抑郁癥的人格類型及其腦功能連接基礎(chǔ)*

李 彧 位東濤 邱 江

抑郁癥的人格類型及其腦功能連接基礎(chǔ)*

李 彧1,2,3位東濤1,2邱 江1,2

(1西南大學(xué)認(rèn)知與人格教育部重點(diǎn)實(shí)驗(yàn)室;2西南大學(xué)心理學(xué)部;3西南大學(xué)教育學(xué)部, 重慶 400715)

本研究采用功能隨機(jī)森林的方法, 將聚類過(guò)程與抑郁癥診斷相結(jié)合, 分別在抑郁癥和控制組中識(shí)別了人格類型(神經(jīng)質(zhì)和外向性的組合), 并進(jìn)一步探究了不同人格類型的靜息態(tài)功能連接差異。聚類分析結(jié)果顯示, 抑郁癥以高神經(jīng)質(zhì)和低外向性趨勢(shì)的個(gè)體為主, 但同樣有低神經(jīng)質(zhì)和高外向性趨勢(shì)的個(gè)體?？刂平M樣本則以低神經(jīng)質(zhì)和高外向性個(gè)體為主。靜息態(tài)功能連接的結(jié)果顯示：在不考慮人格亞型的情況下, 抑郁癥和控制組在杏仁核/海馬/腦島?邊緣網(wǎng)絡(luò)/默認(rèn)網(wǎng)絡(luò)/控制網(wǎng)絡(luò)的功能連接上均無(wú)顯著差異。在納入聚類分析所劃分的亞型進(jìn)行統(tǒng)計(jì)后, 多種人格類型在左側(cè)杏仁核/腦島?邊緣網(wǎng)絡(luò)(以眶額皮質(zhì)區(qū)域?yàn)橹?的功能連接強(qiáng)度上呈現(xiàn)出顯著差異。本研究基于個(gè)人視角識(shí)別的抑郁癥人格類型更符合現(xiàn)實(shí)情況與個(gè)體認(rèn)知模式, 具有潛在的臨床應(yīng)用價(jià)值, 并且其功能連接的差異對(duì)理解抑郁癥異質(zhì)性提供了神經(jīng)層面的參考。

神經(jīng)質(zhì), 外向性, 靜息態(tài)功能連接, 抑郁癥, 以個(gè)體為中心

1 引言

近年來(lái), 心理健康研究中的異質(zhì)性問(wèn)題備受重視。有研究者指出它限制了心理健康與認(rèn)知神經(jīng)方向的研究(Feczko et al., 2019), 需要進(jìn)行一場(chǎng)異質(zhì)性“革命” (Bryan et al., 2021)。與此同時(shí), 強(qiáng)調(diào)樣本異質(zhì)性的以人為中心的人格研究路徑(人格分類的角度)受到了更多的關(guān)注(尹奎等, 2021)。已有研究者結(jié)合人格類型學(xué)的優(yōu)勢(shì), 發(fā)現(xiàn)不同類型的人與心理健康問(wèn)題及其治療效果有著明顯不同的聯(lián)系(Isler et al., 2017)。另一方面, 抑郁癥病人的異質(zhì)性一直是一個(gè)不可忽視的問(wèn)題(多數(shù)研究忽略實(shí)際情況, 直接假定樣本同質(zhì)), 但大量實(shí)證研究聚焦于孤立的人格特質(zhì)與抑郁癥的關(guān)系(Klein et al., 2011), 鮮有研究從人格分類角度考慮個(gè)體差異的影響。

當(dāng)代人格類型學(xué)的研究源于Jack Block。其后, Asendorpf, Robins和Caspi等人相繼重復(fù)出了3種人格類型(RUO人格類型) (Asendorpf et al., 2001)：彈性型(resilients)、控制不足型(undercontrollers)和過(guò)度控制型(overcontrollers)。在一定程度上, RUO人格類型在五因素模型上的趨勢(shì)傾向于：彈性型個(gè)體表現(xiàn)為神經(jīng)質(zhì)低, 其他維度均高; 過(guò)度控制型個(gè)體表現(xiàn)為神經(jīng)質(zhì)高, 外向性低; 控制不足型個(gè)體表現(xiàn)為宜人性和開(kāi)放性均低的特點(diǎn)(Bohane et al., 2017; Donnellan & Robins, 2010)。但是已有不少研究者指出, RUO分類在統(tǒng)計(jì)上受到了廣泛的挑戰(zhàn)(Altman & Krzywinski, 2017)。因?yàn)槭褂貌煌姆椒ê蛿?shù)據(jù)樣本獲得的結(jié)果通常不能完全重復(fù)或識(shí)別出三個(gè)以上的類型(Herzberg & Roth, 2006), 甚至發(fā)現(xiàn)即使是得到RUO分類的研究也顯示出較大的差異(Gerlach et al., 2018), 這都表明在一定程度上, RUO三種人格類型缺乏共識(shí)和可復(fù)制性。

目前, 少量實(shí)證研究基于五因素模型在抑郁癥病人中發(fā)現(xiàn)了2種和5種人格類型。例如, Wardenaar和同事對(duì)146名重度抑郁癥患者使用潛在剖面分析, 確定了2種人格類型。分別是具有高神經(jīng)質(zhì), 低外向性, 低責(zé)任心, 高宜人性趨勢(shì)的易感型和具有中等神經(jīng)質(zhì), 中等外向性, 高責(zé)任心, 高宜人性趨勢(shì)的彈性型(Wardenaar et al., 2014)。另一項(xiàng)針對(duì)急性冠脈綜合征的抑郁癥患者的縱向研究, 采用非層次K-means聚類分析, 對(duì)685名抑郁癥患者進(jìn)行分類, 也發(fā)現(xiàn)了彈性型和易感型兩種人格類型(Kim et al., 2016)。此外還有一項(xiàng)關(guān)于抑郁癥?焦慮癥共病患者的分類研究。該研究采用潛在類別分析定義了5種人格類型, 這5種類型歸納起來(lái)是彈性型的2個(gè)亞型和過(guò)度控制型的3個(gè)亞型(Spinhoven et al., 2012)。綜上, 抑郁癥人格類型研究同樣缺乏一致性, 但大部分研究發(fā)現(xiàn)了彈性型和易感型兩種類型。

其次, 以往研究基本采用無(wú)監(jiān)督聚類的方法來(lái)識(shí)別人格亞型。對(duì)于僅僅探索所有人格特質(zhì)可能形成哪些人格類型的研究而言, 無(wú)監(jiān)督聚類的方法是合適的。但是對(duì)于探索結(jié)合特定問(wèn)題或結(jié)果(比如抑郁癥狀嚴(yán)重程度, 是否診斷為抑郁癥, 是否具有創(chuàng)造性)的人格類型而言, 無(wú)監(jiān)督聚類不能完全滿足需要。并且, 從認(rèn)知功能到臨床疾病, 不一定都與相同的人格特質(zhì)有關(guān); 相反, 它們可能對(duì)應(yīng)于不同人格特質(zhì)的輸入組合。例如, 研究者預(yù)期尋找具有高創(chuàng)造性的個(gè)體, 那么在聚類過(guò)程中, 需要將開(kāi)放性維度作為輸入特征(研究表明開(kāi)放性與創(chuàng)造性存在高度相關(guān)) (Dollinger et al., 2004), 而宜人性不一定需要加入其中。目前神經(jīng)質(zhì), 外向性對(duì)抑郁癥狀中等到強(qiáng)的預(yù)測(cè)效力已被廣泛證明(McDonnell & Semkovska, 2020), 此外也有研究提到責(zé)任心對(duì)預(yù)測(cè)抑郁存在微弱的效力(Miller et al., 2020), 個(gè)別研究也提到開(kāi)放性和宜人性與抑郁癥狀的相關(guān)(Khoo & Simms, 2018), 但后三者的可重復(fù)性很低。因此, 本研究選取與抑郁癥最相關(guān)的人格特質(zhì)(神經(jīng)質(zhì)和外向性)進(jìn)行后續(xù)的聚類分析。聚類分析則采用最近提出的應(yīng)用于臨床精神病學(xué)領(lǐng)域的功能隨機(jī)森林(functional random forest)算法(Chand et al., 2020)算法。綜上, 不同于以往抑郁癥類型研究采用的預(yù)先設(shè)定類別數(shù)目的無(wú)監(jiān)督聚類(非層次K-means聚類分析, 潛在類別分析等)方法, 該方法是一種整合了有監(jiān)督(隨機(jī)森林機(jī)器學(xué)習(xí))和無(wú)監(jiān)督(社區(qū)探測(cè)聚類算法)統(tǒng)計(jì)的新穎的混合方法, 更加契合本研究。

