Keywords endoscopic ultrasound - fine-needle aspiration (FNA) - fine-needle biopsy (FNB) -
adenocarcinoma - pancreatic cancer
Introduction
Pancreatic cancer remains one of the most aggressive malignancies globally, characterized
by poor prognosis and limited treatment options. In India, the reported incidence
ranges from 0.5 to 2.4 per 100,000 women and 0.2 to 1.8 per 100,000 men, although
these figures appear to be rising with the increasing westernization of lifestyles.[1 ] Epidemiological data from tertiary care centers in northern India indicate that
pancreatic cancer is a significant contributor to malignant extrahepatic biliary obstruction,
comprising up to 18% of cases in one study and representing the predominant malignant
etiology of obstructive jaundice (33.6%) in another.[2 ]
[3 ] Given the high mortality associated with pancreatic cancer, timely and precise diagnostic
modalities are essential for improving clinical outcomes.
Endoscopic ultrasound (EUS) has emerged as a pivotal tool for both imaging pancreatic
lesions and procuring tissue samples. It demonstrates superior sensitivity for detecting
small pancreatic tumors and assessing local vascular involvement compared to computed
tomography (CT) and magnetic resonance imaging (MRI), thereby facilitating earlier
disease detection.[4 ] Additionally, EUS offers considerable accuracy in identifying local lymph node involvement.[5 ] Despite the notable specificity of EUS-guided fine needle aspiration (EUS-FNA),
which can approach 100%, its sensitivity varies widely (60–93%) due to factors such
as needle caliber, the availability of rapid on-site cytopathological evaluation (ROSE),
and procedural techniques. Moreover, in patients with chronic pancreatitis, the diagnostic
utility of EUS-FNA may be compromised by an increased risk of false-negative findings.[6 ]
In response to these challenges, advanced EUS techniques have been explored to enhance
diagnostic accuracy, including contrast-enhanced harmonic EUS and real-time elastography
EUS. Both modalities, used alone or together, achieve a high negative predictive value
(NPV), effectively enhancing diagnostic rates in inconclusive EUS-FNA cases. Recently,
EUS-guided fine needle biopsy (EUS-FNB) has been able to obtain more robust histological
cores, potentially reducing reliance on on-site cytopathology and decreasing false-negative
results.[7 ] The overall diagnostic accuracy of EUS-FNA/EUE-FNB for solid pancreatic lesions
is reported to range from 78 to 95%.[7 ] However, even EUS-guided tissue sampling may be less sensitive and specific in patients
with chronic pancreatitis, with sensitivities reported between 50 and 73.9% and specificities
from 73.7 to 100%, depending on the study.[8 ]
[9 ]
While EUS-FNB may offer marginally higher diagnostic yields, it is typically more
expensive than FNA, and the needles are not reusable—constraints that can be prohibitive
in resource-limited health care settings. Consequently, EUS-FNA remains the primary
investigation modality in many developing countries, especially for patients with
a high clinical suspicion of pancreatic malignancy, owing to its relative affordability
and widespread availability.
Against this backdrop, the present study aimed to compare the diagnostic accuracy
of EUS-FNA and EUS-FNB in patients with solid pancreatic masses. Findings from this
investigation may help refine diagnostic strategies and optimize resource allocation
in low- and middle-income settings.
Materials and Methods
This was a single-center, prospective, randomized, double-blinded study (patient and
statistician were blinded) conducted in a tertiary care teaching institute in eastern
India.
Patients
Patients with suspected solid pancreatic lesions, aged between 18 and 80 years, presenting
to our department from July 2022 to March 2024, fulfilling the inclusion criteria,
were recruited for the study (CONSORT diagram; [Fig. 1 ]). Diagnosis of solid pancreatic mass was made based on either abdominal ultrasound,
abdominal CT scan, or MRI findings. Ethical clearance was obtained from the Institutional
Ethics Committee, AIIMS Bhubaneswar (Reference number: IEC/AIIMS BBSR/PG Thesis/2022-23/42,
dated June 13, 2022). The trial was registered with CTRI (CTRI/2022/11/047481). Written
informed consent was obtained from all participants in their local language.
Fig. 1 CONSORT diagram.
Patients either aged under 18 years or more than 80 years, had lesions <1 cm in size,
had a history of previous gastrectomy or any other gastric surgery, were unable or
unwilling to provide informed consent, had uncorrected coagulopathy, had cystic pancreatic
lesions, and were critically ill were excluded from the study. All procedures were
performed by a single experienced endosonographer (H.K.N.) who had performed more
than 500 EUS-FNA before the initiation of this study.
