肝脏 ›› 2025, Vol. 30 ›› Issue (1): 95-100.

• 非酒精性脂肪性肝病 • 上一篇    下一篇

老年非酒精性脂肪肝患者骨代谢特征及并发骨质疏松的绝对风险预测模型构建

朱水津, 王鸿, 赖华梅, 沈丹丹   

  1. 202150 上海健康医学院附属崇明医院消化科
  • 收稿日期:2024-01-31 出版日期:2025-01-31 发布日期:2025-03-10
  • 通讯作者: 王鸿

Establishment of an absolute risk prediction model for bone metabolism and osteoporosis in elderly patients with nonalcoholic fatty liver disease

ZHU Shui-jin, WANG Hong, LAI Hua-Mei, SHEN Dan-dan   

  1. Department of Gastroenterology, Chongming Hospital, Shanghai Health Medical College, Shanghai 202150, China
  • Received:2024-01-31 Online:2025-01-31 Published:2025-03-10
  • Contact: WANG Hong

摘要: 目的 探讨老年非酒精性脂肪肝(non-alcoholic fatty liver disease,NAFLD)患者骨代谢特征及并发骨质疏松(osteoporosis, OP)的绝对风险预测模型的构建方法。 方法 本研究选取上海健康医学院附属崇明医院2020年1月—2023年1月收治的240例NAFLD患者作为研究对象,同时选取120例体检健康人群作为对照组,根据患者是否合并发生OP分为NAFLD组、NAFLD合并OP组,倾向性评分匹配调整均衡性,收集患者的基线资料。利用决策树、Gail绝对风险估算等计算非酒精性脂肪肝并伴有骨质疏松发生的绝对风险,利用R软件绘制ROC(receiver operating characteristics,ROC)曲线寻找最佳截断值。 结果 三组患者在TG、PINP、β-CTX、BMD、GHb、OCN、ALP指标间的差异显著(P<0.05),与对照组相比,NAFLD组和NAFLD合并OP组患者PINP、β-CTX、BMD、OCN水平显著降低,TG、GHb、ALP水平显著升高。决策树模型筛选出PINP、β-CTX、BMD、OCN,4个变量为发生NAFLD合并OP的危险因素。对照组5年发病风险概率和IQR分别为2.4%和0.132%,NAFLD组对应数值分别为23.1%和0.255%,NAFLD合并OP组对应数值分别为42.3%和0.451%,最佳截断值为0.100%。经验证集验证,本模型的AUC为0.826,灵敏度为81.25%,特异度为75.38%,精确率72.94%,准确性78.36%。 结论 本研究构建了老年非酒精性脂肪肝患者骨代谢特征及并发骨质疏松的绝对风险预测模型,可初步预测老年非酒精性脂肪肝患者骨代谢特征及并发骨质疏松发病风险,为疾病的治疗与干预具有指导意义。

关键词: 非酒精性脂肪肝, 骨质疏松, 模型构建, 验证

Abstract: Objective To explore the method of establishing the absolute risk prediction model for elderly patients with non-alcoholic fatty liver disease (NAFLD) and osteoporosis (OP). Methods In this study, 240 patients with NAFLD admitted to our hospital from January 2020 to January 2023 were selected as the research objects, and 120 healthy people were selected as the control group. According to whether the patients were combined with OP, they were divided into the NAFLD group and the NAFLD combined OP group. Baseline data were collected. The absolute risk of nonalcoholic fatty liver disease with osteoporosis was calculated using decision tree and Gail absolute risk estimation, and the receiver operating characteristics (ROC) curve was generated using R software to identify the best cutoff value. Methods There were statistically significant differences in TG, PINP, β-CTX, BMD, GHb, OCN, and ALP among the three groups (P<0.05). Compared with the control group, the levels of PINP, β-CTX, BMD, and OCN in NAFLD group and NAFLD combined OP group were significantly reduced. The levels of TG, GHb, and ALP were significantly increased. Four variables, PINP, β-CTX, BMD, and OCN, were selected by the decision tree model as risk factors for NAFLD and OP. The 5-year risk and IQR of the control group were 2.4% and 0.132%, 23.1% and 0.255% in the NAFLD group, and 42.3%and 0.451% in the NAFLD combined OP group, respectively. The optimal cut-off value was 0.100%. The empirical set verifies that the AUC of this model is 0.826, the sensitivity is 81.25%, the specificity is 75.38%, the accuracy is 72.94%, and the accuracy is 78.36%. Conclusion This study established a prediction model of bone metabolism characteristics and absolute risk of osteoporosis in elderly patients with nonalcoholic fatty liver disease, enabling initial predictions of bone metabolism and osteoporosis risk, thereby providing guidance for disease treatment and intervention.

Key words: Nonalcoholic fatty liver, Osteoporosis, Model construction, Verify