另一方面, 研究表明, 抑郁癥患者的大腦功能與健康對(duì)照組存在顯著差異。尤其是杏仁核、海馬、腦島(Milne et al., 2012; Tang et al., 2018; Yan et al., 2022)等皮層下區(qū)域和邊緣網(wǎng)絡(luò)、默認(rèn)網(wǎng)絡(luò)、控制網(wǎng)絡(luò)等皮層上網(wǎng)絡(luò)(Peters et al., 2016; Rai et al., 2021; Scalabrini et al., 2020)。如下以抑郁癥被試為主的研究或綜述表明, 杏仁核是感知和識(shí)別情緒的中樞(Peluso et al., 2009); 海馬與情景記憶的檢索有關(guān)(Lorenzetti et al., 2009); 腦島與注意監(jiān)測(cè), 情緒感知, 獎(jiǎng)賞系統(tǒng)和決策等功能有關(guān)(Menon & Uddin, 2010; Sprengelmeyer et al., 2011)。它們是感知, 傳遞和整合情緒的關(guān)鍵區(qū)域, 與邊緣網(wǎng)絡(luò)、默認(rèn)網(wǎng)絡(luò)、控制網(wǎng)絡(luò)協(xié)同調(diào)節(jié)一系列復(fù)雜的情緒和生理反應(yīng)(抑郁癥患者在這些腦功能上表現(xiàn)出異常的激活或連接模式) (Alexopoulos, 2002; Scalabrini et al., 2020; Sridharan et al., 2008)。此外, 這些腦功能活動(dòng)還與人格特質(zhì)有關(guān)。例如, 抑郁癥患者在負(fù)性情緒認(rèn)知重評(píng)任務(wù)中, 情緒易感性(神經(jīng)質(zhì)子維度)越高, 背外側(cè)前額葉和杏仁核之間連接越弱(Fournier et al., 2017)。然而, 少有研究者探索抑郁癥患者人格類型的神經(jīng)基礎(chǔ)(Knyazev, 2006), 其神經(jīng)機(jī)制尚不明確。但這方面的研究可從神經(jīng)層面闡釋抑郁癥異質(zhì)性。

綜上, 本研究旨在揭示抑郁癥不同的人格類型, 并從大腦功能連接的角度驗(yàn)證不同類型的差異。不僅可為人格分類的有效性提供證據(jù), 也為未來(lái)的抑郁癥異質(zhì)性研究提供參考。如前所述, 我們選取神經(jīng)質(zhì)和外向性(與抑郁癥最相關(guān), 可重復(fù)性最高的人格特質(zhì))作為輸入特征, 采用功能隨機(jī)森林進(jìn)行聚類分析。然后根據(jù)分類結(jié)果, 驗(yàn)證人格類型在關(guān)鍵(與抑郁癥相關(guān)的)皮層下區(qū)域和皮層上網(wǎng)絡(luò)(杏仁核/海馬/腦島?邊緣網(wǎng)絡(luò)/默認(rèn)網(wǎng)絡(luò)/控制網(wǎng)絡(luò))的靜息態(tài)功能連接是否存在差異。抑郁癥類型研究尚無(wú)一致結(jié)論, 但可在一定程度上假設(shè)抑郁癥中可能包含兩種類型, 一種是低神經(jīng)質(zhì)和高外向性的低風(fēng)險(xiǎn)型(與前人研究中彈性型的神經(jīng)質(zhì)外向性的趨勢(shì)相同), 一種是高神經(jīng)質(zhì)和低外向性的高風(fēng)險(xiǎn)類型(與前人研究中易感型的神經(jīng)質(zhì)外向性的趨勢(shì)相同)。并且假設(shè), 各人格類型在關(guān)鍵的靜息態(tài)功能連接上存在差異。相較于以往研究從統(tǒng)計(jì)上構(gòu)建的, 不一定存在于現(xiàn)實(shí)情境中的人格特質(zhì)的交互效應(yīng), 本研究識(shí)別的抑郁癥人格類型更符合現(xiàn)實(shí)情況以及個(gè)體的認(rèn)知模式。不僅可以更好地揭示高抑郁風(fēng)險(xiǎn)的人格類型, 其所體現(xiàn)出的和標(biāo)簽一樣的便捷與系統(tǒng)性, 也更具有潛在的臨床應(yīng)用價(jià)值。而且, 對(duì)不同類型的大腦功能連接基礎(chǔ)的探索, 不僅可以檢驗(yàn)分類的有效性, 還有助于從神經(jīng)層面更好地認(rèn)識(shí)抑郁癥病人的異質(zhì)性, 更好地幫助未來(lái)抑郁癥腦機(jī)制的研究。

2 方法

2.1 被試

抑郁癥患者來(lái)自重慶醫(yī)科大學(xué)附屬醫(yī)院精神科。本研究中抑郁癥患者選取標(biāo)準(zhǔn)如下：符合精神疾病診斷與統(tǒng)計(jì)手冊(cè)第4版(軸Ⅰ障礙)中抑郁癥的診斷標(biāo)準(zhǔn); 無(wú)嚴(yán)重軀體疾病; 無(wú)嚴(yán)重神經(jīng)系統(tǒng)疾病; 過(guò)去2周, 無(wú)急性或慢性感染, 無(wú)創(chuàng)傷、炎癥、發(fā)熱和過(guò)敏史; 無(wú)酒精和藥物濫用史。此外, 抑郁癥狀嚴(yán)重程度評(píng)估使用17題的漢密爾頓抑郁量表(Hamilton Depression Scale, HAMD) (Hamilton, 1960), 該量表得分小于7分, 便沒(méi)有抑郁癥狀。一共選取了159名患者(平均年齡為38.9歲, 標(biāo)準(zhǔn)差為13.3歲, 其中有98名女性)。所有病人的漢密爾頓抑郁量表得分均大于等于7分?？刂平M被試則從大學(xué)以及周邊社區(qū)招募, 其篩選標(biāo)準(zhǔn)如下; 無(wú)抑郁癥發(fā)作以及抑郁癥病史; 無(wú)其他精神疾病史; 一級(jí)親屬無(wú)精神類疾病; 無(wú)嚴(yán)重軀體疾病和神經(jīng)系統(tǒng)疾病史; 過(guò)去2周, 無(wú)急性或慢性感染, 無(wú)創(chuàng)傷、炎癥、發(fā)熱和過(guò)敏史; 無(wú)酒精及藥物濫用史; 一共選取了156人(平均年齡為41.7歲, 標(biāo)準(zhǔn)差為15.9歲, 其中有102名女性)。所有控制組的HAMD得分均小于7分?？紤]到艾森克人格問(wèn)卷中, 掩飾維度得分過(guò)高的被試會(huì)有掩飾傾向(問(wèn)卷結(jié)果可能失真), 因此選取的所有被試得分均未高于18分。本研究開(kāi)始前, 所有被試均被詳細(xì)告知本研究的目的、具體施測(cè)方法以及潛在的風(fēng)險(xiǎn)和不適。所有程序均在被試自愿參與的原則下進(jìn)行, 并且與被試簽署了知情同意書。抑郁癥和控制組在年齡, 性別和教育年限上無(wú)顯著差異。詳細(xì)的基本人口學(xué)信息見(jiàn)表1。

表1 基本人口學(xué)信息

注：HAMD：漢密爾頓抑郁量表。

2.2 數(shù)據(jù)采集

2.2.1 心理測(cè)量

人格特征的評(píng)估采用成人艾森克人格問(wèn)卷的中文版(王潔等, 2013)。艾森克人格問(wèn)卷最開(kāi)始便是應(yīng)用臨床專業(yè)疾病測(cè)試領(lǐng)域(現(xiàn)今醫(yī)院的精神科和心理科廣泛使用艾森克問(wèn)卷)。該問(wèn)卷一共分為4個(gè)分維度：神經(jīng)質(zhì)(neuroticism), 外向性(extraversion), 精神質(zhì)(psychoticism)和掩飾(lie) (陳仲庚, 1983)。較高的神經(jīng)質(zhì)分?jǐn)?shù)反映出個(gè)體易體驗(yàn)到負(fù)面情緒, 情緒易變, 容易過(guò)度反應(yīng), 且在體驗(yàn)到一種情緒后不容易恢復(fù)到常態(tài)等。較高的外向性分?jǐn)?shù)反映出個(gè)體是開(kāi)朗的, 健談的, 沖動(dòng)的和非抑制的, 擁有廣泛的社交接觸等, 體驗(yàn)到的積極情緒高等。精神質(zhì)與攻擊性, 反社會(huì)行為, 沖動(dòng)性等相關(guān)(Cale, 2006)。掩飾維度反應(yīng)個(gè)體對(duì)真實(shí)情況的掩飾程度, 得分越高掩飾程度越高。這份問(wèn)卷涵蓋了本研究感興趣的人格特質(zhì), 且艾森克人格問(wèn)卷在各個(gè)國(guó)家和地區(qū)廣受歡迎(Bowden et al., 2016), 自1983年被引入中國(guó)以來(lái), 在中國(guó)也得到了廣泛使用(陳仲庚, 1983)。

