2023专家观点:功能磁共振下不同皮层区域间功能连接破裂导致术后谵妄患者的纵向认知解体？

本文由“麻醉新超人”授权转载

该文章选自BJA《英国麻醉杂志》2023年7月刊

翻译：凌子杰 徐医麻醉2023级硕士研究生 审校：赵林林 徐医附院麻醉科

‘Where there’s smoke’: longitudinal cognitive disintegration after postoperative delirium?

“烟从哪里来”：术后谵妄后的纵向认知解体？  

Summary 概要  

Postoperative delirium is an important complication of surgery and is associated with poor long-term cognitive outcomes, although the neural basis underlying this relationship is poorly understood.Neuroimaging studies and network- based approaches play an important role in our understanding of the mechanism by which delirium relates to longitudinal cognitive decline.  

术后谵妄是手术的一种重要并发症，与不良的长期认知结果有关，尽管这种关系的神经基础尚不清楚。神经影像学研究和基于网络的方法在我们理解谵妄与纵向认知能力下降相关的机制方面发挥着重要作用。

A recent resting state functional MRI study is reviewed, which shows reduced global connectivity up to 3 months after delirium, supporting recent models of delirium and opening the door for applying this approach to understanding the complex inter-relationship between delirium and dementia.  

综述了最近的一项静息状态功能性MRI研究，该研究显示，在谵妄发生后3个月内，整体连通性降低，支持了最近的谵妄模型，并为应用这种方法来理解谵妄和痴呆之间的复杂相互关系打开了大门。  

Keywords

关键词: cognitive dysfunction  认知功能障碍; delirium  谵妄; dementia  痴呆; functional connectivity  功能连接; functional magnetic resonance imaging  功能磁共振成像; global connectivity 全球连通性  

Postoperative delirium is an important complication of surgery, with an incidence rate of up to 51% in older patients. Whilst still defined as an acute and transient state of cognitive impairment, delirium is associated with poor long-term cognitive outcomes and dementia.  

术后谵妄是手术的一种重要并发症，老年患者的发病率高达51%。虽然谵妄仍被定义为一种急性和短暂的认知障碍状态，但它与不良的长期认知结果和痴呆症有关。  

Yet, the complex relationship between delirium and cognitive impairment remains poorly understood, with ongoing debate regarding whether delirium represents a marker of premorbid vulnerability to incipient cognitive impairment; whether it unmasks or accelerates cognitive impairment; or itself is a precipitant of the ensuing cognitive decline.Unravelling this complexity, whilst also understanding the pathophysiological basis of delirium, remains an important challenge to the field.

然而，人们对谵妄和认知障碍之间的复杂关系仍知之甚少，关于谵妄是否代表了早期认知障碍的发病前脆弱性标志的争论仍在继续；它是否掩盖或加速了认知障碍；或者其本身就是随之而来的认知能力下降的一个诱因。解开这种复杂性，同时理解谵妄的病理生理基础，仍然是该领域的一个重要挑战。

Neuroimaging studies play an important role in revealing insights into this problem.Much work to date in characterising predisposing factors using neuroimaging have focused on structural measures, showing correlation of delirium with indices of pathology such as global and regional atrophy, white matter hyperintensity, and diffusion imaging abnormalities,although sometimes with mixed results.

神经影像学研究在揭示这一问题方面发挥着重要作用。迄今为止，使用神经成像表征易感因素的许多工作都集中在结构测量上，显示了谵妄与病理指标的相关性，如全身和局部萎缩、白质高信号和扩散成像异常，尽管有时结果喜忧参半。

However, functional MRI (fMRI) measures, such as those used by Ditzel and colleagues, may have a higher sensitivity for understanding the impact of delirium on cognitive function in the acute and short term.In this regard, one of the prevailing paradigms of the delirious state is that it is a ‘cognitive disintegration’, caused by breakdown of functional connectivity between various cortical regions.

