Abstract
The determinants of roxadustat treatment failure in renal anemia remain elusive. This study sought to develop a nomogram for predicting the risk of treatment failure of roxadustat in peritoneal dialysis (PD) with renal anemia. A retrospective cohort analysis from January 1, 2019, to January 31, 2023, included 204 PD patients with renal anemia, stratified by attainment group (Hb ≥ 110 g/L, n = 103) or non-attainment (Hb < 110 g/L, n = 101) within 1 year treatment. Univariate and multivariate Cox proportional hazards regressions were employed to ascertain predictive factors and construct the nomogram. Nomogram efficacy was evaluated via C-index, time-dependent ROC, calibration plots, and decision curve analysis, with internal validation via tenfold cross-validation and 1000 bootstrap resampling iterations. The study identified PD duration, serum transferrin, cardiovascular comorbidities, and stains as significant predictors. The nomogram demonstrated moderate discrimination at 6 months (AUC: 0.717) and enhanced predictive accuracy at 12 months (AUC: 0.741). The predicted and actual risk probabilities were concordant, with clinical net benefits observed at six-month (8 to 53%) and twelve-month (27 to 84%) risk thresholds. This nomogram is a valuable tool for effectively predicting non-attainment risk and facilitating personalized management of renal anemia in PD patients treated with roxadustat.