2.2.2 靜息態(tài)數(shù)據(jù)采集與預(yù)處理

靜息態(tài)功能性磁共振影像數(shù)據(jù)均采集于西門子掃描儀(3.0T Siemens Trio MRI)。掃描過(guò)程中, 所有被試被要求閉上眼睛, 平躺休息, 不做思考和回憶某些特定的事件, 但需要保持清醒。靜息態(tài)掃描使用序列是全腦梯度平面回波成像序列。具體參數(shù)設(shè)置如下：回波時(shí)間 = 30ms; 重復(fù)時(shí)間 = 2000ms; 翻轉(zhuǎn)角= 90°; 矩陣大小= 64×64; 視野大小= 192 mm × 192 mm; 層數(shù)(Slice) = 32; 厚度 = 3 mm; 層間距 = 1 mm; 體素大小 = 3.4 mm × 3.4 mm × 4 mm。整個(gè)掃描過(guò)程持續(xù)8分4秒。

預(yù)處理在中科院嚴(yán)超贛團(tuán)隊(duì)開(kāi)發(fā)的DPARSF (Yan & Zang, 2010) (http:// restfmri.net)工具包中進(jìn)行。首先檢查所獲被試圖像的質(zhì)量, 刪除質(zhì)量不佳的被試。然后按照如下步驟進(jìn)行處理：刪除影像的前10個(gè)時(shí)間點(diǎn)(磁共振觸發(fā)進(jìn)行掃描的初始信號(hào)不穩(wěn)定); 時(shí)間層校正, 用于校正1個(gè)volume中層與層之間掃描時(shí)間的差異; 頭動(dòng)校正, 用于減弱被試頭動(dòng)的影響; 空間標(biāo)準(zhǔn)化, 將個(gè)體的大腦配準(zhǔn)到標(biāo)準(zhǔn)的MNI模板上去, 同時(shí)將體素大小重采樣為3 mm × 3 mm × 3 mm; 圖像平滑(平滑核大小為8 mm);回歸白質(zhì)信號(hào), 腦脊液信號(hào)和Friston提出的24個(gè)頭動(dòng)參數(shù)(Friston et al., 1996); 濾波(0.01～0.1 Hz); 最后, 使用DPARSF的scrubbing功能進(jìn)一步降低頭動(dòng)的影響(Power et al., 2014)。預(yù)處理流程與已發(fā)表文章一致(Cheng, Rolls, Qiu, Xie, Wei et al., 2018), 詳細(xì)信息可參考該文章。

2.3 數(shù)據(jù)分析

2.3.1 聚類分析

新近提出的應(yīng)用于臨床精神病學(xué)領(lǐng)域的功能隨機(jī)森林算法越來(lái)越受到研究者青睞, 本研究將采用此方法劃分抑郁癥病人的人格亞型。該算法運(yùn)用有監(jiān)督的方法(自上而下的, 對(duì)亞群進(jìn)行明確的假設(shè), 然后強(qiáng)制數(shù)據(jù)符合這些假設(shè))和無(wú)監(jiān)督的方法(自下而上的, 根據(jù)數(shù)據(jù)本身的結(jié)構(gòu)或形狀來(lái)識(shí)別亞型, 不需要提前設(shè)定類別數(shù)目)將感興趣的臨床診斷與異質(zhì)性結(jié)合起來(lái), 克服了無(wú)監(jiān)督聚類所得類別與研究者所感興趣問(wèn)題無(wú)關(guān)的缺陷(Chand et al., 2020)。

首先, 將神經(jīng)質(zhì)和外向性(即輸入特征)通過(guò)隨機(jī)森林模型與抑郁癥診斷相擬合, 隨機(jī)種子設(shè)置為1234, 擬合500棵決策樹, 并使用十折交叉驗(yàn)證評(píng)估模型性能。隨機(jī)森林單棵決策樹的預(yù)測(cè)模型基于分類決策樹(預(yù)測(cè)變量為二分變量即是否為抑郁癥病人)。決策樹以一個(gè)倒立的樹形呈現(xiàn), 樹上的內(nèi)部節(jié)點(diǎn)代表自變量(即輸入特征), 節(jié)點(diǎn)之間的連接代表決策, 終端節(jié)點(diǎn)(葉節(jié)點(diǎn))代表一個(gè)結(jié)果(預(yù)測(cè)變量)。隨機(jī)森林分析基于Rstudio平臺(tái)的CORElearn程序包, 采用其內(nèi)置的Breiman的隨機(jī)森林算法。該算法融合了bagging集成算法和CART決策樹。計(jì)算的具體步驟如下：從樣本中隨機(jī)有放回地重采樣個(gè)訓(xùn)練樣本; 對(duì)個(gè)訓(xùn)練樣本建立CART決策樹模型; 重復(fù)前兩個(gè)步驟500次, 即獲得500棵決策樹, 形成隨機(jī)森林; 然后對(duì)來(lái)自每棵樹的結(jié)果進(jìn)行多數(shù)投票的方式輸出預(yù)測(cè)結(jié)果。詳細(xì)計(jì)算公式參見(jiàn)(Breiman, 2001)。其中, 一對(duì)被試被分配到CART決策樹的同一終端節(jié)點(diǎn)中的比率代表兩者的相似性。如圖1中的相似性矩陣, 行和列代表被試, 每個(gè)單元格代表在所有樹中成對(duì)被試落到相同終端節(jié)點(diǎn)的概率。其次, 使用Rstudio中ExplainPrediction軟件包對(duì)生成的隨機(jī)森林模型進(jìn)行解釋。該軟件包可計(jì)算一系列的評(píng)價(jià)指標(biāo), 如通過(guò)生成誤差矩陣(又名混淆矩陣)獲得準(zhǔn)確率, 特異性, 敏感度等評(píng)估指標(biāo)。

然后, 基于相似性矩陣分別對(duì)抑郁癥和控制組進(jìn)行社區(qū)探測(cè)聚類分析, 最后得到與抑郁癥診斷關(guān)聯(lián)的人格類型。其中, 社區(qū)檢測(cè)是一種圖論方法, 迭代地用于識(shí)別子組。本研究使用的社區(qū)檢測(cè)算法是廣義Louvain算法(Generalized Louvain Method), 基于模塊度Q最優(yōu)化的概念(Blondel et al., 2008), 通過(guò)對(duì)社區(qū)結(jié)構(gòu)反復(fù)迭代, 直至獲得最大的模塊度, 即最優(yōu)的分類情況。該算法廣泛應(yīng)用于各種研究, 并提供了可靠的結(jié)果(Cole et al., 2014)。由于該算法涉及到零模型(Newman-Girvan零模型), 社區(qū)探測(cè)結(jié)果會(huì)受到一定程度的擾動(dòng), 并不唯一(具有一定的隨機(jī)性), 因此在每個(gè)被試上該算法都將被執(zhí)行100次。經(jīng)過(guò)100次運(yùn)行的模塊化算法, 可獲得100個(gè)模塊度Q值。經(jīng)過(guò)平均后的質(zhì)量指標(biāo)Q可作為評(píng)估所劃分的社區(qū)團(tuán)塊之間的分離強(qiáng)度。本研究基于抑郁癥病人和控制組被試的相似性矩陣, 分別對(duì)其進(jìn)行社區(qū)探測(cè)分析, 以獲得人格類型。社區(qū)探測(cè)的分析依賴于GenLouvain工具包(https://github. com/GenLouvain/GenLouvain)。參考前人研究(Mantini et al., 2013), 將其中兩個(gè)重要參數(shù)gamma (可調(diào)整社區(qū)規(guī)模)和omega (影響探測(cè)到的社區(qū)的穩(wěn)定性)設(shè)置為1。

2.3.1 靜息態(tài)功能連接分析

為了探索不同分類在杏仁核, 海馬, 腦島與邊緣網(wǎng)絡(luò), 默認(rèn)網(wǎng)絡(luò), 控制網(wǎng)絡(luò)的功能連接強(qiáng)度上是否在存在差異, 也進(jìn)一步驗(yàn)證分類的有效性, 本研究接下來(lái)進(jìn)行功能連接分析。首先, 定義感興趣區(qū)。本研究采用400節(jié)點(diǎn)對(duì)應(yīng)17個(gè)功能網(wǎng)絡(luò)的Schaefer-Yeo腦功能模板(Schaefer et al., 2018)。該模板廣泛應(yīng)用于大腦功能網(wǎng)絡(luò)的研究中, 具有高特異性和高網(wǎng)絡(luò)同質(zhì)性(Kong et al., 2019)。該模板不包含皮層下區(qū)域, 但詳細(xì)劃分了各功能網(wǎng)絡(luò), 例如2個(gè)邊緣網(wǎng)絡(luò)(LimbicA_TempPole, LimbicB_OFC), 3個(gè)控制網(wǎng)絡(luò)(ContA_Cingm, ContB_PFCmp, ContC_ Cingp)和3個(gè)默認(rèn)網(wǎng)絡(luò)(DefaultA_PFCm, DefaultB_ PFCv, DefaultC_PHC)。本研究選取這8個(gè)網(wǎng)絡(luò)涉及的所有節(jié)點(diǎn)作為感興趣區(qū), 具體節(jié)點(diǎn)詳細(xì)信息見(jiàn)網(wǎng)絡(luò)版附錄表S3。其次, 基于AAL模板定義左、右側(cè)杏仁核, 海馬和腦島, 共6個(gè)皮層下感興趣區(qū)。然后, 將選取的皮層下感興趣區(qū)與網(wǎng)絡(luò)所有節(jié)點(diǎn)做皮爾遜相關(guān), 即計(jì)算功能連接強(qiáng)度。再基于節(jié)點(diǎn)間的功能連接, 計(jì)算皮層下感興趣區(qū)與8個(gè)網(wǎng)絡(luò)間的功能連接強(qiáng)度。最后, 基于聚類分析得到的各組別, 對(duì)皮層下感興趣區(qū)與8個(gè)網(wǎng)絡(luò)的功能連接強(qiáng)度進(jìn)行單因素方差分析。該分析使用Gretna工具包, 并采用FDR進(jìn)行多重比較校正。由于有6個(gè)皮層下區(qū)域與皮層上網(wǎng)絡(luò)做相關(guān), 需分別做6次方差分析, 因此方差分析的FDR多重比較校正閾值設(shè)置為0.0083, 小于0.05/6?；谕ㄟ^(guò)多重比較校正的皮層下區(qū)域?皮層上網(wǎng)絡(luò)的功能連接, 再進(jìn)行事后檢驗(yàn)。方差分析和事后檢驗(yàn)中, 均將性別, 年齡和教育年限作為協(xié)變量。

