Hilar malignant strictures are challenging to diagnose. Current guidelines recommend ERCP-based tissue acquisition as the first-line modality for perihilar strictures requiring biliary drainage, yet its sensitivity remains suboptimal and often necessitates multiple procedures. Endoscopic ultrasound-guided fine-needle biopsy (EUS-FNB) can provide direct access to perihilar lesions and regional lymph nodes. It can obtain high-quality histological cores, but its use has historically been limited by concerns about tumour seeding, based on small, outdated studies. With modern FNB needle technology, diagnostic performance has substantially improved, yet contemporary evidence specifically addressing hilar lesions remains limited. To address this gap, we conducted a multicentre study to assess the diagnostic performance and safety profile of EUS-FNB as a first-line tool in the diagnostic approach to suspected hepatic hilum malignancies.

A retrospective analysis was performed on all consecutive patients undergoing EUS-FNB as a first-line tool for perihilar strictures and lymph nodes from October 2014 to April 2025 at four tertiary centres. All patients were monitored after EUS-FNB through clinical and radiological follow-up for ≥6 months or until death. Final diagnosis was based on surgical pathology or a composite reference standard (cytology/histology and clinical–radiological follow-up). Histological adequacy was documented for each biopsy, and descriptive statistical analyses of baseline characteristics, procedural variables, and lesion features were conducted. Diagnostic performance was assessed using intention-to-diagnose (ITD), per-protocol (PP), and full-pathway analyses, including second-look EUS-FNB when performed. Sensitivity, specificity, positive and negative predictive values (PPV, NPV), and accuracy were calculated with 95% confidence intervals. Diagnostic yield, incremental yield after repeat biopsy, and number-needed-to-repeat (NNT-repeat) were also determined. Adverse events were recorded according to ASGE lexicon criteria.

A total of 144 patients were included (mean age 69.1±10.8 years; 62.3% males). Mean lesion size was 29.1±13.6 mm. A biliary stent was present at the time of the first EUS in 35 patients (24.3%). Histological adequacy at first EUS-FNB was achieved in 136/144 patients (94.4%), while 8 samples (5.6%) were inadequate. The final diagnosis was malignant in 121 (84.0%) patients, with cholangiocarcinoma being the most common cause. In the ITD analysis, first-attempt EUS-FNB showed a sensitivity of 81.8% (95% CI 74.0–87.6), specificity of 100% (95% CI 82.2–100.0), and overall accuracy of 84.7% (95% CI 77.8–89.6). Excluding inadequate samples, the per-protocol sensitivity rose to 86.0% (95% CI 78.5–91.1), with an accuracy of 88.2% (95% CI 82.1–92.6). A second EUS-FNB was performed in 27 patients following an initial non-diagnostic or negative result, achieving 100% histological adequacy. Four additional true positives were identified, resulting in an incremental yield of 14.8% and a NNT-repeat of 6.75. When repeat sampling was included, the complete diagnostic pathway showed a sensitivity of 85.1% (95% CI 77.7–90.4) and an accuracy of 87.5% (95% CI 81.4–91.8). Adverse events occurred in 7 patients (4.9%), none of which were severe or procedure-related, and there was no procedure-related mortality. The presence of a biliary stent significantly reduced diagnostic yield (p=0.03), whereas lesion size, needle type and number of passes had no impact.

EUS-FNB provides high diagnostic accuracy and a favourable safety profile for suspected hilar biliary malignancies. These findings support its integration into the diagnostic pathway for hilar lesions.