Prospective randomized study comparing Franseen 22-gauge vs standard 22-gauge needle for endoscopic ultrasound guided sampling of pancreatic solid lesions

Abstract

BACKGROUND

This is a randomized study to compare the diagnostic accuracy of endoscopic ultrasound (EUS)-guided sampling of pancreatic solid lesions obtained with the 22-gauge Franseen (EUS-fine needle biopsy) vs the 22-gauge standard needle (EUS-fine needle aspiration) without rapid onsite evaluation (ROSE), since, in most endoscopy units around the world ROSE is not routinely available.

AIM

To investigate the accuracy of EUS-guided sampling of pancreatic solid lesions obtained between two different needles without ROSE.

METHODS

Patients with a solid pancreatic were included. Patients were biopsied in a randomized order. The primary endpoint was the diagnostic sensitivity for pancreatic malignancy (PM). Secondary outcomes were adequacy of the sample, the mean tissue area, the mean tumor area, and the adverse event rate.

RESULTS

The final diagnosis was pancreatic adenocarcinoma in 38 (76%), neuroendocrine tumor in 4 (8%), chronic pancreatitis in 3 (6%) patients. The sensitivity for PM with Franseen needle was 0.91 [95% confidence interval (CI): 0.80-0.98], vs 0.8 (95%CI: 0.67-0.91) (P = 0.025) with standard needle. The specificity for PM did not differentiate. The accuracy of the standard needle for PM was 0.80 (95%CI: 0.66-0.90), and the Franseen group was 0.90 (95%CI: 0.78-0.97) (P = 0.074). The technical success rates for the standard and Franseen needle groups were 94% (95%CI: 0.83-0.99) and 100% (95%CI: 0.92-1.00), respectively. The mean total tissue area in mm² (SD) was greater in the Franseen group, 2.07 (0.22) vs 1.16 (0.17) (P < 0.01). The mean tumor area in mm² (SD) was not different in Franseen group vs standard group, 0.42 (0.09) vs 0.47 (0.09) (P = 0.80). There were no adverse events.

CONCLUSION

The sensitivity for PM and mean total tissue area, was greater in the as compared with standard needle. The mean tumor area did not differ between the groups.

Key Words: Franseen needle; Standard needle; Endoscopic ultrasound; Pancreatic solid lesions; Rapid onsite evaluation

Core Tip: The main of the study was based in assess the sensitivity of endoscopic ultrasound needles for diagnosing pancreatic malignancy. We found that the diagnostic sensitivity for pancreatic malignancy as well as the mean total tissue area collected was greater with the Franseen needle group compared with the standard needle group. Taking into account, the procedure was done in the absence of an onsite site pathologist for evaluation of the sample collected, bringing important contribution to institutions that do not have pathologist in the examination room.

INTRODUCTION

Endoscopic ultrasound guided fine needle aspiration (EUS-FNA) was initially introduced in the early 1990s and was quickly recognized as the most efficient technique for sampling pancreatic lesions[1].

The use of 22-gauge and 25-gauge needles for solid pancreatic lesions has a sensitivity ranging from 64% to 95%, a specificity ranging from 75% to 100%, and a diagnostic accuracy ranging from 78% to 95%[2,3].

Several strategies have been used to enhance the diagnostic yield of EUS-FNA, such as larger gauge needles[4], rapid onsite evaluation (ROSE)[5], implementation of suction[6], slow withdrawal of the stylet[7], wet suction[8], macroscopic onsite assessment of the material (MOSE)[9], and even the detection of KRAS mutation in the aspirate[10].

The ROSE involves the immediate analysis of the material by the cytopathologist or a cyto technician to guarantee the quality of the aspirated material and even perform a preliminary diagnosis. There is evidence that evaluation at the puncture site by the pathologist increases the diagnostic yield[11], particularly in difficult cases such as lymphoma and association of malignancy with chronic pancreatitis. In a study with 230 patients, the investigators concluded that, in the absence of ROSE, the probability of inconclusive results increased by more than twice (P = 0.03) and by 3 times the number of inappropriate samples for evaluation in block (P < 0.001)[12]. However, availability, logistics, and costs are relevant limitations for the implementation of ROSE in EUS-FNA routine.