Sample Size Calculation
This trial was prospectively designed as a pilot randomized controlled trial (RCT)
to assess feasibility, event rates, and variability for a larger definitive study.
Following Browne's guidance for pilot studies, we aimed for about 20 participants
per arm (total around 40). No formal hypothesis test was powered in advance; instead,
the focus was on the precision (95% confidence intervals) around accuracy estimates
to guide planning for the next trial. Effect sizes and variability observed here will
help inform a comprehensive, adequately powered multicenter RCT.
Randomization and Blinding
Participants were randomly assigned 1:1 using computer-generated permuted blocks of
varying sizes [4 and 6], prepared by an independent coordinator. Allocation was concealed
in sequentially numbered, opaque, sealed envelopes, which were opened in the endoscopy
suite immediately before needle selection. Patients and the study statistician remained
blinded to allocation. The cytopathologist/histopathologist was not blinded due to
operational constraints and the potential recognition of core architecture on H&E;
this is acknowledged as a study limitation.
Technical Procedure
EUS was performed using a linear array echo-endoscope (GF UCT 180; Olympus Ltd., Tokyo,
Japan) connected to an ultrasound unit processor (EUS ME2 Premium plus) by a single
endoscopist (H.K.N.). Patients were placed in the left lateral decubitus position
under conscious sedation using midazolam and pentazocine injections. After reaching
the stomach station, the EUS probe was placed in contact with the gastric wall and
the abdominal aorta was identified in an elongated cross-section. This finding was
confirmed by color Doppler imaging. The scope was then slowly advanced up to the duodenum
when required. EUS evaluation was done from the gastroesophageal junction, duodenal
bulb, and second part of the duodenum to screen the entire pancreas and characterize
pancreatic lesions. The size, shape, location, echotexture, and vascular encasement
of the pancreatic lesion were documented.
A transduodenal approach was used for lesions in the pancreatic head and uncinate
process, while a transgastric approach was used for lesions in the body and tail.
FNA used a 22-gauge conventional bevel needle; FNB used a 22-gauge multi-blade (Franseen-type)
tip. For both arms, we used the same trajectory and fanning technique to limit procedural
bias. After puncturing into the deepest part of the lesion, the stylet was withdrawn,
an inflation system employing a 60-mL syringe with 50-mL high negative pressure was
attached to the proximal end of the needle, and then tissue acquisition was done by
the fanning technique with the standard method of 5–10 to-and-fro movements, by application
of negative suction.
Crossover/rescue: if the initial allocation yielded paucicellular or nondiagnostic
material after the planned passes, a single rescue pass with the alternate needle
was allowed at the endoscopist's discretion (documented as crossover). Reuse/sterilization:
needles were single-use only; no re-use was performed.
Smears (air-dried/alcohol-fixed) and formalin-fixed material were prepared from each
pass; cell blocks were routinely made. All slides/blocks were processed per institutional
protocol and reported by the pathologist. Cellularity was graded as follows: paucicellular
(<10 diagnostic clusters or <100 lesional cells), low (100–500), moderate (500–2,000),
and high (>2,000). Contamination (blood/gastrointestinal epithelium) was semi-quantitatively
graded as minimal (<10% slide area), moderate (10–50%), or heavy (>50%). ROSE was
not available; Macroscopic on-site evaluation (MOSE) was limited to a macroscopic
core assessment, which was not done routinely in our study. In case of scant material
aspirated after the first pass, a second pass for tissue acquisition was performed.
Patients were continuously monitored during the procedure with an electrocardiogram
tracing and pulse oximetry. All patients were followed for 6 months or till surgery
was done, and the surgical specimen biopsy was then correlated with FNA/FNB results.
Lesions diagnosed as positive for malignancy with either FNA or FNB and confirmed
as positive either by surgical resection or the clinical course of the patient were
considered true positives. Lesions diagnosed as benign on EUS-FNA/B and finally confirmed
as benign based on the clinical course of the patient were considered true negatives.
Aspirates initially diagnosed as benign by EUS-FNA/B but finally diagnosed as malignant
either by surgical resection or clinical course of the patient were classified as
false negatives. For analysis, inconclusive samples were considered false negatives
when comparing diagnostic accuracy.
