Study design and ethical consideration
This was a single-center, prospective exploratory study comparing the tissue acquisition rate of EUS-EG-FNB between hard areas and soft areas in pancreatic mass lesions. This study was conducted from November 2018 to April 2021 at Nagoya University Hospital. This study was approved by the Institutional Review Board in Nagoya University Hospital (IRB #2018-0217), registered with the University Hospital Medical Information Network (UMIN-CTR No. UMIN-000033073 Registered date20/06/2018) and conducted according to the provisions of the Declarations of Helsinki.
Patients undergoing EUS-FNB for pancreatic solid mass lesions were candidates for this study. We obtained written informed consent from patients aged 20 years or older who required pathological diagnosis by EUS-FNB and participated in the study before EUS procedures. The exclusion criteria were 1) patients whose pancreatic mass could not be visualized by EUS, 2) patients with pancreatic cystic tumors, 3) patients younger than 20 years old, 4) patients at high risk of bleeding, 5) pregnant patients and 6) patients who refused to participate in the study.
All procedures were performed using a linear-array echoendoscope by experienced endoscopists who were well trained in pancreaticobiliary EUS. Prior to performing EUS-FNB, detailed observation of the pancreatic mass lesion site by EUS was performed. In EUS-EG, the ROI was set to include the entire tumor at the position where the pancreatic mass lesion could be visualized on the maximum diameter on EUS images. We used the strain elastography(SE) method with strain histogram(SH) analysis to measure and quantify the tissue elasticity. EUS-EG images were recorded as moving images, and high-quality still images in which the signal was drawn over the entire ROI were recorded. After evaluating the lesions by EUS-EG, EUS-FNB was performed. We used a GF-UCT260 linear-array echoendoscope (Olympus Co., Tokyo, Japan) and an ARIETTA 850 (Fujifilm healthcare corp, Tokyo, Japan) and EU-ME2 premier plus (Olympus Co., Tokyo, Japan) as ultrasound processors in this study.
A 22-gauge Franseen needle (AcquireTM, BostonScientific Japan) was used to perform EUS-FNB in all patients. At the time of EUS-FNB, the puncture was performed with the stylet slightly removed, and the tissue of the target area was collected by pushing out the stylet before collecting tissue to avoid tissue contamination. After the needle was inserted into the pancreatic mass lesion, the needle was moved back and forth 15 times for sample acquisition using the slow-pull method. The collected specimens were immediately placed in formalin solution to prevent drying of the tissue and sent for histology. The sample was not split for cytology or cell block.
For patients in which the tissue elasticity showed heterogeneity in the mass lesion by EUS-EG (heterogeneous group), one sample each was obtained from the high elasticity region (hard area: presented in blue) and the low elasticity region (soft area: presented in red to green) in the mass lesion. For patients in which the tissue elasticity in the mass lesion was uniform (homogeneous group), a sample was collected from the center of the tumor. The evaluation of heterogeneity based on EUS-EG image findings was diagnosed when the reproducibility was confirmed by at least a 10-second movie and 3 or more still images. For the puncture order for the hard area/soft area, sampling was performed in the region closer to the EUS probe first. When a sufficient amount of tissue was not obtained from the first two passes, additional punctures were performed until a sufficient amount of tissue was obtained based on macroscopic on-site evaluation (MOSE). Specimens that underwent additional puncture were excluded from the evaluation of this study. For the homogeneous group, 2 or 3 samples were obtained from the lesion, and the best sample was used for histological evaluation and comparison with the heterogeneous group.(Figure 1)
Analysis of EUS-elastography images
SH analysis was performed on the EUS-EG images to quantify the tissue elasticity of pancreatic mass lesions. The method used for SH analysis was performed as previously reported (15) . The mean strain value (MSV) was measured for the "hard area" and "soft area" within the ROI and the "whole lesion" in which the entire tumor was set as the ROI in the maximum cross section of the pancreatic mass. A lower mean value indicates greater elasticity, and it has been reported to be a useful method for analysis with the strain method. We extracted three representative EUS-EG images at the puncture sites. The median of three MSVs of each patient was defined as the representative module of the target patient. Histogram analysis was performed using Elasto_ver. 15.1 (Hitachi-Aloka Medical, Ltd., Tokyo, Japan) offline to evaluate the EG images from the two systems using the same method.
