Background
Early diagnosis of pancreatic cancer (PC) remains challenging, particularly in patients with chronic pancreatitis (CP). Currently, imaging is often inaccurate and biopsy is inherently invasive. This is the first network meta-analysis (NMA) comparing blood-based biomarkers for differentiating PC from CP.
Methods
PubMed and EMBASE were searched for studies evaluating blood-based biomarkers for distinguishing PC from CP. Risk of bias was assessed with QUADAS-2. We conducted individual biomarker meta-analysis with generalized bivariate models. For the NMA, we applied a Bayesian inference approach via a Markov-chain Monte Carlo random effects model. The surface under the cumulative ranking curve was used to rank the diagnostic performance of the biomarkers.
Results
Across 139 enrolled studies, 49 biomarkers or panels were analyzed. For differentiating PC from CP, tumor polypeptide antigen (TPA) had the highest area under the summary receiver operating characteristic curve (AUSROC) (0.92). The NMA revealed that microRNA-196b (miR-196b) ranked highest in sensitivity (OR 3.74e+27), while the combination of carbohydrate antigen 19–9 (CA199) and a panel of eight extracellular vesicle long RNAs (exLRs) exhibited the highest specificity (OR 3.78). For early-stage (stage I and II) PC, the eight exLRs showed the highest relative sensitivity (OR 7.00), and carcinoembryonic antigen (CEA) demonstrated the highest specificity (OR 4.70).
Conclusion
CA199 demonstrated only moderate diagnostic discrimination between PC and CP, with reduced efficacy in early-stage PC. Combining CA199 and eight exLRs exhibited promising differential diagnostic efficacy with both high sensitivity and specificity.
扫码关注我们
求助内容:
应助结果提醒方式:
