Introduction
Endoscopic ultrasound-guided fine-needle aspiration (EUS-FNA) is the current standard
method of choice for the tissue diagnosis of pancreatic or peripancreatic neoplasms
[1 ]. A recent meta-analysis reported that the sensitivity and specificity of EUS-FNA
for the diagnosis of pancreatic cancer were 89.9 %–90.8 % and 100 %, respectively
[2 ]. Although high diagnostic yields are reported, EUS-FNA is limited because the diagnostic
yield is impacted by the presence of an on-site pathologist [3 ]. Furthermore, cytology specimens alone are limited in diagnosing diseases for which
the histologic architecture or ancillary studies are necessary, such as lymphomas
and gastrointestinal stromal tumors [4 ]. To overcome these limitations, various designs and needle sizes have been developed
for fine-needle biopsy (FNB) and have been used on pancreatic masses [5 ]
[6 ].
Recently, a 22G needle with Franseen geometry for EUS-FNB (Acquire; Boston Scientific
Corporation, Natick, Massachusetts, USA) was developed. The Franseen design has a
crown tip with three symmetrical surfaces that manifest as three cutting edges. The
needle geometry incorporates a smaller included angle and a larger inclination angle.
This unique geometry contributes to a longer insertion length and area at the crown
tip that facilitates greater tissue acquisition [7 ]. In a pilot study using a 22G Franseen needle, excellent results were shown for
histologic diagnosis[8 ].
The same FNB device has also been introduced in a 25G platform. To date, there have
been no studies comparing the performance of the 22G and 25G Franseen FNB needles.
Therefore, we conducted a prospective parallel-group randomized non-inferiority trial
comparing the histologic core procurement rate between the 22G and 25G Franseen needles
in patients undergoing EUS-guided biopsy for pancreatic and peripancreatic masses.
Methods
Patients
This was a prospective parallel-group randomized non-inferiority trial conducted at
the Asan Medical Center in Korea from November 2018 to May 2019 (Clinical Research
Information Service number: KCT0003834). Patients aged ≥ 18years with suspected pancreatic
or peripancreatic solid masses on computed tomography (CT) or magnetic resonance imaging
(MRI) were eligible for this study. Patients were excluded if their mass had a predominantly
cystic component, or if they had a coagulation disorder (international normalized
ratio > 1.5 or platelets < 50000 /mm3 ) or decompensated cardiopulmonary disease, or were pregnant. The Institutional Review
Board approved this study (S2018–1450–0002), and written informed consent was obtained
from all participants.
Randomization and blinding
Computer-generated randomization assignments were performed before enrollment to the
study in a 1:1 ratio for the two needle types (22G and 25G needles), using block randomization
methods. Subsequently, sequentially numbered needles were placed in opaque sealed
envelopes, and a research nurse confirmed the randomization number if patients met
the inclusion criteria.
Procedural technique
All EUS-guided biopsies were performed using a linear echoendoscope (GF-UCT180; Olympus
Medical, Tokyo, Japan) connected to a processor featuring a color Doppler function
(Pro-Sound Alpha 10; Hitachi Aloka Medical, Ltd., Tokyo, Japan). Three experienced
endosonographers (S.S.L., T.J.S., and D.W.O.) with a current experience of performing
≥ 500 EUS-guided interventions per year (including FNA or FNB cases) performed all
procedures with the patients under conscious sedation (using midazolam and pethidine)
and using a well-established technique [9 ]. The size of the needle was revealed to the endosonographers.
Initially, the mass was identified by EUS, and the area was scanned using color Doppler
to detect any intervening vessels. Subsequently, the endosonographer advanced the
assigned needle into the target lesion under ultrasound guidance. After the mass had
been punctured, the stylet was removed and suction was applied using a manufacturer-provided
10-mL syringe. The needle was moved to and fro within the target lesion at least 10
times in a fanning manner. The current guidelines recommend two to three needle passes
with an FNB needle if rapid on-site evaluation (ROSE) is unavailable [10 ]. Therefore, three needle passes were made. All masses located in the pancreatic
head and uncinate process were approached via the duodenum; masses located in the
body and tail of the pancreas were approached via the stomach.
Preparation and review of specimens obtained by FNA/FNB
After EUS-FNB, the obtained specimen was expressed onto a glass slide by re-introducing
a stylet into the needle assembly. Visible macroscopic cores were placed into formalin
bottles ([Fig.1 ]). A visible macroscopic core was defined as a whitish or yellow piece of tissue
with apparent bulk [11 ]. The specimens were fixed, embedded in paraffin, and cut in serial sections; hematoxylin
and eosin staining was performed for histologic evaluation. Smears were then made
with the remaining specimen by flushing air through the needle assembly onto the previous
slide; these were fixed immediately in 95 % ethyl alcohol for Papanicolaou staining.
A cytopathologist was not present during the EUS procedures.
Fig. 1 Photographs showing: a,b a visible histologic core expressed onto a glass slide; c a histologic core that has been placed into a formalin bottle for fixation before
pathologic assessment.
All biopsy specimens were evaluated by two experienced pathologists (J.K. and S.M.H.)
who were blinded to the needle gauge assignment. The pathologists defined specimens
that contained tissue cores as optimal where such specimens enabled the evaluation
of the histologic architecture of the target lesions. In contrast, if histologic evaluation
was feasible but the tissue core was missing or fragmented, the specimen was defined
as suboptimal. The pathologists assessed the specimen quality according to the scoring
system of Gerke et al. [12 ]. Briefly, the scoring system is as follows: 0, no material; 1–2, sample available
for cytologic diagnosis but not suitable for histology; 3–5, sample that enables histologic
assessment ([Fig.2 ]). In particular, a score of 4 or 5 indicates a sample that enables optimal histologic
interpretation: 4, sufficient material for adequate histologic interpretation but
a low quality sample (length of total material is less than one × 10 power field);
5, sufficient material for adequate histologic interpretation and a high quality sample
(length of total material is more than one × 10 power field) [6 ]
[13 ]
[14 ].
Fig. 2 Representative specimen images for each Gerke score. a Score 1 indicates scanty cells
are present, and the sample is inadequate for cytologic interpretation (hematoxylin
and eosin stain, × 100). b Score 2 indicates cells that are suitable for cytologic
interpretation but not adequate for histologic diagnosis. c Score 3 indicates that
some clusters of tissues are identified, and limited histologic assessment is feasible.
d Score 4 indicates sufficient material to enable the evaluation of tissue architecture;
the total area of the tissue on the slide is no longer than the × 10 power field.
e Score 5 indicates sufficient material to enable the evaluation of tissue architecture;
the total area of the tissue on the slide is longer than the × 10 power field.
If the diagnosis was challenging, additional immunohistochemical or special staining
was performed for differentiation. Cytologic smears were also evaluated using the
Gerke score. For cytologic evaluation, a score of 1 indicates limited cytologic interpretation,
and a score of 2 indicates adequate cytologic interpretation.
Outcome definitions
The primary outcome variable was the quality of the tissue core assessed by the pathologists
using the Gerke score; this score was used to determine whether the quality of the
specimen was optimal for histologic evaluation. The secondary outcome measures were
the sensitivity, specificity, and overall diagnostic accuracy for the diagnosis of
malignancy, the technical failure rate, and the adverse event rate.
The procedure time was measured for EUS-FNB as the time from echoendoscope insertion
to withdrawal of the echoendoscope after successful tissue acquisition. Malignancy
was defined as the definite presence of malignant cells, including adenocarcinoma
and other types of tumor cells, in the EUS-FNA/FNB or surgical specimen, presence
of metastatic lesions, or clinical deterioration during follow-up. Lesions were considered
benign if malignant cells were absent in surgical specimens or no clinical and radiologic
(CT and/or MRI) progression of disease was observed for at least 6months of follow-up
after the index procedure [15 ].
