基于临床和实验室指征的骨折后骨痂过度生长的Nomogram预测模型的建立及临床应用

Establishment and clinical application of a Nomogram model in predicting callus hyper hyperplasia after fracture based on clinical and laboratorial features

目的本研究通过骨折患者的临床和实验室指征建立骨痂过度生长(callus hyperhyperplasia,CH)的Cox风险模型并输出得到Nomogram图。方法选取2018年8月—2022年6月在舟山医院住院骨折患者共127例,其中男性84例,女性43例。患者入院采集临床资料和各项手术参数,并从患者入院开始收集第1天(时间点A)、第7天(时间点B)、第15天(时间点C)的血清,酶联免疫吸附法(ELISA)检测血清人骨成形蛋白-2(bone morphogenetic protein-2,BMP-2)、BMP-4、BMP-7及细胞因子白介素-6(IL-6)、白介素-10(IL-10)、转化生长因子-β(TGF-β)的浓度。按随访是否出现CH分组建立Cox回归模型,得到独立风险因子后输出为Nomogram图。结果共计发现CH结局患者16例。多因素Cox回归模型显示:合并脑外伤(HR=1.865,P<0.01)、手术持续时间(HR=1.004,P<0.01)、术中出血量(HR=1.003,P<0.01)、BMP-2(C)(HR=1.050,P<0.01)、BMP-4(C)(HR=1.005,P<0.05)、BMP-7(C)(HR=1.002,P<0.05)、TGF-β(C)(HR=1.001,P<0.01)7个因素对CH的结局存在显著影响。输出得到第60天和120天的CH发生率预测Nomogram图。结论建立得到一种简单又可操作性强的预测骨折患者HO风险的临床评分系统,有助于临床早期干预降低患者HO的发生率。

Objective This study is aimed to establish a Cox risk model and produce Nomogram diagram to predict callus hyper hyperplasia based on clinical and laboratory features in patients with fractures.Methods Between August 2018 and January 2022,127 patients with fractures were selected including 84 males and 43 females.Clinical data and various surgical parameters were collected after admission,and serum was collected from day 1(time point A),day 7(time point B)and day 15(time point C).Enzyme-linked immunosorbent assay(ELISA)was used to detect human bone morphogenetic protein-2(BMP-2)in serum protein,BMP-4,BMP-7,and the cytokines interleukin-6(IL-6),interleukin-10(IL-10),and transforming growth factor-β(TGF-β).Cox regression model was established according to whether CH appeared in the follow-up group,and independent risk factors were obtained and then output as Nomogram diagram.Results Sixteen patients with CH were observed by clinical follow-up.Multivariable COX analysis showed that combination with brain trauma(HR=1.865,P<0.01),surgery time(HR=1.004,P<0.01),blood loss during surgery(HR=1.003,P<0.01),BMP-2(C)(HR=1.050,P<0.01),BMP-4(C)(HR=1.005,P<0.05),BMP-7(C)(HR=1.002,P<0.05),and TGF-β(C)(HR=1.001,P<0.01)were independent risk factors for CH outcomes.Nomogram of CH incidence prediction on 60 d and 120 d was produced.Conclusion The establishment of a simple and applicable clinical scoring system to predict the risk of HO in fracture patients is helpful to reduce the incidence of HO in early clinical intervention.