# Risk factors for mortality in patients with primary biliary cholangitis: a nomogram to predict 5-year survival

## TL;DR

This study develops a model to predict 5-year survival in primary biliary cholangitis patients using age, NLR, and TBIL, improving clinical decision-making.

## Contribution

A new nomogram for 5-year survival prediction in PBC patients using routine clinical data, outperforming existing models.

## Key findings

- A nomogram incorporating Age, NLR, and TBIL showed good discrimination with AUC-ROC values of 0.7251 and 0.7721 in training and validation cohorts.
- The model outperformed the GLOBE model and demonstrated strong calibration and clinical utility.
- NLR and TBIL were significantly correlated (P=0.0021), suggesting combined impact on prognosis.

## Abstract

The issue of transplant-free survival rate (OS) among patients with primary biliary cholangitis (PBC) remains a persistent concern. In predicting the long-term OS of PBC patients, given the complexity and population specificity of models such as the GLOBE and UK-PBC, our objective is to calculate and assess the risk factors for mortality and 5-year OS among PBC patients based on routine clinical data, ultimately facilitating its clinical application.

This study enrolled 315 patients with PBC from Beijing Ditan Hospital and randomly divided them into a training cohort (n = 189) and a validation cohort (n = 126). Through Cox regression analyses, we identified risk predictors of mortality to develop a 5-year survival nomogram for PBC. The model was evaluated with Receiver Operating Characteristic (ROC) curves, calibration curves, Decision Curve Analysis (DCA).Kaplan-Meier (KM) curves compared OS across risk groups. Additionally, correlations among the indicators were analyzed.

Ultimately, we established a nomogram incorporating Age, NLR, and TBIL. The Area Under the ROC Curve(AUC-ROC) values for the training and validation groups were 0.7251 and 0.7721, respectively, indicating solid consistency and outperforming the GLOBE model. Calibration and DCA curves further underscored the clinical utility of our model.KM curves revealed the model could differentiate OS across risk levels in subgroup. Additionally, a significant correlation between NLR and TBIL (P=0.0021) was observed, potentially impacting patient prognosis.

We have constructed a well-performing prognostic model based on Age, NLR, and TBIL. This model shows good discrimination, consistency, and clinical use. It helps identifying high-risk patients, enabling more frequent follow-ups and tailored interventions, potentially enhancing prognosis and clinical outcomes.