Technical failure was defined as follows: failure of needle retrieval through the
working channel, fracture of the needle during the procedure, and the need for additional
needles to complete the procedure [16 ]. Procedure-related adverse events were defined as immediate or delayed bleeding,
perforation, pancreatitis, or any other cardiopulmonary instability during or after
EUS-FNB, as observed by the operator [17 ].
Statistical analysis
The sample size was calculated to demonstrate the non-inferiority of the 25G Franseen
needle compared with the 22G needle in terms of the histologic core procurement rate.
The null hypothesis was that the difference between the histologic core procurement
rates of the two groups was 15 % or more (non-inferiority margin). The reported tissue
core procurement rate of the 22G FNB needle in the literature is approximately 90 %
[18 ]
[19 ]. A one-tailed sample size calculation was performed with a type I error rate (α)
of 0.025, to obtain 80 % power to show that the difference in tissue core procurement
rate is less than 15 %; the estimated sample size was 62 patients for each needle
group. With a 10 % dropout rate expected, the total recruitment was set at 70 patients
for each group. If the lower limit of the 95 % confidence interval (CI) of the difference
in the procurement rate of the histologic core between the 22G and 25G groups was
found to be < 15 %, the 25G needle would be considered non-inferior to the 22G needle.
Baseline characteristics of the patient population, pancreatic and peripancreatic
lesions, and procedural details were calculated. For the comparison of quantitative
variables, a two-sample t test or a Wilcoxon rank-sum test was performed, depending on distribution normality.
The χ2 or Fisher’s exact test was used, as indicated, to compare qualitative variables.
Statistical significance was determined as a P value of < 0.05.
All statistical analyses were performed using R software (R foundation for Statistical
Computing, Vienna, Austria; http://www.R-project.org, Ver 3.5.3). All authors had
access to the study data and reviewed and approved the final manuscript.
Results
We randomly assigned 140 patients with pancreatic and peripancreatic masses to undergo
EUS-FNB with a 22G (n = 70) or 25G (n = 70) Franseen needle from November 2018 to
May 2019. Without any dropouts, all enrolled patients constituted the study cohort.
Patients demographics and characteristics of their tumors
[Table 1 ] shows the demographic characteristics of the study cohort and the characteristics
of their mass lesions. Except for age, there were no significant differences between
the 22G and 25G groups in terms of sex ratio, location or size of mass, or final diagnosis.
In most cases, masses were located in the pancreas (92.9 % and 97.1 % in the 22G and
25G groups, respectively).
Table 1
Baseline characteristics of the 140 enrolled patients and their tumors.
Characteristic
Type of needle
P value
22 gauge
25 gauge
Sex, n (%)
0.87
40 (57.1 %)
39 (55.7 %)
30 (42.9 %)
31 (44.3 %)
Age, mean (SD), years
61.8 (9.27)
65.6 (9.15)
0.01
Organ, n (%)
0.56
65 (92.9 %)
68 (97.1 %)
3 (4.3 %)
1 (1.4 %)
1 (1.4 %)
0 (0.0 %)
1 (1.4 %)
1 (1.4 %)
Pancreatic tumor location, n (%)
0.94
34 (52.3 %)
36 (52.9 %)
31 (47.7 %)
32 (47.1 %)
Size of mass on EUS, mm
30 (23.3–35)
30 (23.0–40.8)
0.49
Final diagnosis, n (%)
0.82
59 (84.3 %)
58 (82.9 %)
11 (15.7 %)
12 (17.1 %)
SD, standard deviation; EUS, endoscopic ultrasound; IQR, interquartile range
In the 22G group, of 70 mass lesions, 59 were pancreatic adenocarcinomas (84.3 %),
three were pancreatic neuroendocrine tumors (4.3 %), and two were metastatic adenocarcinomas
(2.9 %). The remaining six patients were one each (1.4 %) of the following: pancreatic
acinar cell carcinoma, solid pseudopapillary neoplasm, duodenal gastrointestinal stromal
tumor, mass-forming pancreatitis, retroperitoneal schwannoma, and diffuse large B
cell lymphoma. In the 25G group, of 70 mass lesions, 58 were pancreatic adenocarcinomas
(82.9 %), three were pancreatic neuroendocrine tumors (4.3 %), and two each were metastatic
adenocarcinomas and autoimmune pancreatitis (2.9 % each). The remaining five were
one each (1.4 %) of the following: intraductal papillary mucinous neoplasm, metastatic
adenocarcinoma, pseudocyst, solid pseudopapillary tumor, and retroperitoneal schwannoma.