Primary outcome: diagnostic accuracy for malignancy of index EUS-guided tissue acquisition
(FNA vs. FNB) was assessed, adjusted against surgical histology or ≥6-month clinical
course.
Secondary outcomes: sensitivity, specificity, positive predictive value (PPV)/NPV;
technical success; sample cellularity, adverse events; and procedure total cost (needle + pathology
processing) were assessed.
Statistical Analysis
Categorical variables were expressed as frequencies and percentages, while continuous
variables were reported as means. The chi-square and Fisher's exact tests were used
for categorical variables, and the Student's t -test was employed to compare means between study groups. Continuous variables used
the t -test or Mann–Whitney U-test as appropriate for comparison.
Sensitivity, specificity, diagnostic accuracy, and PPVs and NPVs were calculated.
Sensitivity was defined as the proportion of true positives correctly identified by
the test among cases of malignancy in the study population. Specificity was defined
as the proportion of true negatives correctly identified among patients without tumors.
The positive likelihood ratio was calculated as sensitivity divided by the false-positive
rate, while the negative likelihood ratio was calculated as the false-negative rate
divided by specificity. Statistical analysis was performed using SPSS version 23 (IBM,
Armonk, New York, United States). Group comparisons of proportions (accuracy, sensitivity,
specificity, PPV/NPV) used Fisher's exact test with two-sided α = 0.05; effect sizes
are shown as risk difference with 95% CI.
Given the pilot study with a small sample size, we have not done any exploratory logistic
regression for correct diagnosis (yes/no), including age, lesion site (head vs. body/tail),
size on EUS, bilirubin, CA 1 9-9, and arm (FNA vs. FNB).
The prevalence of solid pancreatic lesions in the Indian subcontinent is not well
established. As this was a prospective, randomized pilot study, the sample size was
set at 40 subjects, with 20 participants in each group, based on the rules of thumb
for pilot studies.[10 ]
Results
We assessed 48 patients with pancreatic head, body, or tail masses for eligibility.
Seven patients were excluded, and 41 patients were randomized into the EUS-FNA and
EUS-FNB groups using computer-generated random number sequences. Twenty patients were
allocated to the FNA group, while 21 patients were assigned to the FNB group. The
patient enrolment and study design are shown in the CONSORT diagram ([Fig. 1 ]).
The baseline characteristics are summarized in [Table 1 ]–[2 ]. The mean age was similar in both groups, with no significant difference. The most
common presenting complaint was abdominal pain in both groups. The head of the pancreas
was the most common site of the pancreatic mass, occurring in 85% of the FNA group
and 61.9% of the FNB group.
Table 1
Baseline characteristics of both groups
EUS-FNA (n = 20)
EUS-FNB (n = 21)
p -Value
Age (mean ± SD)
53.5 ± 11.1
52.9 ± 14.8
0.88
Gender
Male
Female
15 (75%)
5 (25%)
18 (85.7)
3 (14.2%)
Pain abdomen
20 (100%)
19 (90.4%)
0.49
Jaundice
14 (70%)
10 (47.6%)
0.21
Fever
5 (25%)
5 (23.80%)
1.00
Pruritus
9 (45%)
7 (33.3%)
0.530
Weight loss
18 (90%)
10 (47.61%)
0.006
Anorexia
17 (85%)
10 (47.6%)
0.02
Location of lesion
Head
Body–tail
17 (85%)
3 (15%)
13 (61.9%)
8 (38%)
Abbreviations: EUS-FNA, Endoscopic ultrasound-guided fine-needle aspiration; EUS-FNB,
Endoscopic ultrasound-guided fine-needle biopsy; SD, standard deviation.
Table 2
Baseline laboratory and imaging characteristics in both groups
EUS-FNA
EUS-FNB
p- Value
Hemoglobin (gm/dL)
10.8 ± 1.6
11.8 ± 1.9
0.07
Total leucocyte count
9,230 ± 5,891
11,889 ± 5,667
0.14
Platelet count
2.3 ± 1.0
2.8 ± 0.8
0.12
Total bilirubin
8.6 ± 7.6
7.7 ± 6.2
0.33
Direct bilirubin
4.3 ± 3.7
3.8 ± 3.1
0.49
INR
1.3 ± 0.40
1.2 ± 0.2
0.41
Albumin
3.3 ± 0.7
3.7 ± 0.7
0.08
CA 19-9
582 ± 368
429 ± 258
0.54
Head
17 (85%)
13 (62%)
0.16
Body/tail
3 (15%)
8 (38%)
0.09
CCP on imaging
5 (25%)
4 (19%)
0.71
Resectable
5 (25%)
5 (24%)
1.000
Size on CT (mm)
41 ± 15 × 34 ± 12
44 ± 14 × 38 ± 13
0.61
Size on EUS (mm)
32 ± 10 × 29 ± 10
35 ± 10 × 28 ± 8
0.44
Abbreviations: CCP, convoluted cerebriform pattern; CT, computed tomography; EUS-FNA,
Endoscopic ultrasound-guided fine-needle aspiration; EUS-FNB, Endoscopic ultrasound-guided
fine-needle biopsy; INR, international normalized ratio.