圖1 聚類流程

注：隨機(jī)森林：?jiǎn)晤w決策樹上的內(nèi)部節(jié)點(diǎn)代表自變量(即輸入特征), 節(jié)點(diǎn)之間的連接代表決策, 終端節(jié)點(diǎn)(葉節(jié)點(diǎn))代表一個(gè)結(jié)果(即預(yù)測(cè)變量)。相似性矩陣：行和列代表被試, 每個(gè)單元格代表在所有樹中成對(duì)被試落到相同終端節(jié)點(diǎn)的概率。

另外, 我們同樣對(duì)抑郁癥和控制組在6個(gè)皮層下感興趣區(qū)與8個(gè)網(wǎng)絡(luò)的功能連接強(qiáng)度上進(jìn)行了雙樣本檢驗(yàn)。以考察在未進(jìn)行聚類分析, 沒(méi)有劃分亞型的情況下, 抑郁癥和控制組是否在這些功能連接上存在顯著差異。并且, 我們還檢驗(yàn)了神經(jīng)質(zhì)、外向性是否與我們感興趣的功能連接存在相關(guān)。雙樣本檢驗(yàn)和相關(guān)分析均采用FDR進(jìn)行多重比較校正, 閾值設(shè)置為0.0083, 小于0.05/6。

3 研究結(jié)果

3.1 聚類分析結(jié)果

十折交叉驗(yàn)證分析顯示, 隨機(jī)森林預(yù)測(cè)模型的平均準(zhǔn)確性(被正確分類為抑郁癥或控制組的所占比率)為77.50%、特異性(控制組被試被成功預(yù)測(cè)的概率)為74.96%, 靈敏度(抑郁癥被試被成功預(yù)測(cè)的概率)為80.04%。據(jù)此可以看出模型擬合良好, 能夠較為準(zhǔn)確地區(qū)分抑郁癥病人和控制組被試。接下來(lái), 將廣義Louvain社區(qū)探測(cè)分析應(yīng)用于通過(guò)隨機(jī)森林建模生成的相似性矩陣。抑郁癥病人的社區(qū)探測(cè)分析結(jié)果顯示有4種人格類型, 平均質(zhì)量指標(biāo)Q為0.19 (表示網(wǎng)絡(luò)具有模塊性)。其中類型1有41人, 占總?cè)藬?shù)的25.79%; 類型2有44人, 占總?cè)藬?shù)的27.67%; 類型3有42人, 占總?cè)藬?shù)的26.42%; 類型4有32人, 占總?cè)藬?shù)的20.13%。類型1呈現(xiàn)低神經(jīng)質(zhì)和中等偏高的外向性水平; 類型2和4呈現(xiàn)出高神經(jīng)質(zhì)和低外向性水平, 類型3呈現(xiàn)出高神經(jīng)質(zhì)和中等偏高的外向性水平, 詳見(jiàn)圖2和圖S1 (https://chart-studio.plotly.com/～liliyuyu/1/#/)。不同類型的占比相對(duì)均衡, 其中高神經(jīng)質(zhì)和低外向性水平的類型占比最高?？刂平M樣本的社區(qū)探測(cè)分析結(jié)果顯示有5種人格類型, 平均質(zhì)量指標(biāo)Q為0.31 (表示網(wǎng)絡(luò)具有模塊性)。其中類型1有47人, 占總?cè)藬?shù)的30.13%; 類型2有39人, 占總?cè)藬?shù)的25%; 類型3有44人, 占總?cè)藬?shù)的28.21%; 類型4有15人, 占總?cè)藬?shù)的9.62%; 類型5有11人, 占總?cè)藬?shù)的7.05%。類型1與類型3呈現(xiàn)低神經(jīng)質(zhì), 高或者中等偏高的外向性水平, 與前人研究中彈性型的低神經(jīng)質(zhì)高外向性的趨勢(shì)相同(Kim et al., 2016); 類型2的神經(jīng)質(zhì)和外向性都處于中等偏低的程度, 與前人研究中平均型的神經(jīng)質(zhì)和外向性水平相似(Leikas & Salmela-Aro, 2014); 類型4呈現(xiàn)中等神經(jīng)質(zhì)和高外向性的趨勢(shì); 類型5則呈現(xiàn)高神經(jīng)質(zhì)和中等外向性的趨勢(shì), 詳見(jiàn)圖2和圖S1 (https://chart- studio.plotly.com/～liliyuyu/1/#/)。類型4和5占比較少, 類型1, 2, 3占比較高且相對(duì)均衡。整體而言, 抑郁癥以高神經(jīng)質(zhì)和低外向性趨勢(shì)的個(gè)體為主, 但也有低神經(jīng)質(zhì)和高外向性的個(gè)體?？刂平M則以低神經(jīng)質(zhì)和高外向性個(gè)體為主。

圖2 抑郁癥與控制組人格類型的神經(jīng)質(zhì)和外向性的Z分?jǐn)?shù)

注：抑郁癥類型1呈現(xiàn)低神經(jīng)質(zhì)和中等偏高的外向性水平; 類型2和4呈現(xiàn)出高神經(jīng)質(zhì)和低外向性水平, 類型3呈現(xiàn)出高神經(jīng)質(zhì)和中等偏高的外向性水平?？刂平M類型1與類型3呈現(xiàn)低神經(jīng)質(zhì), 高或者中等偏高的外向性水平; 類型2的神經(jīng)質(zhì)和外向性都處于中等偏低的程度; 類型4呈現(xiàn)中等神經(jīng)質(zhì)和高外向性的趨勢(shì); 類型5則呈現(xiàn)高神經(jīng)質(zhì)和中等外向性的趨勢(shì)。N: 神經(jīng)質(zhì)(neuroticism); E: 外向性(extraversion) ; CON: control控制組; DD: depressive disorder 抑郁癥。DD1: 抑郁癥類型1; CON1: 控制組類型1。

同時(shí), 本研究通過(guò)獨(dú)立樣本檢驗(yàn), 考察了抑郁癥和控制組在神經(jīng)質(zhì)和外向性上是否存在差異。結(jié)果顯示, 抑郁癥的神經(jīng)質(zhì)水平顯著高于控制組(Mean difference = 7.77,(313) = 13.98,< 0.001, Cohen’s= 1.575), 外向性顯著低于控制組(Mean difference = ?4.09,(313) = ?7.94,< 0.001, Cohen’s= ?0.895)?；诰垲惙治鏊鶆澐值?個(gè)組, 本研究進(jìn)一步通過(guò)單因素方差分析(性別、年齡和教育年限作為協(xié)變量), 檢驗(yàn)了神經(jīng)質(zhì)和外向性的組間差異。結(jié)果表明, 神經(jīng)質(zhì)((8, 303)= 131.62,< 0.001, η2= 0.771)和外向性((8, 303)= 51.79,< 0.001, η2= 0.575)均存在組間差異。事后檢驗(yàn)詳細(xì)結(jié)果見(jiàn)網(wǎng)絡(luò)版附錄表S1。