Recently, the Franseen needle was designed to obtain samples that allow histological rather than cytological analysis. The puncture performed with these needles has been called endoscopic ultrasound-guided fine needle biopsy (EUS-FNB). The Franseen needle features a crown tip, with three symmetrical surfaces that exhibit three cutting edges. This geometry contributes to a long insertion length and crown tip area that aids better tissue acquisition.

In a pilot study with 30 patients, in which EUS-FNB samples were obtained from pancreatic lesions or other solid masses using Franseen needles, adequate samples with diagnostic capacity were confirmed onsite by the pathologist in 96% of patients, with histological diagnoses by cell block in 96.6% of patients[13]. It is possible that the triple cutting edge could provide more tissue volume than the standard 22-gauge needle.

Some randomized clinical trials (RCT) studies have compared 22-gauge FNB and standard needles for solid pancreatic lesions[14-18]. In the majority of them, ROSE was available.

The objective of this study was to compare the diagnostic accuracy of Franseen 22-gauge needle vs standard 22-gauge needle for EUS-guided sampling of pancreatic solid lesions without ROSE. In most endoscopy units around the world ROSE is not routinely available.

MATERIALS AND METHODS

This was a randomized study conducted at the Instituto do Cancer do Estado de Sao Paulo between December 2019 and January 2023 (registration number at clinicaltrials.gov: NCT04877340). The study protocol was approved by the institutional review board (No. 26962419.3.0000.0065).

Patients with suspected solid pancreatic lesions larger than 15 mm, identified by tomography or magnetic resonance imaging, referred for EUS-guided sampling were eligible for inclusion. Patients willing to participate provided a written informed consent. Patients were submitted to a minimum of two passes with the standard needle and two passes with the Franseen needle. The order of the use of the needles was randomized using permutation blocks.

A function was created with MS Excel, with randomization between blocks of 6, 8, and 10, generating a sequential list of 50 numbers, between 1 and 2 (each number representing a needle type). Randomization process used sealed and sequentially numbered envelopes. Patients were excluded if they had cystic lesions or abnormal coagulation parameters (international normalized ratio > 1.5, platelet count < 50000 cells/mm³).

The primary endpoint was the diagnostic accuracy for the pancreatic malignancy (PM) (pancreatic adenocarcinoma, neuroendocrine tumor, metastatic lesion). Secondary endpoints were adequacy of the sample (technical success rate), the mean tissue area, the mean tumor area, and the adverse event rate.

Pancreatitis (defined as abdominal pain associated with a 3× elevation of serum amylase or lipase), and bleeding (defined as haematemesis or melaena requiring blood transfusion or endoscopic or radiological intervention), occurring within the first 7 days after the endoscopic procedure were considered as adverse events. In the period of 7 days after the procedure, the patients were contacted by phone call in order to evaluate any related adverse event.

Technique

The procedures were performed using a linear echoendoscope (UCT180; Olympus America Corp, Centre Valley, Pa), in propofol-sedated patients, positioned in left lateral decubitus.

The pancreatic mass was firstly punctured with a needle determined by randomization, followed by the other needle in the sequence. Four endoscopists, all with more than 1500 echoendoscopy exams, and fellows under their supervision performed the procedures.

With the first randomized needle, we performed two passes-the first with a slow pull and the second with a 20cc syringe aspiration. Each pass had at least 20 back-and-forth movements. Then, we performed two more passes with the second needle, also using the slow pull and 20cc syringe aspiration sequence. This means that a minimum of four needle passes were performed in all patients. Additional punctures, with both needles, were performed at the discretion of the operator if he or she judged that the material was insufficient. This evaluation was not considered MOSE because we did not evaluate the specimen under a magnifying lens.

Part of the specimen from each pass was smeared onto three slides, and the remaining material was immersed in a formaldehyde solution for cell block analysis.

Cell block preparation, histologic assessement, scanning and measures

Specimens were collected for cell block in a methanol-based preservative solution which was subsequently centrifuged, decanted, and combined to form a tissue clot. After forming a tissue clot, specimens were fixed in formalin, embedded in paraffin, sectioned, and stained using haematoxylin and eosin for histological interpretation. The block was scanned and Panoramic Viewer 1.15.3 (3DHISTECH) was used to photograph the block cuts, always with a measurement reference ruler.