Histology assessment
[Table 2 ] summarizes the histologic outcomes of the enrolled patients. Optimal histologic
cores were obtained from 61 patients (87.1 %) in the 25G group and 68 patients (97.1 %)
in the 22G group, with a rate difference of −10 % (95 %CI−17.35 % to −2.65 %); non-inferiority
test, P = 0.13) ( [Fig.3 ]). Our results demonstrated that the 25G needle was inferior to the 22G needle because
the lower limit of the confidence interval for the histologic procurement rate difference
was below the margin. In addition, according to the two-sided χ2 test, the histologic procurement rate significantly differed between the two groups
(P = 0.03).
Table 2
Comparison of histologic outcomes between the 22G and 25G Franseen needles.
Type of needle
P value
22 gauge
25 gauge
Core procurement, n (%)
0.03
Optimal
68 (97.1 %)
61 (87.1 %)
Suboptimal
2 (2.9 %)
9 (12.9 %)
Specimen quality, n (%)
Cytologic smear (1–2), n (%)
0.60
1 (limited cytologic interpretation)
9 (12.9 %)
7 (10.0 %)
2 (adequate cytologic interpretation)
61 (87.1 %)
63 (90.0 %)
Histology (3–5)
< 0.001
3 (limited histologic interpretation)
2 (2.9 %)
9 (12.9 %)
4 (adequate histologic interpretation, low quality specimen [total material length
less than one × 10 power field])
19 (27.1 %)
41(58.6 %)
5 (adequate histologic interpretation, high quality specimen [total material length
more than one × 10 power field])
49 (70.0 %)
20 (28.6 %)
Fig. 3 Non-inferiority analysis demonstrating the histologic core procurement rate of a
25-gauge Franseen needle compared with that of a 22-gauge needle.
For FNB specimens, high quality specimens with a Gerke score of 5 were more frequently
obtained in the 22G group than in the 25G group (70.0 % [49 /70] vs. 28.6 % [20 /70],
respectively; P < 0.001). In contrast, no difference was observed between the 22G and 25G groups
in the cytologic quality of cytologic smears (87.1 % [61/70] vs. 90.0 % [63/70], respectively;
P = 0.60).
Procedure-related outcomes
Procedure-related outcomes are presented in [Table 3 ]. Tissue acquisition was successful in both groups. In two patients who underwent
EUS-FNB with a 25G needle, EUS-FNB was non-diagnostic because of an insufficient sample;
pancreatic adenocarcinoma was confirmed in these two patients by surgical resection.
Table 3
Comparison of procedural outcomes between the 22G and 25G Franseen needles.
Characteristics
Type of needle
P value
22 gauge
25 gauge
Sensitivity (95 %CI)
100 % (94.3 %–100 %)
98.4 % (91.5 %–99.9 %)
Specificity (95 %CI)
100 % (59.0 %–100 %)
85.7 % (42.1 %–99.6 %)
Overall accuracy (95 %CI)
100 % (94.9 %–100 %)
97.4 % (90.1 %–99.7 %)
Technical failure, n
0
0
Adverse events, n (%)
1 (1.4 %)
0
0.32
1
0
Approach route, n (%)
> 0.99
35 (50.0 %)
36 (51.4 %)
35 (50.0 %)
33 (47.1 %)
Procedure time, mean (SD), minutes
16.5 (5.86)
18.2 (5.89)
0.09
CI, confidence interval; SD, standard deviation.
