Impact of multisession 40Hz tACS on hippocampal perfusion in patients with Alzheimer's disease
多次 40Hz 经颅交流电刺激对阿尔茨海默病患者海马灌注的影响

Keywords: Dementia, Cerebral blood flow, Neuromodulation, Neurostimulation, tES, Gamma band, Gamma activity, Hippocampus, EEG, CBF
关键词：痴呆，脑血流，神经调节，神经刺激，tES，伽马波段，伽马活动，海马，脑电图，脑血流量

Alzheimer’s disease (AD) is the most common cause of dementia, and its prevalence continues to increase [1]. Despite this enormous disease burden and intensive scientific research, therapeutic options are limited. While there are pharmacologic interventions that transiently stabilize cognitive function, no disease-modifying therapy is available. Even promising pharmacological
阿尔茨海默病（AD）是最常见的痴呆原因，其患病率持续上升[1]。尽管疾病负担巨大且科学研究密集，但治疗选择仍然有限。虽然存在能够暂时稳定认知功能的药物干预，但尚无疾病修饰性疗法。即使是有前景的药物治疗
interventions (e.g., aducanumab, an anti-amyloid compound) do not appear to stop cognitive decline [2]. In an effort to develop effective treatments, research has then focused on deepening our understanding of the pathophysiology of AD. Positron emission tomography (PET) as well as single-photon emission computed tomography (SPECT) imaging revealed marked hypometabolism and perfusion deficits in AD patients with respect to healthy controls [3]. More recently, a perfusion-sensitive MRI imaging sequence, arterial spin labeling (ASL), has been developed to study brain perfusion without the need for contrasting agents [4]. ASL has helped reveal a significant reduction of brain perfusion (cerebral blood flow - CBF) in the temporal, parietal, and posterior cingulate cortices in AD patients with respect to healthy subjects [3,5], even though a pathological blood flow increase in preclinical stages of the disease has also been described [6, 7]. Moreover, CBF reduction has been correlated with language impairment in AD patients [8], and its reduction is paralleled by disease progression starting from the prodromal stage [9]. Also, hypoperfusion may predict conversion to AD in mild cognitive impairment (MCI) patients [10], making the quest for approaches to modulate (i.e., increase) perfusion a priority.
干预措施（例如抗淀粉样蛋白化合物 aducanumab）似乎无法阻止认知能力的下降[2]。为了开发有效的治疗方法，研究随后集中于加深对阿尔茨海默病（AD）病理生理学的理解。正电子发射断层扫描（PET）以及单光子发射计算机断层扫描（SPECT）成像显示，与健康对照组相比，AD 患者表现出明显的代谢减退和灌注缺陷[3]。最近，开发了一种灌注敏感的 MRI 成像序列——动脉自旋标记（ASL），用于研究脑灌注，无需使用造影剂[4]。ASL 帮助揭示了 AD 患者在颞叶、顶叶和后扣带皮层的脑灌注（脑血流量，CBF）显著减少，与健康受试者相比[3,5]，尽管也有研究描述了疾病临床前阶段的病理性血流增加[6,7]。此外，CBF 的减少与 AD 患者的语言障碍相关联[8]，其减少与疾病进展同步，从前驱期开始[9]。 此外，低灌注可能预测轻度认知障碍（MCI）患者向阿尔茨海默病（AD）的转变[10]，因此寻找调节（即增加）灌注的方法成为优先任务。

A recent preclinical animal study has shown that gamma ( $\gamma$$\gamma$gamma\gamma ) band oscillatory brain activity is directly responsible for arteriolar vasodilatation and the consequent increase in blood oxygenation [11]. Gamma activity usually refers to cortical oscillations in the $30-80\mathrm{Hz}$$30-80\mathrm{Hz}$30-80Hz30-80 \mathrm{~Hz} frequency band, primarily generated by the interaction between inhibitory interneurons such as parvalbumin (PV)+ interneurons, and pyramidal cells [12]. The investigators found that optogenetic manipulations of $\gamma$$\gamma$gamma\gamma-band electrical power entrain the vasomotor oscillations of corresponding cortical and penetrating arterioles in a unidirectional way, i.e., independently from baseline CBF [11]. In turn, fluctuations in arteriolar diameter coherently drive fluctuations in blood oxygenation [11]. Electrical activity drives the arteriolar vasomotion with a lag of 2 s , which in turn leads to changes in functional MRI signal (detecting increases of oxygenation in the venular component; BOLD signal) one second later [11]. The significance of this electro-arteriolar coupling has been related to the role of vasomotion in the removal of waste and toxic proteins via the paravascular space, i.e., the socalled glymphatic system, that seems to be impaired in AD mouse models and AD patients [13, 14].
一项最新的临床前动物研究表明，伽马（ $\gamma$$\gamma$gamma\gamma ）频段的脑振荡活动直接导致小动脉扩张及随之而来的血氧含量增加[11]。伽马活动通常指的是 $30-80\mathrm{Hz}$$30-80\mathrm{Hz}$30-80Hz30-80 \mathrm{~Hz} 频段的皮层振荡，主要由抑制性中间神经元（如钙结合蛋白（PV）+中间神经元）与锥体细胞之间的相互作用产生[12]。研究人员发现，利用光遗传学操控 $\gamma$$\gamma$gamma\gamma 频段的电功率能够单向地驱动相应皮层及穿透性小动脉的血管运动振荡，即独立于基线脑血流量（CBF）[11]。反过来，小动脉直径的波动与血氧含量的波动相一致[11]。电活动以 2 秒的滞后驱动小动脉血管运动，进而在 1 秒后引起功能性磁共振成像信号的变化（检测静脉成分中氧合水平的增加；BOLD 信号）[11]。 这种电动小动脉耦合的重要性与血管运动在通过血管旁隙清除废物和有毒蛋白质中的作用有关，即所谓的胶淋巴系统，该系统在阿尔茨海默病小鼠模型和患者中似乎受损[13, 14]。

Pivotal preclinical studies have also shown that exoge-nously-induced increase of gamma oscillations (specifically at 40 Hz ) promotes microglial activation and cause subsequent reduction of $\mathrm{A}\beta$$\mathrm{A}\beta$Abeta\mathrm{A} \beta and p-tau depositions in a mouse model of AD [15]. Decreased $\gamma$$\gamma$gamma\gamma activity in AD mouse models is linked to PV+ inhibitory interneuron
关键的临床前研究还表明，外源性诱导的伽马振荡增加（特别是在 40 Hz）促进小胶质细胞激活，并导致阿尔茨海默病小鼠模型中 $\mathrm{A}\beta$$\mathrm{A}\beta$Abeta\mathrm{A} \beta 和 p-tau 沉积的随后的减少[15]。阿尔茨海默病小鼠模型中 $\gamma$$\gamma$gamma\gamma 活性的降低与 PV+抑制性中间神经元相关。
pathology that interferes with fast inhibitory loops in cortical circuits and is associated with a hyperactivation of pyramidal cells leading to global network dysfunction [16]. Remarkably, induction of gamma activity in presymptomatic AD mice-and thus a restoration of the physiological activity of PV interneurons-prevents subsequent neurodegeneration and behavioral deficits [17]. As seen in preclinical models [18], a consistent finding in patients with AD is a relative attenuation and dysregulation of gamma activity [19], therefore gamma induction may represent a novel and powerful therapeutic approach [17].
病理学干扰了皮层回路中快速抑制环路，并与锥体细胞的过度激活相关，导致整体网络功能障碍[16]。值得注意的是，在无症状阿尔茨海默病小鼠中诱导伽马活动——从而恢复 PV 中间神经元的生理活动——可以预防随后的神经退行性变和行为缺陷[17]。正如在临床前模型中所见[18]，阿尔茨海默病患者的一个一致发现是伽马活动的相对减弱和失调[19]，因此伽马诱导可能代表一种新颖且强有力的治疗方法[17]。

Recently, a neuromodulation technique that delivers alternating current stimulation-transcranial alternating current stimulation (tACS)-has received attention for the possibility of translating the aforementioned animal evidence to humans via noninvasive induction of gamma activity [20]. tACS applies low-amplitude alternating (sinusoidal) current to enhance specific oscillations by entraining neurons under specific cortical rhythms, depending on the applied stimulation frequency (e.g., 40 Hz ) [21]. Non-human animal work has demonstrated that tACS entrains neurons in widespread cortical areas [22], with recent non-human primates experiments revealing dose-dependent neural entrainment and increased burstiness as the fundamental response to tACS [23, 24]. Simulations, supported by empirical evidence using electroencephalography (EEG), demonstrated that tACS modulates brain oscillatory activity via network resonance, suggesting that a weak stimulation at a resonant frequency could cause large-scale modulation of network activity [25], and amplify endogenous network oscillations in a frequency-specific manner [18]. Consequently, tACS has been found able to modulate brain oscillation and related behavior in healthy subjects and patients [26-28], with an enhancement of gamma oscillations via tACS leading to transient improvement in motor, working memory, and abstract reasoning tasks on healthy controls [29-31], and effects often lasting beyond the tACS application period [30, 32, 33].
最近，一种神经调节技术——经颅交流电刺激（tACS）因其通过非侵入性诱导伽马活动将上述动物研究证据转化为人类应用的可能性而受到关注[20]。tACS 通过施加低幅度的交流（正弦波）电流，增强特定频率的振荡，通过使神经元在特定皮层节律下同步活动，具体取决于所施加的刺激频率（例如 40 Hz）[21]。非人类动物研究表明，tACS 能够使广泛皮层区域的神经元同步[22]，最近的非人灵长类实验揭示了 tACS 引起的剂量依赖性神经同步和爆发性增加，作为其基本反应[23, 24]。模拟研究结合使用脑电图（EEG）的实证证据表明，tACS 通过网络共振调节大脑振荡活动，提示在共振频率下的弱刺激能够引起大规模的网络活动调节[25]，并以频率特异性的方式放大内源性网络振荡[18]。 因此，研究发现 tACS 能够调节健康受试者和患者的大脑振荡及相关行为[26-28]，通过 tACS 增强伽马振荡可在健康对照组中暂时改善运动、工作记忆和抽象推理任务[29-31]，且这种效应常常持续超过 tACS 施加的时间[30, 32, 33]。