Classification of EUS-elastography images
We hypothesized that the necrotic tissue may be displayed as the soft region, which would be in the central part of the lesion. Therefore, we classified the location of the soft lesion to compare EG findings and histological features. The findings in the pancreatic mass by EUS-EG were classified into less than 10%, 11% to 30%, 31% to 50%, and 51% or more by the soft area occupied rate in the mass lesion. EUS-EG was used to identify patients in which the inside of the mass could be divided into hard and soft areas, defined as the heterogeneous group, and those with uniform hardness in the mass (hardness unevenness of less than 10%), defined as the homogeneous group.(Figure 2)
Preparation and review of specimens obtained by EUS-FNB
All samples obtained by EUS-FNB were subjected to histological analysis. The histological assessment was performed by an experienced pathologist at Nagoya University Hospital (Y.S.) based on the cellularity scoring system .(18) The samples were submitted such that the pathologist was blinded to the tissue elasticity results for each lesion. The cellularity of the samples was scored from 0 to 5. The adequacy of the samples assessed for histological diagnosis was scored as follows: a score of 0-2 was defined as “inadequate”, and a score of 3-5 was defined as “adequate”. “Malignancy” was defined as either positive or suspected to be malignant, and “benign” was defined as either negative or atypical. To evaluate the amount of core tissues obtained for each specimen, the total length of the core tissue core was measured under a photomicroscope using imaging software (CellSense; Olympus Co. Ltd., Tokyo, Japan). The specimen was captured under low magnification so that the whole specimen was included in a single image, and the total length of the core tissues was measured manually. (19) In addition to the qualitative diagnosis (tissue score) and quantitative diagnosis (core tissue length) of EUS-FNB, the proportion of fibrous stroma in the core tissue was evaluated to compare the tissue elasticity with the characteristics of the collected tissue.(Supplementary figure 1-3) Histopathological evaluation was performed using HE-stained specimens. The pathologist assessed the proportion of fibrous stroma as measured the fibrotic tissue length and core tissue length in units of 10% by using images in CellSense.
Patients were diagnosed with malignant disease if metastatic lesions were identified during imaging examinations, if there were signs of disease progression, and/or if malignant EUS-FNB results were obtained. Patients were diagnosed with benign disease if they had a nonresected mass that did not display imaging features of malignancy during at least 6-month follow-up or if they had EUS-FNB results suggestive of a benign lesion with additional needle passes.
The primary endpoint of this study was a comparison of histological diagnostic ability of EUS-EG-FNB between the soft area and the hard area within the pancreatic mass lesions(Study 1). The secondary endpoints were as follows: 1) association of heterogeneity in pancreatic mass lesions with histopathological diagnosis (Study 2), and 2) comparison of EUS-EG findings and histological findings in PDAC patients(Study 3). Any adverse events were recorded and compared according to the lexicon for endoscopic adverse events advocated by the American Society of Gastrointestinal Endoscopy. (20) Quantitative analysis of EUS-EG was blindly analyzed for clinical and pathological diagnosis. Similarly, pathological assessments were performed blinding to EUS-EG findings.
Statistical analysis and sample size calculation
The accuracy rates in EUS-FNA have been reported to range from 65-90%. We hypothesized that the necrotic tissue in the tumor may exhibit characteristics as a soft area. Kamata et al. reported that the sensitivity of EUS-FNA for lesions with and without an avascular area in CH-EUS was 72% and 94%, respectively.(21) Since the avascular area in CH-EUS may reflect intratumoral necrosis, we referred to this result to determine the sample size and the classification of the soft area location in this study. Assuming that the detection rate of tumor tissue when collected from the hard area by elastography is 90% and the detection rate of tumor tissue when collected from the soft area is 65%, the difference in the score is the standard deviation. When it corresponds to 60%, that is, when the effect size of Cohen is a medium effect size (d = 0.6), a total of 42 patients are needed to detect the difference under the conditions of significance level 0.05 and power 80%. The target number of patients was set at 45 at the time of planning the study. In this study, the sampling, sensitivity, and specificity rates were analyzed using the McNemar test. Descriptive statistics are expressed as the median and interquartile range (IQR). Qualitative variables were compared by χ2 tests, and quantitative variables were compared using the Mann-Whitney U test. Data were statistically analyzed using JMP Pro version 12 (SAS Institute, Inc., Cary, NC, USA).