Introduction

Endoscopic ultrasound (EUS) has made significant strides in recent decades, whose current application in liver is mainly concentrated in liver parenchymal diseases, vascular diseases and portal hypertension. Incorporating EUS into the hepatic examination process is advantageous, which requires the endoscopist to have a good grasp of the anatomy of the liver. However, the learning curve for EUS-guided operations is well known to be steep. Thus, in the evaluation of EUS liver, it is necessary to have a system assisting the endoscopist to complete the scan at each station. Based on the anatomical markers in the liver, we construct a system which can perform a comprehensive inspection of the liver segments through the scanning imaging of 8 stations.

Aims & Methods

EUS of the liver can be done from the stations as follows: Station 1: Left hepatic vein (LHV); Station 2: Umbilical potion of portal vein (PV); Station 3: PV trunk and common bile duct (CBD); Station 4: Middle hepatic vein (MHV) and venous ligament; Station 5: Inferior vena cava (IVC) and right hepatic vein (RHV); Station 6: Gallbladder (GB) and left and right branches of PV; Station 7: Anterior and posterior branches of right portal vein; Station 8: Right kidney.
The system was internally validated in images, and subsequently compared with the performance of EUS endoscopists. We retrospectively collected 2743 images from Hospital A for internal validation, and an independent dataset containing 340 images for human-machine competitions. To filter out the white light images, we apply a DCNN1 model primarily. The output ultrasound images are then fed into DCNN2. In this model, they are classified into standard and non-standard categories. And the DCNN3 is then activated with standard images. DCNN3 identifies certain features in the image and groups them into eight stations. We use accuracy (defined as the number of correctly classified images divided by the total number of images) to evaluate the performance of the model. In addition, a T-test with a significance level of 0.05 was performed to assess significant differences between endoscopists and the model.

Results

The accuracy of DCNN1 in white light image and EUS image classification reached 100% [200/200, 95%CI]. The accuracy of DCNN2 in standard and non-standard image classification reached 81.24%. For the classification of standard station images, the overall accuracy of DCNN3 on the internal image verification set was 92.52% [2572/2743, 95%CI]. In the man-machine competition, the system correctly classified the liver standard stations with an accuracy of 88.53% [301/340, 95%CI]. The average accuracy of the two specialists and ten endoscopists was 84.56% [575/680, 95%CI] and 74.99% [2548/3398, 95%CI], respectively. The P-value for this analysis was less than 0.0001, indicating a statistically significant difference between endoscopists and the model.

Conclusion

In this study, we have constructed an intelligence-assisted linear EUS system based on deep learning algorithms. The system allows for real-time localization of liver standard sub-stations and provides valuable guidance to endoscopists during procedures. The system's effectiveness has been extensively validated through internal validations using images and. Moreover, in a competition with experienced endoscopists, our system has demonstrated superior accuracy, comparable to that of EUS specialists. The result shows that the system is expected to overcome the difficulties of ultrasonographic interpretation among trainees in liver ultrasound endoscopy.

Introduction

Liver biopsy is a crucial diagnostic procedure for assessing hepatic pathologies, providing valuable insights into the etiology, severity, and prognosis of liver diseases. In recent years, Endoscopic Ultrasound-Guided Liver Biopsy (EUS-LB) using 22G- fine needle biopsy ( FNB) needle has emerged as a promising alternative, leveraging the superior imaging capabilities of endoscopic ultrasound (EUS) to guide biopsy with increased precision and safety. The 22G- FNB needle features a crown tip with three symmetrical cutting edges that enable swift core tissue procurement for histological analysis. The 22G- FNB needle provide greater than 95% tissue adequacy and an excellent diagnostic accuracy.

Aims & Methods

The aim of our study is to evaluate the diagnostic utility and safety of a 22G-FNB needle using the slow -pull technique for EUS-LB of parenchymal and focal liver lesions.
We performed a retrospective analysis of prospective data basefrom January 2021 to March 2024. Each patient underwent anendoscopic ultrasound by using linear scope (Fujifilm EG-580UT) and fine needle biopsy using 22G - tip needle with the slow-pull aspiration technique:

• For 33 patients we used a 22G Trident™ - tip FNB needle (Micro-Tech Endoscopy);
• For 26 patients we used a 22G Acquire- tip FNB needle (Boston Scientific Corp.);
  
• For 4 patients we used a 22G SonoTip Top Gain - tip FNB needle ( MediGlobe);

The median number of needle passes was approximately 3 (between2 and 5).

