A nomogram based on pretreatment clinical parameters for the prediction of inadequate biochemical response in primary biliary cholangitis

Abstract

Background: Ursodeoxycholic acid (UDCA) has been widely recommended as the first-line drug for primary biliary cholangitis (PBC) in the current guidelines. However, its therapeutic effects are poor in nearly one-third of patients. The early identification and intervention of these patients is crucial for delaying disease progression. Therefore, we explored risk factors for inadequate biochemical response and constructed a nomogram to predict the potential risk.

Methods: We enrolled 356 patients and randomly divided them into training (70%) and validation groups (30%). We defined inadequate biochemical response as the study endpoint. Logistic analysis was used to identify the independent predictors of poor biochemical response. Based on these factors, a predictive nomogram was finally constructed. Then, discrimination and calibration were evaluated by internal validation. Additionally, the association between the model predictions and prognosis was further analyzed.

Results: Female sex, and albumin and bilirubin concentrations were identified as risk factors, and a nomogram was built based on these factors. The areas under the ROC curves of the training and validation groups were 0.809 and 0.791, respectively. Moreover, calibration curves showed that predictions of the nomogram had good concordance with the actual outcomes. The correlation analysis demonstrated that PBC patients with a high probability of a suboptimal biochemical response were more likely to have adverse outcomes.

Conclusion: We constructed a nomogram, which can accurately predict the risk of inadequate biochemical response to UDCA, facilitating the early screening of high-risk patients with PBC who should be prioritized for additional therapy.

Keywords: autoimmune liver disease; biochemical response; nomogram; primary biliary cholangitis; ursodeoxycholic acid.

Figure 1

Nomogram to predict the probability of inadequate biochemical response to ursodeoxycholic acid in PBC patients. ALB, albumin; LLN, lower limit of normal; TBiL, total bilirubin; ULN; upper limit of normal

Figure 2

Receiver operating characteristic (ROC) curves for the predictive model. A, ROC curve in the training group. B, ROC curve in the validation group. (AUC, area under the ROC curve)

Figure 3

Calibration curves for the predictive model. A, Calibration curve in the training group (χ² = 14.539, P > .05). B, Calibration curve in the validation group (χ² = 4.526, P > .05)

Figure 4

Correlations between the nomogram predictions and long‐term prognosis. A, B, and C, Correlations between the probability of inadequate biochemical response and the predicted liver transplant‐free survival by the GLOBE score. (D, E, and F) Correlations between the probability of inadequate biochemical response and the predicted risk of liver transplant or death by the UK‐PBC risk scores