人腦神經纖維束共變性於生命進程之變化

Abstract

研究背景 過去一項研究表明人腦白質纖維在老化的進程中逐漸喪失它們的獨特性，各纖維束之間的協作性增加，這是由於去分化所致。在此研究的基礎之上，我們的目的是探索正常發育中大腦分化的相關關係，並驗證正常衰老中去分化的相關關係。本研究我們使用一種嶄新的大腦白質神經纖維之共變性分析方法(Tract covariance)，用其來表示受試者之間任意一對神經纖維束之間的相關性。為了評估人腦從年輕到高齡的白質神經纖維束共變性變化特徵，我們收集了481名年齡在5到74歲之間的健康人群的擴散頻譜影像(DSI)數據。我們假設神經纖維束共變性從兒童到青少年晚期呈下降趨勢，表明分化，從二十歲開始呈上升趨勢，反映出去分化。 研究方法 481位參與者依照年齡被分為6組。每位受試者均接受核磁共振成像掃描以獲得大腦DSI影像和T1影像。掃描的機台為3T核磁共振(TIM Trio, Siemen),搭配使用32通道相控陣線圈。運用全腦神經纖維自動化分析系統(Tract-based automatic analysis, TBAA)對DSI影像進行分析。透過TBAA技術將收集到的影像資料合併為一個大腦影像範本，此範本再與實驗室已發展之NTU-DSI-122大腦範本及76束大腦白質纖維束進行對位及座標轉換，計算每一位受試者的腦部白質纖維特徵參數。本次研究主要針對綜合非等向性指標(GFA)進行評估。在TBAA分析之後會得到每一年齡層的三維GFA connectogram，隨後將每一條纖維的路徑 GFA值進行平均，得出每一條纖維束的平均GFA值。為了比較每各年齡組白質纖維之間關聯的強弱，首先平均每條神經纖維路徑上的GFA值後進行神經纖維束之共變性分析，共變性分析是使用凈相關(Partial correlation)的方法控制年齡、性別以及DSI遺漏張數之影響後，進行76條神經纖維中任意兩條在同一年齡組之間的共變性分析，最後通過76 x 76之共變性矩陣(Tract covariance matrix)的方式表示出來。為了比較各組共變性矩陣的高低，我們使用排列檢定(Permutation testing)來探討兒童組和其他組之間的任何兩條纖維束的相關性是否有顯著差異。為了確定纖維束共變性之間變化的主要趨勢，我們進一步對排列檢驗得到的所有p值取-log，通過比較-log |p|的大小與數量來確定共變性較高的一組。 結果 白質神經纖維束共變性網路可以表示神經纖維束之間的相互依賴程度。 從兒童到青少年，神經纖維束傾向於具備自己的功能(分化)，並在青春期後逐漸喪失其個性(去分化)。排列堅定的統計結果進一步證實，青少年時期纖維束之間的關聯最少，而五十歲以上的組別纖維束之間的合作最為顯著。值得注意的是，在進入三十歲後，兒童組纖維束相關性較高的主導趨勢被逆轉。 討論與結論 纖維束共變性網路的在進入二十歲之後的變化證實了Cox等人發現的白質從中年到老年的微觀結構產生去分化的假設。使用相同的方法，我們進一步提供了從兒童到年輕成年人之間白質微觀結構出現分化的證據。本次研究的纖維束共變性在十幾歲左右達到最低，這有異於白質結構在三十幾歲達到成熟巔峰的觀念。白質成熟高峰的年齡和最小共變性所在的年齡之間的差異表明，除了大腦發育和衰老的正常影響之外，白質神經纖維束共變性網路的變化也歸因於突觸與軸突的修剪以及大腦白質的可塑性功能。

Introduction An increasing correlation among white matter tracts in older age, which is hypothesized to arise from de-differentiation, has been illustrated by a previous study. Here, we aimed to explore the brain correlates of differentiation in normal development and verify the reported correlates of de-differentiation in normal aging. We proposed a novel metric called tract covariance, which was used to indicate the relationship between any pair of tracts across participants. To evaluate the variation of tract covariance from developmental to aging periods, we collected diffusion spectrum imaging (DSI) data in 481 generally healthy people aged 5 to 74 years. We hypothesized that tract covariance decreased from childhood to late adolescence, indicating differentiation, and increased from 3rd decade, reflecting de-differentiation. Methods The participants were divided into six groups per decade. Each participant received an MRI scan to obtain DSI data and T1 image of the brain on a 3T MRI scanner (TIM Trio, Siemen) with a 32-channel phased array coil. We used tract-based automatic analysis (TBAA) to obtain a 2D mean generalized fractional anisotropy (GFA) connectogram of 76 white matter tract bundles for each participant. We averaged the 100 values of each tract profile to obtain a mean GFA value for each tract. We compared the tract covariance among these six groups, which were defined as the partial correlation between each pair of tracts in variations of GFA values across subjects, with age, gender and dropout number being regressed. For statistical analysis, we used the permutation test to find whether there were significant differences in the association of any two tracts between children group and any other group. In order to identify major trends of the changes between covariance, we further took the -log of all p values derived from the permutation test for visualizing the difference distribution. Results Tract covariance can indicate the variations of inter-dependence of the tracts. Each tract tended to perform its own functions from children to teenagers (differentiation) and lose its individuality gradually after the end of the 2nd decade (de-differentiation). The permutation results further verified that teenagers have the least tract-to-tract associations, whereas the group of 6th-decade-and-beyond exhibits the strongest cooperation among tracts. Particularly, the predominance of higher correlations in the children group is reversed after entering the 4th decade. Discussion and Conclusion The changes in tract covariance after the 3rd decade corroborate the hypothesis of microstructural de-differentiation of white matter in elderly people reported by Cox et al.. By using the same methodology, we further provide evidence of microstructural differentiation from children to young adulthood. Unlike white matter maturation which peaks around the 4th decade, our results of tract covariance reaches its minimum around the 2nd decade. The disparity between the age of peak maturation and the age of least tract covariance suggests that besides normal effects of brain development and aging, the changes of covariance are also modulated in part by pruning and neuroplasticity of the white matter tracts.