Results

We identified 63 patients who underwent EUS-LB of the liver during the study period(28 men and 35 women): 47 (74.6%)of them were with malignant and 16 (25.4%)–with benign disease. The medium age of the patients was 56.9 years. The specimen was considered adequate in all 63 patients.
We classified the EUS-FNB of liver diseases into EUS-FNB of parenchymal and focal liver lesions.

• EUS-FNB of parenchymal liver disease were performed to 13 (20.6%) of the patients: cirrhosis was found in 2 (3.17%), nonalcoholic fatty liver disease – in 5 (7.9%) and cholestasis in 6 (9.5%).
• We performed EUS-FNB of focal liver lesions in 50 ( 79.36%) of the patients.
  
  • Metastasis were found in 39 (61,9%): pancreatic ductal adenocarcinoma was detected in 27 (42.8%), pancreatic acinar cell carcinoma – in 1 (1.58%), pancreatic neuroendocrine carcinoma – in 5 (7.9%), lung adenocarcinoma – in 3 (4.7%), ductal adenocarcinoma of breast – in 2 (3.17%) , colorectal adenocarcinoma – in 1 (1.58%) and Ewing sarcoma – in 1 (1.58%).
  • Primary liver cancer - hepatocellular cancer was diagnosed in 4 (6.34%) patients.
    
  • Cholangiocarcinoma was found in 3 (4.7%) patients.
  • There were 3 (4.7%) patients with negative histology for tumor cells in the specimen: in 2 (3.17%) cases liver abscess was proved and 1 (1.58%) hepatic cyst.
    

No adverse events were observed during the first 48h and after a 30 days follow-up.
There was no death during the study period.

Conclusion

Our study illustrates that ЕUS-guided liver biopsy using a 22 G- FNB needle provides adequate specimen for histologic analysis, lesser postprocedural recovery time and it is a safe, highly effective alternative to other methods of liver biopsy. The use of real time imaging guidance with doppler also helps to reduce the adverse events.

Introduction

Endoscopic ultrasound-guided fine needle biopsy (EUS-FNB) for hepatic solid lesions has emerged as a safe technique to obtain an adequate liver tissue for diagnosis of parenchymal liver disease. However, the optimal FNB techniques have not been evaluated. Out study aimed to compare the diagnostic yield and specimen adequacy for EUS-FNB in hepatic solid lesions using different FNB techniques.

Aims & Methods

This is a single center retrospective study of EUS-FNB for hepatic solid lesions between Mar. 2015 and Jun. 2022. Each one needle pass of suction and slow-pull suction was applied for the same hepatic lesions and the sequence was randomly assigned. The primary outcomes were diagnostic yield to attain a histological diagnosis and rate of adequate specimen acquisition.

Results

A total of 119 patients (40 female, median age 70 years) underwent EUS-FNB for hepatic solid lesions. The mean size of hepatic solid lesion was 34.9+27.3 mm (range 7-150). Solid live lesions were located at S1 in 5 (4.2%), S2/3 in 97 (81.5%), S4 in 13 (10.9%), and S5/6/7/8 in 4 (3.3%). Needle puncture was approached at esophagus in 5 (4.2%), stomach in 108 (90.8%), and duodenum in 6 (5.0%). The use of needle was 20G ProCore in 44 (37.0%) and 22G in 75 (63.0%). The final diagnosis of solid liver lesions was benign in 3 (2.5%) and malignancy in 116 (97.5%) including cholangiocarcinoma (49, 41.2%), metastasis (63, 52.9%), hepatocellular carcinoma (2, 1.7%), and lymphoma (2, 1.7%). The overall diagnostic yield to obtain a tissue diagnosis was 90.8%. There was statistically no difference in diagnostic accuracy between suction and slow-pull suction (79.8% vs. 84.9%, P=0.308). Slow-pull suction obtained more adequate specimen comparing to suction (95.0% vs. 87.4%, P=0.040).

