Comparative diagnostic performance of endoscopic ultrasound-guided fine-needle aspiration (EUS-FNA) versus endoscopic ultrasound-guided fine-needle biopsy (EUS-FNB) for tissue sampling of solid pancreatic and non-pancreatic lesions without ROSE: a prospective multicenter study

Abstract

Background and aims

Endoscopic ultrasound-guided tissue acquisition, including both fine-needle aspiration (EUS-FNA) and fine-needle biopsy (EUS-FNB), has been frequently performed to acquire samples from both pancreatic and non-pancreatic lesions. Still, the impact of the diagnostic yield between FNA and FNB is uncertain. We conducted this study to compare the diagnostic performance and accuracy of the 22-gauge FNA needles with the 22-gauge FNB needles in sampling solid pancreatic and non-pancreatic lesions.

Methods

This is a prospective multicenter study conducted on 465 cases presented with solid pancreatic or non-pancreatic lesions.

Results

Patients were 275 male and 190 females with a mean age of 59 years. Three-hundred twenty-seven patients had solid pancreatic lesions, while 138 had non-pancreatic lesions; 245 cases underwent EUS-FNA, and the remaining 211 cases underwent EUS-FNB. The presence of intact tissue core and sample adequacy was significantly higher in the FNB cases in solid pancreatic and non-pancreatic lesions. Blood contamination was significantly more in cell blocks and smears of EUS-FNA compared to that of EUS-FNB in solid pancreatic and non-pancreatic lesions. Based on histologic assessment of cell block only, EUS-FNB had more diagnostic accuracy (99%) than FNA (61%) (P-value < 0.005). However, cytological diagnosis by smears only showed no significant difference. The combined cytological and histological evaluation had 100% sensitivity, specificity, and accuracy.

Conclusion

EUS-FNA and EUS-FNB have comparable accuracy in diagnosing solid pancreatic and non-pancreatic lesions without ROSE. EUS-FNB is superior to EUS-FNA in acquiring intact tissue core and adequate samples with little blood contamination. Based on histological assessment (cell block/tissue core) only, EUS-FNA has less accuracy than EUS-FNB in diagnosing solid pancreatic lesions.

Core tip

Although endoscopic ultrasound-guided fine-needle aspiration (EUS-FNA) was considered the procedure of choice to obtain samples from pancreatic lesion, however, it still has limitations affecting its diagnostic yield. The advantages of the endoscopic ultrasound-guided fine-needle biopsy (EUS-FNB) needles include more tissue integrity and adequate samples with little blood contamination. Our results were in favor of EUS-FNB sampling, which was found to outperform EUS-FNA for sampling solid pancreatic and non-pancreatic lesions with regard to acquiring intact tissue core and adequate samples with little blood contamination.

Introduction

The newly developed modality for acquisition of cytology from pancreatic masses is both fine-needle aspiration (EUS-FNA) and fine-needle biopsy (EUS-FNB) [1, 2].

As its well known about the cancer pancreas that it is considered as aggressive one, and most of patients usually when presented with it they are presenting present with advanced unresectable tumors. Hence, it becomes imperative to ascertain malignancy via EUS-FNA or EUS-FNB prior to commencing oncological therapy. Additionally, various other pancreatic conditions like focal chronic pancreatitis or autoimmune pancreatitis, alongside pancreatic cystic lesions, necessitate EUS-FNA or EUS-FNB sampling to rule out malignant transformation. Still, some limitations are facing EUS-FNA, despite their great role in the diagnosis of pancreatic lesions, those limitations including small amount of tissue collected and the inability to obtaining intact tissue core, also beside the adequacy of the obtained samples that may be affected by the experience of the endoscopist and cytotechnologist and the presence of rapid on-site cytological evaluation (ROSE) that is not widely available nowadays [3].

For solving these limitations, the tips of FNB needles were designed by cutting edges that provide adequate histologically intact tissue core fragments allowing proper immunohistochemical analysis, molecular profiling, and subsequent targeted therapies [4].

With the growing evidence regarding the diagnostic benefits of FNB over FNA, consensus still not definitive [5], therefore, the aim of this study is to compare the diagnostic performance of EUS-FNA and EUS-FNB in solid pancreatic and non-pancreatic lesions, focusing on one needle size: 22-gauge needles.

