Comparison of Comprehensive Serum miRNA Sequencing and Apolipoprotein A2 Isoforms for Early Detection of Pancreatic Cancer

Backgrounds and Aim: Pancreatic cancer is frequently diagnosed at advanced stages, highlighting the need for biomarkers that are capable of detecting early-stage disease in asymptomatic individuals. Recently, apolipoprotein A2 isoforms (ApoA2-ATQ/AT) have been reported as a new blood biomarker for pancreatic cancer. We recently developed diagnostic models based on 100 highly expressed serum microRNAs (miRNAs) combined with CA19-9; these models achieved high accuracy in terms of distinguishing individuals with pancreatic cancer from healthy individuals. This study aimed to compare the diagnostic performance of these miRNA-based models with that of the ApoA2-ATQ/AT biomarker. Methods: Comprehensive sequencing of serum miRNAs was conducted using samples from 120 pancreatic cancer patients recruited across 14 hospitals, along with 93 healthy controls without cancer. Serum CA19-9 levels, miRNA index values, miRNA+CA19-9 index values, and ApoA2 index values were assessed. miRNA-based indices were derived from classification models built on an automated machine-learning platform. Results: The miRNA model (AUC 0.94; 95% CI 0.91–0.97) and the miRNA+CA19-9 model (AUC 0.99; 95% CI 0.98–1.00) outperformed ApoA2 (AUC 0.89; 95% CI 0.84–0.93) in terms of distinguishing individuals with pancreatic cancer from healthy controls across all stages. In early-stage disease (stages 0–I and 0–II), both miRNA-based models also demonstrated superior performance. Strong negative correlations were observed between the ApoA2 index and both the miRNA model index (r = −0.62) and the miRNA+CA19-9 index (r = −0.63). Conclusions: These findings suggest that miRNA-based diagnostic models, particularly when combined with CA19-9, could serve as powerful tools for the early detection of pancreatic cancer.