然而，功能性MRI（fMRI）测量，如Ditzel及其同事使用的测量，可能对理解谵妄对急性和短期认知功能的影响具有更高的敏感性。在这方面，谵妄状态的一个流行范式是，它是一种“认知解体”，由不同皮层区域之间的功能连接破裂引起。  

This hypothesis is supported by a recent qualitative meta-analysis of studies, showing a consistent signature of delirium of reduced network integration using fMRI.An extension of this hypothesis is whether cognitive disintegration (reduced functional connectivity) occurs beyond the state of delirium and might explain prolonged cognitive impairment.

这一假设得到了最近一项研究的定性meta分析的支持，该研究显示了功能磁共振成像网络整合减少的谵妄的一致特征。这一假设的延伸是，认知解体（功能连接减少）是否发生在谵妄状态之外，并可能解释长期的认知障碍。

Ditzel and colleagues analysed resting state fMRI at baseline and 3 months after elective surgery in a large prospective multicentre observational cohort. Their study included 554 subjects aged 65 yr or older undergoing major elective surgery,including cardiac surgery, and were part of a larger European Biomarker development cohort.Resting state fMRI sequence (rs-fMRI) was captured before and 3 months after surgery, coupled with a simple cognitive test, the Trail Making Test.  

Ditzel及其同事在一个大型前瞻性多中心观察队列中分析了基线和择期手术后3个月的静息状态fMRI。他们的研究包括554名65岁或65岁以上接受包括心脏手术在内的重大选择性手术的受试者，他们是更大的欧洲生物标志物开发队列的一部分。术前和术后3个月采集静息状态fMRI序列（rs-fMRI），并进行简单的认知测试，即追踪测试。  

The authors chose to analyse Part B of the trail making test (TMT-B) as a simple and specific measure of executive function associated with delirium severity and duration.

作者选择分析试验测试（TMT-B）的B部分，作为与谵妄严重程度和持续时间相关的执行功能的一种简单而具体的测量方法。  

Previous work revealed loss of global functional connectivity during and 7 days after delirium, hence Ditzel and colleagues hypothesised that similar changes would be observed at the later time point of 3 months, and that this would correlate with cognitive testing, providing an early ‘smoke’ signal of the ‘fire’ of postoperative delirium on longer term cognition.

先前的研究表明，在谵妄期间和之后7天，整体功能连接丧失，因此Ditzel及其同事假设，在3个月的后期会观察到类似的变化，这将与认知测试相关，为长期认知提供术后谵妄“火焰”的早期“烟雾”信号。  

The authors used accepted graph theoretical approaches to distilling and analysing brain networks . These approaches regard the brain as a set of interconnected ‘nodes’ that are represented by different cortical, subcortical, or both regions.

作者使用公认的图论方法来提取和分析大脑网络。这些方法将大脑视为一组相互连接的“节点”，由不同的皮层、皮层下或两个区域表示。  

In rs-fMRI, the presence or absence of connections between any two nodes, and their relative strengths (weights) are determined by the average correlation of the blood oxygenation level dependent (BOLD) response (taken as a proxy of neural activity) between these regions over the  scanning period. Graph theory approaches allow derivation of parameters that help quantify macroscopic properties of the brain such as ‘whole brain connectivity’ and ‘integration’.  

在rs fMRI中，任何两个节点之间是否存在连接，以及它们的相对强度（权重）由扫描期间这些区域之间的血氧水平依赖性（BOLD）反应（作为神经活动的代表）的平均相关性决定。图论方法允许推导有助于量化大脑宏观特性的参数，如“全脑连接”和“整合”。  

In this study, the authors used a ‘minimum spanning tree’ (MST) analysis. A spanning tree is a connected graph with a path formed between all the available nodes of a network (regions of the brain in this instance), made in such a way that the connections do not form loops.