The areas in the photo fields were measured using ImageJ 1.53k (National Institutes of Health, United States; public domain), always calibrating the scale immediately after each photo, and then measuring the non-tumor areas (fibrosis, non-neoplastic pancreas, non-neoplastic intestinal wall tissue, or any other non-neoplastic pathway tissue) and tumor areas. Areas of necrosis, haemorrhage and mucus were not considered as tissue.

Definitions

Possible histological findings were classified as positive for malignancy, suspicious for malignancy, negative for malignancy, and insufficient material for analysis.

Diagnosis accuracy was defined as a correct diagnosis provided by the needle specimen. The definitive diagnosis of malignancy was based on surgical or clinical assessment (e.g. histology of a surgically resected specimen, or histology of a biopsy of a distant metastasis associated with a clinical course compatible with malignant disease). The definitive diagnosis of benign disease was based on benign histology findings associated with a clinical course compatible with benign disease. The minimum follow-up duration was established as 6 months.

Technical success rate was defined as the presence of samples adequate for cytology or histology. When neither the smears nor the cell-blocks samples allowed a diagnosis, this situation was considered as insufficient material for analysis, and a technical failure.

In the presence of malignancy in the tissue sample, the proportion of the positive area for malignancy was calculated and divided by the total area of the sample. This calculation was performed for specimens obtained from both needles.

Diagnostic sensitivity, specificity, accuracy, positive predictive value and negative predictive value for malignancy were calculated for both needles. Truly positive cases for malignancy were considered based on the association of imaging findings, clinical evolution and histology (see above). True negative cases for malignancy were considered when the association of histology findings, imaging and clinical evolution were inconsistent with malignancy after a minimum follow up of 6 months.

All the specimens were analysed by a senior pathologist who was masked to the type of needle that was used.

Sample size calculation and statistical analysis

In our center, the historical diagnostic accuracy for EUS-FNA for solid pancreatic masses using standard 22-gauge needles prior to this study was 75% (unpublished data). This is a superiority trial, where we hypothesised an increase in diagnostic yield of 25% with the use of the Franseen needle compared to standard needles (75%-95%). Adopting a statistical power of 80% and a P alpha level of 5%, 98 patients should be included. Considering that both needles will be used in all patients, 49 patients would be sufficient. Adopting a loss rate of 10%, we estimated the inclusion of 54 patients. Continuous variables were described as means and SD or medians and interquartile range. The analysis was carried out in two stages. The first stage consisted of calculating descriptive statistics of the variables of interest. The second step, in calculating the comparisons of prediction metrics. The prediction metrics evaluated were sensitivity, specificity, positive value prediction and negative value prediction. For to carry out these comparisons, the McNemar tests were used, for sensitivity and specificity, and the Moskowitz and Pepe test, for prediction of positive value and prediction of negative value. All of these analyzes were performed with the DTComPair package of the R language.

RESULTS

From December 2019 and January 2023, we screened and included 50 consecutive patients with solid pancreatic lesions referred to EUS-guided sampling. There was no screen failures.

Patient demographics and tumor characteristics

The mean age of the study cohort was 64 years (range 36-88 years), and 24 patients (48%) were male. The mean size of the pancreatic mass was 3.47 cm (range 1.7 cm-7.0 cm). Most lesions were located in the pancreatic head (62%) or uncinate process (12%). Vascular invasion was observed in 64% of patients (Table 1). The pathological analysis diagnosis was pancreatic adenocarcinoma in 74%, neuroendocrine tumor in 8 % and chronic pancreatitis in 6% (Table 2).

Table 1 Patient details and pancreatic mass characteristics (n = 50).

Characteristics	n (%)
mean age	64.1
Range	(36-88)
Gender
Female	26
Male	24
Mean pancreatic mass size (cm)
Mean	3.47
Range	1.7-7.0
Pancreatic mass location
Head	31 (62)
Uncinate	6 (12)
Neck	6 (12)
Body	5 (10)
Tail	2 (4)
Vascular invasion	32 (64)
Venous vascular invasion
Portal vein	16 (51.6)
Superior mesenteric vein	19 (61.3)
Splenic vein	13 (41.9)

Table 2 Pathological analysis diagnosis.