Table 1. Comparison of specimen adequacy and diagnostic performances according to EUS-FNB techniques

	Slow-pull suction, n (%)	Suction, n (%)	P value
Histologic diagnosis - Malignancy - Suspicious for malignancy - Atypical - Benign	91 (76.5) 7 (5.9) 11 (9.2) 4 (3.4)	86 (72.3) 6 (5.0) 6 (5.0) 6 (5.0)	0.203
Inadequate specimen	9 (5.0)	15 (12.3)	0.040
Diagnostic accuracy	101 (84.9)	95 (79.8)	0.308
Grossly core tissue acquisition	111 (93.3)	103 (86.6)	0.085

Conclusion

The use of slow-pull suction EUS-FNB demonstrated improved specimen adequacy compared with suction technique with no difference in diagnostic accuracy.

References

Madhok, I. K., et al. (2022). "Endoscopic Ultrasound-Guided Liver Biopsy." Clin Liver Dis 26(1): 127-138.
Gheorghiu, M., et al. (2022). "Endoscopic Ultrasound-Guided Fine-Needle Biopsy versus Fine-Needle Aspiration in the Diagnosis of Focal Liver Lesions: Prospective Head-to-Head Comparison." Diagnostics (Basel) 12(9).

Introduction

Despite the increasing diffusion of non-invasive tests for the diagnosis of liver diseases, histological assessment through liver biopsy (LB) is still often mandatory especially in unclear cases. Actually, percutaneous approach (PC-LB) is the gold standard. Nowadays, endoscopic ultrasound-guided liver biopsy (EUS-LB) has emerged as an established method to achieve liver tissue sampling. EUS-LB has several advantages such as: sampling of both liver lobes, it allows to obtain liver tissue sampling in case of ascites and toavoid damage to nearby organs and vessels.

Aims & Methods

e performed a systematic review with meta-analysis (MA) to evaluate the diagnostic outcomes and safety for liver tissue acquisition of PC-LB and EUS-LB in patients with liver diseases and focal liver lesions. Studies that compare diagnostic outcomes of PC-LB and EUS-LB were identified through literature search usign MEDLINE, Embase and Scopus until January 2024. MA was performed according to a frequentist approach.

Results

Overall, diagnostic adequacy was 94 % (CI 95%, 0.91-0.98, I² = 83.92 %, P < 0.001) for EUS-LB, while 96 % (CI 95%, 0.93-0.99, I² = 91.58 %, P < 0.001) for PC-LB. Moreover, based on eight studies comparison between the two procedures showed an OR 1.89 (95% CI, 0.44 – 8.12, I² = 62.33 %, P=0.003) when comparing PC-LB to EUS-LB. Based on four studies, pooled accuracy was OR 2.45 (95% CI, 0.38 – 15.97, I² = 49.74 %, P=0.0114) when comparing PC-LB to EUS-LB. Regarding safety, based on nine studies, pooled rate of overall adverse events was slightly higher in PC-LB compared to EUS-LB, OR: 1.25 (CI 95%, 0.51 – 3.09 I² = 26.13 %, P=0.212). Technical success for EUS-LB was 99 % (CI 95%, 0.99 - 1.00, I² = 0 %, P = 0.973) while 100 % (CI 95%, 0.99-1.00, I² = 0 %, P =0.973) for PC-LB, with an OR 1.11 (95% CI, 0.26-4.70, I² = 0 %, P=0.988) when comparing PC-LB to EUS-LB. Overall no differences between the two procedures were observed I term of total length specimen with a standardized mean difference of 0.89 (95% CI, -0.87 – 2.64, I² = 98.95 %, P < 0.001) and in term of complete portal tract with a standardized mean difference of 0.50 (95% CI, -0.55 – 1.55, I² = 97.42 %, P < 0.001).

Conclusion

In conclusion, our meta-analysis does not show significant differences between PC-LB and EUS-LB in terms of diagnostic outcomes, albeit EUS-LB is slightly safer than PC-LB.

Introduction

Recent developments in endoscopic ultrasound (EUS) have proven the effectiveness of endoscopic methods for performing liver biopsies, but there is still scarce and conflicting evidence on the comparison between EUS and percutaneous (PC)-guided liver biopsy (LB).

