tDCS是否是安全的？
近年来，在人体经颅上运用弱直流电来无创性的诱导皮质兴奋性调节这一可能性可以通过运用经颅磁刺激来进行系统性的评价。目前数据显示经受刺激后，主要运动皮质的经颅直流电刺激提高或降低皮层兴奋性长达一个小时。兴奋性在方向和持续性上的转变依赖于应用刺激的极性，强度和持续性。同样的影响在视觉皮层上也会出现。因此，这项技术在诱发皮层神经可塑性上是一个有希望的工具。
在他最近发表的文章“人体大脑极化：旧工具来延长大脑兴奋性的无创性调节的再评估”（即第４期第１４卷，《临床神经生理学》）中，Priori指出，Agnew and McCreery(1987)提出为了经颅电磁脑刺激的安全，最大的刺激电极的电荷密度应小于40µC/cm²，直流电刺激几分钟，强度为1mA，电极面积为35cm²，被Nitsche and Paulus (2000,2001)认为是安全的。但是根据Agnew and McCreery(1987)提出的安全限度，Nitsche and Paulus (2000)认为他们的协议安全是不明确的，以5分钟1毫安强度的刺激通过35cm²的电极面积意味着8500µC/cm²的电极密度，这远远超过了40µC/cm²的安全极限。
据此，读者可能会认为我们小组实施的研究是不安全的。我们假定Priori的陈述是依据一个简单的误解，而我们在此想澄清这个误解。因为安全问题对经颅直流电刺激将来在人类上的应用是至关重要的，似乎我们对现有著作进行一个详细的讨论和再评估是有必要的。此外，我们会根据现有准则为安全的直流电刺激提供一个标准。
总的来说，有必要说清楚哪种电刺激机制会对神经或大脑组织带来损害。这在Agnew and McCreery (1987)的文章中有详细描述，其中有重要信息表明我们小组使用的经颅直流电刺激的协议应当被看做是安全的。
首先电流驱动的组织损伤的一般机制不应当仅仅受限于超阈值的脉冲刺激协议，还应当适用于弱连续直流电刺激。根据Agnew and McCreery (1987)，可能导致大脑损害的脑电刺激的重要的可能的特征包括由电极组织界面引起的电化学产生的有毒大脑物质和（金属）电极溶解物质。明显的，正如Agnew and McCreery (1987)指出，这些特征在经颅刺激中不重要，因为刺激电极和大脑组织不是直接接触的。为了减少电极-皮肤界面的化学程序，应当使用特殊电极（如下所示）。根据Agnew and McCreery (1987)，另一种可能的导致界面上的皮肤受损害的途径是电极下的热显影。这种有显示不会出现在我们所使用的经颅直流电刺激的协议中（Nitsche and Paulus ，2000）。
第二，电刺激会通过诱导神经元多动和脑组织加热导致组织损伤（Agnew and McCreery ，1987)。由于皮质多动症导致的损害结果是指数小时的高频超阈值的刺激的后果(Agnew et al., 1983)。但是，使用我们的协议的经颅直流电刺激只会带来皮层兴奋性的适度改变（经颅磁刺激引起的，肌肉组织唤起的，电位波幅变化是40%，基准值1电压，最高可引起的波幅为5电压）（Nitsche and Paulus, 2000, 2001; Nitsche et al., 2003)。而且，在静止膜电位的神经元诱发动作电位下，经颅直流电刺激的影响在阈限以下，不足以引起反应。因此，神经元多动带来损害的结果是不太可能的。鉴于在电极下根本没有这种情况(Nitsche and Paulus, 2000)以及进入大脑的临界电流密度或总电荷只有与电极直接接触进入皮肤的50%，神经元组织过热带来的损害可以被排除（Rush and Driscoll, 1968)。
但是，是否能从现有数据中得到刺激的安全参数仍然是个问题。对于重复的阈上的电刺激，已经通过实验来测试决定刺激的安全限度的因素是：
1.电流密度（刺激强度（A）/ 电极尺寸(cm²)）
2.总电荷（刺激强度（A）/电极尺寸（cm²）x总刺激持续时间（脉冲持续时间x脉冲数）(s) [C/cm²])（与原始物理公式相反，这里电荷指的是刺激区域）
3.每相电荷（刺激强度（A）x每脉冲持续时间(µs)=µc)
4.电荷密度（刺激强度（A）/ 电极尺寸(cm²) x每脉冲持续时间(µs) [µc/cm²])(Agnew and McCreery, 1987; McCreery et al., 1990; Yuen et al., 1981).