Definitive diagnosis (n = 50)	Standard needle (n = 50)		Franseen needle (n = 50)
Adenocarcinoma/pancreatic cancer (n = 38)	n = 31		n = 35
	Benign lesion	Chronic pancreatitis (n = 3)	Benign lesion	Chronic pancreatitis (n = 2)
		Mucinous neoplasia (n = 1)		Mucinous neoplasia (n = 1)
	Inconclusive (n = 3)		Inconclusive (n = 0)
Neuroendocrine tumor	n = 4		n = 4
Metastasis	n = 2		n = 2
Lymphoma	Chromic pancreatitis (n = 1); Lymphoma (n = 0)		Chronic pancreatitis (n = 1); Lymphoma (n = 0)
Plasmocitoma	n = 1		n = 1
Solid pseudopapillary tumor	n = 1		n = 1
Chronic pancreatitis	n = 3		n = 3

Final diagnoses were based on histological examination. No additional passes were necessary. In one case, the diagnosis was established by immunohistochemistry (as a B-cell lymphoma). The sensitivity for a final diagnosis of PM was significantly greater in the Franseen group 0.91 [95% confidence interval (CI): 0.80-0.98] vs standard group 0.81 (95%CI: 0.67-0.91) (P = 0.025). The specificity was 0.67 (95%CI: 0.09-0.99) for PM in the standard and 0.67 (95%CI: 0.09-0.99) for the Franseen needle groups, without difference between the groups. The accuracy of the standard needle for PM was 0.80 (95%CI: 0.66-0.90). In the Franseen needle group, accuracy for PM was 0.90 (95%CI: 0.78-0.97) (P = 0.074). The positive predictive value for the standard group for PM was 0.97 (95%CI: 0.87-1.00) and for the Franseen group, 0.98 (95%CI: 0.88-1.00), (P = 0.36). The negative predictive value for PM in the standard needle group was 0.18 (95%CI: 0.02-0.52), compared to 0.33 (95%CI: 0.04-0.78) in the Franseen needle group, with P = 0.028 (Tables 3-5). Although this study only enrolled patients with solid lesions, one patient was diagnosed with an intraductal papillary mucinous neoplasm. This patient had a solid tumor on both computed tomography scan and EUS imaging.

Table 3 Performance of standard needle for the diagnosis of pancreatic malignancy.

Diagnostic method	Disease
	Pancreatic malignancy1 (n = 47)	Chronic pancreatis (n = 3)	Total
Positive for malignancy	38	1	39
Negative for malignancy	9	2	11
Total	47	3	50
95%CI
Sensitivity		0.81 (0.67-0.91)
Specificity		0.67 (0.09-0.99)
PPV		0.97 (0.87-1.00)
NPV		0.18 (0.02-0.52)
Accuracy		0.80 (0.66-0.90)

¹Pancreatic malignancy included adenocarcinoma, neuroendocrine tumor, distant metastasis, lymphoma, plasmocitoma, solid pseudopapillary tumor.

CI: Confidence interval; PPV: Positive predictive value; NPV: Negative predictive value.

Table 4 Performance of Franseen needle for the diagnosis of pancreatic malignancy.

Diagnostic method	Disease
	Pancreatic malignancy1 (n = 47)	Chronic pancreatis (n = 3)	Total
Positive for malignancy	43	1	44
Negative for malignancy	4	2	6
Total	47	3	50
95%CI
Sensitivity		0.91 (0.83-0.98)
Specificity		0.67 (0.09-0.98)
PPV		0.98 (0.88-1.00)
NPV		0.33 (0.04-0.78)
Accuracy		0.90 (0.78-0.97)

¹Pancreatic malignancy included adenocarcinoma, neuroendocrine tumor, metastasis, lymphoma, plasmocitoma, Frantz tumor.

CI: Confidence interval; PPV: Positive predictive value; NPV: Negative predictive value.

Table 5 Comparison of the diagnostic performance of standard (fine needle aspiration) and Franseen (fine needle biopsy) needles.