Given the evidence of impaired gamma activity in AD patients, and the potential of restoring brain perfusion via gamma entrainment, in the present pilot studies, we aimed to translate to humans the aforementioned preclinical findings on 40 Hz gamma stimulation in animal models of AD by means of tACS applied to a sample of 15 mild to moderate AD patients. We hypothesize that a multiday course of tACS would lead to an increase in CBF in regions targeted by tACS, with a stronger effect for participants receiving longer tACS treatments. Additionally, we hypothesized that CBF changes would show some spatial specificity in relation to the different tACS electrode montages used in the studies, and potentially
鉴于阿尔茨海默病患者伽马活动受损的证据，以及通过伽马同步恢复脑灌注的潜力，在本次初步研究中，我们旨在通过对 15 名轻度至中度阿尔茨海默病患者施加 tACS，将上述在阿尔茨海默病动物模型中关于 40 Hz 伽马刺激的前临床发现转化到人体。我们假设多日 tACS 治疗将导致 tACS 靶向区域的脑血流量增加，且接受较长时间 tACS 治疗的参与者效果更显著。此外，我们假设脑血流量的变化将在空间上与研究中使用的不同 tACS 电极布置呈现一定的特异性，并可能…
covariation with changes in the spectral power of gamma as measured via EEG as well as episodic memory scores indexing temporal lobe/hippocampal function.
与通过脑电图测量的伽马频谱功率变化以及反映颞叶/海马功能的情节记忆评分的协变关系。

Fifteen participants with mild to moderate dementia due to AD were enrolled in total (mean age 72 years, male = 9; Mini-Mental State Examination - MMSE = 23.53, $\mathrm{SD}=3.35$$\mathrm{SD}=3.35$SD=3.35\mathrm{SD}=3.35 ). Participants were enrolled in two separate open-label clinical trials exploring the impact of different tACS doses (i.e., number of stimulation sessions) and targeting approaches (i.e., positioning of tACS electrodes on the scalp and resulting induced electrical field in the brain). Participants received 1 h of daily tACS for 2 or 4 weeks in hospital settings (Monday to Friday), with baseline (pre-tACS) and follow-up (post-tACS) assessments composed of cognitive and memory testing, EEG, and perfusion MRI (ASL) data. Participants underwent additional assessments pre/post tACS not reported in the present manuscript and beyond the scope of the present study, e.g., PET imaging for $\mathrm{A}\beta$$\mathrm{A}\beta$Abeta\mathrm{A} \beta and p-tau, transcranial magnetic stimulation (TMS) measures, combined TMS-EEG recording, voice biomarkers recording, blood biomarkers.
共招募了十五名因阿尔茨海默病导致轻度至中度痴呆的参与者（平均年龄 72 岁，男性 9 人；简易精神状态检查-MMSE=23.53， $\mathrm{SD}=3.35$$\mathrm{SD}=3.35$SD=3.35\mathrm{SD}=3.35 ）。参与者被纳入两个独立的开放标签临床试验，探讨不同 tACS 剂量（即刺激次数）和定位方法（即 tACS 电极在头皮上的位置及其在大脑中诱导的电场）对效果的影响。参与者在医院环境中每天接受 1 小时的 tACS，持续 2 周或 4 周（周一至周五），基线（tACS 前）和随访（tACS 后）评估包括认知和记忆测试、脑电图及灌注 MRI（ASL）数据。参与者还接受了本手稿未报告且超出本研究范围的 tACS 前后额外评估，例如用于 $\mathrm{A}\beta$$\mathrm{A}\beta$Abeta\mathrm{A} \beta 和 p-tau 的 PET 成像，经颅磁刺激（TMS）测量，联合 TMS-EEG 记录，语音生物标志物记录，血液生物标志物等。

Depending on the tACS paradigms, participants can be subdivided into three subgroups: (i) subjects receiving 2 weeks ( 10 sessions $=10\mathrm{h}$$=10\mathrm{h}$=10h=10 \mathrm{~h} ) of unilateral temporo-frontal tACS (Group 1; $n=5$$n=5$n=5n=5 ); (ii) subjects receiving 2 weeks ( 10 sessions $=10\mathrm{h}$$=10\mathrm{h}$=10h=10 \mathrm{~h} ) of bitemporal tACS (Group 2; $n=5$$n=5$n=5n=5 ); (iii) subjects receiving 4 weeks ( 20 sessions $=20\mathrm{h}$$=20\mathrm{h}$=20h=20 \mathrm{~h} ) of bitemporal tACS (Group 3; $n=5$$n=5$n=5n=5 ) (Fig. 1). Common site of stimulation across montages was represented by the right temporal lobe (Fig. 1). Within a 1 -week period before and after the tACS intervention, participants underwent a cognitive assessment battery, 64 channels scalp EEG, and MRI assessments. All participants gave written informed consent prior to participating in the studies, registered separately on ClinicalTrials.gov (NCT03412604, NCT03290326; PI Santarnecchi).
根据 tACS 方案，参与者可细分为三个亚组：（i）接受 2 周（10 次 $=10\mathrm{h}$$=10\mathrm{h}$=10h=10 \mathrm{~h} ）单侧颞额 tACS 的受试者（第 1 组； $n=5$$n=5$n=5n=5 ）；（ii）接受 2 周（10 次 $=10\mathrm{h}$$=10\mathrm{h}$=10h=10 \mathrm{~h} ）双侧颞叶 tACS 的受试者（第 2 组； $n=5$$n=5$n=5n=5 ）；（iii）接受 4 周（20 次 $=20\mathrm{h}$$=20\mathrm{h}$=20h=20 \mathrm{~h} ）双侧颞叶 tACS 的受试者（第 3 组； $n=5$$n=5$n=5n=5 ）（图 1）。所有方案中共同的刺激部位为右侧颞叶（图 1）。在 tACS 干预前后各 1 周内，参与者接受了认知评估、电极帽 64 通道脑电图和 MRI 评估。所有参与者在参加研究前均签署了书面知情同意书，研究分别注册于 ClinicalTrials.gov（NCT03412604，NCT03290326；负责人 Santarnecchi）。

