| 邱娟,何璇.基于 LASSO-logistic回归分析建立多发性骨髓瘤自体造血干细胞移植后早期急性肾损伤预测模型及其价值验证[J].安徽医药,2026,30(9):1836-1842. |
| 基于 LASSO-logistic回归分析建立多发性骨髓瘤自体造血干细胞移植后早期急性肾损伤预测模型及其价值验证 |
| Development and validation of a LASSO-logistic regression-based prediction model for early acute kidney injury after autologous hematopoietic stem cell transplantation in patients with multiple myeloma |
| |
| DOI：10.3969/j.issn.1009-6469.2026.09.026 |
| 中文关键词: 多发性骨髓瘤 自体造血干细胞移植 急性肾损伤 LASSO-logistic回归分析 预测模型 |
| 英文关键词: Multiple myeloma Autologous hematopoietic stem cell transplantation Acute kidney injury LASSO-logistic regres. sion analysis Prediction model |
| 基金项目:四川省医学青年创新科研课题计划( Q20158) |
|
| 摘要点击次数: 243 |
| 全文下载次数: 43 |
| 中文摘要: |
| 目的基于 LASSO-logistic回归分析建立多发性骨髓瘤自体造血干细胞移植后早期发生急性肾损伤的预测模型，并进行验证。方法回顾性研究。选取 2021年 1月至 2023年 12月于四川大学华西医院接受自体造血干细胞移植的 260例多发性骨髓瘤病人，按照 7∶3的比例，采用随机数字表法将其分为建模集( 182例)与验证集( 78例)另根据建模集病人移植后早期是否发生急性肾损伤将其分为损伤组(57例)和未损伤组( 125例)。收集所有研究对象临床资料，通，过 LASSO回归分析筛选自变量，确定最佳 λ值；进一步通过多因素 logistic回归分析确定多发性骨髓瘤病人自体造血干细胞移植后早期发生急性肾损伤的影响因素，使用 R软件构建相关风险预测模型，并以受试者操作特征曲线(ROC曲线)、校准曲线及决策曲线对模型进行评估和验证。结果 260例多发性骨髓瘤病人自体造血干细胞移植后早期发生急性肾损伤共有 85例，发生率为 32.69%。损伤组病人年龄 >50岁占比［38.60%(22/57)比 20.80%(26/125)］、高钙血症占比［33.33%(19/57)比 16.00%(20/125)］、临床分期 Ⅱ~Ⅲ期占比［64.91%(37/57)比 45.60%(57/125)］、血肌酐水平［( 103.25±16.91)μmol/L比( 62.07±10.58)μmol/L］、胱抑素 C(Cys C)水平［( 3.05±0.77)mg/L比( 1.54±0.31)mg/L］、白细胞介素 -18(IL-18)水平［( 214.91±38.73)ng/L比( 116.45±16.22)ng/L］、中性粒细胞明胶酶相关脂质运载蛋白( NGAL)水平［(89.21±14.80)μg/L比( 64.57±10.12)μg/L］均高于未损伤组( P<0.05)； LASSO回归结合 logistic回归分析结果显示，年龄 >50岁、合并高钙血症、临床分期 Ⅱ~Ⅲ期， Cys C、IL-18、NGAL水平升高均为多发性骨髓瘤病人自体造血干细胞移植后早期急性肾损伤的独立危险因素(P<0.05)；基于上述变量构建列线图模型，建模集和验证集中 Hosmer-Lemeshow检验结果均 P>0.05，拟合度良好； ROC曲线结果显示模型在建模集、验证集中预测病人早期发生急性肾损伤的曲线下面积(AUC)及其 95%CI分别为 0.87(0.78，0.94)、 0.84(0.75，0.88)；决策曲线分析结果显示，两组列线图模型均具有良好的临床净收益。结论基于 LASSO-logistic回归分析构建的多发性骨髓瘤自体造血干细胞移植后早期发生急性肾损伤的列线图预测模型具有良好的预测效能和临床应用价值，可为此类病人的早期识别提供参考。 |
| 英文摘要: |
| Objective To develop and validate a prediction model for acute kidney injury after autologous hematopoietic stem celltransplantation in patients with multiple myeloma based on LASSO-logistic regression analysis.Methods This retrospective study in.cluded 260 patients with multiple myeloma who underwent autologous hematopoietic stem cell transplantation at West China Hospitalof Sichuan University from January 2021 to December 2023. Using a random number table, patients were divided into a training set(182 cases) and a validation set (78 cases) at a 7:3 ratio. Patients in the training set were further divided into an injury group (57 cases)and a non-injury group (125 cases) according to whether early acute kidney injury occurred after transplantation. Clinical data were col.lected for all participants. LASSO regression analysis was used to screen independent variables and determine the optimal λ value. Mul. tivariate logistic regression analysis was then performed to identify independent risk factors for early acute kidney injury after autolo.gous hematopoietic stem cell transplantation in patients with multiple myeloma. A risk prediction model was constructed using R soft.ware, and receiver operating characteristic curves (ROC curves), calibration curves, and decision curve analysis were used to evaluateand validate the model. Results Among the 260 patients, 85 developed early acute kidney injury after transplantation, for an inci.dence rate of 32.69%. Compared with the non-injury group, the injury group had a significantly higher proportion of patients aged >50 years [38.60% (22/57) vs. 20.80% (26/125)], a higher proportion of patients with hypercalcemia [33.33% (19/57) vs. 16.00% (20/125)], a higher proportion of patients with clinical stage Ⅱ-Ⅲ disease [64.91% (37/57) vs. 45.60% (57/125)], and higher levels of serum creat. inine [(103.25±16.91) μmol/L vs. (62.07±10.58) μmol/L], cystatin C (Cys C) [(3.05±0.77) mg/L vs. (1.54±0.31) mg/L], interleukin-18 (IL-18) [(214.91±38.73) ng/L vs. (116.45±16.22) ng/L], and neutrophil gelatinase-associated lipocalin (NGAL) [(89.21±14.80) μg/L vs. (64.57±10.12) μg/L] (P<0.05). LASSO regression combined with logistic regression analysis showed that age >50 years, concomitant hy.percalcemia, clinical stage Ⅱ-Ⅲ, and elevated levels of Cys C, IL-18, and NGAL were independent risk factors for early acute kidneyinjury after autologous hematopoietic stem cell transplantation in patients with multiple myeloma (P<0.05). Based on these variables, a nomogram model was constructed. The Hosmer-Lemeshow test indicated good calibration in both the training and validation sets (P> 0.05). ROC curve analysis showed that the area under the curve (AUC) for predicting early acute kidney injury was 0.87 [95% CI: (0.78, 0.94)] in the training set and 0.84 [95% CI: (0.75, 0.88)] in the validation set. Decision curve analysis demonstrated that the nomogrammodel provided good clinical net benefit in both sets.Conclusion The nomogram prediction model for early acute kidney injury afterautologous hematopoietic stem cell transplantation in patients with multiple myeloma, developed using LASSO-logistic regression, dem.onstrates good predictive performance and clinical utility and may serve as a reference for the early identification of high-risk patients. |
|
查看全文
查看/发表评论 下载PDF阅读器 |
| 关闭 |
|
|
|