Aims & Methods

The aim of this study was to compare the routine method and analysis of PC-LB samples against the newly introduced EUS-LB in a series of patients with parenchymal liver lesions in a tertiary hospital between 2022-2024. EUS-LB was performed with a 19-gauge core biopsy needle (Acquire, Boston Scientific) via one pass/three actuation, wet suction technique in Propofol sedation. These samples were analyzed by a dedicated liver pathologist. PC-LB procedures were performed using a 16-gauge Menghini needle via one pass without sedation. Samples from the PC-LB group were analyzed by a non-dedicated liver pathologist. The primary outcome was diagnostic adequacy rate. The secondary outcomes were sample adequacy (assessed based on total specimen length (TSL), number of complete portal tracts (CPTs)), and adverse events. Data were retrospectively assessed by reviewing the hospital information system data.

Results

The study included 56 patients, 35 in the PC-LB group and 21 in the EUS-LB group. The median age of the patients at the time of liver biopsy was 52 years, 55% were female. Predominant diagnoses were nonalcoholic fatty liver disease (25%), primary biliary cholangitis (18%) and autoimmune hepatitis (18%). The rate of establishing histological diagnosis was significantly higher in the EUS-LB group (100% vs. 77%; p=0.019). Adequate biopsy rate (according to European Association for the Study of Liver Disease criteria) and optimal biopsy rate (according to American Association for the Study of Liver Diseases criteria) were 100% and 90.5% in the EUS-LB group vs. 68.6% and 40% in the PC-LB group, crossing the significance threshold (p=0.004 and p=0.0002, respectively). The median number of CPTs was significantly greater in the EUS-LB group, with a median of 35 CPTs (IQR 22-47) compared to 11 CPTs (IQR 8-14) in the PC-LB group (p<<0.05). Among all participants in this study, patients whose histological diagnosis was not established had a lower median number of CPTs, with 10 (5-18), compared to 14 (10-29) for those who were diagnosed. However, this difference was not statistically significant (p=0.11). Both methods had similar safety profiles regarding post-procedural bleeding, which was clinically non-significant (one patient in each group). Additionally, fewer patients in the EUS-LB group required postprocedural analgesics compared to the PC-LB group (5% vs. 26%; p=0.07).

Conclusion

EUS-LB can reliably provide samples suitable for a histopathologic diagnosis in patients with parenchymal liver disease. In our setting EUS-LB proved to be superior to PC-LB in terms of diagnostic yield, tissue adequacy, and occurrence of minor adverse events, respecting the fact that EUS-LB samples were analyzed by a dedicated pathologist. As such, the method was easily incorporated into the workflow of our endoscopy unit.

Introduction

The aim of this study was to evaluate and to establish cut-off values for two new quantitative ultrasound (QUS) parameters, TSI (tissue scatter-distribution imaging) and TAI (tissue attenuation imaging) in early diagnosis of liver steatosis (LS), as compared to controlled attenuation parameter (CAP).

Aims & Methods

A prospective study was conducted in which LS was assessed in the same session by QUS (Samsung Medison RS85,CA1-7A probe) and CAP (FibroScan Compact M530, M/XL probes). Reliable measurements were defined for CAP the median value of 10 measurements with IQR/M<0.3, and for QUS, five consecutive measurements of TAI and TSI acquired by a color-coded map. TAI and TSI were automatically calculated and considered reliable when reliability index, R²>0.6. The CAP cut-off value used for the whole cohort for at least mild steatosis (S1) were 248 dB/m[1]; for ALD (alcoholic liver disease) cohort, 268 dB/mm[2]; for NAFLD(non-alcoholic liver disease) cohort 294dB/m[2]. Demographic and health related datas were recorded.