The first pilot trial NCT03290326 was designed to assess safety and feasibility of performing a two-weeks stimulation tACS treatment in patients with AD. Among the first 10 patients enrolled (Groups 1 and 2 in the manuscript), Group 1 underwent a personalized right-sided stimulation centered on the temporo-frontal lobes. The rationale was to personalize stimulation on the basis of individual $\mathrm{A}\beta$$\mathrm{A}\beta$Abeta\mathrm{A} \beta accumulation maps based on Florbetapir PET imaging. In Group 1 participants, regions with higher $\mathrm{A}\beta$$\mathrm{A}\beta$Abeta\mathrm{A} \beta load on the right temporal and frontal lobes were targeted, with stronger tACS intensity over the temporal lobe given the higher load of $\mathrm{A}\beta$$\mathrm{A}\beta$Abeta\mathrm{A} \beta compared to the frontal lobe (Fig. 1). This approach led to Group 1
第一个试点试验 NCT03290326 旨在评估对阿尔茨海默病患者进行为期两周的 tACS 刺激治疗的安全性和可行性。在最初招募的 10 名患者中（手稿中的第 1 组和第 2 组），第 1 组接受了个性化的右侧刺激，重点针对颞叶-额叶区域。其原理是基于 Florbetapir PET 成像的个体 $\mathrm{A}\beta$$\mathrm{A}\beta$Abeta\mathrm{A} \beta 积累图来个性化刺激。在第 1 组参与者中，针对右侧颞叶和额叶中 $\mathrm{A}\beta$$\mathrm{A}\beta$Abeta\mathrm{A} \beta 负荷较高的区域进行刺激，鉴于颞叶中 $\mathrm{A}\beta$$\mathrm{A}\beta$Abeta\mathrm{A} \beta 负荷高于额叶，因此在颞叶施加了更强的 tACS 强度（见图 1）。这一方法使第 1 组...
subjects receiving stimulation in the right temporo-frontal lobes; however, the electrodes of maximal injected current were slightly different across participant on the basis of the distribution of their amyloid load (i.e., maximal current density on EEG electrode positions F4 and T8 in patient X, F2 and T8 in patient Y). The unilateral personalized stimulation was conceived as a method to (i) test the spatial specificity of tACS stimulation on the individual target regions in the AD brain, (ii) improve the localization and/or recognition of ictal/EEG changes during and/or after the treatment (i.e., more probable in the stimulated right hemisphere). Given that in the first 5 patients (Group 1) a very good spatial localization was achieved and no epileptiform alterations were detected at the electrophysiological as well as clinical assessments, the investigators decided to stimulate Group 2 bilaterally, specifically over the temporal lobes.
受试者接受右侧颞额叶的刺激；然而，最大注入电流的电极因参与者的淀粉样蛋白负荷分布不同而略有差异（即患者 X 在 EEG 电极位置 F4 和 T8 处最大电流密度，患者 Y 在 F2 和 T8 处最大电流密度）。单侧个性化刺激被设计为一种方法，旨在（i）测试 tACS 刺激对阿尔茨海默病大脑中个体目标区域的空间特异性，（ii）改善治疗期间和/或治疗后癫痫发作/脑电图变化的定位和/或识别（即更可能出现在受刺激的右半球）。鉴于前 5 名患者（第 1 组）实现了非常好的空间定位，且在电生理及临床评估中未检测到癫痫样改变，研究人员决定对第 2 组进行双侧刺激，特别是在颞叶区域。
Bitemporal tACS was conceived as a way to induce more focal stimulation over the temporal lobes, given the typical $\mathrm{A}\beta$$\mathrm{A}\beta$Abeta\mathrm{A} \beta and tau protein distribution in the AD brain [34]. Additionally, tau protein, particularly expressed in the temporal lobe, is also significantly correlated with cognitive decline in AD patients, compared to a weak to null association for $\mathrm{A}\beta$$\mathrm{A}\beta$Abeta\mathrm{A} \beta. Therefore, more emphasis on the temporal lobe would guarantee a higher chance of potentially modulating tau accumulation in the future, while also still targeting hypoperfusion present in the same area in AD patients. Finally, in the NCT03412604 trial corresponding to Group 3, bitemporal stimulation was conducted for 4 weeks to enhance the probability of inducing perfusion and protein changes, while at the same time testing the safety and feasibility of a 4 -weeks stimulation protocol. The research proposal and associated methodologies were approved by the local ethics committee (Beth Israel Deaconess Medical Center IRB) in accordance with the principles of the Declaration of Helsinki.
双侧颞叶 tACS 的设计初衷是为了在颞叶区域实现更为局灶的刺激，考虑到阿尔茨海默病（AD）大脑中典型的 $\mathrm{A}\beta$$\mathrm{A}\beta$Abeta\mathrm{A} \beta 和 tau 蛋白分布[34]。此外，tau 蛋白，尤其是在颞叶中表达显著，与 AD 患者的认知下降有显著相关性，而 $\mathrm{A}\beta$$\mathrm{A}\beta$Abeta\mathrm{A} \beta 的相关性则较弱甚至无关。因此，更加侧重于颞叶的刺激将保证未来有更高的可能性调节 tau 蛋白的积累，同时仍能针对 AD 患者该区域存在的低灌注问题。最后，在对应于第 3 组的 NCT03412604 试验中，双侧颞叶刺激持续进行了 4 周，以提高诱导灌注和蛋白质变化的可能性，同时测试 4 周刺激方案的安全性和可行性。该研究方案及相关方法已获得当地伦理委员会（Beth Israel Deaconess Medical Center IRB）的批准，符合《赫尔辛基宣言》的原则。

tACS was delivered via a battery-driven current stimulator (Starstim SS32, Neuroelectrics, Barcelona, Cambridge) through surface circular $\varnothing 20\mathrm{mm}$$\varnothing 20\mathrm{mm}$O/20mm\varnothing 20 \mathrm{~mm} PISTIM electrodes (Neuroelectrics, Barcelona, Spain) with an Ag/ AgCl core and a gel/skin contact area of $3.14{\mathrm{cm}}^2$$3.14{\mathrm{cm}}^2$3.14cm^(2)3.14 \mathrm{~cm}^{2}. The electrodes were placed into holes of a neoprene cap corresponding to the international 10/20 EEG system. Gel (Signa, PARKER LABORATORIES, INC.) was applied to optimize signal conductivity and lower impedance. Electrode impedance was checked before starting each tACS session to assure safety and maximal efficacy of stimulation as well as to ensure familiarization of participants with the tACS-induced scalp sensations (e.g., tingling). For all sessions, 32 electrodes were placed on the scalp to
tACS 通过电池驱动的电流刺激器（Starstim SS32，Neuroelectrics，巴塞罗那，剑桥）施加，使用表面圆形 $\varnothing 20\mathrm{mm}$$\varnothing 20\mathrm{mm}$O/20mm\varnothing 20 \mathrm{~mm} PISTIM 电极（Neuroelectrics，西班牙巴塞罗那），电极具有 Ag/AgCl 核心和 $3.14{\mathrm{cm}}^2$$3.14{\mathrm{cm}}^2$3.14cm^(2)3.14 \mathrm{~cm}^{2} 的凝胶/皮肤接触面积。电极被放置在符合国际 10/20 脑电系统的氯丁橡胶帽的孔中。为了优化信号传导并降低阻抗，涂抹了凝胶（Signa，PARKER LABORATORIES, INC.）。在每次 tACS 会话开始前检查电极阻抗，以确保刺激的安全性和最大效果，同时确保参与者熟悉 tACS 引起的头皮感觉（例如刺痛感）。所有会话中，头皮上放置了 32 个电极以

