A Porcine Model of Endoscopic Ultrasound-Guided Hepatic Gastrostomy for Malignant Obstructive Jaundice

Summary

This article describes a porcine model comparing endoscopic ultrasound-guided hepaticogastrostomy with percutaneous transhepatic biliary drainage for malignant obstructive jaundice, monitoring biochemical, inflammatory, and patency outcomes over 28 days.

Abstract

Current clinical evidence on biliary drainage predominantly focuses on intraoperative outcomes and single-point laboratory assessments, lacking comprehensive longitudinal functional curves and mechanistic insight into the pathways of complications. This study compares endoscopic ultrasound-guided hepaticogastrostomy (EUS-HGS) with percutaneous transhepatic cholangial drainage (PTCD) in terms of hepatobiliary biochemical decline kinetics during days 0–7, drainage efficiency parameters, inflammatory–endotoxin temporal profiles, and 28-day stent patency–survival endpoints. Twenty-four pigs with surgically induced malignant obstructive jaundice were randomized to the EUS-HGS group (n = 12) or the PTCD group (n = 12). Daily monitoring included total bilirubin (TB), direct bilirubin, alkaline phosphatase, and γ-glutamyltransferase. Drainage volume and bile bilirubin concentrations were recorded daily. C-reactive protein (CRP), interleukin-6 (IL-6), and lipopolysaccharide levels were measured on days 0, 1, 3, 5, and 7, with bile cultures performed on days 0, 3, and 7. The EUS-HGS method demonstrated superior TB half-life (2.3 ± 0.4 vs 3.8 ± 0.6 days; β = −0.42; 95% CI: −0.68 to −0.16; P < 0.001) and a higher drainage efficiency index (5.2 ± 1.1 vs 3.1 ± 0.8 µmol/L/mL; P < 0.001). The 28-day stent patency rate was 91.7% for EUS-HGS and 58.3% for PTCD (hazard ratio = 0.31; 95% CI: 0.12–0.78; P = 0.013). The area under the curve for CRP was significantly lower in the EUS-HGS group (420 ± 95 vs 680 ± 120 mg·day/L; P < 0.001); IL-6 peaked earlier but resolved faster in the EUS-HGS group. Bacterial colonization occurred in 25% of EUS-HGS cases versus 50% of PTCD cases (P = 0.041). The EUS-HGS approach demonstrates superior bilirubin clearance kinetics, enhanced drainage efficiency, reduced inflammatory burden, and improved 28-day patency compared with PTCD, supporting the mechanistic advantages of the transgastric approach.

Introduction

Malignant obstructive jaundice (MOJ) represents a critical hepatobiliary emergency requiring prompt intervention to prevent progressive liver dysfunction, septic complications, and mortality¹. The therapeutic landscape has evolved considerably with the introduction of endoscopic ultrasound-guided biliary drainage (EUS-BD) as an alternative to conventional endoscopic retrograde cholangiopancreatography (ERCP) and percutaneous transhepatic biliary drainage (PTCD)^2,3. Recent meta-analyses have demonstrated that EUS-BD achieves technical success rates of 94%–96% with clinical outcomes comparable to or superior to those of PTCD, particularly regarding quality of life and complication profiles^4,5. In a propensity score-matched analysis, Koutlas et al. reported that EUS-guided hepaticogastrostomy (EUS-HGS) after failed ERCP was associated with fewer adverse events and shorter hospital stays compared with PTCD⁶. Similarly, large-scale retrospective analyses have confirmed lower rates of re-intervention and infection-related complications with EUS-BD approaches^7,8.

Despite these advances, critical knowledge gaps persist regarding the temporal dynamics of biliary decompression and inflammatory resolution. Previous studies have primarily focused on binary outcomes such as technical success and 30-day mortality, with limited attention to the continuous functional trajectories that determine clinical recovery^9,10. The kinetics of bilirubin clearance, the relationship between drainage volume and biochemical improvement, and the temporal evolution of inflammatory markers remain poorly characterised^11,12. Furthermore, the mechanistic basis for differential outcomes between EUS-guided and percutaneous approaches has not been systematically investigated using controlled experimental models^13,14. Current evidence, drawn predominantly from retrospective clinical cohorts, cannot adequately isolate procedure-specific effects from patient-level confounders such as tumor stage, nutritional status, and prior biliary interventions¹⁵.

The present study was designed to address these gaps through a comprehensive porcine model of MOJ, chosen for its anatomical and physiological similarities to the human hepatobiliary system¹⁶. As a controlled experimental model, the current design permits isolation of procedure-related variables but does not fully replicate the complexity of human malignant biliary obstruction, including tumor-driven inflammation, coagulopathy, and immune dysfunction. As such, the findings should be interpreted as mechanistic evidence requiring clinical validation. The primary hypotheses were that, compared with PTCD, EUS-HGS would demonstrate accelerated bilirubin clearance kinetics through more efficient drainage mechanics, reduced inflammatory and endotoxin burden due to internalized drainage pathways, and superior 28-day stent patency, translating to improved survival outcomes. This mechanistic approach aims to provide evidence-based parameters for optimizing drainage strategies and predicting treatment response in clinical practice¹⁷.

Protocol

All animal procedures were approved by the Institutional Animal Care and Use Committee (IACUC) of the First Affiliated Hospital of Hebei North University (Approval No. HBNU20240722103) and conducted in accordance with ARRIVE 2.0 guidelines for reporting animal research¹⁸. See the Table of Materials for specific reagents, equipment, and instruments used in this protocol.

Animal preparation and acclimatization
Twenty-four healthy domestic pigs (LY-strain, mixed breed), aged 4–6 months and weighing 35–45 kg, were obtained from a certified facility. The animals were housed in temperature-controlled rooms (22 ± 2 °C) with a 12 h light–dark cycle, and standard porcine feed was provided ad libitum (i.e., without restriction), with water available. The animals underwent a 2-week acclimatization period. Daily health monitoring was performed, including body temperature, appetite, stool consistency, and general behavior. Baseline body weight was recorded on the day of arrival and weekly thereafter. Feed was withheld for 24 h prior to any surgical procedure, while maintaining free access to water. Fasting status was confirmed via visual inspection of the oral cavity and absence of abdominal distension prior to anesthesia induction.

