independent determining factors of the diagnostic performance of EUS-FNA for the mass lesions in the pancreatic head were evaluated by multivariate logistic regression analysis. Based on an anticipated yield of 80% for EUS-FNA for the mass lesion in the pancreatic head, we calculated that 70 patients would be needed to detect significant differences (P < 0.05) with 80% power. Data were collected in Excel 2010 (Microsoft Corp.) and analyzed using the IBM SPSS version 21.0 software package (IBM Corp.).

Baseline clinical-epidemiologic characteristics and diagnostic yield of EUS-FNA procedure for the mass lesions in the pancreatic head

Among the 96 patients enrolled to our study, 18 patients were excluded from the study due to the following causes: 6 patients with a pancreatic cystic lesion in the head, 7 patients without a final clinical diagnosis, and 5 patients who underwent EUS-FNA with both short- and long scope position at the same session. Finally, 78 patients with a total of 390 needle passes were included in the current analyses. The mean age was 66.7 ± 11.1 years, and 42 patients (53.8%) were male. The median number of needle passes was 5, and the maximum number of passes in the current study was 9. A large-caliber FNA needle (22G) was more frequently used in the short scope position than in the long scope position, but the difference in diagnostic performance was not statistically significant (P = 0.09, Table 1).

Table 1 . Baseline Clinical-Epidemiologic Characteristics of Enrolled Patients and Diagnostic Yield of EUS-FNA Procedure for the Mass Lesion in the Pancreatic Head.

Characteristic	Value (n = 78)	P value
Age (yr)	66.7 ± 11.1
Male sex	42 (53.8)
Tumor size (mm)	34.7 ± 10.6
Puncture site (stomach : duodenum)	16 (20.5) : 62 (79.5)
Scope position (short : long)	16 (20.5) : 62 (79.5)
Distance between the mass and scope (< 1 cm : ≥ 1 cm)	52 (66.7) : 26 (33.3)
Correlation between scope position and mass-scope distance		< 0.01
Short (< 1 cm : ≥ 1 cm)	1 (1.3) : 15 (19.2)
Long (< 1 cm : ≥ 1 cm)	51 (65.4) : 11 (14.1)
Median number of needle passes	5 (3–9)
FNA needle used (22G : 25G)	29 (37.2) : 49 (62.8)
Adopted FNA needle according to the scope position		0.09
Short scope position (22G : 25G) (n = 16)	9 (56.3) : 7 (43.8)
Long scope position (22G : 25G) (n = 62)	20 (32.3) : 42 (67.7)
Confirm methods for final diagnosis
Surgical pathology	20 (25.6)
EUS-FNA with compatible follow-up results	48 (61.5)
Clinical follow-up more than 12 months	10 (12.8)
Final diagnosis
Malignancy
Ductal adenocarcinoma	57 (73.1)
Malignant IPMN	5 (6.4)
Neuroendocrine carcinoma	3 (3.8)
Squamous cell carcinoma	1 (1.3)
Poorly differentiated carcinoma	2 (2.6)
Sarcoma	1 (1.3)
Benign
Autoimmune pancreatitis	3 (3.8)
Focal pancreatitis	6 (7.7)
Diagnostic yield of per EUS-FNA procedure
Sensitivity	61/69 (88.4)
Specificity	9/9 (100)
Positive predictive value	61/61 (100)
Negative predictive value	9/17 (52.9)
Accuracy	70/78 (89.7)

The sensitivity, specificity, PPV, NPV, and diagnostic accuracy of EUS-FNA for the mass lesions in the pancreatic head in our series were 61/69 (88.4%), 9/9 (100.0%), 61/61 (100.0%), 9/17 (52.9%), and 70/78 (89.7%), respectively (Table 1).

Determining factors for the diagnostic performance of EUS-FNA for the mass lesion in the pancreatic head

By univariate analyses, the clinical and endoscopic determining factors for the diagnostic performance of EUS-FNA for mass lesions in pancreatic head were the echoendoscope position during needle puncture (long vs. short scope position) and the distance from the echoendoscope to the mass lesion (< 1 cm vs. ≥ 1 cm) (Table 2). In the multivariate logistic regression analysis, the echoendoscope position during needle puncture was the only independent and significant determining factor for the diagnostic performance of EUS-FNA for mass lesions in the pancreatic head (Table 3).

Table 2 . Univariate Analyses for the Clinical and Endoscopic Determinant Factors for the Diagnostic Performance of EUS-FNA for the Mass Lesion in the Pancreatic Head.