Right temporo-frontal tACS
右侧颞额 tACS

Bitemporal tACS  双侧颞叶 tACS

Fig. 1 Experimental protocol. A Study design and relevant pre-post tACS measures. Fifteen participants with mild to moderate dementia due to AD were enrolled in total (mean age 72 years, male $=9$$=9$=9=9; MMSE $=23.53,\mathrm{SD}=3.35$$=23.53,\mathrm{SD}=3.35$=23.53,SD=3.35=23.53, \mathrm{SD}=3.35 ). Participants received 1 h of daily tACS for 2 or 4 weeks in hospital settings (Monday to Friday), with baseline (pre-tACS) and follow-up (post-tACS) assessments composed of cognitive and memory testing, EEG, and perfusion MRI (ASL) data. Participants underwent additional assessments pre/post tACS not reported in the present manuscript and beyond the scope of the present study, e.g. PET imaging for A $\beta$$\beta$beta\beta and p-tau, transcranial magnetic stimulation (TMS) measures, combined TMS-EEG recording, voice biomarkers recording, blood biomarkers. tACS was conducted targeting the normal component of the electric field either to the bilateral temporal lobes (bitemporal tACS hereafter) or unilateral (right) temporal and frontal lobes (temporo-frontal tACS hereafter), thus always impacting the right temporal lobe across all participants (corresponding to T8 on the 10/20 EEG system). Therefore, participants can be subdivided into three subgroups: (i) subjects receiving 2 weeks ( 10 sessions $=10\mathrm{h}$$=10\mathrm{h}$=10h=10 \mathrm{~h} ) of unilateral temporo-frontal tACS (Group 1 ; $n=5$$n=5$n=5n=5 ); (ii) subjects receiving 2 weeks ( 10 session $=10\mathrm{h}$$=10\mathrm{h}$=10h=10 \mathrm{~h} ) of bitemporal tACS (Group 2; $n=5$$n=5$n=5n=5 ); (iii) subjects receiving 4 weeks ( 20 sessions $=20\mathrm{h}$$=20\mathrm{h}$=20h=20 \mathrm{~h} ) of bitemporal tACS (Group 3; $n=5$$n=5$n=5n=5 ). Common site of stimulation across montages was represented by the right temporal lobe. B On the left, normal electrical field (En-field) for representative subject receiving unilateral temporo-frontal tACS (Group 1), on the right En-field for participants receiving bilateral temporal lobe stimulation (Groups 2 and 3)
图 1 实验方案。A 研究设计及相关的 tACS 前后测量。共纳入 15 名因阿尔茨海默病导致轻度至中度痴呆的参与者（平均年龄 72 岁，男性 $=9$$=9$=9=9 ；MMSE $=23.53,\mathrm{SD}=3.35$$=23.53,\mathrm{SD}=3.35$=23.53,SD=3.35=23.53, \mathrm{SD}=3.35 ）。参与者在医院环境中接受每日 1 小时的 tACS，持续 2 周或 4 周（周一至周五），基线（tACS 前）和随访（tACS 后）评估包括认知和记忆测试、脑电图（EEG）及灌注磁共振成像（ASL）数据。参与者还接受了本手稿未报告且超出本研究范围的额外 tACS 前后评估，例如用于 A $\beta$$\beta$beta\beta 和 p-tau 的 PET 成像、经颅磁刺激（TMS）测量、联合 TMS-EEG 记录、语音生物标志物记录、血液生物标志物。tACS 施加于电场的法向分量，靶向双侧颞叶（以下简称双侧颞叶 tACS）或单侧（右侧）颞叶和额叶（以下简称颞额 tACS），因此所有参与者均涉及右侧颞叶（对应 10/20 脑电系统中的 T8 点）。 因此，参与者可以细分为三个亚组：（i）接受 2 周（10 次治疗 $=10\mathrm{h}$$=10\mathrm{h}$=10h=10 \mathrm{~h} ）单侧颞额 tACS 的受试者（第 1 组； $n=5$$n=5$n=5n=5 ）；（ii）接受 2 周（10 次治疗 $=10\mathrm{h}$$=10\mathrm{h}$=10h=10 \mathrm{~h} ）双侧颞叶 tACS 的受试者（第 2 组； $n=5$$n=5$n=5n=5 ）；（iii）接受 4 周（20 次治疗 $=20\mathrm{h}$$=20\mathrm{h}$=20h=20 \mathrm{~h} ）双侧颞叶 tACS 的受试者（第 3 组； $n=5$$n=5$n=5n=5 ）。各组刺激的共同部位为右侧颞叶。B 左侧为接受单侧颞额 tACS（第 1 组）代表性受试者的正常电场（En-field），右侧为接受双侧颞叶刺激（第 2 组和第 3 组）参与者的电场（En-field）。
record EEG before and after each tACS session, although only a subset of the electrodes was used to deliver tACS [35]. tACS at a stimulation frequency of 40 Hz was applied for 1 h with a maximum intensity of 2 mA on each electrode and 4 mA total across all electrodes, preceded by a $30-\mathrm{s}$$30-\mathrm{s}$30-s30-\mathrm{s} ramp up period and followed by a $30-\mathrm{s}$$30-\mathrm{s}$30-s30-\mathrm{s} ramp down period, while research and clinical personnel carefully monitored for side effects for the entire duration of each session. Common site of stimulation across all patients and montages was represented by the right temporal lobe targeted via T8 (10/20 EEG system). Given the long stimulation sessions and the specific patient population, during tACS participants were instructed to watch a series of pre-selected videoclips from a list of selected documentaries freely available on YouTube, with the aim of maintaining a constant brain state while reducing distraction and avoiding constant interaction with operators in the room. The research team selected videos based on their length (i.e., to be approximately 1 h long), thematic subject (i.e., excluding documentaries related to war or other conflictual subjects that could cause excessive arousal and activation in the participants), and language (i.e., excluding those with extremely technical/specific terms). The themes were counterbalanced across genres to provide a nice selection of videos that would engage participants and focused their attention (primary goal of the videoclips) and be palatable for patients with diverse preferences (e.g., documentaries on animals, nature, history, technology, as well as on filmmaking and music). Each day patients were offered to choose from the list, or resume the video presented during the previous tACS session.
在每次 tACS 治疗前后记录脑电图，尽管只有部分电极用于施加 tACS[35]。tACS 以 40 Hz 的刺激频率施加，持续 1 小时，每个电极最大强度为 2 mA，所有电极总强度为 4 mA，刺激前有 $30-\mathrm{s}$$30-\mathrm{s}$30-s30-\mathrm{s} 的上升阶段，刺激后有 $30-\mathrm{s}$$30-\mathrm{s}$30-s30-\mathrm{s} 的下降阶段，研究和临床人员在整个治疗过程中仔细监测副作用。所有患者和电极布置的共同刺激部位是通过 T8（10/20 脑电系统）定位的右颞叶。鉴于刺激时间较长且患者群体特殊，tACS 期间，参与者被指示观看一系列预先选定的 YouTube 上免费提供的纪录片视频片段，目的是保持大脑状态的稳定，同时减少分心，避免与房间内的操作人员频繁互动。 研究团队根据视频的长度（即大约 1 小时）、主题内容（即排除与战争或其他可能引起参与者过度激动和激活的冲突性主题相关的纪录片）以及语言（即排除包含极其专业/特定术语的视频）来选择视频。主题在不同类型之间进行了平衡，以提供一系列能够吸引参与者并集中其注意力的视频（视频剪辑的主要目标），并且适合具有不同偏好的患者（例如，关于动物、自然、历史、技术，以及电影制作和音乐的纪录片）。每天，患者可以从列表中选择视频，或继续观看上一场 tACS 治疗中播放的视频。

Given the expected variability in cortical atrophy among participants, we did not use a fixed montage (electrode positions and currents) across participants of Group 1, but instead, we defined a cortical target designed to keep normal electrical field (En-field) amplitude fixed on the highest $\mathrm{A}\beta$$\mathrm{A}\beta$Abeta\mathrm{A} \beta deposition areas, seeking to ensure that participants received a similar electric field dose. Some variability was still observed due to the constraints on the currents that the stimulator can output and the different sizes of the targets. The resulting montage included 8 stimulation electrodes, delivering tACS at 40 Hz with a maximum intensity of 2 mA on each electrode according to current tACS safety guidelines [36], and with a resulting higher induced field on the temporal lobe due to $\mathrm{A}\beta$$\mathrm{A}\beta$Abeta\mathrm{A} \beta distribution. In Group 1, tACS was not in-phase for all electrodes since obtaining in-phase only stimulation would not be possible (current conservation). The target phase for each area was optimized so that the induced field would be maximal on the PET-defined
鉴于参与者之间皮质萎缩的预期差异，我们没有在第 1 组参与者中使用固定的电极布置（电极位置和电流），而是定义了一个皮质靶点，旨在保持正常电场（En-field）幅度固定在最高 $\mathrm{A}\beta$$\mathrm{A}\beta$Abeta\mathrm{A} \beta 沉积区域，以确保参与者接受类似的电场剂量。由于刺激器输出电流的限制以及靶点大小的不同，仍观察到一些变异。最终的电极布置包括 8 个刺激电极，按照当前 tACS 安全指南[36]，每个电极以最大 2 mA 的强度施加 40 Hz 的 tACS，并且由于 $\mathrm{A}\beta$$\mathrm{A}\beta$Abeta\mathrm{A} \beta 分布，导致颞叶诱发电场较高。在第 1 组中，tACS 并非所有电极均为同相刺激，因为仅实现同相刺激是不可能的（电流守恒）。每个区域的目标相位经过优化，以使诱发电场在 PET 定义的区域达到最大。
targets, leading to an inter-region 180 phase choice. The approach for identifying optimal stimulation targets by fusing PET and MRI data for each patient was developed by the PI of the studies (ES); personalized montages were computed by the PI in collaboration with the Neuroelectrics team using the Stimviewer algorithm and the methods described in [37, 38], adapted to the case of tACS.
靶点，导致区域间 180 度相位选择。由研究负责人（ES）开发了通过融合每位患者的 PET 和 MRI 数据来识别最佳刺激靶点的方法；个性化电极布置由负责人联合 Neuroelectrics 团队使用 Stimviewer 算法及文献[37, 38]中描述的方法计算，并针对 tACS 进行了调整。

The same safety guidelines were followed for the stimulation templates used for Group 2 and Group 3, but stimulation was targeted over the bilateral temporal lobes via 4 fixed stimulating electrodes (P8, T8, P7, and T 7 , right electrodes delivering current with a ${180}^{\circ }$${180}^{\circ }$180^(@)180^{\circ} phase degree respect to the left ones), given the usual pattern of deposition of $\mathrm{A}\beta$$\mathrm{A}\beta$Abeta\mathrm{A} \beta and tau protein commonly involving the bilateral temporal lobes when patients are symptomatic [34] (Fig. 1B). Electrode locations and stimulation intensity for bitemporal tACS used in Groups 2 and 3 were defined by the PI of the studies (ES). Stimulation intensity was titrated for each patient, given the typical discomfort reported for transcranial electrical stimulation (tES) delivered over the temporal regions. Wholescalp 64-channel resting-state EEG was collected in the week before and during the week after the tACS treatment course, along with a comprehensive neurocognitive assessment.
对于第 2 组和第 3 组使用的刺激模板，同样遵循了相同的安全指南，但刺激定位于双侧颞叶，通过 4 个固定的刺激电极（P8、T8、P7 和 T7，右侧电极相对于左侧电极的相位为 0 度），这是基于患者出现症状时β淀粉样蛋白和 tau 蛋白通常沉积于双侧颞叶的常见模式[34]（图 1B）。第 2 组和第 3 组使用的双颞 tACS 的电极位置和刺激强度由研究的主要研究者（ES）确定。鉴于经颅电刺激（tES）在颞区施加时通常会引起不适感，刺激强度针对每位患者进行了调整。在 tACS 治疗课程前一周和治疗后一周期间，收集了全头 64 通道的静息态脑电图（EEG），并进行了全面的神经认知评估。

Neuroimaging acquisition was performed on a GE 3 Tesla MR750 scanner using a 32-channel head array coil from Nova Medical. Participants underwent high resolution T1-weighted structural scan (3D T1-w BRAVO), two runs of resting state functional connectivity, resting perfusion MRI with ASL, diffusion tensor imaging, T2* GRE, and FLAIR sequences (total scan timing 60 min ). ASL studies were performed with 3D pseudo-continuous labeling (1.45s labeling, 2.025s post-labeling delay), background suppression, and 32 centric ordered 4 mm thick slices.
神经影像采集在 GE 3 特斯拉 MR750 扫描仪上进行，使用 Nova Medical 的 32 通道头部阵列线圈。参与者接受了高分辨率 T1 加权结构扫描（3D T1-w BRAVO）、两次静息态功能连接扫描、使用 ASL 的静息灌注 MRI、扩散张量成像、T2* GRE 和 FLAIR 序列扫描（总扫描时间 60 分钟）。ASL 研究采用 3D 伪连续标记（1.45 秒标记，2.025 秒标记后延迟）、背景抑制和 32 个中心顺序排列的 4 毫米厚切片。