在这项研究中，作者使用了“最小生成树”（MST）分析。生成树是一个连通图，其路径在网络的所有可用节点（在本例中为大脑区域）之间形成，以使连接不会形成循环的方式形成。  

When the strongest connections are considered, and the total sum of weights minimised,one can derive the ‘minimum spanning tree’ which is regarded as the ‘backbone’ of the global network.MST analysis is relatively robust against some arbitrary choices that can be made in graph theory (such as how strong a connection should be to be relevant, i.e. the weight of the connection) making it a reproducible measure.Importantly, MST metrics derived from structural and functional MRI have also been shown to capture biologically relevant changes in disease states.

当考虑到最强的连接，并将权重的总和最小化时，可以导出被视为全球网络“骨干”的“最小生成树”。MST分析对图论中可以做出的一些任意选择（例如连接的强度，即连接的权重）相对稳健，使其成为一种可重复的度量。重要的是，源自结构和功能MRI的MST指标也被证明可以捕捉疾病状态的生物学相关变化。

Ditzel and colleagues were concerned with metrics relating to global connectivity and integration. Global connectivity was calculated as the average weight of the connections in the MST (the higher the average strength of connections, the higher the global  connectivity),  meanwhile  a  larger  ‘leaf  fraction’  (the proportion  of  regions/nodes  that  are  connected  to  one  other node  in the  MST)  was  taken  as  a  proxy  of  increased  brain network   integration.    

Ditzel及其同事关注与全球连通性和一体化相关的指标。全局连通性被计算为MST中连接的平均权重（连接的平均强度越高，全局连通性就越高），同时，更大的“叶分数”（连接到MST中另一个节点的区域/节点的比例）被视为脑网络集成增加的代表。

The authors found  that across their sample, global connectivity increased postoperatively.However, in their linear models,  the presence of postoperative delirium (interaction between delirium andtime) reduced this effect. This finding that postoperative delirium was a predictor of  relatively  reduced  global  connectivity  3  months  after  surgery could be seen as an affirmation of their hypothesis that postoperative delirium results in longer term adverse network changes.

作者发现，在他们的样本中，术后大脑连通性增加。然而，在他们的线性模型中，术后谵妄的存在（谵妄与时间之间的相互作用）降低了这种影响。术后谵妄是术后3个月大脑连通性相对降低的预测因素，这一发现可以被视为对他们的假设的肯定，即术后谵妄会导致长期的不良关系网变化。  

Moreover, preoperative global connectivity did not predict development  of  postoperative  delirium  suggesting  that  the differences in connectivity were new, and that these network characteristics on  their  own did  not  reflect a vulnerability  to delirium, as the authors and others have shown previously, and  running  counter  to  an  a  priori  hypothesis  that  impaired connectivity   wouldpredispose   to   delirium.

此外，术前大脑连通性并不能预测术后谵妄的发展，这表明连通性的差异是新的，这些网络特征本身并没有反映出谵妄的脆弱性，正如作者和其他人之前所表明的那样，与连接受损会导致谵妄的先验假设背道而驰。   

Interestingly, increased (not decreased) preoperative functional connectivity in the electroencephalogram predisposed to delirium in our recent study, running counter to findings with structural connectivity (diffusion tensor imaging).Further studies are required to understand the interplay of functional and structural connectivity and predisposition to delirium.

有趣的是，在我们最近的研究中，脑电图的术前功能连接增加（而不是减少）易患谵妄，这与结构连接（扩散张量成像）的发现背道而驰。需要进一步的研究来了解功能和结构连接的相互作用以及谵妄的易感性。

Although unrelated to the initial hypotheses, the very interesting finding of an overall increase in global connectivity in the population overall after major surgery requires explanation. The authors suggest in their study that this may reflect a reactive increase in neuronal activity in response to a major insult.      

