Greater gait variability predicts dementia. However, little is known about the neural correlates of gait variability. The aims of this study were to determine (1) grey matter volume covariance patterns associated with gait variability and (2) whether these patterns were associated with specific cognitive domains. Participants (n = 351; mean age 71.9 ± 7.1) were randomly selected from the Southern Tasmanian electoral roll. Step time, step length, step width and double support time were measured using an electronic walkway. Gait variability was calculated as the standard deviation of all steps for each gait measure. Voxel-based morphometry and multivariate covariance-based analyses were used to identify grey matter patterns associated with each gait variability measure. The individual expressions of grey matter patterns were correlated with processing speed, memory, executive and visuospatial functions. The grey matter covariance pattern of double support time variability included frontal, medial temporal, anterior cingulate, insula, cerebellar and striatal regions. Greater expression of this pattern was correlated with poorer performance in all cognitive functions (p < 0.001). The covariance pattern of step length variability included frontal, temporal, insula, occipital and cerebellar regions and was correlated with all cognitive functions (p < 0.05), except memory (p = 0.76). The covariance pattern of step width variability was limited to the cerebellum and correlated only with memory (p = 0.047). No significant pattern was identified for step time variability. In conclusion, different grey matter covariance patterns were associated with individual gait variability measures. These patterns were also correlated with specific cognitive functions, suggesting common neural networks may underlie both gait and cognition.

较大的步态变异性可预测痴呆。但是，关于步态变异性的神经相关性知之甚少。这项研究的目的是确定（1）与步态变异性相关的灰质体积协方差模式，以及（2）这些模式是否与特定的认知域相关。参与者（n = 351；平均年龄71.9±7.1）是从塔斯马尼亚南部选民名单中随机选择的。使用电子走道测量步长，步长，步长和双重支撑时间。将步态变异性计算为每个步态度量的所有步骤的标准偏差。基于体素的形态计量学和基于多元协方差的分析被用来识别与每个步态变异性度量相关的灰质模式。灰质模式的个体表达与处理速度，记忆，执行力和视觉空间功能相关。双支持时间变异性的灰质协方差模式包括额叶，颞内侧，扣带状，岛状，小脑和纹状体区域。该模式的较高表达与所有认知功能的较差表现相关（p <0.001）。步长可变性的协方差模式包括额叶，颞叶，岛状，枕叶和小脑区域，并且与所有认知功能相关（p <0.05），但记忆力除外（p = 0.76）。步宽可变性的协方差模式仅限于小脑，并且仅与记忆相关（p = 0.047）。没有发现明显的步进时间变异性模式。综上所述，不同的灰质协方差模式与个体步态变异性测度相关。这些模式还与特定的认知功能相关，表明常见的神经网络可能是步态和认知的基础。