Circulating six-microRNA signature for diagnosis, prognosis, and surgical monitoring in extrahepatic cholangiocarcinoma: a multicenter, multi-cohort study

Robust blood biomarkers for the preoperative diagnosis and perioperative assessment of extrahepatic cholangiocarcinoma (eCCA) are lacking. We aimed to develop and validate plasma microRNA signatures for diagnosing eCCA and to explore their perioperative and prognostic relevance. This multicenter biomarker study included 474 participants and 514 plasma specimens across one derivation, one discovery, three internal validation, and two independent external validation cohorts. Candidate microRNAs were prioritized using small-RNA sequencing and random-forest analysis. Reverse-transcription quantitative polymerase chain reaction and logistic regression were used to develop a weighted six-microRNA model (Signature I) and an unweighted score (Signature II). Discovery-derived formulas and thresholds were locked before validation. Paired preoperative and postoperative-day-7 samples were used to assess perioperative changes. Multivariable Cox models adjusted for pathological T and N stages, resection-margin status, and tumor differentiation evaluated prognostic associations. Signature I achieved an area under the receiver operating characteristic curve (AUC) of 0.90 (95% confidence interval [CI], 0.84–0.97) for distinguishing eCCA from common bile duct stones in both internal and external disease-control cohorts, compared with AUCs of 0.77 and 0.76, respectively, for CA19-9 (DeLong P = 0.041 and P = 0.033, respectively). At the prespecified threshold, its sensitivity and specificity in the external disease-control cohort were 82.5% and 80.0%, respectively. In the external healthy-control cohort, Signature I achieved an AUC of 0.89 (95% CI, 0.81–0.96), while the corresponding AUC for Signature I plus CA19-9 was 0.91. Signature II showed comparable discrimination. The perioperative decrease in Signature I was greater after R0 than after R1 resection ( P = 0.045). In the internal prognostic cohort, high preoperative Signature I was independently associated with shorter overall survival (HR, 2.15; 95% CI, 1.02–4.54) and recurrence-free survival (HR, 2.29; 95% CI, 1.14–4.60). High Signature I also stratified overall and recurrence-free survival in the pooled external cohort (log-rank P = 0.007 and P = 0.004, respectively). Signature I showed consistent diagnostic discrimination across centers and clinically distinct control groups and warrants further evaluation as an adjunct to preoperative assessment of suspected, potentially resectable eCCA. Its perioperative and prognostic associations require prospective confirmation. The incremental value of CA19-9 was context dependent.