ASL data preprocessing was performed via an ad-hoc pipeline implemented in MATLAB (MATLAB 2016b, MathWorks) and SPM12 (https://www.fil.ion.ucl.ac. uk/spm/) developed by the laboratory of one of the coauthors at BIDMC (DCA, inventor of the pseudo-continuous ASL technique). Segmentation into three classes (grey matter, white matter, and CSF) and normalization of 3D T1w BRAVO images were obtained via DARTEL [39]. ASL subtraction images were co-registered to the grey matter map and normalized to MNI152 space. Normalized CBF maps were masked with Intracranial Volume (ICV) mask from SPM12 and globally normalized for the CBF values. Manual masking of every CBF map with the
ASL 数据预处理通过一个专门的流程在 MATLAB（MATLAB 2016b，MathWorks）和 SPM12（https://www.fil.ion.ucl.ac.uk/spm/）中实现，该流程由 BIDMC 一位合著者实验室开发（DCA，伪连续 ASL 技术的发明者）。通过 DARTEL [39]对 3D T1 加权 BRAVO 图像进行三类分割（灰质、白质和脑脊液）及归一化。ASL 减法图像与灰质图进行共注册，并归一化到 MNI152 空间。归一化的脑血流量（CBF）图使用 SPM12 的颅内体积（ICV）掩膜进行掩膜处理，并对 CBF 值进行全局归一化。每个 CBF 图的手动掩膜处理与
corresponding normalized grey matter mask obtained during the segmentation process was performed for each patient for the two timepoints (pre-stimulation and poststimulation). Grey matter CBF maps were smoothed with a full width at half maximum (FWHM) of 6 mm .
在分割过程中获得的相应归一化灰质掩模针对每位患者的两个时间点（刺激前和刺激后）进行了处理。灰质脑血流量（CBF）图使用 6 毫米的半高宽（FWHM）进行了平滑处理。

In order to check for perfusion MRI data quality and check that that participants aligned the expected values of CBF in the temporal lobes (usually around $30\mathrm{mL}/$$30\mathrm{mL}/$30mL//30 \mathrm{~mL} / $\min /100\mathrm{g}$$\min /100\mathrm{g}$min//100g\mathrm{min} / 100 \mathrm{~g} for grey matter in AD patients [3,5]), individual CBF values from the temporal lobes were extracted using the REX toolbox (https://www.nitrc.org/proje cts/rex/) embedded in SPM12. Longitudinal statistical analyses assessing the impact of tACS were conducted in SPM12 on normalized grey matter CBF maps via a paired $t$$t$tt test on all participants ( $n=15$$n=15$n=15n=15; single voxel level $p<0.001$$p<0.001$p < 0.001p<0.001, cluster-level $p<0.05$$p<0.05$p < 0.05p<0.05, FDR corrected) as well as on each group of participants separately. Analyses were carried out at whole-brain level, to ensure observed changes in CBF were not amplified by the selection of a specific region-of-interest. In the case a significant change in perfusion was identified, CBF changes within significant clusters were correlated with changes ( $\mathrm{\Delta }=$$\mathrm{\Delta }=$Delta=\Delta= post minus pre) at episodic memory and language tests, as well as changes in gamma spectral power after tACS
为了检查灌注 MRI 数据的质量，并确认参与者在颞叶的脑血流量（CBF）值是否符合预期（阿尔茨海默病患者灰质通常约为 $30\mathrm{mL}/$$30\mathrm{mL}/$30mL//30 \mathrm{~mL} / $\min /100\mathrm{g}$$\min /100\mathrm{g}$min//100g\mathrm{min} / 100 \mathrm{~g} [3,5]），使用嵌入在 SPM12 中的 REX 工具箱（https://www.nitrc.org/projects/rex/）提取了颞叶的个体 CBF 值。通过 SPM12 对归一化的灰质 CBF 图进行纵向统计分析，评估 tACS 的影响，采用配对 $t$$t$tt 检验对所有参与者（ $n=15$$n=15$n=15n=15 ；单体素水平 $p<0.001$$p<0.001$p < 0.001p<0.001 ，簇水平 $p<0.05$$p<0.05$p < 0.05p<0.05 ，FDR 校正）以及各组参与者分别进行分析。分析在全脑水平进行，以确保观察到的 CBF 变化不是由于选择特定感兴趣区域而被放大。如果发现灌注有显著变化，则将显著簇内的 CBF 变化与情景记忆和语言测试的变化（ $\mathrm{\Delta }=$$\mathrm{\Delta }=$Delta=\Delta= 术后减术前）以及 tACS 后通过 EEG 测量的伽马频谱功率变化进行相关分析。
measured via EEG. Finally, the SPM Anatomy toolbox was used to label significant clusters extracted via SPM via a probabilistic atlas (see Table 1).
最后，使用 SPM 解剖工具箱通过概率图谱对 SPM 提取的显著簇进行标注（见表 1）。

Whole-scalp 64-channel resting-state EEG was collected in the week before and in the week after the tACS treatment course via an actiCHamp EEG amplifier system (Brain Products GmbH). EEG recording was obtained while subjects sat in a semi-reclined armchair. During recordings, participants were instructed to remain quiet with their face muscles relaxed. Given the specific study population, particular care was put into ensuring participants understood the importance of staying still and quiet during recording. Both participant and EEG were monitored for signs of drowsiness, at which point the participant was asked to blink their eyes a few times and reminded to stay awake. Recording was done at a sampling rate of 1 Khz and impedances were maintained below $5\mathrm{k}\mathrm{\Omega }$$5\mathrm{k}\mathrm{\Omega }$5kOmega5 \mathrm{k} \Omega during recording.
在 tACS 治疗课程前后一周内，使用 actiCHamp 脑电放大器系统（Brain Products GmbH）采集了全头皮 64 通道的静息态脑电图。记录时，受试者坐在半躺式扶手椅中。记录过程中，要求参与者保持安静，面部肌肉放松。鉴于特定的研究人群，特别注意确保参与者理解在记录期间保持静止和安静的重要性。参与者和脑电图均被监测是否有瞌睡迹象，若出现，要求参与者眨几次眼并提醒保持清醒。记录采样率为 1 kHz，记录期间阻抗保持在 $5\mathrm{k}\mathrm{\Omega }$$5\mathrm{k}\mathrm{\Omega }$5kOmega5 \mathrm{k} \Omega 以下。

Data were preprocessed using EEGLAB 2020 [40], Fieldtrip toolbox for EEG/MEG-analysis (Donders Institute for Brain, Cognition and Behaviour, Radboud University, the Netherlands, see http://fieldtriptoolbox.org),
数据预处理使用了 EEGLAB 2020 [40]和 Fieldtrip 脑电/脑磁分析工具箱（荷兰拉德堡德大学 Donders 脑、认知与行为研究所，见 http://fieldtriptoolbox.org），

Table 1 Significant clusters of changes in perfusion across all subjects. Probability anatomical mapping and cluster coordinates for significant CBF changes detected when analyzing whole-brain cortical CBF changes (post>pre) across all subjects (upper panel), and for Group 3 participants who received the 4 weeks tACS intervention
表 1 所有受试者灌注变化的显著簇。对所有受试者全脑皮层脑血流量（CBF）变化（干预后>干预前）分析中检测到的显著 CBF 变化的概率解剖映射和簇坐标（上半部分），以及接受 4 周 tACS 干预的第 3 组参与者的结果

the Brainstorm suite [41], and in-house scripts in Matlab R2017b (MathWorks Inc.). Data were initially reduced into 60 dimensions by using principal component analyses (PCA) to minimize overfitting and noise components. Band pass filter was performed using a forward-backward 4th order Butterwoth filter from 1 to 100 Hz , a notch filter between 58 and 62 Hz was applied, and the data were subsequently referenced to a global average. Subsequently, independent component analysis (ICA) was run to manually remove all remaining artifact components including eye movement/blink, muscle noise (EMG), single electrode noise, cardiac beats (EKG), as well as auditory evoked potentials. Finally, the data were interpolated for missing/removed channels using a 32 spherical interpolation.
使用 Brainstorm 套件[41]和 Matlab R2017b（MathWorks Inc.）的内部脚本进行处理。数据最初通过主成分分析（PCA）降维至 60 维，以减少过拟合和噪声成分。采用前向-后向 4 阶巴特沃斯带通滤波器对数据进行 1 至 100 Hz 的滤波，应用了 58 至 62 Hz 的陷波滤波器，随后数据被重新参考到全局平均。接着，运行独立成分分析（ICA）手动去除所有剩余的伪影成分，包括眼动/眨眼、肌肉噪声（EMG）、单电极噪声、心跳（EKG）以及听觉诱发电位。最后，使用 32 点球面插值对缺失或移除的通道进行插值。