3.2 靜息態(tài)功能連接分析結(jié)果

經(jīng)過(guò)預(yù)處理, 抑郁癥樣本有125名被試的數(shù)據(jù)可用, 其中類型1有32人, 類型2有32人, 類型3有33人, 類型4有28人?？刂平M樣本有122名被試的數(shù)據(jù)可用, 其中類型1有34人, 類型2有28人, 類型3有37人, 類型4有12人, 類型5有11人?？刂平M類型4和5人數(shù)過(guò)少, 因此未納入后續(xù)的統(tǒng)計(jì)分析。首先, 在所有組別(7種類型)上的單因素方差分析(性別, 年齡和教育年限作為協(xié)變量)的結(jié)果顯示：經(jīng)過(guò)FDR多重比較校正(閾值為0.05/6), 左側(cè)杏仁核與邊緣網(wǎng)絡(luò)(LimbicB_OFC,(6, 214) = 4.273,= 0.0004), 左側(cè)腦島與邊緣網(wǎng)絡(luò)(LimbicB_ OFC,(6, 214) = 4.177,= 0.0005)的功能連接強(qiáng)度存在組間差異; 杏仁核, 海馬, 腦島與默認(rèn)網(wǎng)絡(luò), 控制網(wǎng)絡(luò)的功能連接均不存在顯著差異。然后, 對(duì)經(jīng)過(guò)多重比較校正的左側(cè)杏仁核?邊緣網(wǎng)絡(luò)(LimbicB_OFC)進(jìn)行事后檢驗(yàn)(兩兩比較采用“Holm”法校正)的結(jié)果顯示, 抑郁癥類型1 (= ?3.47,= 214,= 0.013, Cohen’s= ?0.977)和類型2 (= ?3.34,= 214,= 0.018, Cohen’s= ?0.851)的左側(cè)杏仁核?邊緣網(wǎng)絡(luò)(LimbicB_OFC)功能連接強(qiáng)度顯著低于控制組類型3; 控制組類型2的左側(cè)杏仁核?邊緣網(wǎng)絡(luò)(LimbicB_OFC)功能連接強(qiáng)度顯著低于類型1 (= ?3.38,= 214,= 0.016, Cohen’s= ?0.893)和類型3 (= ?3.49,= 214,= 0.013, Cohen’s= ?1.180), 詳見(jiàn)表2和圖3。同樣地, 對(duì)經(jīng)過(guò)多重比較校正的左側(cè)腦島與邊緣網(wǎng)絡(luò)(LimbicB_OFC)的事后檢驗(yàn)結(jié)果顯示, 抑郁癥類型2的左側(cè)腦島與邊緣網(wǎng)絡(luò)(LimbicB_OFC)的功能連接強(qiáng)度顯著低于控制組類型3 (= ?4.06,= 214,= 0.001, Cohen’s= ?1.034); 抑郁癥類型3的左側(cè)腦島與邊緣網(wǎng)絡(luò)(LimbicB_OFC)的功能連接強(qiáng)度高于類型2 (邊緣顯著,= 3.04,= 214,= 0.053, Cohen’s= 0.757); 控制組類型2和類型3連接強(qiáng)度差異也呈邊緣顯著(= ?2.94,= 214,= 0.070, Cohen’s= ?0.995), 詳見(jiàn)表2和圖4。上述差異顯著的功能連接強(qiáng)度的效應(yīng)量絕對(duì)值均大于0.8, 屬于較大的效應(yīng)量。

表2 左側(cè)杏仁核/腦島?邊緣網(wǎng)絡(luò)功能連接強(qiáng)度的事后檢驗(yàn)結(jié)果

注：CON: control控制組; DD: depressive disorder 抑郁癥。DD1: 抑郁癥類型1; CON1: 控制組類型1。

圖3 左側(cè)杏仁核與邊緣網(wǎng)絡(luò)的功能連接強(qiáng)度的組間差異

注：CON: control控制組; DD: depressive disorder 抑郁癥。DD1: 抑郁癥類型1; CON1: 控制組類型1。

此外, 抑郁癥和控制組的雙樣本檢驗(yàn)顯示, 經(jīng)過(guò)FDR多重比較校正, 抑郁癥和控制組在6個(gè)皮層下感興趣區(qū)與8個(gè)網(wǎng)絡(luò)的功能連接強(qiáng)度上不存在顯著差異。但相關(guān)分析顯示, 外向性與左/右側(cè)腦島?邊緣網(wǎng)絡(luò)(LimbicB_OFC), 左/右側(cè)腦島?默認(rèn)網(wǎng)絡(luò)(DefaultA_PFCm), 左/右側(cè)腦島?默認(rèn)網(wǎng)絡(luò)(DefaultB_PFCv)功能連接強(qiáng)度顯著相關(guān), 詳細(xì)信息見(jiàn)網(wǎng)絡(luò)版附錄表S2。

圖4 左側(cè)腦島與邊緣網(wǎng)絡(luò)的功能連接強(qiáng)度的組間差異

注：CON: control控制組; DD: depressive disorder 抑郁癥。DD1: 抑郁癥類型1; CON1: 控制組類型1。

4 討論

本研究從個(gè)人中心視角出發(fā), 借助功能隨機(jī)森林, 將人格類型的分類過(guò)程與抑郁癥的診斷相結(jié)合, 分別在抑郁癥和控制組中識(shí)別出了與抑郁診斷關(guān)聯(lián)的人格類型, 并基于此探索了人格類型的靜息態(tài)功能連接基礎(chǔ)。聚類分析結(jié)果顯示, 抑郁癥以高神經(jīng)質(zhì)和低外向性趨勢(shì)的個(gè)體為主, 但同樣有低神經(jīng)質(zhì)和高外向性趨勢(shì)的個(gè)體。控制組樣本則以低神經(jīng)質(zhì)和高外向性個(gè)體為主。靜息態(tài)功能連接的結(jié)果顯示：在不考慮人格亞型的情況下, 抑郁癥和控制組在杏仁核/海馬/腦島?邊緣網(wǎng)絡(luò)/默認(rèn)網(wǎng)絡(luò)/控制網(wǎng)絡(luò)的功能連接上均無(wú)顯著差異。在納入聚類分析所劃分的亞型進(jìn)行統(tǒng)計(jì)后, 多種人格類型在左側(cè)杏仁核/腦島?邊緣網(wǎng)絡(luò)(LimbicB_OFC)的功能連接強(qiáng)度上呈現(xiàn)出顯著差異。

人格特質(zhì)常與抑郁癥的易感性聯(lián)系在一起(Jones et al., 2010; Ormel et al., 2013; Zuroff et al., 2004), 其中神經(jīng)質(zhì)與抑郁易感性的研究尤為豐富(Duggan et al., 1995; 龐雅靜, 2020)。研究表明, 個(gè)體神經(jīng)質(zhì)越高, 其抑郁情緒調(diào)節(jié)能力、心理復(fù)原力越弱, 負(fù)性思維和生活壓力程度越高, 其抑郁癥狀嚴(yán)重程度也更高(Kendler et al., 2004; Kercher et al., 2009; McDonnell & Semkovska, 2020; Yoon et al., 2013)。外向性同樣與情緒調(diào)節(jié)和復(fù)原力等特質(zhì)相關(guān), 其與抑郁的相關(guān)也可被這些特質(zhì)影響(Kokkonen & Pulkkinen, 2001; McDonnell & Semkovska, 2020)。眾多直接考察人格特質(zhì)與抑郁關(guān)系的研究也證實(shí)高神經(jīng)質(zhì)、低外向性是抑郁癥的“風(fēng)險(xiǎn)因素” (Abbasi et al., 2018; Banjongrewadee et al., 2020; McDonnell & Semkovska, 2020; Murray & O'Neill, 2019)。并且, 研究者在考察人格特質(zhì)對(duì)抑郁的影響時(shí), 還發(fā)現(xiàn)了神經(jīng)質(zhì)和外向性存在交互作用(Allen et al., 2018)。本研究中識(shí)別的抑郁癥人格類型以高神經(jīng)質(zhì)和低外向性趨勢(shì)的個(gè)體為主, 控制組則以低神經(jīng)質(zhì)和高外向性的個(gè)體為主, 這側(cè)面體現(xiàn)了高神經(jīng)質(zhì)和低外向性的“高抑郁風(fēng)險(xiǎn)”。此外, 抑郁癥中低神經(jīng)質(zhì)和高外向性的類型, 與Kim在抑郁癥中發(fā)現(xiàn)的彈性型的神經(jīng)質(zhì)外向性趨勢(shì)相同(Kim et al., 2016)。這表明抑郁癥中存在神經(jīng)質(zhì)和外向性處于低風(fēng)險(xiǎn)水平的個(gè)體, 而他們可能表現(xiàn)出更好的臨床治療效果(Wardenaar et al., 2014)?？刂平M其他人格類型傾向于至少有一個(gè)人格特質(zhì)處于中/低風(fēng)險(xiǎn)的水平, 體現(xiàn)出健康人群大部分處于中低抑郁風(fēng)險(xiǎn)的特點(diǎn)。

