Work

Staging laparoscopy for radiographically non-metastatic pancreatic ductal adenocarcinoma: cytology-driven upstaging and the SENDAI-SL risk-enrichment score

Year: 2026

Type: article

Abstract: BACKGROUND: Cross-sectional imaging may miss occult metastatic disease in pancreatic ductal adenocarcinoma and cannot detect positive peritoneal lavage cytology. This study quantified the yield of staging laparoscopy (SL) with routine lavage cytology and developed the SENDAI-SL score for risk-based prioritization. METHODS: Consecutive patients with radiographically non-metastatic pancreatic ductal adenocarcinoma who underwent SL at one centre from January 2013 to June 2025 were analysed retrospectively. SL positivity was defined as liver metastasis, macroscopic peritoneal metastasis, and/or positive cytology. Multivariable logistic regression identified predictors; internal validation used 2000 bootstrap resamples. RESULTS: Of 504 patients, 141 (28.0%) had positive SL. Positivity rates were 21.3%, 16.0%, 32.3%, and 41.7% in resectable, borderline resectable with portal vein involvement, borderline resectable with arterial involvement, and unresectable locally advanced disease, respectively. Gross liver or peritoneal metastases were detected in 51 patients (10.1%); cytology was positive in 125 (24.8%) and the sole finding in 90 of 141 positive procedures (63.8%). Among 499 patients with quantifiable serum carbohydrate antigen 19-9, unresectable locally advanced disease, body and/or tail location, tumour size ≥ 37 mm, and serum carbohydrate antigen 19-9 ≥ 180 U/ml were independently associated with SL positivity. The model's area under the receiver operating characteristic curve was 0.732 (optimism-corrected, 0.718). Score-specific positivity ranged from 8.6% to 83.3%. A score ≥ 2 prioritized 149 patients; 74 (49.7%) were positive, capturing 52.9% (74 of 140) of positives in the analysis cohort. CONCLUSIONS: SL upstaged 28.0% of patients, predominantly through cytology-only disease. The SENDAI-SL score may support prioritization when resources are limited but requires external validation and should not justify omitting SL.

Source: BJS Open

DOI: 10.1093/bjsopen/zrag176