Given the longitudinal CBF changes involving primarily the bilateral temporal lobes and the right anterior temporal lobe in particular, changes in gamma spectral power were focused on an array of tACS electrodes indexing the bilateral temporal lobes (i.e., T8, P8, P7, T7), as well as on electrode T8 as a proxy to the right anterior temporal lobe and the common stimulation electrode across tACS montages. Moreover, considering the documented slowing of EEG activity in AD patients [16, 42], with increasing spectral power for activity in the theta and delta band associated with a decrease of fast oscillations such as beta and gamma, statistical analysis was centered on detecting potential changes in gamma spectral power as well as signs of a change in spectral frequency distribution (e.g., restoration of gamma and/or decrease of slower oscillatory activity). Also, considering the limited sample size and exploratory nature of the study, we opted for a simpler statistical framework rather than a full-blown repeated measured ANOVA. Specifically, longitudinal changes in each frequency band were quantified by subtracting baseline absolute spectral power values from post-tACS ones (e.g., baseline theta minus post-tACS theta). A one-way ANOVA with a single factor “Frequency” was computed comparing the pre-posts differences in each frequency band (alpha level $=0.05$$=0.05$=0.05=0.05 ). Once a main effect was found, post hoc comparisons between pairs of frequency bands were computed as well. EEG bands were defined as follows: delta ( $1-4\mathrm{Hz}$$1-4\mathrm{Hz}$1-4Hz1-4 \mathrm{~Hz} ), theta ( $4-8$$4-8$4-84-8 Hz), alpha ( $9-13\mathrm{Hz}$$9-13\mathrm{Hz}$9-13Hz9-13 \mathrm{~Hz} ), beta ( $14-30\mathrm{Hz}$$14-30\mathrm{Hz}$14-30Hz14-30 \mathrm{~Hz} ), low gamma ( $35-$$35-$35-35- 45 Hz ), narrow gamma ( $38-42\mathrm{Hz}$$38-42\mathrm{Hz}$38-42Hz38-42 \mathrm{~Hz}; centered around the stimulation frequency of 40 Hz ), mid gamma ( $45-60\mathrm{Hz}$$45-60\mathrm{Hz}$45-60Hz45-60 \mathrm{~Hz} ), and high gamma ( $60-90\mathrm{Hz}$$60-90\mathrm{Hz}$60-90Hz60-90 \mathrm{~Hz} ). One participant of Group 2 did not complete the post-tACS EEG assessment; analyses were conducted on 14 participants.
鉴于纵向脑血流变化主要涉及双侧颞叶，尤其是右前颞叶，伽马频谱功率的变化集中在一组标记双侧颞叶的 tACS 电极（即 T8、P8、P7、T7）上，以及作为右前颞叶代理和所有 tACS 电极阵列共用刺激电极的 T8 电极。此外，考虑到阿尔茨海默病患者脑电活动减慢的已知情况[16, 42]，即θ波和δ波频段活动的频谱功率增加，同时快速振荡如β波和γ波减少，统计分析重点在于检测伽马频谱功率的潜在变化以及频谱频率分布的变化迹象（例如伽马波的恢复和/或较慢振荡活动的减少）。同时，鉴于样本量有限且研究具有探索性质，我们选择了较为简单的统计框架，而非完整的重复测量方差分析。 具体来说，每个频段的纵向变化通过用 tACS 后绝对频谱功率值减去基线值来量化（例如，基线 theta 减去 tACS 后 theta）。采用单因素“频率”的单因素方差分析（ANOVA）比较各频段的前后差异（显著性水平 $=0.05$$=0.05$=0.05=0.05 ）。一旦发现主效应，还计算了频段之间的事后配对比较。脑电频段定义如下：δ波（ $1-4\mathrm{Hz}$$1-4\mathrm{Hz}$1-4Hz1-4 \mathrm{~Hz} ），θ波（ $4-8$$4-8$4-84-8 Hz），α波（ $9-13\mathrm{Hz}$$9-13\mathrm{Hz}$9-13Hz9-13 \mathrm{~Hz} ），β波（ $14-30\mathrm{Hz}$$14-30\mathrm{Hz}$14-30Hz14-30 \mathrm{~Hz} ），低γ波（ $35-$$35-$35-35- 45 Hz），窄γ波（ $38-42\mathrm{Hz}$$38-42\mathrm{Hz}$38-42Hz38-42 \mathrm{~Hz} ；以 40 Hz 刺激频率为中心），中γ波（ $45-60\mathrm{Hz}$$45-60\mathrm{Hz}$45-60Hz45-60 \mathrm{~Hz} ），高γ波（ $60-90\mathrm{Hz}$$60-90\mathrm{Hz}$60-90Hz60-90 \mathrm{~Hz} ）。第 2 组的一名参与者未完成 tACS 后的脑电评估；分析基于 14 名参与者进行。

Participants underwent specific tests evaluating global cognition (Alzheimer’s Disease Assessment
参与者接受了评估整体认知功能的特定测试（阿尔茨海默病评估

Scale-Cognitive Subscale (ADAS-cog) [43]; MMSE [44], Montreal Cognitive Assessment (MoCA) [45], activities of daily living (ADL) [46]) to assess any potential change in overall cognitive functioning after tACS. Additionally, tasks assessing cognitive functions relevant for the brain regions stimulated by tACS were also used, using the National Alzheimer’s Coordinating Center Uniform Data Set (NACC UDS) Neuropsychological Battery: the Craft Story 21 Recall Immediate and Delayed addressing episodic memory [47], and the Category Fluency task (animals), a widely used measure of verbal fluency and language [48].
认知量表子量表（ADAS-cog）[43]；MMSE [44]，蒙特利尔认知评估（MoCA）[45]，日常生活活动能力（ADL）[46]，用于评估 tACS 后整体认知功能的任何潜在变化。此外，还使用了评估与 tACS 刺激脑区相关的认知功能的任务，采用国家阿尔茨海默病协调中心统一数据集（NACC UDS）神经心理学电池：Craft Story 21 即时和延迟回忆，针对情景记忆[47]，以及类别流畅性任务（动物），这是一种广泛使用的言语流畅性和语言测量方法[48]。

All 15 participants completed the study and tolerated the intervention with only minor side effects commonly reported in the tACS literature: tingling (10/15) rated as mild; scalp irritation (7/15) rated as mild-moderate; visual changes (8/15) rated as mild-moderate, and headache (5/15, rated as mild-moderate) induced by mechanical pressure from the stimulation cap. Participants attended 95% of the study visits (190/200 daily tACS visits, 10 sessions in total missed distributed across 7 patients), showing excellent treatment compliance. No epileptiform alterations were detected at the electrophysiological as well as clinical assessments.
所有 15 名参与者均完成了研究，并且对干预的耐受性良好，仅出现了 tACS 文献中常见的轻微副作用：刺痛感（10/15），评为轻微；头皮刺激（7/15），评为轻度至中度；视觉变化（8/15），评为轻度至中度；以及由刺激帽机械压力引起的头痛（5/15，评为轻度至中度）。参与者出席了 95%的研究访问（200 次每日 tACS 访问中出席了 190 次，7 名患者共缺席了 10 次分散的疗程），显示出极佳的治疗依从性。电生理和临床评估均未检测到癫痫样改变。

