文献速递脊柱脊髓外科神经重症

年龄相关差异:创伤性脊髓中央综合征手术结局的多机构因果机器学习分析

Age-related differences in surgical outcomes for traumatic central cord syndrome: a multi-institutional causal machine learning analysis.
PubMed 文献 · J Neurosurg Spine · 2026-05-08

该多机构回顾性研究分析2006至2021年SCIMS数据库中890例创伤性脊髓中央综合征患者,采用全最优匹配比较手术与非手术组,并用因果森林算法检测异质性治疗效应。匹配队列中手术与PFCS均值改善6.6分(95%CI -4.5至17.8)未达统计学显著,但年龄是最强效应修饰因子(相对重要性42.8%,p<0.001),分段回归识别约64岁为拐点,>64岁患者手术改善34.7分(95%CI 12.2-57.2),年轻患者无显著获益(-4.1分,95%CI -16.8至8.6)。

要点

  • 匹配队列中手术对TCCS患者1年PFCS改善6.6分,未达统计学显著。
  • 年龄是最强效应修饰因子(相对重要性42.8%,p<0.001),拐点约64岁。
  • >64岁患者手术PFCS改善34.7分(95%CI 12.2-57.2),年轻患者无显著获益。

核心结论

  • TCCS手术疗效可能受年龄影响,>64岁成人获得差异更大的功能与神经获益。

研究设计与样本

  • 多机构回顾性分析SCIMS数据库2006至2021年890例TCCS患者。
  • 全最优匹配后手术组698例、非手术组179例。
  • 采用因果森林算法检测异质性治疗效应,分段回归识别年龄拐点。

主要发现

  • 匹配队列手术与PFCS均值改善6.6分(95%CI -4.5至17.8),未达统计学显著。
  • 年龄为最强效应修饰因子(相对重要性42.8%,p<0.001)。
  • 分段回归识别拐点约64岁。
  • >64岁患者PFCS改善34.7分(95%CI 12.2-57.2)。
  • ≤64岁患者手术获益-4.1分(95%CI -16.8至8.6),无统计学显著。

关键数值与推荐

  • TCCS定义为ASIA C或D级颈髓损伤且下肢运动评分至少高于上肢5分。
  • 主要结局为PFCS(范围0-300),反映躯体独立、活动与职业功能。
  • 敏感性分析基于ASIA运动评分评估神经恢复并采用更严格TCCS标准。

临床怎么用

  • 对>64岁TCCS患者,手术可能带来更大功能获益,可支持个体化决策。
  • 对年轻TCCS患者,手术获益不显著,需结合病情审慎评估。

局限性

  • 回顾性设计,需前瞻性研究验证观察结果与机制。
  • 入院ASIA缺失数据经随机森林插补处理。

以上解读依据论文摘要(全文未开放获取或许可不允许翻译转载)。

原文

OBJECTIVE: Traumatic central cord syndrome (TCCS) is the most common incomplete spinal cord injury, yet the optimal management strategy remains controversial, particularly for older adults who often present with worse outcomes. The authors aimed to determine whether surgical intervention confers different benefits across age groups, focusing on 1-year functional and neurological recovery. METHODS: The authors retrospectively analyzed 890 patients with TCCS from the multi-institutional Spinal Cord Injury Model Systems (SCIMS) Database between 2006 and 2021. TCCS was defined as an American Spinal Injury Association (ASIA) Impairment Scale grade C or D cervical injury and at least a 5-point discrepancy favoring lower extremity motor score over upper extremity motor score. Missing admission ASIA data were addressed via validated random forest imputation. The authors compared surgical versus nonsurgical groups using full optimal matching to reduce confounding, achieving balanced cohorts (698 patients in the surgical group vs 179 in the nonsurgical group). The authors then applied a causal forest algorithm to detect heterogeneous treatment effects and used segmented regression to identify age-related inflection points. The primary outcome was the Physical Function Composite Score (PFCS) (range 0-300), which captures physical independence, mobility, and occupational function. Sensitivity analyses further evaluated neurological recovery on the basis of ASIA motor scores and stricter criteria for TCCS identification after the exclusion of imputed data. RESULTS: Across the matched cohort, surgery was associated with mean PFCS improvement of 6.6 points (95% CI -4.5 to 17.8), which did not reach statistical significance. However, subgroup analysis revealed that age was the strongest effect modifier of surgical benefit (relative importance 42.8%, p < 0.001). Segmented regression identified an inflection at approximately 64 years of age. Patients older than 64 years demonstrated a significant improvement of 34.7 points (95% CI 12.2-57.2), whereas younger individuals showed no statistically significant surgical benefit (-4.1 points, 95% CI -16.8 to 8.6). All sensitivity analyses were consistent with the results of our primary findings. CONCLUSIONS: The authors' findings suggest that the effectiveness of surgical intervention for TCCS may be influenced by age, with adults older than 64 years experiencing differentially greater functional and neurological benefit. These results may inform patient-specific treatment decisions and clinical guidelines. Prospective research is needed to validate these observations, elucidate underlying mechanisms, and guide evidence-based TCCS management.

来源:J Neurosurg Spine,查看原文。版权归原作者所有。

引用

  1. Aude CA, Vattipally VN, Jillala R, et al. Age-related differences in surgical outcomes for traumatic central cord syndrome: a multi-institutional causal machine learning analysis. J Neurosurg Spine. 2026. PMID: 42102409. doi:10.3171/2025.12.SPINE25166

供执业医师工作参考,不替代临床判断;数值与分级以指南原文为准。