Abstract:

Objective Post-renal transplantation anemia（PTA）occurs frequently in kidney transplant recipients，significantly impacting their quality of life and graft loss. Currently，effective methods to predictthe risk of persistent PTA early post-transplantation are lacking. This study aimed to develop a nomogram prediction model for early persistent PTA specifically tailored to kidney transplant recipients. Methods Using the electronic medical record system of Southern Hospital of Southern Medical University，patient data from January 1，2020 to December 31，2022 were obtained，and 245 subjects were ultimately selected as the research subjects. Among these，85% were randomly selected as the training set for model development，and the remaining 15% constituted the testing set. Using the Least Absolute Shrinkage and Selection Operator（Lasso）regression model，variables potentially affecting early persistent PTA were screened to identify predictive factors.A logistic regression analysis was employed to establish the prediction model. Model performance was assessed using Receiver operating characteristic（ROC）curves，area under the curve（AUC），Calibration plots，and decision Curve Analysis（DCA）. Results Identified predictive factors after screening included recipient's preoperative body mass index，preoperative serum albumin level，preoperative hemoglobin level，preoperative mean corpuscular volume，perioperative use of angiotensin-converting enzyme inhibitors or angiotensin receptor blockers，exogenous iron supplementation，and exogenous erythropoietin supplementation. The model demonstrated good discriminativeability with an AUC of 0.87 for the training set and 0.75 for the testing set，indicating robust predictive performance. Calibration and DCA further confirmed the accuracy and clinical utility of the model. Conclusion This nomogram prediction model utilizes early recipient information，including demographic characteristics，laboratory data，and medication regimens，to accurately predict individualized risk of early persistent PTA in kidney transplant recipients. This provides a basis for early clinical intervention，potentially improving patient prognosis and quality of life. 

Key words:

Post-renal transplantation anemia , Nomogram , Binary logistic regression , Clinical prediction model

摘要：

目的 肾移植术后贫血（post-renal transplantation anemia，PTA）在肾移植受者中的患病率较高。长期持续的 PTA 不仅显著影响受者的生活质量，还对移植肾的存活产生负面影响。然而，目前尚缺乏有效预测受者术后早期持续性 PTA 风险的方法。本研究旨在开发针对肾移植受者群体的术后早期持续性 PTA 的列线图预测模型。 方法 利用南方医科大学南方医院电子病历系统，获取 2020 年 1 月 1 日至2022 年 12 月 31 日期间的患者资料，最终选取了 245 例受者作为研究对象。在这些受者中，85% 的受者作 为训练集开发模型，余下 15% 的受者作为验证集。运用最小绝对值收缩和选择算法回归模型（least absoluteshrinkage and selection operator，Lasso）对可能影响早期持续性 PTA 发生的变量进行筛选而获得预测因素，并使用二元 Logistic 回归分析建立预测模型，使用受试者工作特征曲线（receiver operating characteristic，ROC）曲线、ROC 曲线下面积（area under curve，AUC）、Calibration 校准曲线及决策曲线分析（decision curve analysis，DCA）评估模型性能。 结果 经过 Lasso 回归筛选后获得的预测因素包括 ：受者术前体重指数（body mass index，BMI）、受者术前血清白蛋白水平、受者术前血红蛋白水平、受者术前平均红细胞体积、受者围手术期是否使用血管紧张素转换酶抑制剂或血管紧张素受体阻滞剂类降血压药物、受者是否外源补充铁剂、受者是否外源补充促红细胞生成素。该模型显示出良好的区分度，训练集的 AUC 值为 0.87，验证集的 AUC 值为 0.75，表明模型的预测性能较好，且 Calibration 校准曲线和 DCA 曲线进一步证明了模型具有准确性和临床实用性。 结论 该列线图预测模型利用早期可获得的受者信息，包括受者特征、实验室检验数据及用药方案，能够准确预测肾移植受者术后早期持续性 PTA 的个体化风险。这为早期临床干预提供了重要依据，有助于改善患者的预后及生活质量。 

关键词:

肾移植术后贫血 , 列线图 , 二元 Logistic 回归 , 临床预测模型