Mean CBF values in the right temporal lobe calculated at baseline (pre tACS intervention) across all the participants were $32.3\mathrm{mL}/\min /100\mathrm{g}$$32.3\mathrm{mL}/\min /100\mathrm{g}$32.3mL//min//100g32.3 \mathrm{~mL} / \mathrm{min} / 100 \mathrm{~g} ( $\mathrm{SD}=6.8$$\mathrm{SD}=6.8$SD=6.8\mathrm{SD}=6.8 ), consistent with literature on hypoperfusion in AD patients [3, 5] and validating image acquisition and CBF extraction procedures. Post intervention CBF values in the right temporal lobe increased significantly from 32.3 to $34\mathrm{mL}/\min /100\mathrm{g}$$34\mathrm{mL}/\min /100\mathrm{g}$34mL//min//100g34 \mathrm{~mL} / \mathrm{min} / 100 \mathrm{~g} $(\mathrm{SD}=8.5)(t=2.01,p<0.05$$(\mathrm{SD}=8.5)(t=2.01,p<0.05$(SD=8.5)(t=2.01,p < 0.05(\mathrm{SD}=8.5)(t=2.01, p<0.05; Cohen’s $d=0.22)$$d=0.22)$d=0.22)d=0.22). CBF values of the left temporal lobe were extracted for patients of Group 2 and $3(n=10)$$3(n=10)$3(n=10)3(n=10) who received bitemporal tACS, showing a mean value of $33\mathrm{mL}/\min /100\mathrm{g}(\mathrm{SD}=6.5)$$33\mathrm{mL}/\min /100\mathrm{g}(\mathrm{SD}=6.5)$33mL//min//100g(SD=6.5)33 \mathrm{~mL} / \mathrm{min} / 100 \mathrm{~g}(\mathrm{SD}=6.5) at baseline and $34\mathrm{mL}/\min /100\mathrm{g}(\mathrm{SD}=6.4)$$34\mathrm{mL}/\min /100\mathrm{g}(\mathrm{SD}=6.4)$34mL//min//100g(SD=6.4)34 \mathrm{~mL} / \mathrm{min} / 100 \mathrm{~g}(\mathrm{SD}=6.4) after the intervention ( $t=1.24,p=0.11$$t=1.24,p=0.11$t=1.24,p=0.11t=1.24, p=0.11 ). CBF values of the right frontal lobe were extracted for Group 1 that received personalized right temporo-frontal stimulation, revealing a baseline CBF of $35.5\mathrm{mL}/\min /100\mathrm{g}(\mathrm{SD}=9.7)$$35.5\mathrm{mL}/\min /100\mathrm{g}(\mathrm{SD}=9.7)$35.5mL//min//100g(SD=9.7)35.5 \mathrm{~mL} / \mathrm{min} / 100 \mathrm{~g}(\mathrm{SD}=9.7) and 39 $\mathrm{mL}/min/100\mathrm{g}$$\mathrm{mL}/min/100\mathrm{g}$mL//min//100g\mathrm{mL} / \min / 100 \mathrm{~g} ( $\mathrm{SD}=14$$\mathrm{SD}=14$SD=14\mathrm{SD}=14 ) after the intervention ( $t=2.36,p$$t=2.36,p$t=2.36,pt=2.36, p $=0.35$$=0.35$=0.35=0.35; Cohen’s $d=0.29$$d=0.29$d=0.29d=0.29 ).
所有参与者在基线（tACS 干预前）右侧颞叶的平均脑血流量（CBF）值为 $32.3\mathrm{mL}/\min /100\mathrm{g}$$32.3\mathrm{mL}/\min /100\mathrm{g}$32.3mL//min//100g32.3 \mathrm{~mL} / \mathrm{min} / 100 \mathrm{~g} （ $\mathrm{SD}=6.8$$\mathrm{SD}=6.8$SD=6.8\mathrm{SD}=6.8 ），与阿尔茨海默病患者低灌注的文献报道一致[3, 5]，验证了图像采集和 CBF 提取程序的有效性。干预后右侧颞叶的 CBF 值显著从 32.3 升至 $34\mathrm{mL}/\min /100\mathrm{g}$$34\mathrm{mL}/\min /100\mathrm{g}$34mL//min//100g34 \mathrm{~mL} / \mathrm{min} / 100 \mathrm{~g} $(\mathrm{SD}=8.5)(t=2.01,p<0.05$$(\mathrm{SD}=8.5)(t=2.01,p<0.05$(SD=8.5)(t=2.01,p < 0.05(\mathrm{SD}=8.5)(t=2.01, p<0.05 ；Cohen’s $d=0.22)$$d=0.22)$d=0.22)d=0.22) 。对接受双侧颞叶 tACS 的第 2 组和 $3(n=10)$$3(n=10)$3(n=10)3(n=10) 组患者提取了左侧颞叶的 CBF 值，基线平均值为 $33\mathrm{mL}/\min /100\mathrm{g}(\mathrm{SD}=6.5)$$33\mathrm{mL}/\min /100\mathrm{g}(\mathrm{SD}=6.5)$33mL//min//100g(SD=6.5)33 \mathrm{~mL} / \mathrm{min} / 100 \mathrm{~g}(\mathrm{SD}=6.5) ，干预后为 $34\mathrm{mL}/\min /100\mathrm{g}(\mathrm{SD}=6.4)$$34\mathrm{mL}/\min /100\mathrm{g}(\mathrm{SD}=6.4)$34mL//min//100g(SD=6.4)34 \mathrm{~mL} / \mathrm{min} / 100 \mathrm{~g}(\mathrm{SD}=6.4) （ $t=1.24,p=0.11$$t=1.24,p=0.11$t=1.24,p=0.11t=1.24, p=0.11 ）。对接受个性化右侧颞额叶刺激的第 1 组提取了右侧额叶的 CBF 值，基线 CBF 为 $35.5\mathrm{mL}/\min /100\mathrm{g}(\mathrm{SD}=9.7)$$35.5\mathrm{mL}/\min /100\mathrm{g}(\mathrm{SD}=9.7)$35.5mL//min//100g(SD=9.7)35.5 \mathrm{~mL} / \mathrm{min} / 100 \mathrm{~g}(\mathrm{SD}=9.7) ，干预后为 39 $\mathrm{mL}/min/100\mathrm{g}$$\mathrm{mL}/min/100\mathrm{g}$mL//min//100g\mathrm{mL} / \min / 100 \mathrm{~g} （ $\mathrm{SD}=14$$\mathrm{SD}=14$SD=14\mathrm{SD}=14 ）（ $t=2.36,p$$t=2.36,p$t=2.36,pt=2.36, p $=0.35$$=0.35$=0.35=0.35 ；Cohen’s $d=0.29$$d=0.29$d=0.29d=0.29 ）。

Apart from standard regional CBF assessment, voxelwise whole brain analyses with no prespecified masks were performed to guarantee a more unbiased result. When comparing post-tACS CBF maps with pre-tACS CBF maps across the entire brain in all subjects, a significant CBF increase was detected in multiple anatomical clusters primarily located in the right temporal
除了标准的区域 CBF 评估外，还进行了无预设掩模的体素级全脑分析，以保证结果更加客观。在所有受试者中比较 tACS 后与 tACS 前的全脑 CBF 图时，发现多个解剖簇中 CBF 显著增加，主要位于右侧颞叶。

Fig. 2 Perfusion results. A CBF increase after $tACS$$tACS$tACSt A C S. Paired $t$$t$tt test (post $>pre,p<0.05$$>pre,p<0.05$> pre,p < 0.05>p r e, p<0.05, FDR-corrected) revealed an increase of CBF selectively involving the right temporal lobe, representing the common site of stimulation across participants ( $n=15$$n=15$n=15n=15 ). B Whole cortical brain CBF analyses of participants who received the highest dose of tACS (20h of bilateral temporal lobe stimulation over 4 weeks, $n=5$$n=5$n=5n=5, Group 3 ) revealed a selective increase in CBF in the bilateral temporal lobes, accordingly to the stimulation template ( $p<0.05$$p<0.05$p < 0.05p<0.05, FDR-corrected). C Examples of CBF variations in two representative participants belonging to Group 2 (pt #8) and 3 (pt #14)
图 2 灌注结果。A CBF 在 $tACS$$tACS$tACSt A C S 后增加。配对 $t$$t$tt 检验（ $>pre,p<0.05$$>pre,p<0.05$> pre,p < 0.05>p r e, p<0.05 后，FDR 校正）显示 CBF 选择性增加，涉及右侧颞叶，代表所有参与者共同的刺激部位（ $n=15$$n=15$n=15n=15 ）。B 对接受最高剂量 tACS 的参与者（4 周内双侧颞叶刺激 20 小时， $n=5$$n=5$n=5n=5 ，第 3 组）进行的全皮层脑 CBF 分析显示，双侧颞叶 CBF 选择性增加，与刺激模板一致（ $p<0.05$$p<0.05$p < 0.05p<0.05 ，FDR 校正）。C 两名分别属于第 2 组（患者#8）和第 3 组（患者#14）的代表性参与者 CBF 变化示例。
lobe ( $n=15$$n=15$n=15n=15, Fig. 2A) ( $p<0.05$$p<0.05$p < 0.05p<0.05, FDR-corrected), specifically involving the right medial temporal pole, fusiform gyrus, and entorhinal cortex (see Table 1 for probability anatomical mapping and clusters’ coordinate). Of note, results are consistent with the right temporal lobe being the only region consistently stimulated across all 15 participants.
叶（ $n=15$$n=15$n=15n=15 ，图 2A）（ $p<0.05$$p<0.05$p < 0.05p<0.05 ，FDR 校正），具体涉及右侧内侧颞极、梭状回和内嗅皮层（概率解剖映射及簇坐标见表 1）。值得注意的是，结果与右侧颞叶是所有 15 名参与者中唯一持续接受刺激的区域一致。

Dividing participants on the basis of their respective tACS montage and looking at pre-post tACS CBF changes, both temporo-frontal (Group 1) and bitemporal (Groups 2 and 3) tACS montages resulted in significant CBF changes, however displaying different topographies of local CBF increase reflecting the two montages: Group 1 (right temporo-frontal tACS; Fig. 3A); Group 2 - 3 (bitemporal tACS; Fig. 3B).
根据各自的 tACS 电极布置对参与者进行分类，并观察 tACS 前后的脑血流量（CBF）变化，发现颞额叶（第 1 组）和双颞叶（第 2 组和第 3 组）tACS 电极布置均导致了显著的 CBF 变化，但表现出不同的局部 CBF 增加拓扑结构，反映了两种电极布置的差异：第 1 组（右侧颞额叶 tACS；图 3A）；第 2 组-第 3 组（双颞叶 tACS；图 3B）。

Additionally, when looking specifically at Group 3 who received the longest intervention targeting the bilateral temporal lobes (i.e., 20 h of tACS over 4 weeks compared to 10 h in Group 1 and 2), a significant pattern of CBF increase mainly involving the bilateral temporal lobes
此外，专门观察接受最长干预的第 3 组，该组针对双侧颞叶进行了 tACS（即 4 周内 20 小时的 tACS，相较于第 1 组和第 2 组的 10 小时），发现了主要涉及双侧颞叶的显著 CBF 增加模式，
was found, including medial temporal poles, fusiform gyri, bilateral entorhinal cortices and hippocampi (see Table 1 for probability anatomical mapping and clusters’ coordinate), matching the bitemporal tACS montage ( $t=$$t=$t=t= 2.13, $p<0.05$$p<0.05$p < 0.05p<0.05, FDR-corrected; Fig. 2B). For more detailed results on CBF changes in the left and right temporal lobe please see Supplementary Materials.
包括内侧颞极、梭状回、双侧内嗅皮层和海马（详见表 1 的概率解剖映射和簇坐标），与双颞叶 tACS 电极布置相匹配（ $t=$$t=$t=t= 2.13， $p<0.05$$p<0.05$p < 0.05p<0.05 ，FDR 校正；图 2B）。有关左侧和右侧颞叶 CBF 变化的更详细结果，请参见补充材料。

