皮質紋狀體徑與皮質脊髓徑於慢性中風病患與健康成人短期腳踝追蹤動作學習之角色：擴散頻譜造影研究

Abstract

背景與目的：動作學習的能力在個體一生中學習新技巧與因應生活中環境的變化都相當重要。許多功能性磁振造影的研究發現在不同的視覺動作學習階段，大腦額葉皮質區和紋狀體會有功能性活化，然而，在不同的腳踝追蹤（tracking）任務時，這些動作學習區域的結構性連結，尤其是皮質紋狀體徑（corticostriatal tract）與皮質脊髓徑（corticospinal tract）對動作學習的貢獻卻極少被研究。 方法：21位（年齡：62.2±8.5歲；男：16，女：5）慢性中風病患與26位（年齡：62.0±8.1歲；男：7，女：19）年齡對照健康成人參與腳踝追蹤學習之研究，受試者使用客製化腳踝追蹤評估及訓練裝置進行連續五天之練習，包含重複（repeated sequence）及隨機次序（random sequence）的追蹤，追蹤表現會以方均根誤（root mean squared error，RMSE）計算與呈現，隨後進入學習保留休息2天，以及第一週之保留測試（retention test）。受試者於訓練開始前與一週學習結束後進行臨床評估測驗、腳踝追蹤表現測驗以及腦部擴散頻譜影像（diffusion spectrum imaging，DSI）。本研究以特定神經束之神經徑路追蹤分析（tract-specific tractography analysis）重建大腦雙側之外背側前額葉-尾核徑（dorsolateral prefrontal cortex-caudate tract）、輔助動作皮質區-殼核徑（supplementary motor area-putamen tract）與皮質脊髓徑且此神經束之完整性會量化為普擴散不等向性（generalized fractional anisotropy，GFA）表示。淨相關分析（partial correlation analysis）會用以探究訓練前後對大腦的三條神經束在受試者與一週學習之行為或學習進步上的關係。 結果：無論健康或中風受試者，其腳踝重複與隨機次序追蹤之準確度均隨訓練而進步（p< 0.001），而大腦雙側之三條神經束的完整性在訓練一週後均無顯著改變（p> 0.05）。淨相關分析結果顯示，健康人訓練前之對側皮質脊髓徑完整性（GFAB_CST_contra）與訓練前之動作表現有顯著相關（重複次序（RMSEB_rep）: r= 0.423, p= 0.035 ;隨機次序（RMSEB_ran）: r= 0.456, p= 0.022）；訓練後對側輔助動作皮質區-殼核徑之完整性（GFAW1_SMA_contra）與學習的進步有關（重複次序（∆RMSEB-W1_rep）: r= -0.393, p=0.052 ;隨機次序（∆RMSEB-W1_ran）: r= -0.411, p=0.041）；而在訓練前之對側外背側前額葉-尾核徑完整性與學習的進步無顯著相關（重複次序: r= -0.189, p= 0.386 ;隨機次序: r= -0.157, p= 0.453）。病人組的結果則顯示，訓練前（GFAB_CST_contra）（r= 0.536, p= 0.018）與後（GFAW1_CST_contra）（r= 0.520, p= 0.023）之對側皮質脊髓徑完整性與隨機學習的進步（∆RMSEB-W1_ran）皆有顯著相關。 討論與結論：不論健康人或中風病患在接受一週的腳踝追蹤動作訓練後都有表現上的進步。兩組受試者在腦部結構性之白質神經纖維的完整性在此短期學習後並未有顯著變化，不過特定神經纖維的完整性與學習的表現或進步有密切相關，且健康人與中風病患在學習此視覺動作任務上呈現不同的腦部結構性機轉。輔助動作皮質區-殼核徑與健康人之此動作學習有關，而皮質脊髓徑則在慢性中風病患的此種動作學習扮演較重要的角色。

Background and Purpose: Motor learning ability is crucial for individuals to learn new skills in order to adapt to environmental changes throughout the lifespan. Many fMRI studies have found that functional brain activations in different frontal cortical regions and the striatum are relevant to different stages of visuomotor learning. However, little is known about how the structural connectivity between these learning-related regions, in particular, the corticostriatal tracts and the corticospinal tracts, contributes to learning an ankle tracking task. Methods: Twenty-one patients with chronic stroke (age= 62.2±8.5 yr, male: 16, female: 5) and 26 age-matched healthy adults (age= 62.0±8.1 yr, male: 7, female: 19) participated in this study. Using a custom-built ankle tracking assessment and training device, all participants underwent a short-term ankle tracking learning paradigm for 5 consecutive practice sessions within 5 days, followed by a 2-day retention interval and a Week 1 retention test. Repeated and random sequences were both practiced in the 5 days. Tracking performance was measured by using root mean squared error (RMSE). Clinical assessments, ankle tracking performance, and diffusion spectrum MR image (DSI) of the brain were obtained at Baseline test and Week 1 retention test. Tract-specific tractography analysis were used to reconstruct bilateral dorsolateral prefrontal cortex-caudate (dlPFC-caudate), supplementary motor area-putamen (SMA-putamen), and corticospinal tracts (CSTs). Tract integrity was indexed by using generalized fractional anisotropy (GFA) of DSI. Separate partial correlation analyses were performed to evaluate relationships between white matter tract integrity and tracking performance or improvement after learning. Results: Both healthy and stroke subjects significantly improved tracking accuracy over time, regardless of sequences (p< 0.001 both). Among the investigated white matter tracts, no change in tract integrity was found for each tract from baseline to Week 1 retention test (p> 0.05 of each tract). Separate partial correlations showed that, in the healthy group, GFAB_CST_contra was associated with RMSEB_rep (r= 0.423, p= 0.035) and RMSEB_ran (r= 0.456, p= 0.022); GFAW1_SMA_contra was associated with ∆RMSEB-W1_ran at a significant level (r= -0.411, p= 0.041) and with ∆RMSEB-W1_rep at a marginal level (r=-0.393, p=0.052). However, there were no significant correlations between baseline integrity of the contralateral dlPFC-caudate tract and the performance improvement under repeated (r= -0.189, p= 0.386) and random sequence tracking conditions (r= -0.157, p= 0.453) after short-term learning for healthy subjects. In the stroke group, GFAB_CST_contra was associated with ∆RMSEB-W1_rep (r= 0.536, p= 0.018); GFAW1_CST_contra was associated with ∆RMSEB-W1_ran (r= 0.520, p= 0.023). Discussion and conclusions: Both healthy adults and hemiparetic patients with chronic stroke could learn this ankle tracking task. Although we did not find motor learning-related structural changes of the investigated white matter tracts after such as short-term learning, the integrity of specific white matter tracts were found to be closely linked to performance outcomes or gains. In particular, different structural brain mechanisms were found to be related to learning a novel ankle visuomotor task between healthy adults and patients with chronic stroke. For healthy adults, the SMA-putamen tract was closely associated with ankle tracking learning, whereas in patients with stroke, the CST played an important role in such learning.