本研究結(jié)果中, 邊緣網(wǎng)絡(luò)的主要節(jié)點(diǎn)位于眶額皮質(zhì)(orbitofrontal cortex, OFC), 其位置在解剖學(xué)上與腹內(nèi)側(cè)前額葉皮質(zhì)一致。杏仁核是感知和識(shí)別情緒的中樞, 與自然的情緒和生理喚醒有關(guān)(Peluso et al., 2009)?？纛~皮質(zhì)是情緒處理(整合, 評(píng)估, 獎(jiǎng)賞和決策)的核心區(qū)域(Rolls, 2015), 自上而下調(diào)控杏仁核等其他區(qū)域的活動(dòng)(Spielberg et al., 2015)。杏仁核-OFC的功能連接在眾多研究中被討論, 其功能主要涉及情緒加工, 恐懼消退(Etkin et al., 2011; Kim et al., 2011), 以及抑制/控制杏仁核在面對(duì)負(fù)性刺激時(shí)的過(guò)度激活(Quirk et al., 2000)。研究顯示, 面對(duì)情緒刺激時(shí), 個(gè)體的神經(jīng)質(zhì)程度與杏仁核激活強(qiáng)度呈正相關(guān)(Haas et al., 2007), 但與OFC激活強(qiáng)度呈負(fù)相關(guān)(Kehoe et al., 2012)。而在情緒加工任務(wù)中, 杏仁核激活(積極情緒?消極情緒)與外向性水平正相關(guān)(Canli, 2004); 在獎(jiǎng)賞相關(guān)任務(wù)(獲得獎(jiǎng)賞?未獲得獎(jiǎng)賞)中, 杏仁核和OFC的激活都與外向性正相關(guān)(Cohen et al., 2005)。

由上可知, 低神經(jīng)質(zhì)、高外向性可能與杏仁 核-OFC的高功能連接強(qiáng)度(兩者激活的高同步性)有關(guān), 可能對(duì)應(yīng)著更強(qiáng)的調(diào)節(jié)控制能力, 與我們的結(jié)果相呼應(yīng)。例如, 控制組類型2的左側(cè)杏仁核?邊緣網(wǎng)絡(luò)(LimbicB_OFC)的功能連接強(qiáng)度低于控制組類型1和3。控制組類型2的神經(jīng)質(zhì)和外向性處于中等水平, 而控制組類型1和3處于低神經(jīng)質(zhì)和偏高的外向性水平。其次, 研究表明靜息態(tài)下, 抑郁癥狀越嚴(yán)重, OFC的激活越弱(Rocca et al., 2015), 杏仁核-OFC功能連接也越弱(Cheng, Rolls, Qiu, Xie, Lyu et al., 2018)。側(cè)面支持我們?cè)诳刂平M類型3與抑郁癥類型1上發(fā)現(xiàn)的差異：即使兩者都屬于低神經(jīng)質(zhì)、高外向性的類型, 抑郁癥的患者的杏仁核-OFC連接強(qiáng)度確更低。此外, 我們還發(fā)現(xiàn)抑郁癥類型2的杏仁核?邊緣網(wǎng)絡(luò)(LimbicB_OFC)的功能連接強(qiáng)度顯著低于控制組類型3。結(jié)合上述內(nèi)容, 這種差異可能是相反趨勢(shì)的神經(jīng)質(zhì)和外向性的體現(xiàn), 也可能是樣本差異(是否為抑郁癥)的體現(xiàn)。

腦島是個(gè)體注意監(jiān)測(cè), 情緒感知, 獎(jiǎng)賞系統(tǒng)和決策等功能有關(guān)(Menon & Uddin, 2010; Sprengelmeyer et al., 2011)。研究顯示, 面對(duì)情緒刺激時(shí), 個(gè)體神經(jīng)質(zhì)越高, 腦島激活越強(qiáng)(Paulus et al., 2003); 面對(duì)負(fù)性刺激時(shí), 抑郁癥患者腦島的激活也比控制組更高(Surguladze et al., 2010; Suslow et al., 2010); 而OFC在情緒加工中的激活與神經(jīng)質(zhì)和抑郁癥狀嚴(yán)重程度負(fù)相關(guān)(Kehoe et al., 2012; Rocca et al., 2015)。面對(duì)愉悅物體時(shí), 腦島和OFC的激活均與外向性負(fù)相關(guān)(Hooker et al., 2008); 面對(duì)幽默材料時(shí), 腦島和OFC的激活均與外向性正相關(guān)(Mobbs et al., 2005)?？煽闯? 神經(jīng)質(zhì)與腦島-OFC的功能協(xié)同性可能負(fù)相關(guān), 外向性可能與其正相關(guān)。此外靜息態(tài)下, 抑郁癥患者的腦島-OFC的功能連接強(qiáng)度比控制組弱, 且其功能連接強(qiáng)度與抑郁癥狀嚴(yán)重程度負(fù)相關(guān)(Zhang et al., 2021)。我們的研究結(jié)果也支持以上觀點(diǎn)：抑郁癥類型2 (高神經(jīng)質(zhì)低外向性)的腦島-OFC功能連接強(qiáng)度低于控制組類型3 (低神經(jīng)質(zhì)高外向性)和抑郁癥類型3 (高神經(jīng)質(zhì)高外向性, 邊緣顯著); 控制組類型2 (中等神經(jīng)質(zhì)和外向性)低于控制組類型3 (邊緣顯著)。這可能也是不同的神經(jīng)質(zhì)和外向性組合或者樣本在腦功能連接上差異的體現(xiàn)。

本研究在納入人格亞型進(jìn)行統(tǒng)計(jì)后發(fā)現(xiàn)了以上功能連接的差異, 但在不考慮人格亞型的情況下, 抑郁癥和控制組在選取的功能連接上均無(wú)顯著差異。不僅從側(cè)面說(shuō)明了分類的有效性, 還有助于從神經(jīng)層面更好地認(rèn)識(shí)抑郁癥的異質(zhì)性, 更好地幫助未來(lái)抑郁癥腦機(jī)制與分類干預(yù)治療的研究。此外, 本研究仍存在以下三點(diǎn)局限性：第一, 本研究的聚類分析傾向于數(shù)據(jù)驅(qū)動(dòng), 未來(lái)有待挖掘更加有實(shí)質(zhì)意義的科學(xué)問(wèn)題, 提升創(chuàng)新性。第二, 影像數(shù)據(jù)的分析只進(jìn)行了基于種子點(diǎn)的功能連接分析, 未來(lái)研究可以嘗試進(jìn)行全腦層面的分析, 并結(jié)合多種腦結(jié)構(gòu)和功能的指標(biāo), 深入考察不同人格類型的神經(jīng)基礎(chǔ)。第三, 本研究并沒(méi)有考慮抑郁癥患者的其他癥狀(如焦慮癥狀)和共病情況(如共病焦慮障礙)對(duì)人格分類及其腦功能的影響。未來(lái)研究可考慮在數(shù)據(jù)采集和分析過(guò)程中, 對(duì)其他癥狀和共病情況進(jìn)行詳細(xì)探討。

5 結(jié)論

本研究將聚類過(guò)程與抑郁癥診斷相關(guān)聯(lián), 分別在抑郁癥和控制組中識(shí)別出了人格類型, 并探討了不同類型靜息態(tài)功能連接差異, 加深了對(duì)抑郁癥異質(zhì)性的認(rèn)知。具體發(fā)現(xiàn)：抑郁癥以高神經(jīng)質(zhì)和低外向性趨勢(shì)的類型為主, 但同樣有低神經(jīng)質(zhì)和高外向性趨勢(shì)的類型?？刂平M則以低神經(jīng)質(zhì)和高外向性的類型為主。多種人格類型在左側(cè)杏仁核/腦島?邊緣網(wǎng)絡(luò)的功能連接強(qiáng)度上存在差異顯著：高神經(jīng)質(zhì)和低外向性的類型功能連接可能更弱; 同為低神經(jīng)質(zhì)和高外向性趨勢(shì)的類型, 抑郁癥的功能連接強(qiáng)度仍然更弱。

Abbasi, M., Mirderikvand, F., Adavi, H., & Hojati, M. (2018). The relationship between personality traits (neuroticism and extraversion) and self-efficacy with aging depression.,(4), 458–466.

Alexopoulos, G. S. (2002). Frontostriatal and limbic dysfunction in late-life depression.,(6), 687–695.

Allen, T. A., Carey, B. E., McBride, C., Bagby, R. M., deYoung,C. G., & Quilty, L. C. (2018). Big Five aspects of personality interact to predict depression.,(4), 714–725.

Altman, N., & Krzywinski, M. (2017). Points of significance: Clustering.,(6), 545–547.

Asendorpf, J. B., Borkenau, P., Ostendorf, F., & van Aken, M. A. (2001). Carving personality description at its joints: Confirmation of three replicable personality prototypes for both children and adults.,(3), 169–198.

Banjongrewadee, M., Wongpakaran, N., Wongpakaran, T., Pipanmekaporn, T., Punjasawadwong, Y., & Mueankwan, S. (2020). The role of perceived stress and cognitive function on the relationship between neuroticism and depression among the elderly: A structural equation model approach.,(1), 1–8.

Blondel, V. D., Guillaume, J. -L., Lambiotte, R., & Lefebvre, E. (2008). Fast unfolding of communities in large networks.,(10), P10008.

Bohane, L., Maguire, N., & Richardson, T. (2017). Resilients, overcontrollers and undercontrollers: A systematic review of the utility of a personality typology method in understanding adult mental health problems.,, 75–92.