In the EUS-FNA group, 11 patients (55%) were diagnosed with adenocarcinoma based on
cytology, while 7 patients (35%) were diagnosed with an inflammatory mass. In the
EUS-FNB group, adenocarcinoma was diagnosed in 10 patients (47.6%) on biopsy, while
8 patients (38.1%) had an inflammatory mass (summarized in [Table 3 ]).
Table 3
Histopathological diagnosis in both groups
Diagnosis
EUS-FNA (n = 20)
EUS-FNB (n = 21)
Adenocarcinoma
11 (55%)
10 (47.6%)
Squamous carcinoma
1 (5%)
0
Inflammatory mass
7 (35%)
8 (38.09%)
Neuroendocrine tumor
1 (5%)
1 (4.76%)
Undifferentiated adenocarcinoma
0
1 (4.76%)
SPEN
0
1 (4.76%)
Abbreviations: EUS-FNA, Endoscopic ultrasound-guided fine-needle aspiration; EUS-FNB,
Endoscopic ultrasound-guided fine-needle biopsy; SPEN, solid pseudopapillary neoplasm.
Two FNA-allocated cases underwent single-rescue FNB due to initial nondiagnostic/discordant
results; one converted to adenocarcinoma, and one remained inflammatory. Analyses
were performed by intention-to-treat, counting inconclusive results as false negatives
for accuracy.
A total of five patients underwent surgery (Whipple procedure) in the EUS-FNA group,
in which adenocarcinoma was the final diagnosis in four patients, while in one patient,
an inflammatory mass was seen with no evidence of malignancy.
Similarly, five patients in the FNB group underwent surgery. Four cases had adenocarcinoma
as the histological diagnosis, and one case was initially diagnosed with an inflammatory
mass. In all five cases, the final histological diagnosis after surgery was adenocarcinoma.
The diagnostic accuracy was 90% in the EUS-FNA group and 95.2% in the EUS-FNB group
([Table 4 ]). In both groups, most of the patients had moderately cellular samples ([Table 5 ]).
Table 4
Diagnostic accuracy in both groups
EUS-FNA (n = 20)
EUS-FNB (n = 21)
Positive for malignancy
13 (65%)
13 (61.9%)
Negative for malignancy
7 (35%)
8 (38.1%%)
Diagnostic accuracy (%)
18/20 (90%)
20/21 (95.2%)
Sensitivity (%)
12/13 (92.3%)
13/14 (92.8%)
Specificity (%)
6/7 (85.7%)
7/7 (100%)
Positive predictive value (%)
12/13 (92.3%)
13/13 (100%)
Negative predictive value (%)
6/7 (85.7%)
7/8 (87.5%)
Technical success (%)
100%
100%
Average cellularity (moderate to high)
60%
71%
Total cost (needle + pathology processing cost per patient (INR)
22,200
27,400
Abbreviations: EUS-FNA, Endoscopic ultrasound-guided fine-needle aspiration; EUS-FNB,
Endoscopic ultrasound-guided fine-needle biopsy.
Note: Values are n /N (%). Group comparisons used Fisher's exact test. CI = 95% confidence interval. No
between-group differences were statistically significant.
Table 5
Cellularity in both groups
Cellularity
EUS-FNA (n = 20)
EUS-FNB (n = 21)
Paucicellular (scant)
1 (5%)
0
Cellular (low)
7 (35%)
6 (28.6%)
Moderately cellular (moderate)
9 (45%)
9 (42.8%)
Highly cellular (high)
3 (15%)
6 (28.6%)
Abbreviations: EUS-FNA, Endoscopic ultrasound-guided fine-needle aspiration; EUS-FNB,
Endoscopic ultrasound-guided fine-needle biopsy.
