Development and Exploratory Evaluation of an ANXA1 and ANXA2 Expression Score in Intrahepatic Cholangiocarcinoma

Background/Objectives: We assessed whether an ANXA1–ANXA2 expression score stratifies survival in intrahepatic cholangiocarcinoma (iCCA). Methods: We assessed the discovery model’s internal performance and examined independent cohorts, single-cell expression and immunohistochemistry (IHC). Results: The discovery cohort included 244 patients and 99 deaths. Bootstrap-corrected 1- and 3-year areas under the curve (AUCs) were 0.712 and 0.629, respectively, for the three-candidate procedure; broader screening retained an additional gene. TCGA-CHOL was inconclusive (36 tumours, 18 deaths; hazard ratio per score standard deviation 1.00, 95% CI 0.57–1.77). Cohort-standardised microarray analysis of 81 iCCA patients with 40 deaths supported a continuous association (hazard ratio 1.85, 95% CI 1.30–2.63). Median-split survival separation was not significant. Single-cell analysis included 19,278 cells from ten libraries and nine patients, with immune cells present in the initial epithelial-enriched selection. Within-patient ANXA2–oxidative-phosphorylation associations were positive, but none of the 16 patient-pseudobulk tests passed false discovery rate (FDR) correction. IHC showed descriptive differences in three matched tumour–adjacent pairs per marker, using different patient sets and no inferential testing. Conclusions: The score remains exploratory. Interpretation depends on candidate selection, differences between assays and patient-level uncertainty. Clinical utility and causal mechanisms require further evaluation.