Anesthesia induction and maintenance
Intramuscular azaperone (4.8 mg/kg) and midazolam (0.6 mg/kg) were administered for sedation. The animals were then left for approximately 10–15 min until they were recumbent and non-responsive to tactile stimulation. Intravenous (IV) access was established via a marginal ear vein using a 20-gauge catheter. Etomidate (0.4 mg/kg IV) was administered for induction, and orotracheal intubation was performed using a cuffed endotracheal tube (size: 7.0–8.0, selected according to tracheal diameter). Anesthesia was maintained with continuous IV infusion of ketamine (10 mg/kg/h), midazolam (0.6 mg/kg/h), and pancuronium (0.12 mg/kg/h). Mechanical ventilation was initiated, with a tidal volume of 10–12 mL/kg, a respiratory rate of 12–16 breaths/min, and end-tidal CO₂ maintained at 35–45 mmHg. Continuous physiological monitors were attached, including pulse oximetry (SpO₂ target >95%), electrocardiography, capnography, and invasive arterial blood pressure via a femoral artery catheter. Vital signs were recorded every 5 min throughout the procedure. Body temperature was maintained between 37.5 °C and 39.0 °C using a forced-air warming blanket. Hypothermia (<36.5 °C) may alter hepatic metabolism and bile flow.

Malignant obstructive jaundice model establishment
The animal was positioned in dorsal recumbency, and the abdomen was prepared with povidone-iodine solution. A sterile surgical field was established using the sterile technique with surgical drapes. A midline laparotomy incision (approximately 15 cm) was made from the xiphoid process caudally. Electrocautery was used to achieve hemostasis of the subcutaneous tissues and the linea alba. The common bile duct (CBD) was identified by retracting the duodenum laterally and inspecting the hepatoduodenal ligament. The CBD was carefully dissected from the surrounding portal vein and hepatic artery using blunt dissection with right-angle forceps, taking care to preserve the vascular pedicle. The duodenal (distal) portion of the CBD was ligated using 3-0 silk sutures placed at two points 5 mm apart. The duct between the ligatures was transected using fine scissors to prevent recanalization. Complete transection was confirmed visually. A 6-French silicone catheter was inserted into the hepatic (proximal) side of the transected duct. The catheter was secured with double purse-string sutures using a 4-0 absorbable monofilament. Catheter position was verified by gentle aspiration of bile (expected: clear golden-yellow fluid). The catheter was tunneled subcutaneously from the abdominal incision to exit at the dorsal neck region using a trocar-guided subcutaneous passage. The external end of the catheter was clamped with a hemostat to allow bile accumulation and progressive cholestasis. The abdominal wall was closed in layers: the peritoneum and fascia with continuous 2-0 absorbable sutures, subcutaneous tissue with interrupted 3-0 absorbable sutures, and skin with interrupted 3-0 nylon sutures or skin staples. The animals were monitored continuously for 2 h post-procedure in a recovery area. Buprenorphine (0.02 mg/kg intramuscularly [IM] every 8 h) and meloxicam (0.4 mg/kg orally [PO] daily) were administered for multimodal analgesia. Pain was assessed using a validated porcine grimace scale every 4 h for the first 48 h. Development of obstructive jaundice was confirmed by measuring serum total bilirubin (TB) levels at 48 and 72 h post-ligation. Randomization was commenced when TB exceeded 100 µmol/L. The expected time to jaundice is 48–72 h. Animals failing to develop TB > 100 µmol/L within 96 h should be excluded and evaluated for incomplete duct ligation or catheter-related bile leak. Humane endpoints include uncontrolled pain (grimace score ≥6/10), persistent anorexia for >48 h, or signs of peritonitis (abdominal rigidity, fever >41 °C).

Randomization and group allocation
Computer-generated block randomization with variable block sizes (4 and 6) was performed to assign animals to the EUS-HGS group (n = 12) or the PTCD group (n = 12). Allocation concealment was maintained using sequentially numbered, sealed opaque envelopes, opened immediately before the drainage procedure. Blinding of all laboratory personnel and statisticians was ensured through coded specimen labeling. Procedural operators could not be blinded because of the nature of the interventions.

Endoscopic ultrasound-guided hepaticogastrostomy procedure
The anesthetized animal was placed in the left lateral decubitus position. A linear echoendoscope was advanced through the oral cavity into the gastric body under endoscopic visualization. Systematic ultrasonographic evaluation of the left liver lobe was performed through the gastric wall. Dilated intrahepatic bile ducts (target diameter >8 mm) in segments 2 or 3 were identified. Color Doppler mode was activated, and all visible vascular structures were mapped to select an avascular puncture trajectory. The absence of interposed vessels between the gastric wall and the target duct was confirmed. Under real-time EUS guidance, a 19-gauge fine-needle aspiration needle was advanced through the gastric wall into the selected bile duct. Intraductal position was confirmed via aspirating bile (the aspirate should appear dark golden-green and viscous). If no bile was obtained, the needle was repositioned under continuous imaging before reattempting.

Contrast medium (5–10 mL of iohexol, 300 mg I/mL) was injected through the needle under fluoroscopic guidance. A cholangiogram documenting the biliary anatomy was obtained, and the level and completeness of obstruction were confirmed. A 0.035-inch hydrophilic guidewire was advanced through the needle into the dilated intrahepatic system. Guidewire positioning was confirmed fluoroscopically, ensuring adequate purchase within the biliary tree (minimum 5 cm intrahepatic). The needle was removed while maintaining the guidewire position. The hepaticogastric tract was dilated by advancing a 6-French cystotome over the guidewire and applying electrocautery at 30 W pure-cut mode. Tract creation was confirmed by tactile decreased resistance and visible bile flow. A fully covered self-expanding metal stent (FCSEMS) (8 × 60 mm) was placed over the guidewire under combined endoscopic and fluoroscopic guidance. The stent was released gradually, confirming that the proximal (biliary) end was positioned within the intrahepatic duct and the distal (gastric) end protruded 1–2 cm into the gastric lumen. Bile flow through the stent was verified endoscopically (visible bile drainage into the stomach) and fluoroscopically (contrast clearance). The 8 × 60 mm FCSEMS was selected based on the anatomical characteristics of the porcine model, in which the transgastric distance from the gastric wall to segment 2/3 ducts typically measures 3–5 cm. A 60 mm length provides adequate bridging with approximately 1–2 cm of intragastric protrusion for anchoring while minimizing excessive intrahepatic extension. It should be noted that the larger diameter of the metal stent used in EUS-HGS compared with the 10-French PTCD catheter may have contributed to the observed differences in drainage efficiency, and future studies employing matched-diameter stents are warranted to isolate the effect of the drainage route from the stent caliber.

