Cyst Detection Ratio, a Quality Indicator for EUS-Based Pancreatic Cancer Screening: A Systematic Review and Meta-Analysis to inform Quality Improvement Benchmark

Aims Endoscopic ultrasound (EUS) is a safe and effective surveillance tool for pancreatic cancer in high-risk individuals (HRIs). However, variability in its diagnostic yield underscores the need for quality improvement through standardized screening and reporting. Pancreatic cysts, considered precursors to pancreatic cancer, make the cyst detection rate (CDR)—the proportion of EUS procedures identifying at least one pancreatic cystic lesion (PCL)—a potential quality indicator (QI). This systematic review and meta-analysis assess the feasibility of using CDR as a QI for EUS-based pancreatic cancer screening.

Methods We identified and assessed the suitability of pre-malignant pancreatic lesions noted on EUS screening examinations to serve as intermediate-outcome QIs. Using an iterative process, we determined that a pancreatic cyst-based QI would be optimal. We then conducted a systematic search (CRD42018117189) of Cochrane Library, Embase, Google Scholar, Medline, PubMed, Scopus, and Web of Science of pancreatic cancer screening studies to determine the prevalence, characteristics, and malignant potential of pancreatic cystic lesions (PCLs) in high-risk individuals (HRIs). Cumulative effect sizes were calculated through the meta-analysis of proportions using a random-effects model and expressed as pooled proportions (95% confidence interval).

Results Nineteen cohort studies included 3,281 high-risk individuals (HRIs), with a pooled prevalence of pancreatic cystic lesions (PCLs) at 32% (95% CI: 25%-39%; I²=95.6). Among these, PCLs with worrisome features—defined as cyst size≥3 cm, thickened or enhancing cyst wall, mural nodule, abrupt pancreatic ductal cutoff, rapid cyst growth, or main pancreatic duct caliber≥5 mm—had a prevalence of 3.4% (95% CI: -2.0%-8.8%; I²=0). PCLs with worrisome features were twice as likely to progress to advanced neoplasia compared to those without (odds ratio 2.02; 95% CI: 1.08-2.96; I²=0). The overall prevalence of advanced neoplasia arising from PCLs was 4% (95% CI: -1.7%-9.7%; I²=0), accounting for a substantial proportion [40% (95% CI: 16%-64%; I²=0)] of all advanced neoplasia.

Conclusions PCLs are bonafide precursor lesions for advanced neoplasia in HRIs. Given its prevalence of 32% in an ideal sample, we propose Cyst Detection Ratio,, with a target benchmark of 10%, as an EUS-based QI for pancreatic cancer screening. Incorporating CDR as a QI could standardize reporting, enhance screening quality, and improve outcomes in pancreatic cancer surveillance programs.

Publication History

Article published online:
27 March 2025