Between-group differences in diagnostic accuracy (FNA 90.0% vs. FNB 95.2%) and sensitivity
(92.3% vs. 92.9%) were not statistically significant (Fisher's exact, p -values >0.05).
Costs were calculated from the provider perspective and included needle + pathology
processing (smear, cell-block, histology). The estimated cost per procedure was INR
22,200 for FNA and INR 27,400 for FNB. FNB remained costlier by approximately INR
5,200.
No needle malfunction or mechanical failure was seen, and technical success was 100%.
No severe adverse events were registered. Abdominal pain was seen in two patients
(10%) in the EUS-FNA group and three (15%) in the EUS-FNB group. No procedural bleeding
was seen in either group. All of the complications were managed conservatively.
Discussion
The diagnostic accuracy of EUS-guided tissue acquisition, whether using FNA or FNB
needles, is affected by several factors. These include the sampling technique, the
type and size of the needle, the availability of rapid on-site evaluation (ROSE),
and the expertise of both the endosonographer and the cytopathologist or histopathologist.[11 ] There remains a lack of consensus regarding the optimal needle type, emphasizing
the need for further studies to clarify how FNA and FNB needles influence diagnostic
accuracy and tissue quality.
In this study, diagnostic accuracy was slightly higher in the FNB group (95.2%) compared
to the FNA group (90%). Sensitivity was similar in both groups, but specificity was
greater with FNB (100% vs. 85.7%). In the absence of ROSE, a common scenario in low-resource
settings, achieving high diagnostic accuracy becomes more challenging. EUS-FNA achieved
90% diagnostic accuracy and sensitivity, reinforcing its utility as a frontline tool
when performed by experienced endosonographers. Despite EUS-FNB providing marginally
better diagnostic accuracy and specificity, EUS-FNA remains a robust and cost-efficient
alternative, particularly in settings lacking ROSE.
Various previously published studies have evaluated the diagnostic accuracy of EUS-FNA
compared to EUS-FNB.[12 ]
[13 ] Notably, Wong et al conducted a retrospective study involving 151 patients, which
showed a higher diagnostic accuracy for EUS-FNB (94.6%) compared to EUS-FNA (89.6%),
although the difference was not statistically significant.[14 ]
Similarly, a study by Yousri et al and Kuraoka et al observed a slight diagnostic
advantage of FNB over FNA, but the difference was not statistically significant.[15 ]
[16 ] Similarly, another multicenter prospective study with a relatively adequate sample
size observed no differences in diagnostic accuracy when smears alone were used.[17 ]
Cellularity is another critical factor influencing diagnostic success. Both FNA and
FNB needles provided sufficient histological cores for diagnosing solid pancreatic
lesions. In our study, both arms yielded predominantly moderate to high cellularity
samples: 60% in the FNA group and 71% in the FNB group. Although FNB showed a slight
advantage, the difference was not statistically significant. This suggests that when
the optimal technique is applied, FNA can yield diagnostically sufficient material
in the majority of cases, even without typical histological core or ROSE support.
From a cost-effectiveness standpoint, the contrast was more pronounced. The needle
cost per procedure for FNB was more than that of FNA (INR 27,400 vs. INR 22,200),
resulting in an additional expense for the FNB arm. This cost differential must be
carefully weighed, especially in public or resource-limited settings.
Although FNB remains valuable, especially for histologic subtyping or challenging-to-diagnose
lesions, our results show that EUS-FNA offers nearly comparable diagnostic performance
at a lower cost. These findings are especially important for countries with limited
access to advanced cytopathology support or funding for repeat procedures.
The strengths of our study include its prospective, randomized design and focus on
real-world limitations, such as cost and lack of ROSE. Limitations include the pilot
sample size, single-center design, absence of ROSE, lack of pathologist blinding,
procedures performed by a single operator, and the exploratory nature of regression
analyses. Cost analysis was limited to direct provider costs and did not account for
downstream costs (e.g., repeat procedures). Findings should be validated in adequately
powered multicenter studies.
In conclusion, EUS-FNB offers a marginally better diagnostic yield but comes at a
higher cost, making EUS-FNA a cost-effective and reliable first-line option in resource-constrained
settings. Larger multicenter trials are warranted to establish clearer guidelines
for optimal needle selection and sampling strategies, with careful consideration of
cost–benefit balance in resource-limited settings.