Percutaneous transhepatic cholangial drainage procedure
The anesthetized animal was placed in the left lateral decubitus position. The right subcostal region was prepared using an antiseptic solution. Using real-time ultrasound guidance, dilated peripheral bile ducts in segments 6 or 7 of the right liver lobe were identified. A target duct with a diameter >5 mm was selected, and a clear percutaneous path was created, avoiding the pleural space. Under continuous ultrasound imaging, an 18-gauge needle with a stylet was advanced through the skin and liver parenchyma into the target bile duct. Intraductal position was confirmed by aspirating bile (dark golden-green fluid) and injecting 5–10 mL of iodinated contrast under fluoroscopy. The stylet was exchanged for a 0.035-inch hydrophilic guidewire, which was advanced centrally towards the hilum under fluoroscopic guidance. Serial tract dilation was performed using fascial dilators (6, 8, and 10-French sequentially). Each dilator was passed over the guidewire with gentle rotational advancement, pausing between sizes to allow tissue to accommodate. A 10-French multipurpose drainage catheter with side holes was placed over the guidewire, positioning the pigtail tip and side holes spanning the obstructed segment. Positioning was confirmed fluoroscopically. The external catheter was secured with skin sutures (2-0 nylon) and a sterile adhesive dressing. The catheter was then connected to a closed gravity drainage bag. The PTCD catheter was flushed daily with 10 mL sterile normal saline to maintain patency. If daily drainage output falls below 20 mL or the patient exhibits clinical deterioration (elevated bilirubin, fever), they should be examined for catheter obstruction or dislodgement with fluoroscopic imaging, and catheter exchange should be considered.

Post-procedural care
For EUS-HGS animals, nasogastric decompression was maintained for 24 h post-procedure. A proton pump inhibitor (pantoprazole, 40 mg IV daily) was administered, as well as broad-spectrum antibiotic prophylaxis (ceftriaxone, 1 g IV daily) for 72 h. A graduated refeeding protocol was commenced for EUS-HGS animals, including clear liquids at 24 h, soft diet at 48 h, and standard feed at 72 h if tolerated. For PTCD animals, daily catheter care included flushing with 10 mL of normal saline, inspecting the insertion site for erythema or discharge, and performing dressing changes. Drainage output volume and color were monitored. Multimodal analgesia (buprenorphine, 0.02 mg/kg IM every 8 h for 72 h; meloxicam, 0.4 mg/kg PO daily for 7 days) was continued for all animals. Pain was assessed using the porcine grimace scale every 8 h for the first 7 days.

Data collection and sample acquisition
Venous blood samples (10 mL) were collected daily from day 0 (pre-intervention baseline) to day 7, and on days 14, 21, and 28 from the jugular vein using an aseptic technique, into serum separator and ethylenediaminetetraacetic acid tubes. The hepatobiliary panel (TB, direct bilirubin [DB], indirect bilirubin, conjugated bilirubin, alkaline phosphatase [ALP], γ-glutamyl transferase [γ-GT], aspartate aminotransferase, and alanine aminotransferase) and albumin were analyzed using an automated clinical chemistry analyzer. All samples were processed within 2 h of collection. The residual serum was aliquoted and stored at −80 °C for batch analysis. Inflammatory markers were measured (C-reactive protein [CRP] via immunoturbidimetric assay, interleukin-6 [IL-6] via porcine-specific enzyme-linked immunosorbent assay, and lipopolysaccharide [LPS] via limulus amoebocyte lysate chromogenic assay) on days 0, 1, 3, 5, and 7. Drainage volume was recorded every 24 h using graduated collection systems with 1 mL precision. Bile bilirubin concentration was measured spectrophotometrically at a 454-nm wavelength. Bile cultures were obtained on days 0, 3, and 7 by sterile aspiration of 5 mL of bile. They were then inoculated onto blood agar, MacConkey agar, and chocolate agar plates and incubated aerobically and anaerobically at 37 °C for 48–72 h. Bacteria were identified using MALDI-TOF mass spectrometry, and quantitative colony counts were reported (CFU/mL). Complete blood count with differential, coagulation profile (PT, aPTT, INR), and renal function tests (creatinine, blood urea nitrogen) were performed at each sampling time for safety monitoring.

Endpoint assessment and follow-up
The primary endpoints were defined as follows: (a) TB half-life (t½), calculated from the log-linear phase of decline using mixed-effects modeling of daily measurements; and (b) 28-day stent patency, defined as maintained bile flow without clinical, biochemical (TB increase >50% from nadir), or imaging evidence of obstruction requiring reintervention. Secondary endpoints included clinical success (>50% TB reduction within 7 days without procedure-related mortality), complication rates, as classified by the Clavien–Dindo system, inflammatory burden, quantified as the area under the curve (AUC) for CRP, IL-6, and LPS, and 28-day survival. All animals were monitored daily for 28 days. Focused clinical assessment was performed every 12 h for the first 7 days, then daily thereafter. Activity level, feeding behavior, stool character, and body temperature were recorded. Necropsy was performed on animals reaching humane endpoints or at study termination (day 28). Cause of death, stent position, fistula tract integrity, and pathological findings were documented. Humane endpoints warranting euthanasia include unresponsive pain (grimace score ≥8/10 despite rescue analgesia), persistent anorexia >72 h with >15% weight loss, signs of uncontrolled sepsis (temperature >41.5 °C with hemodynamic instability), or biliary peritonitis confirmed by abdominal paracentesis.

Statistical analysis
Statistical analyses were conducted using R (version 4.3.1) with the lme4, survival, and ggplot2 packages. Logarithmic transformation was applied to right-skewed biochemical variables (TB, DB, ALP, γ-GT, LPS). Missing values (<10% overall) were handled via multiple imputation (m = 20) using predictive mean matching. Outliers were detected using Cook’s distance (threshold >4/n). Linear mixed-effects models were constructed for repeated hepatobiliary measures using the following specification: log(TB_ij) = β₀ + β₁ × Day_ij + β₂ × Group_i + β₃ × (Day × Group)_ij + b_oj + b_1j × Day_ij + ε_ij, with random intercepts and slopes. Group-specific half-lives were derived as t½ = ln(2)/|slope|, with 95% CIs from parametric bootstrapping (n = 1,000). Drainage efficiency was analyzed using the following formula: drainage efficiency index (DEI) = θ₀ + θ₁ × Volume + θ₂ × Group + θ₃ × (Volume × Group) + u_i + e_ij, with robust standard errors and quantile regression at the 25th, 50th, and 75th percentiles. Time-to-event analyses were performed using Kaplan–Meier estimation with log-rank tests and Cox proportional hazards regression with the following formula: h(t) = h₀(t) exp(γ₁ × Group + γ₂ × BaselineTB + γ₃ × ALP + γ₄ × BileCulture + γ₅ × ProcedureTime). Proportional hazards assumptions were verified using scaled Schoenfeld residuals. Fine–Grey competing risks analysis was performed, treating death as a competing event. All tests were two-sided, with α = 0.05.

Results

Baseline characteristics
Twenty-four pigs underwent MOJ model creation and were subsequently randomly assigned to the EUS-HGS group (n = 12) or the PTCD group (n = 12). Baseline characteristics were well balanced between the groups (Table 1). The mean age was 4.8 ± 0.6 months, and the mean weight was 40.2 ± 3.4 kg, with a 58% female distribution. Pre-intervention TB levels were comparable (EUS-HGS: 332.8 ± 73.2 µmol/L vs PTCD: 341.2 ± 76.5 µmol/L; P = 0.782). The median time from MOJ creation to intervention was 72 h (range: 68–76 h), with no significant differences in baseline liver function parameters or systemic inflammatory markers between groups.