The One-way ANOVA on T7/8-P7/8 cluster, corresponding to the bitemporal stimulation, revealed an effect of tACS on spectral power around the stimulation frequency (i.e., narrow gamma, $38-42\mathrm{Hz}$$38-42\mathrm{Hz}$38-42Hz38-42 \mathrm{~Hz} ), compared to the rest of the spectrum ( $F_{(7,78)}=4.12,p<0.05;\eta 2=0.03$$F_{(7,78)}=4.12,p<0.05;\eta 2=0.03$F_((7,78))=4.12,p < 0.05;eta2=0.03F_{(7,78)}=4.12, p<0.05 ; \eta 2=0.03 ) (Fig. 4). Post hoc analysis showed that narrow gamma spectral power displays a post tACS increase higher than activity in the theta band ( $t=3.43,p<0.01$$t=3.43,p<0.01$t=3.43,p < 0.01t=3.43, p<0.01 Cohen’s $d=0.031$$d=0.031$d=0.031d=0.031 ), beta band ( $t=2.37,p<0.05$$t=2.37,p<0.05$t=2.37,p < 0.05t=2.37, p<0.05; Cohen’s $d=$$d=$d=d= 0.026 ) and high gamma band ( $t=1.84,p<0.05$$t=1.84,p<0.05$t=1.84,p < 0.05t=1.84, p<0.05 Cohen’s $d=0.024$$d=0.024$d=0.024d=0.024 ) (Fig. 4A). The same analysis performed on
对 T7/8-P7/8 簇（对应双侧颞叶刺激）进行的一元方差分析显示，与频谱的其他部分（ $F_{(7,78)}=4.12,p<0.05;\eta 2=0.03$$F_{(7,78)}=4.12,p<0.05;\eta 2=0.03$F_((7,78))=4.12,p < 0.05;eta2=0.03F_{(7,78)}=4.12, p<0.05 ; \eta 2=0.03 ）相比，tACS 对刺激频率附近的频谱功率（即窄带伽马， $38-42\mathrm{Hz}$$38-42\mathrm{Hz}$38-42Hz38-42 \mathrm{~Hz} ）有影响（图 4）。事后分析表明，窄带伽马频谱功率在 tACS 后显示出比θ波段（ $t=3.43,p<0.01$$t=3.43,p<0.01$t=3.43,p < 0.01t=3.43, p<0.01 Cohen’s $d=0.031$$d=0.031$d=0.031d=0.031 ）、β波段（ $t=2.37,p<0.05$$t=2.37,p<0.05$t=2.37,p < 0.05t=2.37, p<0.05 ；Cohen’s $d=$$d=$d=d= 0.026）和高伽马波段（ $t=1.84,p<0.05$$t=1.84,p<0.05$t=1.84,p < 0.05t=1.84, p<0.05 Cohen’s $d=0.024$$d=0.024$d=0.024d=0.024 ）活动更高的增加（图 4A）。对以下部分进行的相同分析

Fig. 3 Comparison of CBF increases between tACS montages. A When looking at significant CBF increase after tACS in Group #1 of participants who received right temporo-frontal stimulation, a pattern of predominant right temporo-frontal CBF increases was found, matching with tACS targeting. B CBF increase in participants from Group #2 + #3 ( $n=10$$n=10$n=10n=10 ) who received bilateral temporal lobe stimulation showed a significant increase of CBF predominantly localized in bilateral temporal regions
图 3 tACS 电极配置间脑血流量（CBF）增加的比较。A 在观察接受右侧颞额叶刺激的第 1 组参与者 tACS 后显著 CBF 增加时，发现主要的右侧颞额叶 CBF 增加模式，与 tACS 的靶向区域相符。B 接受双侧颞叶刺激的第 2 组和第 3 组参与者（ $n=10$$n=10$n=10n=10 ）的 CBF 增加显示，CBF 显著增加主要局限于双侧颞叶区域
electrode T8, the common site of stimulation across groups, produced a similar distribution of pre-post tACS changes across frequencies, with a significant difference between post tACS changes in the narrow gamma and theta bands ( $t=2.06,p<0.05$$t=2.06,p<0.05$t=2.06,p < 0.05t=2.06, p<0.05; Cohen’s d 0.025). Even though other gamma frequency sub-bands displayed a similar trend of narrow gamma, no other comparisons reached significance.
电极 T8，作为各组共同的刺激部位，在各频率段的 tACS 前后变化分布相似，窄伽马波段与θ波段的 tACS 后变化存在显著差异（ $t=2.06,p<0.05$$t=2.06,p<0.05$t=2.06,p < 0.05t=2.06, p<0.05 ；Cohen’s d 0.025）。尽管其他伽马频率子波段显示出与窄伽马类似的趋势，但没有其他比较达到显著性。

Narrow gamma spectral power changes observed on T8 were found to significantly correlate with increase in CBF in the right anterior temporal lobe. In details, narrow gamma spectral power changes on T8 significantly correlated with the cluster of significant CBF increase extracted among all available participants (total = 12; one participant did not complete the post tACS EEG assessment, 2 outliers were removed) ( $r=0.57;p=0.05$$r=0.57;p=0.05$r=0.57;p=0.05r=0.57 ; p=0.05; $R^2=33\mathrm{\%}$$R^2=33\mathrm{\%}$R^(2)=33%R^{2}=33 \% Fig. 4B).
在 T8 上观察到的窄伽马频谱功率变化与右前颞叶脑血流量（CBF）增加显著相关。具体而言，T8 上的窄伽马频谱功率变化与所有可用参与者中提取的显著 CBF 增加簇显著相关（共 12 人；1 名参与者未完成 tACS 后 EEG 评估，剔除 2 个异常值）（ $r=0.57;p=0.05$$r=0.57;p=0.05$r=0.57;p=0.05r=0.57 ; p=0.05 ； $R^2=33\mathrm{\%}$$R^2=33\mathrm{\%}$R^(2)=33%R^{2}=33 \% 图 4B）。

No significant changes ( $p>0.05$$p>0.05$p > 0.05p>0.05 ) in overall cognition were found after tACS (ADAS-Cog baseline mean $=$$=$== 18.27, $\mathrm{SD}=7.68$$\mathrm{SD}=7.68$SD=7.68\mathrm{SD}=7.68, post $=18.11,\mathrm{SD}=7.69$$=18.11,\mathrm{SD}=7.69$=18.11,SD=7.69=18.11, \mathrm{SD}=7.69, Cohen’s $d$$d$dd $=0.02$$=0.02$=0.02=0.02; ADL baseline mean $=68.5,\mathrm{SD}=4.68$$=68.5,\mathrm{SD}=4.68$=68.5,SD=4.68=68.5, \mathrm{SD}=4.68, post $=68.3,\mathrm{SD}=6.23$$=68.3,\mathrm{SD}=6.23$=68.3,SD=6.23=68.3, \mathrm{SD}=6.23, Cohen’s $d=0.03$$d=0.03$d=0.03d=0.03; MMSE baseline mean $=23.53,\mathrm{SD}=3.35$$=23.53,\mathrm{SD}=3.35$=23.53,SD=3.35=23.53, \mathrm{SD}=3.35, post $=22.77,\mathrm{SD}:3.68$$=22.77,\mathrm{SD}:3.68$=22.77,SD:3.68=22.77, \mathrm{SD}: 3.68,
tACS 后整体认知未见显著变化（ $p>0.05$$p>0.05$p > 0.05p>0.05 ）（ADAS-Cog 基线均值 $=$$=$== 18.27， $\mathrm{SD}=7.68$$\mathrm{SD}=7.68$SD=7.68\mathrm{SD}=7.68 ，后测 $=18.11,\mathrm{SD}=7.69$$=18.11,\mathrm{SD}=7.69$=18.11,SD=7.69=18.11, \mathrm{SD}=7.69 ，Cohen’s $d$$d$dd $=0.02$$=0.02$=0.02=0.02 ；ADL 基线均值 $=68.5,\mathrm{SD}=4.68$$=68.5,\mathrm{SD}=4.68$=68.5,SD=4.68=68.5, \mathrm{SD}=4.68 ，后测 $=68.3,\mathrm{SD}=6.23$$=68.3,\mathrm{SD}=6.23$=68.3,SD=6.23=68.3, \mathrm{SD}=6.23 ，Cohen’s $d=0.03$$d=0.03$d=0.03d=0.03 ；MMSE 基线均值 $=23.53,\mathrm{SD}=3.35$$=23.53,\mathrm{SD}=3.35$=23.53,SD=3.35=23.53, \mathrm{SD}=3.35 ，后测 $=22.77,\mathrm{SD}:3.68$$=22.77,\mathrm{SD}:3.68$=22.77,SD:3.68=22.77, \mathrm{SD}: 3.68 ，