Metrics	FNA	FNB	Statistical difference	P value1
Sensitivity	0.809	0.915	5.000	0.025
Specificity	0.667	0.667	-	-
PPV	0.974	0.977	0.913	0.361
NPV	0.182	0.333	2.202	0.028
Accuracy	0.800	0.900	3.200	0.074

¹McNemar (for sensitivity and specificity comparisons), Moskowitz and Pepe (for positive predictive value and negative predictive value comparisons) tests.

FNA: Fine needle aspiration; FNB: Fine needle biopsy; PPV: Positive predictive value; NPV: Negative predictive value.

The technical success rates for standard and Franseen needle groups were 94% (95%CI: 0.83-0.99) and 100% (95%CI: 0.92-1.00), respectively.

Histology assessment

The mean total tissue area mm² (± SD) was significantly higher for the Franseen than for the standard needle group, 2.07 ± 0.22 vs 1.16 ± 0.17 (P < 001). The mean total tumor area, mm² (± SD) did not differ between the Franseen, and in the standard needle groups 0.42 ± 0.09 vs 0.47 ± 0.09, P = 0.8 (Table 6). In Figure 1, there are examples of pathology images from samples obtained with the standard and Franseen needles, respectively.

Figure 1 Pathology images from samples obtained of pancreatic adenocarcinoma. Both specimens showing desmoplastic fibrosis and malignant ductal epithelium in cell block. A: Standard needle (100×); B: Franseen needle (100×).

Table 6 Comparison of procured histologic tissue areas (in mm²) between standard needle versus Franseen needle.

	Standard needle	Franseen needle	P value
Mean total tissue area, mm2 (SD)	1.16 (0.17)	2.07 (0.22)	0.001
Median	0.45	1.09
75% IQR	1.53	2.89
Range	0-9.76	0-9.22
Mean total tumor area, mm2 (SD)	0.42 (0.09)	0.47 (0.09)	0.8
Median	0.05	0.05
75% IQR	0.51	0.50
Range	0-7.03	0-4.88

IQR: Interquartile range

Adverse events

In this study, we observed mild abdominal pain in only 5 patients, that resolved with simple analgesia and did not required hospitalization, three in the FNA and two in the FNB group. There were no cases of pancreatitis or bleeding related to the procedure.

DISCUSSION

We conducted a randomized study to compare the diagnostic accuracy of EUS-guided sampling of pancreatic solid lesions obtained with the 22-gauge Franseen vs the 22-gauge standard needle with no ROSE. The justification of the study design relies on the fact that ROSE is not routinely available in many centers in our country and around the world.

Our findings suggest a better diagnostic performance with the Franseen needle translated in a better sensitivity and negative predictive value for the diagnosis of PM. In fact, with the Franseen needle it was possible to correctly diagnose pancreatic adenocarcinoma in four misdiagnosed patients (inconclusive diagnosis in three and chronic pancreatitis in one patient) with the standard needle. The technical success rates for the standard and Franseen needle groups were 94% (95%CI: 0.83-0.99) and 100% (95%CI: 0.92-1.00) respectively, which reflects the greater amount of tissue samples obtained by the Franseen needle (mean total tissue area in mm², standard vs Franseen needles, 1.16 vs 2.07, P = 0.001). Recently, Kovacevic et al[19] also used a Franseen needle (TopGain; Medi-Globe GmbH, Grassau, Germany), as in our study, and found a larger area of tissue with the FNB vs the FNA needle (2.74 mm²vs 0.44 mm², P < 0.001).

Facciorusso et al[20], in a network meta-analysis, observed no significant difference in diagnostic accuracy between EUS-guided tissue acquisition for sampling pancreatic masses using different needle models. However, this systematic review was based mainly on studies that used first-generation reverse-bevel FNB needles because only a limited number of RCTs that tested newer end-cutting FNB needles were available at that time.

We found five RCTs comparing FNA vs FNB needles. Their results are summarized in Table 7.

Table 7 Randomized clinical trials comparing fine needle aspiration vs fine needle biopsy needles for endoscopic ultrasound-guided sampling of solid pancreatic tumors.