No significant differences were noted between the 22G and 25G groups in sensitivity
(100 % [95 %CI94.3 %–100 %] vs. 98.4 % [95 %CI91.5 %–99.9 %], respectively), specificity
(100 % [95 %CI59.0 %–100 %] vs. 85.7 % [95 %CI42.1 %–99.6 %], respectively), or overall
accuracy (100 % [95 %CI94.9 %–100 %] vs. 97.4 % [95 %CI90.1 %–99.7 %], respectively)
in differentiating malignancies. When confined to cytologic smears, the sensitivity
to detect malignancies did not statistically differ between the two groups (96.8 %
vs. 95.1 %, respectively; P = 0.98).
Adverse events
The incidence of adverse events was similar in both groups (1.4 % [1/70] with the
22G needle vs. 0 % [0/70] with the 25G; P = 0.32). One patient in the 22G group who experienced mild acute pancreatitis recovered
completely with conservative treatment within 2days.
Discussion
In this prospective parallel-group randomized non-inferiority trial, the 25G Franseen
needle was inferior to the 22G needle with respect to the quality of the histologic
core. Although there were no statistical differences in the technical and diagnostic
success rates, quality of the cytologic smears, and diagnostic accuracy rate between
the two needles, the 22G needle was superior to the 25G needle in the ability to procure
high quality FNB specimens.
With the widespread use of EUS-FNA, it is becoming an essential modality for the diagnosis
of pancreatic diseases. In recent years, EUS-FNB needles have garnered increased attention
in the field of EUS. They are useful for obtaining core biopsy specimens, which is
important for histologic diagnosis[20 ].
Currently, few studies have reported the clinical outcomes of the novel Franseen needles.
Franseen needles are three-plane symmetric needles; their geometry enables tissue
puncturing with a reduced penetration force and allows for deeper insertion to obtain
a greater amount of specimen [8 ]. Bang et al. first reported the efficacy of the 22G Franseen needle in 30 patients,
with a histologic core present in 29 of 30 patients (96.7 %) and only one technical
failure in a patient who underwent transduodenal sampling, as a result of stylet dysfunction
[8 ]. In a recent randomized trial comparing a 22G Franseen needle and a 22G standard
FNA needle, the Franseen needle demonstrated a higher histologic core procurement
rate than the standard FNA needle [21 ]. In a more recent prospective study by Sugiura et al., the 25G Franseen needle showed
an adequate specimen acquisition rate of 82.0 % [22 ]. In our study, the histologic core procurement rates of the 22G and 25G needles
were 97.1 % (68 /70) and 87.1 %, (61/70), respectively. These results were similar
to the results of the aforementioned studies.
Theoretically, larger needles allow the collection of larger samples, but they may
increase the rate of adverse events. Moreover, they may cause some technical problems,
mostly owing to the higher stiffness of the device, the likelihood of bloody contamination,
or the presence of cellular debris in the sample [23 ]. The theoretical advantages of the 25G FNB needle are its high flexibility because
of its small caliber (making it easy to manipulate when the duodenal scope is angulated),
fewer blood contaminations, and easier puncture of calcified masses [24 ]. Two studies have compared the performance of 22G and 25G FNB needles with a reverse
bevel design; there were no differences in the diagnostic accuracy and histologic
core procurement [25 ]
[26 ].
In our series, the Gerke scoring system was used to evaluate the quality of the histologic
core. Recently, several reports have advocated the use of a software imaging program
for histologic evaluation [8 ]
[27 ]. The advantage of using a software program is that it can be more objective; however,
it can also be more complicated to apply in clinical practice. Although histologic
interpretation depends on the subjective opinion of pathologists using the scoring
system, in clinical practice, it has the advantage of not requiring a software imaging
program.
