Introduction Cholangiocarcinoma (CC) is a malignant neoplasm of the bile duct. Diagnosis of CC is hampered by the inadequate performance of current plasma markers of disease, particularly in patients with preexisting primary sclerosing cholangitis (PSC). We aimed to identify potential new protein biomarkers of CC

Methods In an initial discovery study, blood plasma samples from 18 subjects with CC, 17 with PSC and 10 healthy controls were subjected to SELDI-TOF MS. Comparisons of m/z peak intensity were made between groups using the Mann-Whitney U test. Differentiating m/z peaks were then confirmed in a further validation study of 81 subjects with CC, 54 with PSC and 90 healthy controls. Pearson´s correlation was used to investigate the relationship of each m/z peak´s intensity to routine laboratory indices. Diagnostic performance was investigated using receiver operator characteristic area-under-the curve (ROC-AUC) analyses. Multiple linear regression was used to investigate the performance of combinations of differentiating m/z peaks, as well as the combination of m/z peaks with routine laboratory markers (including CA19–9).

Results Seven differentially expressed m/z peaks were identified in the CC group and these were subsequently confirmed in the validation study (p = 2.6 × 10–4 to 9.4 × 10–13). The intensity of the seven m/z peaks of interest did not correlate with creatinine, ALP, bilirubin, CRP, white cell count or CA19–9. A panel of three peaks discriminated CC from PSC subjects with ROC-AUC of 0.76 (sensitivity 75%, specificity 64%). A panel of five peaks discriminated CC subjects from healthy controls with ROC-AUC of 0.90 (sensitivity 95%, specificity 74%). Addition of routine laboratory indices did not change the diagnostic performance of these models significantly.

Conclusion SELDI-TOF has been used to successfully identify seven m/z peaks that are differentially intense in CC subjects (total n = 99), when compared to PSC subjects (n = 64) and healthy controls (n = 107). These peaks appear to be independent of standard markers of renal impairment, cholestasis, sepsis and inflammation, as well as CA19–9. Individually, and more so in combination, these peaks exceed the expected diagnostic performance of CA19–9, particularly in discriminating CC from PSC. Work to identify the proteins represented by these m/z peaks is ongoing.

Disclosure of Interest None Declared