Longitudinal hepatobiliary biochemical dynamics (days 0–7)
Linear mixed-effects modeling revealed significant group × time interactions for all hepatobiliary markers, indicating differential clearance kinetics between EUS-HGS and PTCD (Table 2, Figure 1A–D). The EUS-HGS group demonstrated more rapid TB decline, with a calculated half-life of 2.3 ± 0.4 days, compared with 3.8 ± 0.6 days for PTCD (β for interaction = −0.42; 95% CI: −0.68 to −0.16; P < 0.001). This accelerated clearance was consistent across bilirubin fractions, with DB showing the most pronounced difference (t½: 2.1 ± 0.3 vs 3.5 ± 0.5 days; P < 0.001). Alkaline phosphatase normalization followed similar patterns, achieving a 50% reduction by day 4 in the EUS-HGS group and by day 6 in the PTCD group (P = 0.003). Individual trajectory analysis revealed less variability in the EUS-HGS group (random-slope SD: 0.08 vs 0.14), suggesting a more predictable treatment response.

Drainage efficiency and bile composition analysis
The DEI, representing biochemical improvement per unit drainage volume, was significantly higher in the EUS-HGS group throughout the study period (Table 3, Figure 2A). The mean DEI was 5.2 ± 1.1 µmol/L/mL for the EUS-HGS group and 3.1 ± 0.8 µmol/L/mL for the PTCD group (P < 0.001). Volume–response analysis demonstrated a steeper slope for EUS-HGS (β = 0.68; 95% CI: 0.52–0.84) compared with PTCD (β = 0.41; 95% CI: 0.28–0.54), indicating superior efficiency of internalized drainage. Bile bilirubin concentrations were initially similar but diverged by day 3, with EUS-HGS showing progressive dilution (420 ± 95 to 180 ± 45 µmol/L), whereas PTCD remained elevated (445 ± 102 to 310 ± 78 µmol/L; P = 0.002) (Figure 2B). Quantile regression confirmed these differences across the distribution, with the greatest effects at the 75th percentile.

Inflammatory-endotoxin dynamics and microbiological profiles
Temporal analysis of inflammatory markers revealed distinct patterns between groups (Table 4, Figure 3A–D). Here, CRP peaked earlier in the EUS-HGS group (day 1: 145 ± 28 mg/L) but declined more rapidly, resulting in a lower 7-day AUC (420 ± 95 vs 680 ± 120 mg·day/L; P < 0.001). Interleukin-6 trajectories showed similar patterns, with peak values of 285 ± 65 pg/mL in the EUS-HGS group and 410 ± 88 pg/mL in the PTCD group at 24 h post-procedure. Lipopolysaccharide levels, indicative of bacterial translocation, remained lower in the EUS-HGS group throughout the monitoring period (median day 3: 0.42 vs 0.78 EU/mL; P = 0.008). Bile cultures yielded positive results in 3/12 (25%) EUS-HGS animals, compared with 6/12 (50%) in PTCD animals (P = 0.041). Isolated organisms were predominantly Gram-negative (Escherichia coli 40%, Klebsiella spp. 25%, Enterobacter spp. 20%), with higher bacterial loads in PTCD specimens (median 10⁴ vs 10² CFU/mL). Positive cultures correlated with elevated inflammatory markers (r = 0.62 for CRP, r = 0.71 for IL-6) and were associated with subsequent complications (odds ratio [OR] = 4.2; 95% CI: 1.3–13.5).

Twenty-eight-day patency and survival outcomes
Kaplan–Meier analysis showed superior 28-day stent patency for EUS-HGS (91.7%, 11/12) compared with PTCD (58.3%, 7/12), with median times to dysfunction of >28 days versus 18 days, respectively (log-rank P = 0.031) (Table 5, Figure 4A). After adjusting for baseline factors, Cox regression confirmed a reduced hazard of patency loss with EUS-HGS (hazard ratio [HR] = 0.31; 95% CI: 0.12–0.78; P = 0.013). Factors independently associated with maintained patency included EUS-HGS technique (HR = 0.35), absence of bacterial colonization (HR = 0.42), and lower peak CRP (HR = 1.02 per 10 mg/L increase). The overall 28-day survival was 100% in the EUS-HGS group and 83.3% in the PTCD group (P = 0.226), with deaths attributed to septic complications (n = 1) and biliary peritonitis (n = 1) (Figure 4B). Time to clinical recovery, defined as return to baseline activity and feeding behavior, was shorter for the EUS-HGS group (median: 5 vs 8 days, P = 0.018).

Complications and procedural safety
The overall complication rate was significantly lower in the EUS-HGS group (16.7%, 2/12) than in the PTCD group (41.7%, 5/12) (P = 0.037) (Table 6). Here, EUS-HGS complications included minor bleeding managed conservatively (n = 1) and transient fever (n = 1); PTCD complications comprised bile leak requiring drainage (n = 2), catheter dislodgement (n = 1), and cholangitis (n = 2). The temporal distribution showed early complications (<72 h) predominantly with PTCD, correlating with peak inflammatory markers.

DATA AVAILABILITY:
All data generated or analyzed during this study are included in the supplementary files submitted with this manuscript.

Figure 1: Longitudinal hepatobiliary biochemical dynamics (Days 0–7). (A) Total bilirubin. (B) Direct bilirubin. (C) Alkaline phosphatase. (D) γ-Glutamyl transferase. Panels show group means ± 95% CIs for total bilirubin, direct bilirubin, alkaline phosphatase, and γ-glutamyl transferase, with individual trajectories in light grey. The insets display model-estimated half-lives (t½) for each group. Please click here to view a larger version of this figure.

Figure 2: Drainage efficiency analysis. (A) Daily drainage volume versus total bilirubin (TB) reduction scatter plot with linear mixed-effects model fitted lines and 95% CI bands for each group. (B) Drainage efficiency index (DEI = ΔTB/drainage volume) over time by group (means ± 95% CI). Please click here to view a larger version of this figure.

Figure 3: Inflammatory–endotoxin dynamics and microbiological profiles. The left panels show time-course curves for (A) C-reactive protein (mg/L), (B) interleukin-6 (pg/mL), and (C) lipopolysaccharide (EU/mL) by group. (D) Stacked bar chart showing bacterial culture results and the distribution of organisms for each group. Positive cultures: EUS-HGS 25%, PTCD 50% Please click here to view a larger version of this figure.

Figure 4: Kaplan–Meier survival analysis. (A) Twenty-eight-day stent patency curves, log-rank P values, and Cox hazard ratios. (B) Overall survival curves. Numbers at risk and cumulative event counts are displayed below each panel. Please click here to view a larger version of this figure.

