CA19-9/DBil: a promising indicator to distinguish between CA19-9-elevated pancreatic head-type autoimmune pancreatitis and pancreatic head cancer.

BACKGROUND

Pancreatic head-type autoimmune pancreatitis (PH-AIP) with elevated CA19-9 is sometimes difficult to distinguish from pancreatic head cancer (PHC) with elevated CA19-9. At times, IgG4 proves inadequate in offering assistance. The study aimed to elucidate the performance of CA19-9/DBil in distinguishing between the two conditions.

METHODS

This was a retrospective study. We collected serologic indicators from participants in PH-AIP and PHC Group. Three logistic regression equations were established ranging from non-adjustment (Model 1, only CA19-9/DBil included) to adjusting for sex, age, and CEA (Model 2 and Model 3) to explore the relationship between CA19-9/DBil and PH-AIP probability. ROC, Decision Curve Analysis (DCA), calibration curve were conducted. P for AUCs and net reclassification improvements (NRI) were computed to evaluate differences in discrimination and the improvement in risk reclassification between models.

RESULTS

The study included 90 PHC and 35 PH-AIP patients, all with elevated CA19-9. The ORs for CA19-9/DBil in three models were similar (0.915 to 0.921). ROC revealed that Model 1 had an AUC of 0.772. The sensitivity, specificity, and accuracy at the best threshold were all > 0.7. Model 1, although simple, was not inferior in its discriminative ability compared to complex models: the difference in discrimination between Model 1 and each of two adjusted models was not statistically significant (P > 0.05, both AUC and NRI). Additionally, calibration curve and DCA suggested that Model 1 had good calibration and clinical utility.

CONCLUSIONS

CA19-9/DBil exhibited promising diagnostic performance in differentiating between CA19-9-elevated PH-AIP and PHC.