Characteristic	Congruous results between EUS-FNA and final diagnosis (n = 70)	Incongruous results between EUS-FNA and final diagnosis (n = 8)	P value
Age (yr)	67.0 ± 11.1	63.9 ± 11.0	0.451
Sex, male	36 (51.4)	6 (75.0)	0.275
Malignancy	61 (87.1)	8 (100)	0.586
Tumor size (mm)	34.4 ± 10.7	37.0 ± 10.0	0.520
Scope position (short : long)	11 (15.7) : 59 (84.3)	5 (62.5) : 3 (37.5)	< 0.01
Distance from scope to mass lesion (< 1 : ≥ 1 cm)	50 (71.4) : 20 (28.6)	2 (25.0) : 6 (75.0)	0.015
No. of needle passes	5.2 ± 1.1	5.0 ± 1.3	0.674
FNA needle used (22G : 25G)	27 (38.6) : 43 (61.4)	2 (25.0) : 6 (75.0)	0.703

Table 3 . Multivariate Logistic Regression Analyses to Determine the Independent and Significant Determining Factor for the Diagnostic Performance of EUS-FNA for Mass Lesions in the Pancreatic Head.

Variable	Category	OR	95% CI	P value
Age (yr)	> 65	0.896	0.172–4.655	0.896
Sex	Male	3.583	0.584–21.993	0.168
Echoendoscope position	Long scope position	8.939	1.861–42.944	< 0.01
Distance from the scope to the mass lesion	≥ 1 cm	0.308	0.028–3.413	0.338

To the best of our knowledge, this is the first study that assessed the clinical and endoscopic factors determining the diagnostic yield of EUS-FNA for mass lesions in the pancreatic head. The sensitivity, specificity, PPV, NPV, and diagnostic accuracy of EUS-FNA for mass lesions in the pancreatic head in our series were 88.4%, 100.0%, 100.0%, 52.9%, and 89.7%, respectively. In addition, the echoendoscope position during needle puncture was the only independent and significant determinant for the diagnostic performance of EUS-FNA for mass lesions in the pancreatic head.

EUS-FNA has already been an established diagnostic tool for pancreatic mass lesions as observed in previous studies, with 54% to 95% sensitivity, 71% to 100% specificity, and 65% to 96% overall accuracy rate.^8–10 However, most previous studies have examined the diagnostic yield of EUS-FNA for pancreatic mass lesions, irrespective of the tumor location. Data on the diagnostic performance of EUS-FNA for mass lesions located in the pancreatic head are very sparse. To date, only few studies have investigated this issue.^3,4,11 One retrospective study from Japan including 56 subjects with suspected pancreatic head tumors reported that the sensitivity, specificity, and diagnostic accuracy of EUS-FNA for pancreatic head tumor were 94% (33/35), 100% (21/21), and 96% (54/56), respectively.⁴ In another two studies from Korea, the diagnostic accuracy of EUS-FNA for pancreatic head mass was somewhat lower than that of the abovementioned study, ranging from 64% to 76%.^3,11 One possible reason for this discrepancy is the presence or absence of an on-site cytopathologist to actively assess the adequacy of samples during the procedure. Many studies have revealed that on-site cytopathology can increase the diagnostic yield of EUS-FNA along with reducing the number of required needle passes and possible adverse events associated with the procedures.^4,12,13 However, the diagnostic discrimination values and yield of EUS-FNA without ROSE has been found to be comparable to those with ROSE by a cytopathologist if the attending endosonographer has received sufficient and qualified training for proper assessment of the gross specimens from EUS-FNA.⁷ Indeed, the diagnostic accuracy of EUS-FNA for pancreatic head tumor in our study, in which a trained endosonographer determined the macroscopic specimen adequacy according to our previously suggested scoring system, was approximately 90%. These results are comparable to that of previous studies with an on-site cytopathologist. Meanwhile, in the present study, the NPV of EUS-FNA for pancreatic head mass was low at 52.9%, which is consistent with the results of previous studies showing low NPV of 46% to 64%.^8,10 Compared with other tumors, pancreatic cancer is associated with more severe inflammation and fibrosis in the surrounding tissue, causing the malignant cells to be undetected even in a cytologically adequate specimen.^11,14,15 Therefore, when physicians encounter the negative results of EUS-FNA for mass lesions in the pancreatic head, other information such as EUS findings, radiological imaging, or clinical course must also be considered before concluding that the lesion is truly not malignant.¹² If malignancy could not be ruled out, repeating the EUS-FNA and careful clinical follow-up is necessary.^1,11,16

