PC is sometimes accompanied by necrosis, mostly in the central area of the tumor (Figure 1). A previous study using transabdominal ultrasound-guided FNA reported that sampling from the peripheral area of the pancreatic mass improved the diagnostic accuracy[63]. The same is also true with EUS-FNA[60]. In this sense, the fanning method is considered effective, as it collects greater numbers of viable tumor cells. The needle movements within the multiple marginal areas of the mass using the “up-down” dial of the endoscope releases more cells when compared to the standard method that targets one peripheral area of the mass[23] (Figure 1). Bang et al[23] demonstrated the efficacy of a fanning method that targeted four marginal sites of the tumor; they needed significantly fewer passes to establish diagnosis than with the standard method [by randomized control trial (RCT) median 1 (interquartile range: 1-1) vs 1 (1-3), P = 0.02] and found a significantly higher rate of achieving a diagnosis with a single pass (85.7% vs 57.7%; P = 0.02).

The standard EUS-FNA is done with a needle controlled under negative pressure, usually applied with a 10-20 mL syringe[29] (COOK: 10 mL, Boston Scientific and Medi-Globe: 20 mL). However, the suction has been altered to determine its effect on FNA; i.e., by the no suction method[32,36,64,65], slow pull method[22] and high-negative pressure (HNP) method[29]. The no suction method is performed without suction[32,36,64,65] (Figure 1). The slow pull technique applies 10-20 to-and-fro needle movements with simultaneous minimum negative pressure provided by slow and continuous pulling of the stylet from the needle[22]. The HNP method is conducted under a vacuum provided by a 50 mL syringe [in contrast to the normal-negative pressure (NNP) method that uses a 10 mL syringe for vacuum[29].

Recent studies have confirmed that higher amounts of tissue are acquired and that blood contamination increases when the suction level is increased for EUS-FNA of solid pancreatic lesions[22,29,32]. An RCT by Puri et al[32] demonstrated a higher sensitivity by adding suction (86% in suction and 67% in non-suction, P = 0.05), but subsequent studies[22,65] showed no diagnostic superiority for the suction method. Interestingly, a retrospective study by Nakai et al[22] revealed a higher accuracy of the slow-pull method than with the ordinary suction method, but only when using 25G needles (91% vs 70%, P = 0.004); no difference was noted with a 22G needle. A comparison between HNP and NNP for EUS-FNA of a pancreatic mass using 25G needles confirmed that HNP was superior in terms of adequate tissue acquisition and accurate histological diagnosis compared to NNP (adequate tissue: 90% vs 72%, P = 0.0003; diagnostic accuracy: 82% vs 73%, P = 0.06). A high level of blood contamination was recognized in the HNP samples (P = 0.004), but the numbers of blood cells did not affect the histological diagnosis. A concern was noted for highly vascular lesions such as pancreatic neuroendocrine tumors (PNETs), as only limited cases have been examined[29]. For the FNA of lymph nodes, the quantity of tissue acquired is usually good and suction is not recommended, in order to reduce blood contamination[66].

The stylet is believed to prevent a contamination of the sample with tissue that does not originate from the target lesion; however, procedures for pushing out and withdrawing the stylet are time consuming. Three RCTs[67-69] found no superiority arising from the use of a stylet in terms of tissue contamination and diagnostic yield, and conversely found the adequacy of sample acquisition to be inferior (stylet: 75% vs non-stylet: 87%, P = 0.01) and saw an increase in blood contamination (75% vs 52%, P < 0.0001)[68]. However, as mentioned, a slow pull of the stylet during the pass could improve the quality of these FNA samples[22]. A stylet is also useful for pushing the tissues out from the needle onto slides or into a medium (Figure 1).

An on-site pathologist can provide the endosonographer with helpful information; for example, if the samples obtained by EUS-FNA contain tissues from the targets or whether additional procedures are needed[15]. However, this system is not feasible at all institutions and not in every tertiary hospital[22,24,70]. In the absence of rapid on-site evaluation (ROSE), knowing the optimum number of passes is critical information[24,25].