Material and methods

Patients

This is a prospective multicentric study conducted at five tertiary referral centers in Egypt on four-hundred sixty-five (465) patients between March 2020 and March 2023. Patients enrolled in this study had radiological evidence of solid pancreatic, other intra-abdominal or mediastinal lesions and required EUS-guided tissue sampling. Patients with anemia (Hb < 8 mg/dl), uncorrected coagulopathy, or cardiorespiratory dysfunction were excluded. Patients were then divided into two groups: Group (A) underwent EUS-FNA using 22G needles (Cook Medical), including 254 cases. and group (B) underwent EUS-FNB using 22G Franseen-tip acquire needle with three symmetric cutting edges (Boston Scientific), including 211 cases. The primary outcome of our study was diagnostic accuracy. Secondary outcomes were sample adequacy, diagnostic sensitivity, specificity, and adverse event rate.

Technique of sampling

EUS examination was performed under intravenous propofol sedation using a linear echoendoscope Pentax EG3870UTK attached to Hitachi Avius ultrasound machine. The endoscopist used color Doppler to identify the optimal position for puncture without intervening vessels between the needle and target lesion. The endoscopist choose either the FNA 22-gauge needle (Cook Medical) or the 22-gauge FNB Franseen-tip needle (Acquire; Boston Scientific, USA). The needle was then inserted into the lesion with back-and-forth movements done, while simultaneous minimal negative pressure was applied by pulling the needle stylet slowly and continuously (slow-pull or capillary technique). Continuous suction was then applied (dry-suction technique) using a 10-ml syringe while the needle moving back-and-forth for at least 20 times within the lesion. Then, suction was released, and the needle was withdrawn from the lesion. The tissue samples were expelled from the needle in a standardized manner. The stylet was introduced until the tip of the needle to expel the tissue samples on the glass slide, followed by flushing the needle with air or saline to expel the retained tissue on slides and in 10% neutral buffered formalin. ROSE was not available during the procedure, and one to three passes were carried out for each lesion. Smears were prepared from the material obtained by either FNA or FNB and immediately fixed in 95% ethyl alcohol for 20 min and then stained with Papanicolaou stain. Moreover, abundant material obtained from FNA and FNB (blood clots containing cellular aggregates and/or tissue fragments) was directly fixed in 10% neutral buffered formalin and then routinely processed. Tissue blocks were cut into 4-μm slides and stained with hematoxylin and eosin.

Cytological and histological evaluation

A three-class grading system was adopted for assessment of the cytological smears (cellularity and blood contamination) and the cell blocks/tissue cores (tissue integrity and blood contamination) obtained by each needle.

• 1. Tissue integrity:

  • Grade A: An architecturally intact tissue fragment at least 550 µm in the greatest axis (= the diameter of a high-power field of Olympus CX31 microscope).

  • Grade B: The tissue fragments do not meet the criteria of Grade A, but a diagnosis can still be made based on cellular morphology.

  • Grade C: No lesion cells present, and a diagnosis cannot be made.

• 2. Smear cellularity:

  • Grade A: > 4 cell clusters, with at least 10 cells in each.

  • Grade B: 2 to 4 cell clusters, with a minimum of 10 cells in each.

  • Grade C: Unsatisfactory, no lesion cells present.

• 3. Blood cell contamination (for both smears and tissue blocks):

  • Grade A: Little blood contamination (< 25% of the slide surface area).

  • Grade B: Medium blood contamination (25–50% of the slide surface area).

  • Grade C: Much blood contamination (> 50% of the slide surface area).

Immunocytochemistry

Immunostaining was performed on cell block/tissue core material following the standard immunoperoxidase methods using Ventana BenchMark. The antibodies used according to the differential diagnosis included CK, CK7, CK20, P63, Ki67, synaptophysin, chromogranin, CD10, CD68, LCA, TTF1, CEA, ER, PR, HER2, GATA-3, SMA, CD117, DOG-1, beta-catenin, CD34, S100, glypican-3, and HepPar-1.

Final diagnosis

The diagnosis of pancreatic cases followed “The Papanicolaou Society of Cytopathology System for Reporting Pancreaticobiliary Cytology; standardized terminology and nomenclature for pancreaticobiliary cytology” [6].

The diagnosis of non-pancreatic lesions followed five categories: “unsatisfactory,” “positive for malignancy,” “suspicious for malignancy,” “atypia,” and “negative for malignancy” [6].

The diagnostic accuracy of pancreatic and non-pancreatic lesions was compared to the gold standard postsurgical pathology for those who underwent surgery, the presence of distant metastasis, or 6-month follow-up without changes in the size of the benign lesions or progression in the size or newly developed metastatic deposits in malignant lesions.

