Emerging Technologies in Pancreatic Cyst Diagnosis

Pancreatic cystic lesions (PCLs) are emerging as a common incidental finding, fueled by the routine use of imaging in aging populations. Differentiating benign from premalignant or malignant cysts remain a major challenge, as conventional imaging and cytology provide limited sensitivity and specificity. We reviewed emerging diagnostic modalities including radiomics, contrast-enhanced harmonic endoscopic ultrasound (CH-EUS), confocal laser endomicroscopy (CLE), EUS-guided microforceps biopsy, and cyst fluid biomarker analysis. Evidence from multicenter trials, validation studies, and meta-analyses was synthesized to evaluate diagnostic performance, reproducibility, and limitations. Current first-line tools—cross-sectional imaging and EUS-FNA cytology—are limited (EUS morphology sensitivity ~50%; cytology ~60%; cyst fluid CEA >192 ng/mL, sensitivity 58%, specificity 87%). Radiomics models integrating CT/MRI features with clinical data achieved 76–92% accuracy for cyst subtype classification and malignant risk stratification, though external validation is lacking. CH-EUS improved mural nodule detection (sensitivity up to 97%, specificity 75%), outperforming CT (77%) and MRI (79%). Needle-based CLE provided real-time histology-like imaging; multicenter trials reported >90% sensitivity and specificity for mucinous vs. nonmucinous cysts, with accuracy superior to cyst fluid CEA. Interobserver agreement varied (κ 0.66–0.82), but deep learning networks applied to CLE videos achieved >90% classification accuracy. EUS-guided microforceps biopsy showed pooled sensitivity of 76.6% and specificity of 98.9%. Cyst fluid biomarkers demonstrated promise: glucose <50 mg/dL (93% sensitivity, 89% specificity), GNAS/KRAS mutations (90% sensitivity, 100% specificity), and combined molecular panels (88%/98% for advanced neoplasia). DNA methylation, miRNA panels, and glycoproteomics further enhanced risk stratification but remain investigational. Emerging technologies are reshaping PCL diagnostics. Radiomics, CH-EUS, CLE, microforceps biopsy, and molecular biomarkers provide complementary strengths. Their integration into standardized algorithms, supported by multicenter validation, will be essential to optimize patient outcomes.