Earlier reports during 2000-2004 recommended 5-7 passes[33,71] for cases with a pancreatic mass. Erickson et al[71] reported in 2002 that cytological diagnosis of malignancy was obtained in 104 (95%) of 110 cases with PC. The average number of needle passes was 3.4 ± 2.2 (range: 1-10) with ROSE, and the number of passes was affected by the differentiation level of the cancer (well differentiated cancer: 5.5 ± 2.7, moderately differentiated: 2.7 ± 1.2, moderately to poorly differentiated: 3.4 ± 2.1, poorly differentiated: 2.3 ± 1.1) (P <0.001)[71]. LeBlanc et al[33] reported in 2004 that 7 passes were needed to achieve a sensitivity of 83% and specificity of 100% from pancreatic and miscellaneous lesions. However, recent studies demonstrated excellent sensitivity (90%-94%)[16,24,25] and specificity (96%-100%)[16,24,25] with fewer numbers of passes. Suzuki et al[25] reported an almost equal sensitivity and specificity of 25G needle EUS-FNA for the solid pancreatic lesion between a fixed number of 4 passes (93% and 100%) and a ROSE dependent procedure with a mean of 2.3 passes (94% and 100%). The sensitivity of the first method was increased by adding passes; 53% by 1 pass, 73% by 2 passes, 87% by 3 passes, and 93% by 4 passes. The most recent RCT by Ramesh et al[20] reported a similar high sensitivity by cytology obtained until the 2^nd pass, with either a 19G flexible needle (92%) or a 25G needle (90%). When on-site cytological information is not available, gross inspection of the whitish component in the obtained materials is simple and useful for judgment of good sampling[24] (Figure 1).

The pancreatic neuroendocrine tumor (PNET) accounts for < 3% of all pancreatic neoplasms[72], and is often defined as a well-demarcated, highly vascular tumor based on clinical images. These image characteristics sometimes resemble solid pseudopapillary tumors (SPTs), acinar cell carcinomas (ACCs), and solid-type serous cystic tumors[73]; the required therapy varies among these tumors. Even within the category of PNET, therapeutic strategy varies by the histological grade (G1, G2 and G3), which is defined by the mitotic index and Ki-67 labeling index[74]. Therefore, EUS-FNA must be able to provide a differential diagnosis as well as accurate grading for treatment of PNET.

PNET has been correctly diagnosed by EUS-FNA at a rate of 77%[75]-90%[76]. Chen et al[75] reported that 54 (77%) of 70 histologically confirmed PNETs were preoperatively diagnosed as PNET by EUS-FNA. A cytological diagnosis of the remaining 16 cases of showed suspected PNET (4), no atypical cells (4), atypical cells (3), unsatisfactory material (2), adenocarcinoma (2), and suspected carcinoma (1), suggesting an overall difficulty in the cytological diagnosis of PNET. Chatzipantelis et al[77] reported the most helpful cytological findings of PNETs were a richly cellular sample with a monotonous, poorly cohesive population of small or medium-sized cells with granular chromatin (salt and pepper) and plasmacytoid morphology; helpful immunochemical diagnostic markers were neuron-specific enolase, synaptophysin, and chromogranin A.

Figueiredo et al[76] reported a large discrepancy between FNA and surgical materials (κ: 0.38, P = 0.003) when using a previous grading system of WHO criteria (2004). However, many recent reports demonstrated high agreement with this grading system. Comparison between EUS-FNA cytology and histology of the resected material showed a concordance of WHO grade and percent agreement of the Ki-67 index of 86% (19/22)[78] and 89% (κ: 0.78)[79], respectively. Histology of EUS-FNA samples and surgical materials also showed high agreement (79%[80]-89%[81]) with the WHO grade based on the Ki-67 index. Hasegawa et al[80] analyzed the intratumoral dispersion of the Ki-67 index of PNET and demonstrated a higher level of dispersion in G2 PNET (0.78) than in G1 PNET (0.03) (P < 0.001). In their study, grading concordance increased up to 90% when the FNA sample contained ≥ 2000 tumor cells, suggesting a necessity for a large amount of sample for accurate grading or prediction of patient’s survival[80].

ACC[82-84], mixed acinar carcinoma[83-85], SPT[86,87], and intraductal tubullopapillary neoplasm (ITPN)[88,89] can also be diagnosed using EUS-FNA samples. These tumors are histologically similar and often require additional immunostainings for differentiation markers: BCL10, lipase, and trypsin are useful markers specific to ACC[83,84]; ITPN is often positive for cytokeratin 7 and 19, MUC1, and MUC6[88,89], but negative for MUC2 or MUC5AC, which are usually stained in intraductal papillary mucinous neoplasms (IPMNs)[88]. The cytological features of SPT are papillary structures, cercariform cells, large cytoplasmic vacuoles, foam cells, and giant cells. Immunostaining of SPT is usually positive for nuclear β-catenin, but usually negative for chromogranin or lipase[86].