Statistical methods

The collected data were entered on the computer using Microsoft Office Excel Software Program 365. Pre-coded data were then entered into the Statistical Package of Social Science Software program, version 26 (SPSS). For quantitative variables, they were described as mean, standard deviation, median, minimum, and maximum. For qualitative variables, they were described as frequency and percentage and compared using the chi-square test, where the P-value is significant if less than 0.05.

Results

A total of 465 patients were enrolled in this study: 275 males and 190 females. The mean age was 59 years. Three-hundred twenty-seven cases had solid pancreatic lesions (70.3%), and most of them (201 cases) were in the pancreatic head. The non-pancreatic lesions included abdominal lymph nodes (65 cases), stomach (21 cases), liver (16 cases), ampulla of Vater (8 cases), common bile duct (7 cases), duodenum (6 cases), rectum (5 cases), esophagus (3 cases), mediastinum (3 cases), retroperitoneum (2 cases), gall bladder (1 case), and suprarenal (1 cases). The mean size of both pancreatic and non-pancreatic lesions was 4 cm (Table 1).

Table 1 Patient’s variables of FNA and FNB groups

The FNA group included 254 patients; 180 cases had solid pancreatic lesions, and the remaining 74 cases had non-pancreatic lesions. The FNB group included 211 patients; 147 had solid pancreatic lesions, and the remaining 64 patients had non-pancreatic lesions. In both groups, the median number of needle passes was 2. The baseline characteristics for all patients are summarized in Table 1. Technical success occurred in all cases without complications (Figs 1, 2, 3, 4, 5, 6, 7 and 8).

Fig. 1

EUS-FNA of gastric body GIST

Fig. 2

EUS-FNB of pancreatic adenocarcinoma

Fig. 3

Case of mediastinal lymph node sarcoidosis; EUS-FNB (22 gauge). A Cell block showing grade A tissue integrity and grade A blood contamination (hematoxylin & eosin × 100). B Higher power of the same section showing epithelioid cells along with lymphocytes forming noncaseating granulomas (hematoxylin & eosin × 400)

Fig. 4

Case of metastatic adenocarcinoma in mediastinal lymph node; EUS-FNB (22 gauge). A Cell block showing irregular crowded tubular glands set within desmoplastic stroma with focal mucinous differentiation (hematoxylin & eosin × 100). B Higher power shows irregular glands lined by columnar epithelial cells with occasional intracytoplasmic vacuolization (hematoxylin & eosin × 400)

Fig. 5

Case of moderately differentiated duodenal adenocarcinoma; EUS-FNB (22 gauge). A Cell block showing grade A tissue integrity and grade A blood contamination (hematoxylin & eosin × 100). B Higher power shows malignant irregular glands lined by columnar mucous secreting cells with evident perineural invasion (hematoxylin & eosin × 400)

Fig. 6

Case of moderately differentiated pancreatic adenocarcinoma; EUS-FNA (22 gauge). A Smear shows grade A cellularity and grade A blood contamination (Papanicolaou × 200). B Higher power shows clusters of malignant epithelial cells having pleomorphic irregular hyperchromatic nuclei with squamoid cytoplasm (Papanicolaou × 400). C Low-power view of the cell block of the same case showing grade A tissue integrity and grade C blood contamination (hematoxylin & eosin × 40). D Higher power of the same cell block showing irregular glands lined by pleomorphic columnar cells within desmoplastic stroma (hematoxylin & eosin × 400)

Fig. 7

Case of solid pseudopapillary tumor of the pancreas; EUS-FNB (22 gauge). A Cell block showing grade A tissue integrity and grade A blood contamination (hematoxylin & eosin × 100). B Higher power shows sheets of monotonous round cells with nuclear grooves separate by delicate vascular stroma (hematoxylin & eosin × 400). C Tumor cells show diffuse strong nuclear and cytoplasmic reaction to beta-catenin (× 400). D Tumor cells show diffuse strong nuclear reaction to PR (× 400)

Fig. 8

Case of pancreatic diffuse large B-cell non-Hodgkin’s lymphoma; EUS-FNB (22 gauge). A Cell block showing grade A tissue integrity and grade A blood contamination (hematoxylin & eosin × 40). B Higher power shows monotonous atypical large lymphoid cells (hematoxylin & eosin × 400). C Tumor cells are diffusely positive for CD20 (× 400). D CD3 stains very reactive T lymphocytes (× 100)