Ref.	FNA (n)	FNB (n)	ROSE	Randomization	Technique of sampling	Sensitivity for malignancy (95%CI)-FNA	Sensitivity for malignancy (95%CI)-FNB	P value	Accuracy malignancy (95%CI)-FNA	Accuracy malignancy (95%CI)-FNB	P value
Bang et al[14]	22-gauge (28)	Reverse bevel (28)	Yes	Sequence of the needle	Capillarity and dry suction				100	83.3	0.26
Bang et al[15]	22-gauge (46)	Franseen 22-gauge (46)	Yes	Sequence of the needle					82.6	97.8	0.03
Noh et al[16]	22-gauge (30)	Reverse bevel 22-gauge (30)	Yes	Sequence of the needle	Dry suction				95	93.3	0.564
Vanbiervliet et al[17]	22-gauge (39)	Reverse bevel 22-gauge (41)	Yes	First needle FNA	Dry suction				92.5	90	0.68
Mavrogenis et al[18]	22-gauge (19)	Reverse bevel 25-gauge (19)	No	Sequence of the needle	Capillarity and dry suction	89.5 (66.82-98.39)	89.5 (66.82-98.39)		84.8 (67.3-94.2)	84.8 (67.3-94.2)
Kovacevic et al[19]	22-gauge (33)	22-gauge (31)	No	Sequence of the needle	Capillarity	65.5%	89.7%	> 0.5	69.7 (51.3-84.4)	90.3 (74.2-98%)
Our study	22-gauge (50)	Franseen 22-gauge (50)	No	Sequence of the needle	Capillarity and dry suction	0.83 (0.69-0.92)	0.91 (0.80-0.98)		0.84 (0.71-0.93)	0.92 (0.81-0.98)

CI: Confidence interval; FNA: Fine needle aspiration; FNB: Fine needle biopsy; ROSE: Rapid onsite evaluation.

In a more recent network meta-analysis of different FNB needles, Gkolfakis et al[21] found that Franseen and fork-tip needles, particularly those of 22-gauge size, showed the highest performance for tissue acquisition. One observation was that the availability of ROSE had a great impact on the comparative efficacy of different techniques for tissue sampling of pancreas.

In a retrospective study, Wong et al[22] found that the diagnostic yield of solid pancreatic mass was higher with FNB using the Franseen needle than in FNA using the conventional FNA needle in a center where ROSE was unavailable[22]. Therefore, it seems that a main advantage of EUS-FNB needles, particularly with newer end-cutting needles, is to obviate the use of ROSE by performing the EUS-FNB sampling without an on-site pathologist[16]. Our findings are in line with this concept.

In times of personalised medicine, molecular profiling of pancreatic cancer and application of next-generation sequencing may provide an opportunity to advance the development of targeted therapies. Asokkumar et al[23] demonstrated that the Franseen EUS-FNB device can obtain better nucleic acid yield than EUS-FNA, with quality and quantity sufficient for downstream genomics applications. Therefore, the FNB can become a convenient and safe method for obtaining tumor material for precision genomics. This could have implications for the outcome of pancreatic cancer treatment in the near future.

There are a few limitations of this study. First, we included only pancreatic masses, and therefore the performance of the Franseen and standard needles for evaluating other solid mass lesions could not be evaluated. A second limitation was the small number of patients. Maybe we could have detected a greater diagnostic accuracy in the Franseen needle group with a larger number of patients. Third, we did not compare the performance of the strategies of sampling, i.e., stylet retraction vs suction. This comparison was not among the study aims, and the sample size was not gauged for it. Finally, the pathologist was masked to the type of needle. However, the pathologist evaluated all the processed slides and cell blocks of a specific patient at once. It is possible that the results of the interpretation of the specimen obtained with one needle model influenced the interpretation of the specimen obtained with the other needle.

CONCLUSION

In conclusion, for the EUS-guided tissue sampling of solid pancreatic lesions, the Franseen needle obtained a greater tissue area compared to the standard needle. The diagnostic sensitivity and negative predictive value of the Franseen needle was greater for the diagnosis of any PM.

ACKNOWLEDGEMENTS

We are grateful to Peetermans J and Rousseau M for their insightful comments on the manuscript text and statistics.