Results

A total of 285 patients, with a mean age of 56.1±12.4, 114 female and 171 male, were included in the study. According to aetiology, 164 (57.5%) patients had NAFLD, 61(21.4%) ALD, and 60 (21.1%) were other etiologies (viral, cardiac, autoimmune). The obtained TSI and TAI cut-off values for the diagnosis of at least mild steatosis (S1) are presented in Table 1.
Table 1. Cut-off values of TAI and TSI for S1

Variable		Overall	NAFLD	ALD
TSI	S1	>96.2 AUC=0.74,p<0.0001, Se=87.9%,Sp=53.0%;	>96.5 AUC=0.73, p<0.0001, Se=90.9%, Sp=47.9%;	>94.9 AUC=0.70, p=0.003 Se=72.9%,Sp=75.0%;
TAI	S1	>0.73 AUC=0.82,p<0.0001, Se=78.3%,Sp=71.6%;	>0.75 AUC=0.81, p<0.0001, Se=57.5%, Sp=90.8%;	>0.66 AUC=0.74, p<0.0001, Se=84.8%,Sp=57.1%;

There were no differences between the performance of TAI and TSI according to the DeLong test, p=0.18 and p=0.24. A strong direct correlation was observed between TAI and CAP r=0.701, moderate between TSI and CAP r=0.56 for the all cohort, but for NAFLD and ALD subgroups moderate correlations were found between TAI and CAP r=0.66, and r=0.66 respectively and TSI and CAP r=0.56 and 0.66, respectively.
In univariate regression analysis, the factors associated with TSI were hypertension, Diabettes mellitus(DM), and obesity, and all p-values were <0.0001, but in multivariate analysis, only hypertension(p=0.001) and obesity(p=0.0005) were associated. For TAI, in univariate analysis, the same factors: hypertension(p=0.004), obesity(p=0.0006) and DM (p=0.002) were associated, but in multivariate analysis, DM was the only factor associated, p=0.02. Age and gender were not correlated with any method, TSI or TAI.

Conclusion

TAI and TSI are feasible non-invasive methods for screening and diagnosis of liver steatosis, with good accuracy. Patients with NAFLD presented highest cut-off values, probably by the presence of DM and obesity, conditions independently associated with TAI and TSI values.

References

1.Karlas T, Petroff D, Sasso M, Fan JG, Mi YQ, de Lédinghen Vet al. Individual patient data meta-analysis of controlled attenuation parameter (CAP) technology for assessing steatosis. J Hepatol. 2017 May;66(5):1022-1030. doi: 10.1016/j.jhep.2016.12.022. Epub 2016 Dec 28. PMID: 28039099.
2. Petroff D, Blank V, Newsome PN, Shalimar, Voican CS, Thiele M, et al. Assessment of hepatic steatosis by controlled attenuation parameter using the M and XL probes: an individual patient data meta-analysis. Lancet Gastroenterol Hepatol. 2021 Mar;6(3):185-198. doi: 10.1016/S2468-1253(20)30357-5. Epub 2021 Jan 16. PMID: 33460567.

Introduction

Endoscopic ultrasound-guided fine-needle aspiration (EUS-FNA) is an important diagnostic tool for suspected parenchymal lesions within the gastrointestinal tract and adjacent organs. However, the amount of literature on EUS-FNA for sampling focal liver lesions (FLLs) is limited.

Aims & Methods

This study aims at evaluating the safety and effectiveness of EUS-FNA in FLLs. Data from 88 patients who were diagnosed with FLLs by imaging and underwent EUS-FNA was reviewed in this retrospective study at the Second Affiliated Hospital of Soochow University and Ruijin Hospital of School of Medicine of Shanghai Jiao Tong University from 1 January 2017 to 31 August 2022.The EUS-FNA biopsy results were compared with the final diagnosis to evaluate the diagnostic value. Meanwhile, the relevant factors were analyzed to determine their influences on the EUS-FNA biopsy results.

Results

A final diagnosis was obtained (86 malignant and 2 benign) from the 88 cases. The overall diagnostic accuracy of EUS-FNA in FLLs was 93.18% (82／88; 95% confidence interval [CI ],87.9–98.5), with a sensitivity, specificity, positive predictive value, and negative predictive value of 93.02% (80／86; 95%CI, 87.6–98.4), 100% (2／2; 95%CI, 100–100), 100% (80／80; 95%CI, 100–100), and 25% (2／8; 95%CI, -5–55.0), respectively . The parameters related to lesion and procedure were not statistically significant between these two groups (p> 0.05) (Table 1). The number of puncture needles in the groups showed a statistically significant difference between multiple and single punctures (p = 0.001).

Conclusion

It is revealed in this study that EUS-FNA is a safe and excellent diagnostic method for FLLs with high accuracy.