Obtaining an intact tissue core (Grade A tissue integrity) was successful in 167 cases (79.1%) of the FNB group compared to only 21 cases of the FNA group with statistically significant P-value as shown in Table 2. FNB failed to obtain intact tissue fragments only in two cases (Grade C tissue integrity). On the other hand, FNA failed in 26 cases (Grade C tissue integrity), and most of them were solid pancreatic cases (25 cases). The majority of cell blocks of FNA group (81.5%) contain tissue fragments less than 550 µm (Grade B tissue integrity). There was no significant difference in smears cellularity between both groups. Smears were inadequate only in three cases of the FNA group and in nine cases of the FNB group. Blood contamination was significantly more in the material sampled by FNA (smears and cell blocks) with statistically significant P-value as shown in Table 2. No technical difficulties were recorded during EUS procedure using either FNA or FNB. All patients were kept under observation after tissue acquisition for at least 6 h and followed up for 2 days thereafter. Minor adverse events occurred in 3.3% of the patients in the form of mild abdominal pains, transient fever, or small hemodynamically insignificant blood collection not necessitating hospital admission. No major complications occurred.

Table 2 Comparison of the sampling outcome between both groups in pancreatic and non-pancreatic lesions

Comparison of sampling outcomes between solid pancreatic and non-pancreatic lesion in both groups is demonstrated in Table 2.

The diagnostic categories of pancreatic and non-pancreatic lesions are shown in Table 3. Three-hundred eighty-seven cases (83.2% of all cases) were diagnosed malignant: 281 solid pancreatic cases and 106 non-pancreatic cases. Sixty-four cases (13.8% of all cases) were negative for malignancy, divided equally between both pancreatic and non-pancreatic cases. Pancreatic ductal adenocarcinoma represents about 84% of the pancreatic cases, and about 10% had chronic pancreatitis. Various diagnoses are demonstrated in Tables 4 and 5.

Table 3 The diagnostic categories of pancreatic and non-pancreatic lesions

Table 4 Different diagnoses of pancreatic lesions

Table 5 Different diagnoses of non-pancreatic lesions

Immunocytochemistry was done to confirm the diagnoses of neuroendocrine tumors, solid pseudopapillary tumors of the pancreas, acinar cell carcinoma, GIST, hepatocellular carcinoma, undifferentiated carcinoma, and lymphoma.

Interestingly, there were nine cases of GIST in the FNB group in which core tissue samples played a major role in confirming the diagnosis by applying the specific immunohistochemistry.

Based on smear examination only, reaching a diagnosis failed in nine cases of the FNB group in contrast to three cases only in the FNA group.

Based on cell block/tissue core only, the FNA needles failed to establish diagnosis in 26 cases (most of them are pancreatic, 25 cases) in contrast to 2 cases only (one solid pancreatic and one non-pancreatic) in the FNB group. Based on examination of both smears and cell blocks, a diagnosis could be established in all cases.

The combined evaluation of both cytological material (smears) and histological material (cell blocks/tissue cores) obtained either by FNA or FNB needles resulted in 100% sensitivity, specificity, PPV, NPV, and accuracy. However, the sensitivity and accuracy based on cell blocks only decreased in the FNA group to 88% and 90%, respectively. Based on cell blocks/tissue cores only, the FNB group had more diagnostic accuracy 99% and sensitivity 98% compared to 90% and 88% for the FNA group, respectively. There was no statistically significant difference in establishing a conclusive diagnosis based on smears only or cell blocks/tissue cores only (Table 6).

Table 6 Comparison of the diagnostic outcomes between both groups based on smears only and cell blocks/tissue cores only

Follow-up of cases was done for at least 6 months, and correlation with radiological data was also done. All cases diagnosed as malignant, suspicious, or atypical were matched with the radiological data and follow-up results.

Discussion

EUS-TA considers the best modality for cytopathological and histological characterization of pancreatic masses and beside several abdominal lesions [7].

There are several factors that could influence the diagnostic efficacy of EUS-TA such as different techniques of sampling, fanning technique, use of rapid on-site cytopathology evaluation [ROSE], and endosonographer and cytopathologist training and volume. Also, the design of needle (FNA and FNB) and needle size (19 gauge vs 22 gauge vs 25 gauge) consider the two mainstay factors [8].

Safety of the EUS-FNA in diagnosis of solid pancreatic lesions, with high diagnostic accuracy, which usually reaching up to 85%, is raised when accompanied by ROSE [9].

Insufficient sample represents the most important and popular limitation of EUS-FNA in diagnosing pancreatic lesions [10].