Bowden, S., Saklofske, D., van de Vijver, F., Sudarshan, N., & Eysenck, S. (2016). Cross-cultural measurement invariance of the Eysenck Personality Questionnaire across 33 countries.,, 53–60.

Breiman, L. (2001). Random forests.,(1), 5–32.

Bryan, C. J., Tipton, E., & Yeager, D. S. (2021). Behavioural science is unlikely to change the world without a heterogeneity revolution.,(8), 980–989.

Cale, E. M. (2006). A quantitative review of the relations between the “Big 3” higher order personality dimensions and antisocial behavior.,(3), 250–284.

Canli, T. (2004). Functional brain mapping of extraversion and neuroticism: :Learning from individual differences in emotion processing.,(6), 1105–1132.

Chand, G. B., Dwyer, D. B., Erus, G., Sotiras, A., Varol, E., Srinivasan, D., ... Dazzan, P. (2020). Two distinct neuroanatomical subtypes of schizophrenia revealed using machine learning.,(3), 1027–1038.

Chen, Z. -G. (1983). Item analysis of Eysenck Personality Questionnaire tested in Beijing-district.,(2), 85–92.

[陳仲庚. (1983). 艾森克人格問(wèn)卷的項(xiàng)目分析.,(2), 85–92.]

Cheng, W., Rolls, E. T., Qiu, J., Xie, X., Lyu, W., Li, Y., ... Feng, J. F. (2018). Functional connectivity of the human amygdala in health and in depression.,(6), 557–568.

Cheng, W., Rolls, E. T., Qiu, J., Xie, X., Wei, D., Huang, C. -C., ... Feng, J. F. (2018). Increased functional connectivity of the posterior cingulate cortex with the lateral orbitofrontal cortex in depression.,(1), 1–10.

Cohen, M. X., Young, J., Baek, J. -M., Kessler, C., & Ranganath, C. (2005). Individual differences in extraversion and dopamine genetics predict neural reward responses.,(3), 851–861.

Cole, M. W., Bassett, D. S., Power, J. D., Braver, T. S., & Petersen, S. E. (2014). Intrinsic and task-evoked network architectures of the human brain.,(1), 238–251.

Dollinger, S. J., Urban, K. K., & James, T. A. (2004). Creativity and openness: Further validation of two creative product measures.,(1), 35–47.

Donnellan, M. B., & Robins, R. W. (2010). Resilient, overcontrolled, and undercontrolled personality types: Issues and controversies.,(11), 1070–1083.

Duggan, C., Sham, P., Lee, A., Minne, C., & Murray, R. (1995). Neuroticism: A vulnerability marker for depression evidence from a family study.,(3), 139–143.

Etkin, A., Egner, T., & Kalisch, R. (2011). Emotional processing in anterior cingulate and medial prefrontal cortex.,(2), 85–93.

Feczko, E., Miranda-Dominguez, O., Marr, M., Graham, A. M., Nigg, J. T., & Fair, D. A. (2019). The heterogeneity problem: Approaches to identify psychiatric subtypes.,(7), 584–601.

Fournier, J., Jones, N., Chase, H., Cummings, L., Graur, S., & Phillips, M. (2017). Personality dysfunction in depression and individual differences in effortful emotion regulation.,(10), S336–S337.

Friston, K. J., Williams, S., Howard, R., Frackowiak, R. S., & Turner, R. (1996). Movement-related effects in fMRI time-series.,(3), 346– 355.

Gerlach, M., Farb, B., Revelle, W., & Amaral, L. A. N. (2018). A robust data-driven approach identifies four personality types across four large data sets.,(10), 735–742.

Haas, B. W., Omura, K., Constable, R. T., & Canli, T. (2007). Emotional conflict and neuroticism: Personality-dependent activation in the amygdala and subgenual anterior cingulate.,(2), 249–256.

Hamilton, M. (1960). A rating scale for depression.,(1), 56–62.

Herzberg, P. Y., & Roth, M. (2006). Beyond resilients, undercontrollers, and overcontrollers? An extension of personality prototype research.,(1), 5–28.

Hooker, C. I., Verosky, S. C., Miyakawa, A., Knight, R. T., & D’Esposito, M. (2008). The influence of personality on neural mechanisms of observational fear and reward learning.,(11), 2709–2724.

Isler, L., Fletcher, G. J., Liu, J. H., & Sibley, C. G. (2017). Validation of the four-profile configuration of personality types within the Five-Factor Model.,, 257–262.

Jones, L., Scott, J., Cooper, C., Forty, L., Smith, K. G., Sham, P., ... Jones, I. (2010). Cognitive style, personality and vulnerability to postnatal depression.,(3), 200–205.

Kehoe, E. G., Toomey, J. M., Balsters, J. H., & Bokde, A. L. (2012). Personality modulates the effects of emotional arousal and valence on brain activation.,(7), 858–870.

Kendler, K. S., Kuhn, J., & Prescott, C. A. (2004). The interrelationship of neuroticism, sex, and stressful life events in the prediction of episodes of major depression.,(4), 631–636.

Kercher, A. J., Rapee, R. M., & Schniering, C. A. (2009). Neuroticism, life events and negative thoughts in the development of depression in adolescent girls.,(7), 903–915.

Khoo, S., & Simms, L. J. (2018). Links between depression and openness and its facets.,(3), 203–215.

Kim, M. J., Loucks, R. A., Palmer, A. L., Brown, A. C., Solomon, K. M., Marchante, A. N., & Whalen, P. J. (2011). The structural and functional connectivity of the amygdala: From normal emotion to pathological anxiety.,(2), 403–410.

Kim, S. Y., Stewart, R., Bae, K. Y., Kim, S. W., Shin, I. S., Hong, Y. J., ... Kim, J. M. (2016). Influences of the Big Five personality traits on the treatment response and longitudinal course of depression in patients with acute coronary syndrome: A randomised controlled trial.,, 38–45.

Klein, D. N., Kotov, R., & Bufferd, S. J. (2011). Personality and depression: Explanatory models and review of the evidence.,, 269–295.

Knyazev, G. G. (2006). EEG correlates of personality types.,(2), 78–87.

Kokkonen, M., & Pulkkinen, L. (2001). Extraversion and neuroticism as antecedents of emotion regulation and dysregulation in adulthood.,(6), 407–424.

Kong, R., Li, J., Orban, C., Sabuncu, M. R., Liu, H., Schaefer, A., ... Eickhoff, S. B. (2019). Spatial topography of individual- specific cortical networks predicts human cognition, personality, and emotion.,(6), 2533– 2551.

Leikas, S., & Salmela-Aro, K. (2014). Personality types during transition to young adulthood: How are they related to life situation and well-being?,(5), 753–762.

Lorenzetti, V., Allen, N. B., Fornito, A., & Yücel, M. (2009). Structural brain abnormalities in major depressive disorder: a selective review of recent MRI studies.,(1-2), 1–17.

Mantini, D., Corbetta, M., Romani, G. L., Orban, G. A., & Vanduffel, W. (2013). Evolutionarily novel functional networks in the human brain?,(8), 3259–3275.

McDonnell, S., & Semkovska, M. (2020). Resilience as mediator between extraversion, neuroticism, and depressive symptoms in university students.,(1), 26–40.

Menon, V., & Uddin, L. Q. (2010). Saliency, switching, attention and control: A network model of insula function.,(5-6), 655–667.

Miller, C., Brady, M., & Dean, S. (2020). Is Conscientiousness a proxy for Depression? A study of Big Five Con-scientiousness predictive power above depression.,(1), 112.

Milne, A. M., MacQueen, G. M., & Hall, G. B. (2012). Abnormal hippocampal activation in patients with extensive history of major depression: An fMRI study.,(1), 28–36.

Mobbs, D., Hagan, C. C., Azim, E., Menon, V., & Reiss, A. L. (2005). Personality predicts activity in reward and emotional regions associated with humor.,(45), 16502–16506.

Murray, L. E., & O'Neill, L. (2019). Neuroticism and extraversion mediate the relationship between having a sibling with developmental disabilities and anxiety and depression symptoms.,, 232–240.

Ormel, J., Bastiaansen, A., Riese, H., Bos, E. H., Servaas, M., Ellenbogen, M., ... Aleman, A. (2013). The biological and psychological basis of neuroticism: Current status and future directions.,(1), 59–72.

Pang, Y. J. (2020).(Unpublished doctorial dissertation). University of Electronic Science and Technology of China.

[龐雅靜. (2020).(博士學(xué)位論文). 電子科技大學(xué), 成都.]

Paulus, M. P., Rogalsky, C., Simmons, A., Feinstein, J. S., & Stein, M. B. (2003). Increased activation in the right insula during risk-taking decision making is related to harm avoidance and neuroticism.,(4), 1439–1448.

Peluso, M. A., Glahn, D. C., Matsuo, K., Monkul, E. S., Najt, P., Zamarripa, F., ... Soaress, J. C. (2009). Amygdala hyperactivation in untreated depressed individuals.,(2), 158–161.