Supplementary Figure 1: Intraoperative view during malignant obstructive jaundice model establishment. The common bile duct is identified and dissected from the surrounding structures within the hepatoduodenal ligament using blunt dissection with right-angle forceps.Please click here to download this file.

Supplementary Figure 2: Experimental setup for EUS-guided hepaticogastrostomy. The anesthetized pig is positioned in the left lateral decubitus position with orotracheal intubation and continuous physiological monitoring. An ultrasound system is positioned adjacent to the operating table.Please click here to download this file.

Characteristic	EUS-HGS (n=12)	PTCD (n=12)	P value
Demographics
Age, months	4.7 ± 0.5	4.9 ± 0.6	0.382
Weight, kg	39.8 ± 3.2	40.6 ± 3.6	0.571
Female, n (%)	7 (58.3)	7 (58.3)	1.000
Pre-intervention Laboratory Values
Total bilirubin, μmol/L	332.8 ± 73.2	341.2 ± 76.5	0.782
Direct bilirubin, μmol/L	191.5 ± 55.6	188.5 ± 40.2	0.878
Indirect bilirubin, μmol/L	112.0 ± 39.0	115.4 ± 40.7	0.832
Conjugated bilirubin, μmol/L	218.3 ± 54.7	222.4 ± 43.5	0.842
ALP, U/L	542.8 ± 41.6	529.4 ± 55.1	0.513
γ-GT, U/L	644.2 ± 128.2	628.7 ± 122.5	0.761
ALT, U/L	126.5 ± 37.1	120.2 ± 40.3	0.692
AST, U/L	110.1 ± 24.4	112.3 ± 27.9	0.834
Albumin, g/L	28.2 ± 3.1	27.8 ± 2.9	0.748
Hematologic Parameters
WBC, ×10⁹/L	12.4 ± 2.8	11.9 ± 3.1	0.681
Hemoglobin, g/dL	10.8 ± 1.2	11.0 ± 1.0	0.658
Platelet count, ×10⁹/L	285 ± 68	292 ± 75	0.812
INR	1.18 ± 0.12	1.20 ± 0.14	0.708
Model Characteristics
Time to intervention, hours	72.3 ± 2.4	71.8 ± 2.7	0.633
CBD diameter at obstruction, mm	12.4 ± 1.8	12.7 ± 1.6	0.667
Intrahepatic duct diameter, mm	9.2 ± 1.1	9.0 ± 1.3	0.689

Table 1: Baseline characteristics of study animals. Data are presented as mean ± SD or n (%).

Parameter	Coefficient (β)	Standard Error	95% CI	P value	Half-life (days)
Total Bilirubin
Intercept (β₀)	5.82	0.14	5.54 to 6.10	<0.001	-
Day (β₁)	-0.18	0.02	-0.22 to -0.14	<0.001	-
Group (EUS-HGS) (β₂)	-0.08	0.19	-0.45 to 0.29	0.672	-
Day × Group (β₃)	-0.12	0.03	-0.18 to -0.06	<0.001	-
EUS-HGS slope	-0.30	0.04	-0.38 to -0.22	<0.001	2.3 ± 0.4
PTCD slope	-0.18	0.03	-0.24 to -0.12	<0.001	3.8 ± 0.6
Direct Bilirubin
Intercept	5.28	0.12	5.04 to 5.52	<0.001	-
Day	-0.20	0.02	-0.24 to -0.16	<0.001	-
Group (EUS-HGS)	-0.10	0.17	-0.43 to 0.23	0.554	-
Day × Group	-0.13	0.03	-0.19 to -0.07	<0.001	-
EUS-HGS slope	-0.33	0.04	-0.41 to -0.25	<0.001	2.1 ± 0.3
PTCD slope	-0.20	0.03	-0.26 to -0.14	<0.001	3.5 ± 0.5
ALP
Intercept	6.31	0.08	6.15 to 6.47	<0.001	-
Day	-0.08	0.01	-0.10 to -0.06	<0.001	-
Group (EUS-HGS)	0.03	0.11	-0.19 to 0.25	0.788	-
Day × Group	-0.06	0.02	-0.10 to -0.02	0.003	-
EUS-HGS slope	-0.14	0.02	-0.18 to -0.10	<0.001	4.9 ± 0.8
PTCD slope	-0.08	0.02	-0.12 to -0.04	<0.001	8.7 ± 1.4
γ-GT
Intercept	6.48	0.10	6.28 to 6.68	<0.001	-
Day	-0.10	0.01	-0.12 to -0.08	<0.001	-
Group (EUS-HGS)	0.02	0.14	-0.25 to 0.29	0.884	-
Day × Group	-0.05	0.02	-0.09 to -0.01	0.012	-
EUS-HGS slope	-0.15	0.02	-0.19 to -0.11	<0.001	4.6 ± 0.6
PTCD slope	-0.10	0.02	-0.14 to -0.06	<0.001	6.9 ± 1.0
Random Effects (SD)
Individual intercept	0.24	-	-	-	-
Individual slope (EUS-HGS)	0.08	-	-	-	-
Individual slope (PTCD)	0.14	-	-	-	-
Residual	0.16	-	-	-	-

Table 2: Linear mixed-effects model coefficients for hepatobiliary markers. Coefficients are based on log-transformed values; half-lives are calculated as ln(2)/|slope|. Abbreviations; TB = total bilirubin; DB = direct bilirubin; ALP = alkaline phosphatase; γ-GT = γ-glutamyl transferase; CI = confidence interval.