Usually, EUS-FNA samples do not provide adequate tissue characterization for diagnosis of some lesions such as subepithelial (stromal) tumors, lymphomas, autoimmune pancreatitis, metastasis, and other uncommon lesions. Therefore, achieving an adequate and proper amount of intact tissue cores that are suitable for immunohistochemical staining is crucial for confirming the diagnosis of such lesions and for allowing subsequent molecular studies that could help in choosing the proper way for management of some tumors [11].

Several novel needle designs are achieving a significant stride in the field of EUS-TA, together with tissue core needles, and provide a more histology-grade tissue compared to the regular FNA needles owing to the higher number of their cutting edges [5].

There are still some limitations in providing intact tissue cores with these needles, beside the diagnostic value of the obtained cytological smears versus histological tissue cores [12]. Therefore, further studies are needed to ensure the impact of the needle type (FNA or FNB) on the diagnostic yield and tissue quality. Hence, the aim of this study was to assess sample quality obtained by EUS-FNA and EUS-FNB techniques and to compare the diagnostic accuracy of both techniques in pancreatic and non-pancreatic lesions.

Considering the safety of the EUS-TA technique, the adverse effects associated are rare; in most reported cases, sampling was successful with no recorded complications either in FNA or FNB groups [13,14,15].

In our study, there was no significant statistical difference between the FNA and FNB groups regarding age, sex, size, and lesion site. These findings are similar to those obtained from studies done in the last few years [14, 16].

Most of published studies that compare FNA and FNB showed that FNB typically requires fewer needle passes to achieve adequate samples [14, 17].

The main benefits of a lower number of passes are shorter procedure time and less risk of adverse events. In our study, the median number of needle passes was two in both groups. The randomized controlled trial of Bang et al. [18] also showed no significant difference in the number of passes required to establish a diagnosis between both types of needles. This may be attributed to the experience of the endoscopist.

The FNB Franseen biopsy needle is capable of extracting a tissue core necessary for conducting immunocytochemical and molecular investigations. These studies are more useful for the diagnosis of some non-pancreatic lesions (as GIST, lymphoma, and metastasis) and also for pancreatic malignancies, especially in the era of individualized therapy [19].

When comparing the integrity of tissue cores that are obtained by FNB needles (either from solid pancreatic or non-pancreatic lesions) with those obtained by FNA needles, it is shown that the first one is better with similar overall smear cellularity [20,21,22].

The diagnostic outcome based on the obtained smears and cell blocks/tissue cores was assessed and compared between FNA and FNB groups in many studies, like the study of Cheng et al. [21] that was found FNB achieved better histological yield in solid pancreatic masses, while there was no difference between both needles in non-pancreatic lesions. This is unlike the study of Aadam et al. [23] who concluded that FNB was superior to FNA in the diagnostic yield of the non-pancreatic lesions.

Our study results agreed with recent network meta-analysis conducted by Gkolfakis et al. (2022) and Han et al. (2021) who showed higher performances with 22-gauge newer FNB needles as compared with FNA needles [24, 25].

Several prior studies have demonstrated variances in diagnostic capabilities of EUS-FNA and EUS-FNB, which are suggesting an overall superiority of FNB needles [26, 27]. Our study does not show significant difference in the diagnostic accuracy of Franseen acquire needles and standard FNA needles in either solid pancreatic or non-pancreatic lesions. This result was also recorded by other studies [28,29,30,31].

Finally, we may referred this to the experienced cytopathologist who can made a diagnosis based on cytomorphology of cells present in smears and cell block, unlike the traditional surgical pathologist who depends on tissue architecture to establish the diagnosis. In contrast, other studies [32, 33] found that FNB was inferior to FNA in diagnosing pancreatic masses. This might be attributed to the smaller number of cases, and the different number of passes used for FNA and FNB, in addition that these studies were using first-generation reverse-bevel FNB needles.

Conclusion

EUS-guided FNA and FNB are safe procedures for sampling pancreatic and non-pancreatic lesions with comparable diagnostic outcomes using the same number of needles passes and in the absence of ROSE. EUS-FNB is superior to EUS-FNA for sampling solid pancreatic and non-pancreatic lesions as regard the integrity of tissue cores and adequate samples with little blood contamination.

Development of new generations of EUS-FNB needles and recent advances in EUS-TA are of interest given the proposed advantages concerning improved diagnostic performance, preserved tissue architecture, thus allowing for immunohistochemistry required for certain diagnoses, and obviating ROSE.