尽管与最初的假设无关，但大手术后总体人群的全球连通性总体增加这一非常有趣的发现需要解释。作者在他们的研究中认为，这可能反映了神经元活动对重大损伤的反应性增加。

However, overall reduced global connectivity (irrespective of delirium status), did correlate with reduced TMT-B scores over time compared with those who experienced an increase or no change in their global connectivity.This would favour increased connectivity changes being more adaptive than maladaptive.

然而，与那些大脑连通性增加或没有变化的人相比，随着时间的推移，全球连通性的总体降低（无论谵妄状态如何）确实与TMT-B评分的降低相关。这将有利于增加连通性变化，使其更具适应性，而不是不适应性。

Indeed, given most patients undergoing major surgery are doing so as a result of significant disease that might have negative systemic consequences, an alternative explanation is that increased global connectivity reflects removal of the inciting pathology (e.g. improvement of pain after orthopaedic surgery).

事实上，考虑到大多数接受大手术的患者都是由于严重疾病导致的，这些疾病可能会产生负面的系统后果，另一种解释是，大脑连通性的增加反映了煽动性病理的消除（例如整形外科手术后疼痛的改善）。

However, increases in activity and connectivity have been suggested to precede frank neurodegeneration in biological models, emphasising that further information is required about how we interpret functional connectivity in the spectrum of disease.

然而，在生物学模型中，活性和连接性的增加被认为是在直接的神经退行性变之前发生的，这强调了需要进一步的信息来解释我们如何在疾病谱中解释功能连接性。

The study has a number of strengths including a relatively large cohort, with an impressive protocol including baseline (preoperative) and longitudinal imaging with good retention. Some limitations of the study included exclusion of patients with low cognitive baseline function with mini-mental state examination (MMSE) scores <24, resulting in 75% of their population having an MMSE score of ≤28.   

该研究具有许多优势，包括相对较大的队列，具有令人印象深刻的方案，包括基线（术前）和具有良好保留的纵向成像。该研究的一些局限性包括排除了认知基线功能低、微小精神状态检查（MMSE）评分<24的患者，导致75%的患者的MMSE评分≤28。  

The higher cognitive baseline could have conferred greater resilience to changes in brain network measures over time or obscured any differences in performance in TMT-B (ceiling effect). Further limiting statistical power, after strict motion correction, the authors ended up with 246 patients left for analysis, of whom only 38 (16%) experienced postoperative delirium.

较高的认知基线可能会赋予大脑网络测量随时间变化的更大弹性，或掩盖TMT-B表现的任何差异（上限效应）。进一步限制了统计能力，在严格的运动矫正后，作者最终留下246名患者进行分析，其中只有38人（16%）经历了术后谵妄。  

Another consideration is the analytical method itself. Metrics, such as leaf fraction, is one measure of the integrated capacity of the brain, but also reflects to some extent the specific configuration of the brain network. For instance, a higher leaf fraction can represent more reliance on a few central hubs (highly interconnected nodes).

另一个考虑因素是分析方法本身。树叶分数等指标是衡量大脑综合能力的一种指标，但在一定程度上也反映了大脑网络的具体配置。例如，较高的树叶分数可以表示更多地依赖于几个中心集线器（高度互连的节点）。

The MST analysis considers changes on a global or macroscopic level, but lacks the resolution, sensitivity, or both to detect more focal disturbances occurring within and between key putative networks, such as the default mode network, dorsal attentional networks, salience network, and frontoparietal control network.

MST分析考虑了全局或宏观层面的变化，但缺乏检测关键假定网络内部和之间发生的更多局灶性干扰的分辨率、灵敏度或两者兼有，如默认模式网络、背侧注意网络、显著性网络和额顶控制网络。

For instance, one rs-fMRI study of 22 patients during an episode of delirium and a subset after delirium, found that there was enhanced connectivity within regions of the default mode network, but decreased connectivity between subcortical regions.