Interestingly, our data revealed that the echoendoscope position (long scope position) during needle puncture was the only independent and significant determining factor for the diagnostic accuracy of EUS-FNA for mass lesions in the pancreatic head. Obtaining a specimen suitable for distinguishing malignancy with EUS-FNA in the pancreatic head is actually considered to be technically more difficult than the other parts of the pancreas (body/tail).^5,6 However, no studies have been conducted on the factors affecting the diagnostic accuracy of EUS-FNA for mass lesions in the pancreatic head. In general, EUS-FNA can be best performed when the echoendoscope is straight in a short loop position, which allows an easier needle passage.¹⁷ However, when the tumor is in the head of pancreas, it is often difficult to perform the EUS-FNA in a short scope position because it is relatively unstable and the echoendoscope tends to recoil into the stomach during FNA without obtaining an adequate specimen.^18,19 In this case, a long scope position with advancement of the echoendoscope into the bulb or 2nd portion of the duodenum, allowing a loop to form against the greater curvature of the stomach, could be a better option. Particularly, when attempting to puncture a very firm lesion located in the pancreatic head, a long scope position is more stable than a short scope position, with the mechanical advantage of making the opposite gastric and duodenal walls as a hinge during FNA procedures.⁵ Meanwhile, more resistance may be encountered to advance the needle out due to the tortuosity and tip angulation of the scope.^4,5 However, selecting thinner gauge needles or reorientation of the scope can reduce the difficult passage of the needle in the long scope position.¹⁷ Our data indicate that a long scope position may be helpful for increasing the diagnostic yield of EUS-FNA for mass lesions located in the pancreatic head.

In the present study, the short distance from the scope to the mass (< 1 cm) was associated with an increased diagnostic accuracy of EUS-FNA for pancreatic head mass in univariate analysis; however, when further adjusted for confounders, the association was markedly attenuated and no apparent association between the distance from the scope to the mass lesions and the diagnostic accuracy of EUS-FNA was observed.

The needle size in EUS-FNA also may be a decisive factor influencing the diagnostic accuracy of EUS-FNA for pancreatic lesions. Larger needles can get more cellular materials which increase cellular yield, but at the same time increase stiffness, making it difficult to pass the needle through the echoendoscope and limiting the flexibility of the echoendoscope, especially when the tip is angulated.⁴ Hence, transduodenal puncture to obtain samples from the pancreatic head lesions may demand more pliable needles due to the flexed tip of the echoendosope.¹⁷ In the present study, no significant difference was noted between the diagnostic accuracy of EUS-FNA and the needle sizes (22G vs. 25G). Unlike our results, one prospective study showed a clear benefit of the 25G needle over the 22G needle in terms of the technical success rate and overall diagnostic accuracy for mass lesions in the pancreatic head.²⁰ In two randomized controlled trials, EUS-FNA with 25G needle tended to be superior to 22G when lesions are in the pancreatic head, although they were not statistically significant.^21,22 However, the sample size of previous studies was relatively small (n = 12–42). In addition, selection of the optimal needle size for each patient by an endosonographer in the present study may be the cause of inconsistent results. Further prospective studies with a large sample size are required for optimal choice between 25G and 22G needles for pancreatic head lesions.

Obtaining cytology specimens by EUS-FNA in patients with metastatic or locally advanced pancreatic cancer on imaging is usually desirable since a definitive tissue diagnosis provides the information to make appropriate treatment strategies including chemotherapy, while it is not recommended prior to surgery in patients with potentially resectable pancreatic tumors on imaging studies.^1,6 However, there may be cases demanding a conclusive diagnosis of cancer for neoadjuvant therapy before surgery and unusual cases with non-malignant pancreatic lesions or lymphoma that do not require surgery which is still an extremely complex intervention with significant periprocedural morbidity and mortality.^23–26 Therefore, accurate diagnosis with EUS-FNA is crucial in all patients with solid pancreatic masses. Our data provided a better understanding on the performance EUS-FNA and the factors associated with the diagnostic accuracy of EUS-FNA for pancreatic mass lesions located in the head.

The current study has several limitations. First, this was a retrospective cohort study and our cohort was recruited from a single referral center in urban area (Seoul) of Korea. Therefore, there was likely some degree of selection bias. Second, all EUS-FNA procedures were performed by one highly experienced endosonographer with more than 10 years of EUS-FNA experience for pancreatic mass lesions in our institution. Hence, care must be taken when generalizing our results. However, on the other hand, we could minimize the possibility of operator dependent bias including technical discrepancy which may arise from different endosonographers or facilities. Finally, our study could not evaluate molecular analysis such as genetic alterations in Kras, p53 tumor suppressor, or epidermal growth factor receptor genes which are commonly found in pancreatic cancer.^27,28 They may provide additional diagnostic information especially in cases of indeterminate EUS-FNA results and enhance the diagnostic yield of EUS-FNA.^27,28

In conclusion, EUS-FNA is a feasible and effective technique for diagnosing pancreatic mass lesions, even when the tumor is in the pancreatic head. Additionally, the echoendoscope position (long scope position) during needle puncture is an independent determinant affecting the diagnostic accuracy of EUS-FNA for mass lesions in the pancreatic head. Our results suggest that a long scope position may be helpful for increasing the diagnostic yield of EUS-FNA for the mass lesions in the pancreatic head.

None.

The data that support the findings of this study are available from the corresponding author upon reasonable request.

Hong Joo Kim has been an editor (Deputy Editor) of the International Journal of Gastrointestinal Intervention (IJGII) since 2017; however, Hong Joo Kim has not been involved in the peer reviewer selection, evaluation, or decision process of this article. No other potential conflicts of interest relevant to this article were reported.