Day	EUS-HGS (n = 12)	PTCD (n = 12)	Difference (95% CI)	P value
Drainage Volume (mL/24 h)
Day 1	145 ± 22	168 ± 28	-23 (-44 to -2)	0.033
Day 2	132 ± 18	155 ± 25	-23 (-41 to -5)	0.014
Day 3	118 ± 15	142 ± 22	-24 (-40 to -8)	0.005
Day 4	105 ± 12	128 ± 20	-23 (-37 to -9)	0.002
Day 5	92 ± 11	115 ± 18	-23 (-35 to -11)	0.001
Day 6	84 ± 10	102 ± 16	-18 (-28 to -8)	0.002
Day 7	78 ± 9	95 ± 14	-17 (-26 to -8)	0.001
TB Reduction (μmol/L/24 h)
Day 1	78.2 ± 14.5	52.3 ± 11.2	25.9 (15.3 to 36.5)	<0.001
Day 2	68.4 ± 12.1	45.6 ± 9.8	22.8 (13.6 to 32.0)	<0.001
Day 3	58.7 ± 10.3	38.2 ± 8.4	20.5 (12.6 to 28.4)	<0.001
Day 4	48.2 ± 8.8	31.4 ± 7.2	16.8 (10.1 to 23.5)	<0.001
Day 5	38.5 ± 7.2	25.8 ± 6.1	12.7 (7.2 to 18.2)	<0.001
Day 6	30.1 ± 5.8	20.3 ± 5.0	9.8 (5.3 to 14.3)	<0.001
Day 7	24.2 ± 4.5	16.5 ± 4.2	7.7 (4.1 to 11.3)	<0.001
Drainage Efficiency Index (μmol/L/mL)
Day 1	5.4 ± 1.2	3.1 ± 0.8	2.3 (1.5 to 3.1)	<0.001
Day 2	5.2 ± 1.0	2.9 ± 0.7	2.3 (1.6 to 3.0)	<0.001
Day 3	5.0 ± 0.9	2.7 ± 0.6	2.3 (1.7 to 2.9)	<0.001
Day 4	4.6 ± 0.8	2.5 ± 0.5	2.1 (1.5 to 2.7)	<0.001
Day 5	4.2 ± 0.7	2.2 ± 0.5	2.0 (1.5 to 2.5)	<0.001
Day 6	3.6 ± 0.6	2.0 ± 0.4	1.6 (1.2 to 2.0)	<0.001
Day 7	3.1 ± 0.5	1.7 ± 0.4	1.4 (1.0 to 1.8)	<0.001
Mean DEI (Days 1-7)	5.2 ± 1.1	3.1 ± 0.8	2.1 (1.3 to 2.9)	<0.001
Bile Bilirubin Concentration (μmol/L)
Day 1	445 ± 102	450 ± 98	-5 (-87 to 77)	0.901
Day 3	320 ± 78	385 ± 85	-65 (-132 to 2)	0.057
Day 5	240 ± 58	348 ± 80	-108 (-166 to -50)	0.001
Day 7	180 ± 45	310 ± 78	-130 (-182 to -78)	<0.001

Table 3: Drainage efficiency index (DEI) analysis (DEI = ΔTB/drainage volume per 24 h). Abbreviations; DEI = drainage efficiency index; TB = total bilirubin; CI = confidence interval.

Marker	EUS-HGS (n = 12)	PTCD (n = 12)	Difference (95% CI)	P value
C-Reactive Protein (mg/L)
Baseline (Day 0)	42.3 ± 8.5	40.8 ± 9.2	1.5 (-5.7 to 8.7)	0.675
Day 1	145.2 ± 28.4	168.5 ± 32.6	-23.3 (-48.2 to 1.6)	0.065
Day 3	78.5 ± 18.2	142.3 ± 28.9	-63.8 (-83.8 to -43.8)	<0.001
Day 5	45.2 ± 12.3	98.6 ± 22.4	-53.4 (-68.2 to -38.6)	<0.001
Day 7	28.4 ± 8.6	72.3 ± 18.5	-43.9 (-56.1 to -31.7)	<0.001
AUC₀₋₇ (mg·day/L)	420 ± 95	680 ± 120	-260 (-350 to -170)	<0.001
Interleukin-6 (pg/mL)
Baseline (Day 0)	85.2 ± 18.3	82.6 ± 20.1	2.6 (-13.1 to 18.3)	0.738
Day 1	285.4 ± 65.2	410.3 ± 88.4	-124.9 (-189.2 to -60.6)	0.001
Day 3	142.3 ± 38.5	285.6 ± 72.3	-143.3 (-192.1 to -94.5)	<0.001
Day 5	98.4 ± 28.2	198.5 ± 52.4	-100.1 (-135.8 to -64.4)	<0.001
Day 7	72.3 ± 22.1	145.2 ± 38.6	-72.9 (-99.8 to -46.0)	<0.001
AUC₀₋₇ (pg·day/mL)	1185 ± 285	1920 ± 420	-735 (-1030 to -440)	<0.001
Lipopolysaccharide (EU/mL)
Baseline (Day 0)	0.18 ± 0.05	0.20 ± 0.06	-0.02 (-0.07 to 0.03)	0.385
Day 1	0.35 ± 0.12	0.52 ± 0.18	-0.17 (-0.30 to -0.04)	0.012
Day 3	0.42 ± 0.15	0.78 ± 0.25	-0.36 (-0.53 to -0.19)	<0.001
Day 5	0.38 ± 0.14	0.85 ± 0.28	-0.47 (-0.66 to -0.28)	<0.001
Day 7	0.32 ± 0.12	0.68 ± 0.22	-0.36 (-0.51 to -0.21)	<0.001
AUC₀₋₇ (EU·day/mL)	2.45 ± 0.68	4.32 ± 1.25	-1.87 (-2.68 to -1.06)	<0.001
Bile Culture Results
Positive cultures, n (%)	3 (25.0)	6 (50.0)	-	0.041*
Bacterial load (log₁₀ CFU/mL)†	2.1 ± 0.4	4.2 ± 0.8	-2.1 (-2.8 to -1.4)	<0.001
Isolated Organisms, n (%)‡
E. coli	1 (33.3)	3 (50.0)	-	-
Klebsiella spp.	1 (33.3)	2 (33.3)	-	-
Enterobacter spp.	1 (33.3)	1 (16.7)	-	-

Table 4: Inflammatory marker values and area under the curve (days 0–7). Abbreviations; CRP = C-reactive protein; IL-6 = interleukin-6; LPS = lipopolysaccharide; AUC = area under the curve.

Variable	Events/Total	Median Patency (days)	Univariate Analysis		Multivariate Analysis
			HR (95% CI)	P value	HR (95% CI)	P value
Treatment Group
PTCD	5/12	18	Reference	–	Reference	–
EUS–HGS	1/12	>28	0.18 (0.02–1.54)	0.031	0.31 (0.12–0.78)	0.013
Baseline TB (per 50 μmol/L)	–	–	1.42 (0.98–2.06)	0.062	1.28 (0.85–1.93)	0.234
Baseline ALP (per 100 U/L)	–	–	1.18 (0.82–1.70)	0.371	1.15 (0.78–1.69)	0.482
Peak CRP (per 10 mg/L)	–	–	1.08 (1.02–1.14)	0.008	1.02 (1.00–1.04)	0.048
Bile Culture
Negative	2/15	>28	Reference	–	Reference	–
Positive	4/9	16	3.85 (1.28–11.6)	0.016	2.38 (1.05–5.40)	0.037
Procedure Time
<60 min	3/14	>28	Reference	–	Reference	–
≥60 min	3/10	22	1.92 (0.64–5.76)	0.242	1.45 (0.42–5.00)	0.554
DEI (per unit increase)	–	–	0.58 (0.38–0.88)	0.010	0.72 (0.45–1.15)	0.168
IL–6 Peak (per 100 pg/mL)	–	–	1.35 (0.95–1.92)	0.093	–	–
LPS Day 3 (per 0.5 EU/mL)	–	–	2.15 (1.12–4.13)	0.021	1.68 (0.82–3.44)	0.155
Model Statistics
Concordance index	–	–	–	–	0.82	–
Schoenfeld test P value	–	–	–	–	0.384	–

Table 5: Cox regression analysis of 28-day stent patency. Abbreviations; HR = hazard ratio; CI = confidence interval; TB = total bilirubin; ALP = alkaline phosphatase; CRP = C-reactive protein.