To the best of our knowledge, this is the first prospective parallel-group randomized
non-inferiority trial to compare the histologic core procurement rates of 22G and
25G Franseen needles. In this study, our results showed that the 25G needle was inferior
to the 22G needle.
There were no statistical differences in the quality of the cytologic smear between
the 22G needle and the 25G needle (87.1 % vs. 90.0 %, respectively; P = 0.60), sensitivity (100 % [95 %CI94.3 %–100 %] vs. 98.4 % [95 %CI91.5 %–99.9 %],
respectively), specificity (100 % [95 %CI59.0 %–100 %] vs. 85.7 % [95 %CI42.1 %–99.6 %],
respectively), and overall accuracy (100 % [95 %CI94.9 %–100 %] vs. 97.4 % [95 %CI90.1 %–99.7 %],
respectively). The 25G group showed a relatively low specificity compared with the
22G group. There is a possibility that the lower quality of the histologic core of
the 25G needle may play a part in these differences; however, with no statistical
difference between the two groups in terms of specificity, a larger scale comparative
study is required.
The histologic procurement rate (Gerke scores 4 and 5) was statistically higher in
the 22G group than in the 25G group (97.1 % vs. 87.1 %, respectively; P = 0.03). In addition, the rate of FNB samples of high quality (Gerke score 5) was
significantly higher in the 22G group than in the 25G group (70 % vs. 28.6 %, respectively;
P < 0.001). The diagnostic performance of our results was similar to that of other
studies showing a diagnostic accuracy of > 90 % [7 ]
[8 ]
[28 ]
[29 ].
These results suggested that both needle sizes would be suitable for diagnosing malignant
pancreatic neoplasms. However, in the era of personalized medicines, a 22G needle
should be considered preferentially when the suspected disease is lymphoma or autoimmune
pancreatitis for example, where tissue architecture or ancillary staining are essential
for accurate pathologic assessment, or when molecular profiling is warranted in anticancer
treatment.
As the Franseen needle provides an acceptable diagnostic yield within three needle
passes, these results are important for units where ROSE is not always feasible. ROSE
was not used in any of our cases. ROSE during EUS-FNA is not always available in all
institutes. According to a recent survey, ROSE was available for 48 % of responders
from Europe and 55 % of responders from Asia [30 ]. It is not well established whether ROSE is necessary for EUS-FNB, although it may
theoretically enhance sampling efficiency. In a systematic review, EUS-FNB without
ROSE was comparable to EUS-FNB with ROSE in diagnostic adequacy and accuracy [31 ]. In this study, the diagnostic accuracy in the entire cohort was 98.5 % (95 %CI94.4 %–99.7 %).
This study had several limitations. First, when evaluating tissue quality, a software
imaging program was not used for the quantification of tissue. However, two experienced
pathologists validated the quality of tissues qualitatively using the scoring system.
Second, the desmoplastic stroma of specimens was not evaluated. Desmoplasia is a cellular
reaction to a neoplastic process; recently, it has attracted attention for its role
in drug resistance [32 ]. Assessing desmoplastic stroma may be useful to predict the treatment response and
prognosis. In addition, considering the relative rarity of pancreatic tumors and varying
levels of proficiency in performing EUS-FNB, the non-inferiority margin was 15 % in
this study. There is a possibility that the results may change depending on the non-inferiority
margin and the sample size. However, even with a relatively large non-inferiority
margin of 15 %, our results failed to show that the 25G Franseen needle is not inferior
to the 22G Franseen needle.
In conclusion, the 25G Franseen needle failed to demonstrate similar efficacy in histologic
core procurement compared with the 22G Franseen needle. Therefore, the 22G needle
may be a better device to diagnose diseases that require a large volume specimen or
those that require molecular profiling. Without ROSE, Franseen needles may provide
an acceptable diagnostic accuracy.
