骨质疏松性椎体压缩骨折患者发生Kümmell病的风险预测

Predictive analysis and risk assessment of Kümmell's disease in patients with osteoporotic vertebral compression fractures

目的探讨骨质疏松性椎体压缩骨折(osteoporotic vertebral compression fractures,OVCFs)患者发生Kümmell病(Kümmell's disease)的风险预测指标。方法采用1∶1频数匹配的病例对照研究设计,选取2021年1月至2023年6月在柳州市工人医院脊柱外科就诊并发或不并发Kümmell病的OVCFs患者为研究对象,纳入Kümmell病组和非Kümmell病组。详细收集患者的人口学资料、合并症及血清学指标,比较两组患者的基线特征。通过单因素分析筛选与目标变量明显相关的预测变量,并构建相关性热图评估目标变量与预测变量间的共线性,采用lasso回归模型进一步筛选潜在的预测指标,采用二分类logistic回归模型确定Kümmell病的风险预测指标。结果单因素分析筛选出与Kümmell病明显相关的预测变量包括年龄、骨密度、后凸Cobb角、双节段及以上椎体骨折等。Lasso回归分析确定了系数非零的关键预测指标包括年龄、骨密度、后凸Cobb角、双节段及以上椎体骨折、血小板计数(platelet count,PLT)、天冬氨酸转氨酶/谷氨酸转氨酶比值(aspartate aminotransferase/alanine aminotransferase,AST/ALT)、白蛋白(albumin,Alb)、白蛋白/球蛋白比值(albumin/globulin ratio,Alb/Glb)、碱性磷酸酶(Alkaline phosphatase,ALP)、血清尿素(serum urea,UREA)、血清尿酸(serum uric acid,SUA)、纤维结合蛋白(fibrinogen,Fn)、血糖(blood glucose,BG)及C反应蛋白(C-reactive protein,CRP)。相关性分析显示ALT与AST/ALT有较高的相关性(r=0.750),PLT与Alb表现出较低相关性(r=-0.110)。Logistic回归分析显示存在双节段及以上椎体骨折[OR=2.078,95%CI(1.072,4.025),P=0.030]、后凸Cobb角增大[OR=1.033,95%CI(1.008,1.058),P=0.009]、ALP水平升高[OR=1.013,95%CI(1.004,1.023),P=0.006]、SUA水平升高[OR=1.004,95%CI(1.000,1.007),P=0.043]、Fn水平降低[OR=0.996,95%CI(0.992,0.999),P=0.008]是OVCFs患者发生Kümmell病的预测指标。结论双节段及以上椎体骨折、后凸Cobb角增大、ALP和SUA水平升高、Fn水平降低,可作为OVCFs患者是否发展为Kümmell病的预警指标,监测这些指标对早期发现和干预有一定的作用。

Objective To analyze predictive risk indicators associated with the development of Kümmell's disease(KD)in patients with osteoporotic vertebral compression fractures(OVCFs).Methods A 1∶1 frequency-matched case-control study design was employed,selecting patients who visited the Department of Spine Surgery at Liuzhou Workers'Hospital from January 2021 to June 2023.Patients were divided into case and control groups based on whether they progressed to Kümmell's disease(KD).Detailed demographic information,comorbidities,and laboratory data were collected,and baseline characteristics of the two groups were compared.Initial predictive variables significantly associated with the target variable were preliminarily screened through univariate analysis.A correlation heatmap was then constructed to assess collinearity among these variables,followed by further selection of potential predictors using the Lasso regression model.Finally,a multivariable logistic regression model was used for the prediction and analysis of KD-related risk indicators.Results Univariate analysis identified significant predictors of Kümmell's disease,including patient age,bone mineral density,kyphotic Cobb angle,and multiple vertebral fractures.These were included in the subsequent Lasso regression analysis,which identified key predictors with non-zero coefficients:age,bone density,Cobb angle,multiple vertebral fractures,platelet count(PLT),aspartate aminotransferase/alanine aminotransferase(AST/ALT),albumin(Alb),albumin/globulin ratio(Alb/Glb),alkaline phosphatase(ALP),urea(UREA),serum uric acid(SUA),fibrinogen(Fn),blood glucose(BG),and C-reactive protein(CRP).The correlation heatmap revealed the correlation and collinearity risks between these variables,with ALT and AST/ALT showing a high correlation(r=0.750)and PLT and Alb showing a low correlation(r=-0.110).Multivariable logistic regression indicated that the presence of multiple vertebral fractures[OR=2.078,95%CI(1.072,4.025),P=0.030],increased Cobb angle[OR=1.033,95%CI(1.008,1.058),P=0.009],elevated levels of ALP[OR=1.013,95%CI(1.004,1.023),P=0.006],and SUA[OR=1.004,95%CI(1.000,1.007),P=0.043]were associated with an increased risk of KD in patients with OVCFs.Conversely,decreased levels of Fn[OR=0.996,95%CI(0.992,0.999),P=0.008]were linked to an increased risk of KD.Conclusion Multiple vertebral fractures,increased Cobb angle,elevated levels of ALP and SUA,along with decreased levels of Fn,can be used as early-warning indicators to predict whether patients with OVCFs will develop KD.Monitoring these indicators is crucial for the early detection and intervention in these patients.