Complication	EUS-HGS (n = 12)	PTCD (n = 12)	P value
Overall Complications, n (%)	2 (16.7)	5 (41.7)	0.037*
Clavien-Dindo Classification, n (%)
Grade I	1 (8.3)	1 (8.3)	1.000
Grade II	1 (8.3)	2 (16.7)	0.590
Grade IIIa	0 (0)	1 (8.3)	0.500
Grade IIIb	0 (0)	1 (8.3)	0.500
Grade IV	0 (0)	0 (0)	-
Grade V	0 (0)	2 (16.7)	0.226
Specific Complications, n
Hemorrhage
Minor (no intervention)	1	0	-
Major (requiring transfusion)	0	0	-
Bile Leak
Confirmed by imaging	0	2	-
Requiring drainage	0	2	-
Infection
Fever only	1	0	-
Cholangitis	0	2	-
Sepsis	0	1	-
Mechanical
Stent/catheter migration	0	1	-
Stent/catheter occlusion	1	3	-
Timing of Complications
<24 h, n	1	2	-
24-72 h, n	1	2	-
3-7 days, n	0	1	-
>7 days, n	0	0	-
Management, n
Conservative only	2	1	-
Antibiotics	1	3	-
Percutaneous drainage	0	2	-
Revision procedure	0	1	-
Associated Factors (mean ± SD)
Peak CRP in complicated cases, mg/L	125 ± 18	215 ± 42	0.018
Peak IL-6 in complicated cases, pg/mL	245 ± 55	485 ± 95	0.012
LPS in complicated cases, EU/mL	0.45 ± 0.15	1.12 ± 0.35	0.008
Procedure time in complicated cases, min	48 ± 8	72 ± 15	0.042

Table 6: Complications by group and timing. This table summarizes the incidence and distribution of procedure-related complications in the EUS-HGS and PTCD groups according to timing of occurrence. Early complications are events that occur shortly after the procedure, whereas late complications are events identified during follow-up. Abbreviations; EUS-HGS = endoscopic ultrasound-guided hepaticogastrostomy; PTCD = percutaneous transhepatic cholangial drainage.

Discussion

This porcine model study provides mechanistic insights into the differential outcomes of EUS-HGS versus PTCD for biliary obstruction, demonstrating the endoscopic approach’s favorable performance across multiple functional domains. The accelerated bilirubin clearance kinetics observed with EUS-HGS, characterized by a 40% shorter half-life compared with PTCD, represent a potentially meaningful difference that may translate into earlier recovery of hepatic synthetic function and reduced exposure to cholestatic hepatotoxicity. These findings extend beyond previous studies reporting binary outcomes^9,10 by quantifying the temporal dynamics of biochemical improvement, providing parameters that may be useful for monitoring treatment response and identifying non-responders who might benefit from early reintervention.

From a procedural standpoint, several critical steps most strongly influence the success of EUS-HGS. Accurate identification of dilated intrahepatic ducts with adequate diameter (>8 mm) and selection of an avascular transgastric puncture trajectory are essential; failure at this step risks hemorrhage or non-target puncture. Confirmation of intraductal needle position via bile aspiration prior to guidewire insertion is a key checkpoint that prevents false-tract formation. During stent deployment, correct positioning with adequate intrahepatic purchase and appropriate intragastric protrusion (1–2 cm) is critical to prevent migration and maintain long-term patency. For PTCD, the most critical steps include selecting a peripheral duct that avoids the pleural reflection, performing adequate serial tract dilation before catheter placement, and securing external fixation to minimize dislodgement risk.

The mechanistic advantages of EUS-HGS appear to derive from fundamental differences in drainage anatomy and physiology. The transgastric approach creates a short, direct conduit between the dilated intrahepatic system and the gastrointestinal lumen, minimizing dead space where bile stasis and bacterial colonization can occur. This anatomical configuration, combined with the pressure gradient between the biliary tree and the stomach, facilitates more efficient bile clearance, as reflected in DEI measurements. The DEI concept introduced here provides a novel metric for comparing drainage procedures that accounts for both volume and biochemical response. Previous studies have focused primarily on absolute drainage volumes^19,20, but the present analysis suggests that the relationship between volume and biochemical improvement is nonlinear and procedure dependent, with EUS-HGS achieving greater TB reduction per unit volume drained. It should be noted, however, that the larger stent diameter used in EUS-HGS (an 8-mm FCSEMS) compared with PTCD (10-French catheter, approximately 3.3 mm) may have partially contributed to the observed drainage efficiency differences, and the present design does not allow complete separation of the effects attributable to the drainage route and those attributable to the stent calibre²¹.

Several common procedural problems and corresponding modifications deserve attention. During EUS-HGS, if the initial transgastric puncture fails to access a bile duct (i.e., no bile return upon aspiration), the needle should be withdrawn and repositioned under real-time EUS guidance, targeting an alternative bile duct. Multiple puncture attempts (>3) should prompt consideration of aborting the procedure to avoid hemorrhagic complications. Guidewire coiling within the biliary tree can be managed by gentle withdrawal and re-advancement under fluoroscopy or by using a stiffer guidewire. If stent deployment is incomplete or mispositioned, immediate endoscopic repositioning should be attempted using grasping forceps. For PTCD, catheter obstruction (identified by drainage output <20 mL/day) should prompt saline flushing followed by fluoroscopic evaluation; persistent obstruction requires over-the-wire catheter exchange. Catheter dislodgement can be minimized by using a suture-and-adhesive dual fixation technique and applying a protective external dressing.

The differences in inflammatory profiles between procedures offer important insights into the pathophysiology of procedure-related complications. The rapid resolution of CRP and IL-6 following EUS-HGS, despite initial elevation from tissue trauma, suggests that internalized drainage minimizes ongoing inflammatory stimulation. In contrast, the sustained elevation observed with PTCD likely reflects continued peritoneal irritation from the transcutaneous tract and external drainage system. The correlation between LPS levels and bacterial colonization rates supports the hypothesis that external drainage systems may provide a route for ascending contamination, consistent with clinical observations of higher infection rates with PTCD^22,23. Recent studies have highlighted the prognostic significance of inflammatory markers in biliary obstruction^24,25, and the present temporal analysis suggests that the trajectory of these markers, rather than their absolute values, may better predict outcomes. Mechanistic analysis further revealed that complications clustered with specific pathophysiological patterns: bile leaks coincided with persistently elevated intra-abdominal pressure and CRP > 200 mg/L, whereas infectious complications followed positive bile cultures and LPS elevation > 1.0 EU/mL.