例如，一项针对谵妄发作期间和谵妄后一个子集的22名患者的功能磁共振成像研究发现，默认模式网络区域内的连通性增强，但皮质下区域之间的连通性降低。 

They also found loss of the normal ‘anti- correlation’ between the dorsolateral prefrontal cortex (part of the frontoparietal network) and the posterior cingulate cortex (a hub of the default mode network) during delirium.

他们还发现，在谵妄期间，背外侧前额叶皮层（额顶网络的一部分）和后扣带皮层（默认模式网络的中枢）之间的正常“反相关性”丧失。  

This phenomenon persisted beyond the resolution of the delirium episode and was speculated to underlie the mild inattention that can follow delirium. These findings suggest these mesoscale relationships are also important to consider  with respect to long-term network changes after delirium.

这种现象在谵妄发作后持续存在，并被推测为谵妄后轻度注意力不集中的原因。这些发现表明，这些中尺度关系对于谵妄后的长期网络变化也很重要。

Delirium is associated with global disruption of the spatiotemporal properties of the brain. Therefore, in addition to considering analyses at different spatial scales as above, another can also consider other methods, such as dynamic functional connectivity analysis which considers the timevarying properties of the brain.

谵妄与大脑时空特性的全局破坏有关。因此，除了如上所述考虑不同空间尺度的分析外，另一种方法还可以考虑其他方法，例如考虑大脑时变特性的动态功能连接分析。  

Temporal variability has been shown to be an important property of consciousness,and the balance between integration and segregation over the time course of an individual scan correlates with cognitive performance and relates to specific neuromodulators.Altered dynamic functional connectivity characterises a range of neurodegenerative conditions and can predict the development of dementia in at-risk populations.

时间变异性已被证明是意识的一个重要特性，在个体扫描的时间过程中，整合和分离之间的平衡与认知表现相关，并与特定的神经调节剂有关。改变的动态功能连接是一系列神经退行性疾病的特征，可以预测高危人群痴呆症的发展。

Recently, dynamic connectivity changes have been shown to correlate with attentional fluctuations in Lewy body dementia which have close similarity to delirium.Understanding temporal disruptions of connectivity, and the balance of integration and segregation at varying timescales, might confirm the dysconnectivity hypotheses, and demonstrate longer term changes after an acute episode of delirium.

最近，动态连接变化已被证明与路易体痴呆的注意频率波动相关，路易体痴呆与谵妄非常相似。 了解连接的时间中断，以及不同时间尺度上整合和分离的平衡，可能会证实连接障碍假说，并证明急性谵妄发作后的长期变化。  

Perhaps the most exciting possibility is using imaging metrics to ‘fingerprint’ vulnerability to particular mechanisms and consequences of delirium to allow precision medicine targeting of therapies. However, for this to become a reality, both advances in bedside imaging and analytic techniques are required.In the meantime, further functional imaging studies are required with wide ranging analytic approaches to leverage the immense depth of information that these studies yield.  

也许最令人兴奋的可能性是使用成像指标来“指纹”谵妄的特定机制和后果的脆弱性，从而实现精确的药物靶向治疗。然而，要实现这一点，需要在床边成像和分析技术方面取得进展。与此同时，需要用广泛的分析方法进行进一步的功能成像研究，以利用这些研究产生的巨大信息深度。  

Taking all the above into account, the study by Ditzel and colleagues highlights that the impact of delirium by way of functional network changes (the ‘smoke’) can still be seen long after the initial episode of delirium (the ‘fire’).

Hence, delirium may not simply reflect an underlying vulnerability, but may itself participate in the pathological process and induce long- term maladaptive changes.

考虑到以上所有因素，Ditzel及其同事的研究强调，在最初的谵妄发作（“火灾”）很久之后，仍然可以看到谵妄通过功能网络变化（“烟雾”）产生的影响。 因此，谵妄可能不仅仅反映了潜在的脆弱性，它本身也可能参与病理过程并诱发长期的适应不良变化。  

Perhaps more importantly, the study brings light to the utility of rs-fMRI and opens the door towards similar and other network-based approaches that can be used to help probe the complex interrelationship between delirium and long-term cognitive decline.