Mass-forming pancreatitis includes a progressive form of the ordinary chronic pancreatitis or a specific etiology such as focal-type of autoimmune pancreatitis (AIP). The former histologically shows various level of acinar atrophy associated with a progression of the fibrous stroma, sometimes accompanied with ductal dilation, calculi, and squamous metaplasia of the ductal epithelia[72,90]. Autoimmune pancreatitis (AIP) is a unique form of chronic pancreatitis, sometimes resembling pancreatic malignancies on clinical images; it is histologically classified into type 1 (lymphoplasmacytic sclerosing pancreatitis: LPSP) and type 2 (idiopathic duct-centric pancreatitis: IDCP)[91].

Imai et al[16] described that EUS-FNA specimens obtained from 21 cases of AIP using 22G needles showed no histology meeting the criteria of LPSP or IDCP, but was beneficial only for eliminating PC. In contrast, Ishikawa et al[92] performed EUS-FNA using 22G needles on 47 cases of AIP, and obtained level 1 histological findings of LPSP in 9 cases (19%), level 2 LPSP in 5 cases (11%), and level 2 IDCP in 3 cases (6%). Iwashita et al[28] also reported that EUS-FNA using 19G needles supplied adequate tissue for the histological diagnosis of AIP in 43% (19/44) of the cases, suggesting the importance of the amount or size of EUS-FNA samples for proper diagnosis of AIP.

Several studies have been reported the diagnostic ability of EUS-FNA for biliary strictures. Byrne et al[99] reported that EUS-FNA performed using 22G needles in 23 resected cases with biliary stricture or masses revealed 11 cases of confirmed malignancy in the surgically materials. The sensitivity and specificity of EUS-FNA were 100%, although a sensitivity of EUS observation alone was 45%. DeWitt et al[100], Eloubeidi et al[101] and Ohshima et al[102] reported on the effects of EUS-FNA in cases with biliary strictures (22-28 cases); most underwent ERCP but gave negative or non-diagnostic results by brush cytology. In their studies, the sensitivity, specificity, PPV, NPV, and accuracy for detecting malignancy in 71%-82% of the cases were 77%-100%, 100%, 100%, 29%-57%, and 79%-88%, respectively[100-102]. No complications were reported in these studies.

Needle tract seeding is also a concern for the biliary cancers. El Chafic et al[103] analyzed factors associated with survival in 119 patients with biliary cancer that underwent curative-intent surgery with or without preoperative EUS-FNA (EUS-FNA done in 39 cases), and patient’s age, tumor size, and lymph node metastasis were listed as significant prognostic factors but not preoperative EUS-FNA (HR = 1.09, 95%CI: 0.69-1.73, P value = 0.7).

However, indications for EUS-FNA in cases with biliary stricture may be limited, as transpapillary biliary sampling (biopsy and brushing cytology) can be done at the same time as biliary drainage, with safety and high sensitivities [forceps biopsy (77%-92%)[1,6] and brush cytology (75%-79%)[2,7]].

Gallbladder cancer (GBC) is difficult to diagnose with pathological evidence, and sometimes mimics xanthogranulomatous cholecystitis or acute/chronic cholecystitis. Transpapillary approaches such as naso-gallbladder drainage cytology is feasible and highly diagnostic[104,105], but is sometimes technically difficult and is associated with several complications (mild to moderate acute pancreatitis, cholecystitis and cholangitis)[104,106]. EUS-FNA of the gallbladder lesions has been attempted and provided excellent diagnostic rates for detecting GBC (sensitivity: 80%-90% and specificity: 100%)[107-109]. Kim et al[108] reported that all 14 cases of lymphadenopathy accompanied with suspected GBC were confirmed as lymph node metastasis by EUS-FNA. Interestingly, Hijioka et al[109] reported that 40 (87%) out of 46 cases of GBC was accompanied with lymphadenopathy, of which 36 cases (90%) were confirmed as lymph node metastases by EUS-FNA, in contrast to null lymphadenopathy in 5 cases with xanthogranulomatous cholecystitis. EUS-FNA of the gallbladder wall may have a risk for dissemination in cases with GBC; hence, the first attempt is safer on the regional lymph nodes[109], if swollen.

Tumor of the papilla of Vater is usually easy to diagnose if it originates from the duodenal mucosa, as it is exposed to the intestinal lumen and is detectable by the endoscopy. However, some of the ampullary carcinomas originate from ampullary bile duct, ampullary pancreatic duct, or common duct[110] and are not exposed to duodenal lumen at the early stage. In these cases, endoscopic sphincterotomy of the papilla and subsequent forceps biopsy may increase the chance of obtaining cancer tissues[111], but this may not always effective. Ogura et al[112] attempted EUS-FNA in 10 patients and diagnosed 3 intra-ampullary carcinomas without complications, after the diagnostic failure of cytology and/or forceps biopsy under ERCP.