基于常规胸腹部CT的L1椎体骨衰减–椎旁肌肉退变耦合指标辅助识别老年骨质疏松性椎体压缩骨折的价值

Value of an L1 Vertebral Bone Attenuation-Paraspinal Muscle Degeneration Coupling Index Derived from Routine Chest and Abdominal CT for Aiding Identification of Osteoporotic Vertebral Compression Fractures in Older Adults

摘要: 目的：探讨常规胸腹部计算机体层成像(computed tomography, CT)中L1椎体松质骨CT值及椎旁肌肉参数对老年骨质疏松性椎体压缩骨折(osteoporotic vertebral compression fracture, OVCF)的辅助识别价值，并评价联合模型的增量效能。方法：回顾性纳入2023年1月至2025年12月于西部战区总医院接受胸部或腹部平扫CT且扫描范围包含L1椎体的老年患者196例，其中OVCF组72例、非OVCF组124例。测量L1椎体CT值、椎旁肌肉总横截面积、身高校正肌肉指数、椎旁肌肉密度及肌肉脂肪浸润指数，构建骨–肌耦合评分。采用多因素Logistic回归、受试者工作特征(receiver operating characteristic, ROC)曲线、校准曲线、决策曲线及重分类分析评价模型效能。结果：OVCF组L1椎体CT值、椎旁肌肉总横截面积、身高校正肌肉指数及椎旁肌肉密度低于非OVCF组，肌肉脂肪浸润指数高于非OVCF组(均P < 0.01)。多因素Logistic回归分析结果显示，年龄 ≥ 75岁、既往脆性骨折史、L1椎体CT值、椎旁肌肉总横截面积及椎旁肌肉密度是OVCF的影响因素(均P < 0.05)。联合模型ROC曲线下面积为0.805 (95% CI: 0.741~0.869)，高于临床变量 + L1椎体CT值模型(DeLong检验P = 0.026)；Hosmer-Lemeshow检验P = 0.312，Bootstrap校正后校准斜率为0.94。阈值概率为15%~65%时，联合模型具有较高净获益；连续净重分类改善指数为0.284，综合判别改善指数为0.061。结论：基于常规胸腹部CT的L1椎体骨衰减与椎旁肌肉退变联合评估可辅助识别老年OVCF，并为进一步骨密度及骨代谢评估提供影像学提示，但仍需多中心、前瞻性研究验证。

Abstract: Objective: To investigate the value of L1 vertebral cancellous bone CT attenuation and paraspinal muscle parameters measured on routine chest and abdominal computed tomography (CT) for aiding identification of osteoporotic vertebral compression fracture (OVCF) in older adults, and to evaluate the incremental performance of a combined model. Methods: This retrospective study included 196 older patients who underwent non-contrast chest or abdominal CT at the General Hospital of Western Theater Command from January 2023 to December 2025, with the scan range including the L1 vertebra. There were 72 patients in the OVCF group and 124 in the non-OVCF group. L1 vertebral CT attenuation, total cross-sectional area of the paraspinal muscles, height-adjusted muscle index, paraspinal muscle density, and muscle fat infiltration index were measured, and a bone-muscle coupling score was constructed. Multivariable logistic regression, receiver operating characteristic (ROC) curves, calibration curves, decision curve analysis, and reclassification analyses were used to evaluate model performance. Results: The OVCF group had lower L1 vertebral CT attenuation, total paraspinal muscle cross-sectional area, height-adjusted muscle index, and paraspinal muscle density, and a higher muscle fat infiltration index than the non-OVCF group (all P < 0.01). Multivariable logistic regression showed that age ≥ 75 years, history of fragility fracture, L1 vertebral CT attenuation, total paraspinal muscle cross-sectional area, and paraspinal muscle density were associated with OVCF (all P < 0.05). The area under the ROC curve of the combined model was 0.805 (95% CI: 0.741~0.869), which was higher than that of the model including clinical variables plus L1 vertebral CT attenuation (DeLong test, P = 0.026); the Hosmer-Lemeshow test yielded P = 0.312, and the bootstrap-corrected calibration slope was 0.94. At threshold probabilities of 15%~65%, the combined model showed higher net benefit; the continuous net reclassification improvement was 0.284 and the integrated discrimination improvement was 0.061. Conclusion: Combined assessment of L1 vertebral bone attenuation and paraspinal muscle degeneration on routine chest and abdominal CT can aid in identifying OVCF in older adults and provide imaging clues for further bone mineral density and bone metabolism evaluation; however, multicenter prospective validation is still required.