Banerjee Anjishnu, Yang Yushan, Wang Marjorie C, Vedantam Aditya
Division of Biostatistics, Institute for Health and Equity, Medical College of Wisconsin, Milwaukee, WI.
Department of Neurosurgery, Medical College of Wisconsin, Milwaukee, WI.
Clin Spine Surg. 2025 Mar 1;38(2):E69-E74. doi: 10.1097/BSD.0000000000001662. Epub 2024 Jul 22.
Retrospective study.
The aim of this study was to identify recovery trajectory clusters after surgery for degenerative cervical myelopathy (DCM), as well as to determine clinical and imaging characteristics associated with functional recovery trajectories.
Accurate prediction of postsurgical neurological recovery for the individual patient with DCM is challenging due to varying patterns of functional recovery. Latent class Bayesian models can model individual patient patterns and identify groups of patients with similar phenotypes for personalized prognostication.
A prospective single-center study of 70 consecutive patients with DCM undergoing elective cervical spine decompression for DCM between 2010 and 2017 was performed. Outcomes were recorded using the modified Japanese Orthopedic Association (mJOA), Neck Disability Index (NDI), and the Short Form-36 Physical Component Score (SF-36 PCS) at 3, 6, 12, and 24 months. Recovery trajectories were constructed based on unsupervised Bayesian latent class modeling. Clinical and imaging predictors of recovery trajectories were also determined.
Recovery after surgery for DCM showed 3 distinct recovery trajectory clusters for each outcome. The commonest recovery trajectory was sustained improvement for the mJOA (41.1%), stagnation for the NDI (60.3%), and stability for the SF-36 PCS (46.6%). Age, duration of symptoms, and baseline disability were the strongest predictors of each recovery trajectory. Degree of cord compression, neck pain, and intramedullary T2-hyperintensity were predictive of NDI and SF-36 PCS but not mJOA recovery trajectory. Sex was associated with the NDI recovery trajectory but not SF-36 PCS and mJOA recovery trajectories.
Using prospective data and a data-driven approach, we identified 3 distinct recovery trajectory clusters and associated factors for mJOA, NDI, and SF-36 PCS in the first 24 months after surgery for DCM. Our results can enhance personalized clinical prognostication and guide patient expectations at different time points after surgery for DCM.
回顾性研究。
本研究旨在确定退行性颈椎病（DCM）手术后的恢复轨迹集群，并确定与功能恢复轨迹相关的临床和影像学特征。
由于功能恢复模式不同，准确预测DCM个体患者的术后神经功能恢复具有挑战性。潜在类别贝叶斯模型可以对个体患者模式进行建模，并识别具有相似表型的患者群体以进行个性化预后评估。
对2010年至2017年间连续70例因DCM接受择期颈椎减压手术的患者进行前瞻性单中心研究。在3、6、12和24个月时使用改良日本骨科协会（mJOA）评分、颈部残疾指数（NDI）和简明健康调查36项身体成分评分（SF-36 PCS）记录结果。基于无监督贝叶斯潜在类别建模构建恢复轨迹。还确定了恢复轨迹的临床和影像学预测因素。
DCM手术后的恢复在每个结果方面显示出3个不同的恢复轨迹集群。最常见的恢复轨迹是mJOA持续改善（41.1%）、NDI停滞（60.3%）和SF-36 PCS稳定（46.6%）。年龄、症状持续时间和基线残疾程度是每个恢复轨迹的最强预测因素。脊髓压迫程度、颈部疼痛和脊髓内T2高信号可预测NDI和SF-36 PCS，但不能预测mJOA恢复轨迹。性别与NDI恢复轨迹相关，但与SF-36 PCS和mJOA恢复轨迹无关。
使用前瞻性数据和数据驱动方法，我们在DCM手术后的前24个月内确定了mJOA、NDI和SF-36 PCS的3个不同恢复轨迹集群及相关因素。我们的结果可以加强个性化临床预后评估，并指导DCM手术后不同时间点的患者预期。