The present model employed surgical CBD ligation with external catheter placement to induce obstructive jaundice, which differs from alternative porcine models in which biliary obstruction is achieved by endoscopic clipping of the major papilla²⁶. The papilla-clipping approach offers the advantage of being less invasive and preserving the peritoneal cavity, but may result in variable obstruction severity and potential spontaneous resolution if the clip dislodges. The surgical ligation model used here produces reliable and complete obstruction with consistent biochemical derangement, enabling standardised comparison between drainage techniques; however, it requires a laparotomy that introduces surgical inflammation as a confounder. Both models differ from the clinical scenario of tumor-induced obstruction, and the specific research question should guide the choice of model.

Several limitations must be acknowledged. First, the porcine model, although anatomically similar to humans, differs in bile composition, metabolic rate, and immune responses, which may influence drainage outcomes¹⁶. Second, the 28-day follow-up period, although sufficient for identifying early complications and patency issues, cannot capture long-term outcomes such as tumor ingrowth or delayed stricture formation observed in clinical practice^27,28. Third, although the sample size (n = 12 per group) was adequate for detecting large effect sizes in this controlled experimental setting, it limited subgroup analyses and may not have captured rare complications. Additionally, the use of healthy animals with surgically induced obstruction does not fully replicate the complexity of malignant obstruction in patients with cancer complicated with underlying hepatic dysfunction, altered coagulation, and immunosuppression^29,30. As noted above, the difference in stent diameter between the two groups represents an inherent confounder that should be addressed in future comparative studies using matched-caliber devices. Finally, as a protocol development study, representative procedural photographs are added (Supplementary Figures 1–2).

Future research directions should focus on validating these mechanistic findings in clinical populations, with attention to patient-reported outcomes and quality-of-life measures. The development of predictive models incorporating baseline characteristics, early biochemical responses, and inflammatory trajectories could enable personalized treatment selection and the timing of reintervention^31,32. Investigation of adjunctive therapies, such as biliary stent coatings, antibiotic prophylaxis regimens, and choleretic agents, may further improve outcomes. The potential for combining EUS-HGS with local tumor therapies, including photodynamic therapy or radiofrequency ablation, deserves exploration, given the direct access to the biliary tree³³. Standardization of training protocols using validated animal models could accelerate the learning curve for this technically demanding procedure and improve patient safety during the adoption phase. Additionally, operator experience, which was controlled in the current study by having a single experienced endoscopist perform all EUS-HGS procedures, represents an important variable that multicentre studies with independent operators should address to establish generalisability.

Materials

List of materials used in this article

Name	Company	Catalog Number	Comments
Automated clinical chemistry analyzer	Beckman Coulter	AU5800	Serum biochemistry analysis
Azaperone	Elanco	NDC 58198-4510	Sedation, 4.8 mg/kg IM
Blood agar plates	Thermo Fisher Scientific	R01202	Aerobic bacterial culture
Buprenorphine	Reckitt Benckiser	NDC 12496-0757	Analgesia, 0.02 mg/kg IM
Ceftriaxone	Roche	NDC 0004-1963	Antibiotic prophylaxis, 1 g IV
Chiba needle, 18-gauge	Cook Medical	DTHN-18-15.0	Percutaneous liver puncture
Chocolate agar plates	Thermo Fisher Scientific	R01206	Bacterial culture
Cystotome, 6-French	Cook Medical	CRST-6	Tract dilation for EUS-HGS
Drainage catheter, 10-French multipurpose	Boston Scientific	5523130	PTCD external drainage
EDTA collection tubes	BD Vacutainer	367835	Blood sample collection
Electrocautery unit	ERBE	VIO 300D	Electrocautery at 30 W pure-cut
Endotracheal tube, cuffed 7.0-8.0	Medtronic	86050	Orotracheal intubation
Etomidate	Par Pharmaceutical	NDC 42023-179	Induction, 0.4 mg/kg IV
Fascial dilators (6, 8, 10-French set)	Cook Medical	G00560	Serial tract dilation for PTCD
Fine-needle aspiration needle, 19-gauge	Cook Medical	ECHO-19	EUS-guided bile duct puncture
Forced-air warming blanket	3M	Bair Hugger 775	Temperature maintenance
Fully covered self-expanding metal stent (8 mm x 60 mm)	Boston Scientific	M00556270	EUS-HGS biliary stent
Graduated drainage collection system	Bard Medical	154004	Drainage volume measurement
Guidewire, 0.035-inch hydrophilic	Boston Scientific	M00556600	Biliary access
IL-6 ELISA kit, porcine-specific	R&D Systems	P6000B	Interleukin-6 measurement
Iohexol contrast medium (300 mg I/mL)	GE HealthCare	Omnipaque 300	Fluoroscopic cholangiography
IV catheter, 20-gauge	BD	381134	Peripheral venous access
Ketamine	Dechra	NDC 17033-010	Maintenance anesthesia, 10 mg/kg/h
LAL chromogenic assay kit	Lonza	QCL-1000	Lipopolysaccharide measurement
Linear echoendoscope	Olympus	GF-UCT260	EUS-guided procedures
MacConkey agar plates	Thermo Fisher Scientific	R01552	Gram-negative bacterial culture
MALDI-TOF mass spectrometer	Bruker	MALDI Biotyper	Bacterial identification
Mechanical ventilator	Draeger	Primus	Intraoperative ventilation
Meloxicam	Boehringer Ingelheim	NDC 0010-4491	Analgesia, 0.4 mg/kg PO
Midazolam	Hikma	NDC 0641-6024	Sedation/maintenance, 0.6 mg/kg
Nylon suture, 3-0	Ethicon	663G	Skin closure and catheter fixation
Pancuronium	Organon	NDC 0052-0441	Muscle relaxation, 0.12 mg/kg/h
Pantoprazole	Pfizer	NDC 0008-0841	Proton pump inhibitor, 40 mg IV
Patient monitor (SpO2, ECG, capnography, BP)	Philips	IntelliVue MX800	Continuous physiological monitoring
Povidone-iodine solution	Purdue Products	NDC 67618-150	Antiseptic skin preparation
R statistical software, version 4.3.1	R Foundation	N/A	Statistical analysis (packages: lme4, survival, ggplot2)
Rotational viscometer	Brookfield	DV2T	Bile viscosity assessment
Serum separator tubes	BD Vacutainer	367986	Blood sample collection
Silk suture, 3-0	Ethicon	SA85G	Bile duct ligation
Silicone catheter, 6-French	Cook Medical	PERC-6-38	External bile drainage catheter
Spectrophotometer	Thermo Fisher Scientific	NanoDrop 2000	Bile bilirubin at 454 nm
Ultrasound system	Canon Medical	Aplio i800	PTCD imaging guidance