Peters, A. T., Burkhouse, K., Feldhaus, C. C., Langenecker, S. A., & Jacobs, R. H. (2016). Aberrant resting-state functional connectivity in limbic and cognitive control networks relates to depressive rumination and mindfulness: A pilot study among adolescents with a history of depression.,, 178–181.

Power, J. D., Mitra, A., Laumann, T. O., Snyder, A. Z., Schlaggar, B. L., & Petersen, S. E. (2014). Methods to detect, characterize, and remove motion artifact in resting state fMRI.,, 320–341.

Quirk, G. J., Russo, G. K., Barron, J. L., & Lebron, K. (2000). The role of ventromedial prefrontal cortex in the recovery of extinguished fear.,(16), 6225–6231.

Rai, S., Griffiths, K. R., Breukelaar, I. A., Barreiros, A. R., Chen, W., Boyce, P., ... Korgaonkar, M. S. (2021). Default- mode and fronto-parietal network connectivity during rest distinguishes asymptomatic patients with bipolar disorder and major depressive disorder.,(1), 547.

Rocca, M. A., Pravata, E., Valsasina, P., Radaelli, M., Colombo, B., Vacchi, L., ... Filippi, M. (2015). Hippocampal- DMN disconnectivity in MS is related to WM lesions and depression.,(12), 5051–5063.

Rolls, E. T. (2015). Limbic systems for emotion and for memory, but no single limbic system.,, 119–157.

Scalabrini, A., Vai, B., Poletti, S., Damiani, S., Mucci, C., Colombo, C., ... Northoff, G. (2020). All roads lead to the default-mode network-global source of DMN abnormalities in major depressive disorder.,(12), 2058–2069.

Schaefer, A., Kong, R., Gordon, E. M., Laumann, T. O., Zuo, X. -N., Holmes, A. J., ... Yeo, B. T. (2018). Local-global parcellation of the human cerebral cortex from intrinsic functional connectivity MRI.,(9), 3095–3114.

Spielberg, J. M., Forbes, E. E., Ladouceur, C. D., Worthman, C. M., Olino, T. M., Ryan, N. D., & Dahl, R. E. (2015). Pubertal testosterone influences threat-related amygdala- orbitofrontal cortex coupling.,(3), 408–415.

Spinhoven, P., de Rooij, M., Heiser, W., Smit, J. H., & Penninx, B. W. (2012). Personality and changes in comorbidity patterns among anxiety and depressive disorders.,(4), 874–884.

Sprengelmeyer, R., Steele, J. D., Mwangi, B., Kumar, P., Christmas, D., Milders, M., & Matthews, K. (2011). The insular cortex and the neuroanatomy of major depression.,(1-2), 120–127.

Sridharan, D., Levitin, D. J., & Menon, V. (2008). A critical role for the right fronto-insular cortex in switching between central-executive and default-mode networks.,(34), 12569– 12574.

Surguladze, S. A., El-Hage, W., Dalgleish, T., Radua, J., Gohier, B., & Phillips, M. L. (2010). Depression is associated with increased sensitivity to signals of disgust: A functional magnetic resonance imaging study.,(14), 894–902.

Suslow, T., Konrad, C., Kugel, H., Rumstadt, D., Zwitserlood, P., Sch?ning, S., ... Dannlowski, U. (2010). Automatic mood- congruent amygdala responses to masked facial expressions in major depression.,(2), 155–160.

Tang, S., Lu, L., Zhang, L., Hu, X., Bu, X., Li, H., ... Huang, X. (2018). Abnormal amygdala resting-state functional connectivity in adults and adolescents with major depressive disorder: A comparative meta-analysis.,, 436–445.

Wardenaar, K. J., Conradi, H. J., Bos, E. H., & de Jonge, P. (2014). Personality modulates the efficacy of treatment in patients with major depressive disorder.,(9), e916–923.

Wang, J., Wang, N. N., & Zhang, Y. (Eds). (2013).. Beijing Normal University Press.

[王潔, 王寧寧, 張艷. (主編). (2013).. 北京：北京師范大學(xué)出版社.]

Yan, C., & Zang, Y. (2010). DPARSF: A MATLAB toolbox for "pipeline" data analysis of resting-state fMRI.,, 13.

Yan, R., Geng, J. T., Huang, Y. H., Zou, H. W., Wang, X. M., Xia, Y., ... Lu, Q. (2022). Aberrant functional connectivity in insular subregions in somatic depression: A resting-state fMRI study.,(1), 146.

Yin, K., Zhao, J., Zhou, J., & Nie, Q. (2021). The big-five personality profiles: A person-centered approach.,(10), 1866–1877.

[尹奎, 趙景, 周靜, 聶琦. (2021). “大五”人格剖面: 以個(gè)體為中心的研究路徑.(10), 1866–1877.]

Yoon, K. L., Maltby, J., & Joormann, J. (2013). A pathway from neuroticism to depression: Examining the role of emotion regulation.,(5), 558–572.

Zhang, T., Bai, T., Xie, W., Wei, Q., Lv, H., Wang, A., ... Wang, K. (2021). Abnormal connectivity of anterior-insular subdivisions and relationship with somatic symptom in depressive patients.,(4), 1760–1768.

Zuroff, D. C., Mongrain, M., & Santor, D. A. (2004). Conceptualizing and measuring personality vulnerability to depression: Comment on Coyne and Whiffen (1995).,(3), 489–511.

Personality subtypes of depressive disorders and their functional connectivity basis

LI Yu1,2,3, WEI Dongtao1,2, QIU Jiang1,2

(1Key Laboratory of Cognition and Personality, Ministry of Education, Southwest University;2Department of Psychology, Southwest University;3Faculty of Education, Southwest University, Chongqing 400715, China)

Heterogeneity among mental health issues has always attracted considerable attention, thereby restricting research on mental health and cognitive neuroscience. Additionally, the person-centred approach to personality research, which emphasizes population heterogeneity, has received more attention. On the other hand, the heterogeneity among depressive patients has been a problem that cannot be ignored (most studies ignored the actual situation and directly assumed sample homogeneity). A large number of empirical studies have provided evidence that isolated personality traits are often associated with depression. Only a few studies have considered the probable effect from a taxonomy perspective. Moreover, the neural mechanisms of personality types in depression remain unclear. This study aimed to reveal different personality subtypes of depressive disorders and elucidate subtypes from the perspective of resting-state functional connectivity.

Personality and resting-state functional imaging data of 135 depressive patients and 133 controls were collected. First, combined with “depression diagnosis”, the personality types in depressive patients and controls were identified through functional random forest. Specifically, neuroticism and extraversion (input features) were fitted with the diagnosis of depression by a random forest model. The random seeds were set to 1234, and 500 decision trees were fitted. The performance of the model was evaluated by tenfold cross-validation. Subsequently, the random forest algorithm generated a proximity matrix that represented the similarity between paired participants. Then, based on the proximity matrix, community detection clustering analysis was conducted on depressive patients and controls, and personality types associated with depression diagnosis were obtained. Finally, we selected nodes of the subcortical network as regions of interest according to the power-264 template and calculated the functional connectivity map of the region of interest to the whole brain. Based on the functional connectivity map, the differences in resting-state functional connectivity between the main types were compared.

Personality and resting-state functional imaging data of 159 depressive patients and 156 controls were collected. First, combined with “depression diagnosis”, the personality types in depressive patients and controls were identified through functional random forest. Specifically, neuroticism and extraversion (input features) were fitted with the diagnosis of depression by a random forest model. The random seeds were set to 1234, and 500 decision trees were fitted. The performance of the model was evaluated by tenfold cross-validation. Subsequently, the random forest algorithm generated a proximity matrix that represented the similarity between paired participants. Then, based on the proximity matrix, community detection clustering analysis was conducted on depressive patients and controls, and personality types associated with depression diagnosis were obtained. Finally, we selected the amygdala, hippocampus, insula (AAL atlas) and limbic network, default network, and control network (Schaefer-Yeo template) as regions of interest and calculated the functional connectivity of the subcortical regions to the networks. ANOVA was used to compare resting-state functional connectivity between the personality types.

The results showed the following. (1) Depression was more common among individuals with high neuroticism and low extraversion tendencies, but there were also individuals with low neuroticism and high extraversion tendencies. The controls were more likely to be individuals with low neuroticism and high extraversion. (2) The results of resting-state functional connectivity showed no significant difference between depression and controls. (3) The functional connectivity strength of the left amygdala/insula-limbic network was significantly different across personality subtypes.

In summary, the personality subtypes of depression identified by person-centred perspectives are more in line with reality and individual cognitive patterns, and they have potential clinical adaptive value. The findings of this study enhance the understanding of heterogeneity among depressive disorders.

neuroticism, extraversion, resting-state functional connectivity, depressive disorders, person-centred

B845; R395

2022-01-28

*重慶市自然科學(xué)基金(cstc2015jcyjA10106); 中國(guó)博士后科學(xué)基金面上資助(2021M702705); 重慶市博士后創(chuàng)新人才支持計(jì)劃(A33600125)。

邱江, E-mail: qiuj318@swu.edu.cn