也许更重要的是，这项研究揭示了rs功能磁共振成像的实用性，并为类似和其他基于网络的方法打开了大门，这些方法可用于帮助探索谵妄和长期认知能力下降之间的复杂相互关系。  

Authors’ contributions 作者的贡献  

Wrote the first draft of the editorial, which was iteratively edited by both authors: EM.Approved the final submission: both authors.

撰写社论初稿，由两位作者反复编辑：EM。批准最终提交：两位作者。  

Declaration of interest 利益申报  

RDS is an editor of the British Journal of Anaesthesia. EM declares that they have no conflict of interest.  

RDS是《英国麻醉学杂志》的编辑。EM声明他们没有利益冲突。

Funding 资金  

EM is supported by a National Health and Medical Research Council (NHMRC) Emerging Leadership fellowship (2008565). RDS is supported by National Institutes of Health (NIH) (R01 AG063849-01, 1R01NS117901-01).

EM得到了国家卫生和医学研究委员会（NHMRC）新兴领导力研究金（2008565）的支持。RDS得到美国国立卫生研究院（NIH）的支持（R01 AG063849-01，1R01NS117901-01）。

图表

Using graph theory to understand network changes associated with delirium. Initial steps include preprocessing of raw data, followed by anatomical parcellation of the brain into cortical and subcortical regions. In functional MRI (fMRI), the correlation of the signal across the period of the scan (or a prespecifified scanning window of time in the case of dynamic fMRI) is taken as a strength or ‘weight’ of the connection between regions or ‘nodes’ of the network (red and green in the top right panel).

使用图论来理解与谵妄相关的网络变化。最初的步骤包括对原始数据进行预处理，然后将大脑解剖划分为皮层和皮层下区域。在功能性MRI（fMRI）中，信号在扫描周期内的相关性（或者在动态fMRI的情况下是预先指定的扫描时间窗口）被视为网络的区域或“节点”之间连接的强度或“权重”（右上角的红色和绿色）。  

The weights or presence/absence of a connection (binarised if desired) can be seen as a connectivity matrix which is a representation of the brain network. This can be analysed using graph theoretical methods, such as derivation of the minimum spanning tree (MST) of the network (see text).

连接的权重或存在/不存在（如果需要的话进行二值化）可以被视为连接矩阵，该连接矩阵是大脑网络的表示。这可以使用图论方法进行分析，例如网络的最小生成树（MST）的推导（见正文）。  

Quantitative measures that characterise different aspects of the network are then extracted, such as leaf fraction (Lf ¼ proportion of nodes that have only one connection; viewed as a measure of the ‘centralisation’ or ‘integration’ of the network) and diameter (D ¼ largest distance between any two nodes; viewed as a measure of the ‘effificiency’ of the network).

然后提取表征网络不同方面的定量度量，如叶分数（Lf¼只有一个连接的节点的比例；被视为网络“集中”或“集成”的度量）和直径（D¼任意两个节点之间的最大距离；被视是网络“效率”的度量。  

These parameters can be compared within and between subjects or groups of subjects under different conditions (e.g. healthy vs delirious state).Combining changes in these measures with other findings (such as focal elevation of slow wave activity) can give mechanistic insight into the pathological changes occurring in the brain during an episode, after an episode, or both of delirium.

这些参数可以在不同条件下（例如，健康状态与谵妄状态）在受试者或受试者组内部和之间进行比较。 将这些测量结果的变化与其他发现（如慢波活动的局灶性升高）相结合，可以从机制上深入了解谵妄发作期间、发作后或两者中大脑发生的病理变化。

免责声明：

文中所涉及药物使用、疾病诊疗等内容仅供医学专业人士参考。

—END—

编辑：Michel.米萱

校对：MiLu.米鹭

打赏