知道电流密度独立于刺激持续时间，总电荷反映了整个刺激持续过程中的产品的电流密度和刺激持续时间，每相电荷和电荷密度指一系列的应用数小时的高频阈上的刺激中的一个脉冲很重要。并且，了解安全限值指的是只在重复性的高频刺激给定数小时下应用的每相电荷和电荷密度也是至关重要的。这就是为什么每相电荷和电荷密度不适用于经颅直流电刺激的唯一原因，因为在经颅直流电刺激里，一个整个刺激持续时间内，只给定一个（连续的）刺激物。不幸的是，Priori指的这个安全限度就是电荷密度。McCreery认为（与其个人交流）得出经颅直流电刺激的安全限度的正确参数应当是电流密度。正如McCreery et al. (1990)指出，低于25 mA/cm²的电流密度，就算给予数小时的高频刺激也不能造成大脑组织损伤。在我们的协议中，我们最多用低于此安全限度千分之一0.02857 mA/cm²的电流密度来刺激。因为刺激持续时间是造成组织损伤的另外一个重要的因素，总电荷是经颅直流电刺激安全标准的其它重要参数。最小总电荷216 C/cm²(Yuen et al., 1981)下就能检测到组织损伤。迄今为止，我们的协议包括总电荷0.022 C/cm²，这又一次远远低于这一临界值。这说明如果考虑到电诱导组织损伤的可能机制和安全限值的适用性，我们的研究协议没有超过Agnew and McCreery的小组所说的安全限值。
虽然由Agnew and McCreery的小组研究中得出来的安全限值大体上适用于经颅直流电刺激，由于他们的研究和经颅直流电刺激所实验的刺激协议上的技术性区别，关于安全限值的精确值的可比性可能会受到一定限制。我们建议除非目前有更多数据可利用，电流密度和总电荷不应该超出我们小组所使用的协议限值太多。针对这些协议的一些其它研究也正在开展，并无证据显示这些协议是有害的：证据显示我们协议下的经颅直流电刺激在电极下并不导致热效应(Nitsche and Paulus, 2000)，并不提高血清神经元特异性烯醇化酶水平(Nitsche and Paulus,2001; Nitsche et al., 2003)（一个神经损伤的敏感标志(Steinhoff et al., 1999)也并不导致弥散加权或对比增强的核磁共振成像的变化，或病态的脑电图变化（未发表的观察）。另外，由于多动性，与基线相比，已完成的４０％的兴奋性变化不应该导致神经损伤，其影响的受局限性的持续时间不会导致稳定的（数天或数周）功能的或构造的皮层的变异，这些在健康的人中都是不受欢迎的。在我们的实验室，我们已经对５００个人进行这项协议的实验，迄今为止，没有发现任何副作用，除了在开始数几秒的刺激中电极下一点轻微的刺痛的感觉或由于可能刺激开始和结束的太过迅速而有的短暂的闪光的感觉。针对后一点，我们现在采用了楔形的电流开关。因为似乎超过0.02857 mA/cm²的电流密度（即1 mA/35cm²）可能会比较痛（未发表的观察），我们建议最好不要超过这个值。另外，电极及导联可能会导致脑干及心脏神经刺激，比较危险，应当避免。Lippold and Redfearn(1964)曾描述了一例对脑干进行刺激后呼吸紊乱，言语中断和精神错乱的例子，不能完全排除电流可以调节心率。因此，根据现有知识，不仅皮层刺激电极，远程的电极都应当被定位来防止电流经过脑干。刺激装备应该保证恒电流密度，因为电流密度，而非电压是诱导神经元损伤的相关参数(Agnew and McCreery, 1987)，如果电阻不稳定，恒压装置可能导致不必要的电流密度的变化。为了减小电极和皮肤界面的化学反应，经颅直流电刺激应该采用完全被盐水浸泡的海绵包裹的非金属导电橡胶电极(Nitsche and Paulus 2000)。这样一来，海绵就是唯一直接与皮肤接触的东西，化学反应就能减小了。如果刺激应用在小孔之上，电流就能集中，有效电极尺寸就能减小了(Agnew and McCreery, 1987; Rush and Driscoll, 1968)。因此，化学反应就能被避免了。持续性刺激可能会导致超过一个小时的兴奋性变化，应当被谨慎地应用在健康人体上，这是因为兴奋性变化需要如此长时间地巩固和稳定(Abraham et al., 1993),而且可能会导致功能紊乱。同样因此，长期的兴奋性变化不能应用多于一周一次，因为用在动物身上重复性的每日刺激导致兴奋性变化稳定在数周甚至数月(Weiss et al., 1998)。
我们知道由刺激持续性延长而很大可能性诱导的后遗症的扩展是这项技术可能能进行临床应用的先决条件。但是，我们认为，在刺激持续性被延长之前，我们必须进行额外的系统的安全性研究。这些研究目前正在我们的研究室进行。
如果所有这些先决条件得到满足，并组成了一个有效但安全的经颅直流电刺激疗程，那就再没有理由怀疑经颅直流电刺激是有害的了。
Safety criteria for transcranial direct current stimulation (tDCS) in humans In recent years, the possibility of inducing cortical excitability modulations in humans non-invasively by the transcranial application of weak direct currents (tDCS) was systematically evaluated by using transcranial magnetic stimulation (TMS) for evaluation. Present data demonstrate that tDCS of the primary motor cortex increases or decreases cortical excitability for up to an hour after the end of stimulation. The direction and duration of these shifts in excitability depend on the polarity, intensity and duration of the applied stimulation (Nitsche and Paulus, 2000, 2001;Nitsche et al., 2003). Similar effects can be achieved in the visual cortex (Antal et al., 2001, 2003). This technique is thus evolving as a promising tool to induce cortical neuroplasticity.
In his recently published article “Brain polarization in humans: a reappraisal of an old tool for prolonged noninvasive modulation of brain excitability” (Clinical Neurophysiology, Volume 14, Issue 4) Priori stated that“Interpreting the criteria of Agnew and McCreery (1987) who proposed a maximum charge density at the stimulating electrode of ,40 µC/cm² for the safety of transcranial electric and magnetic brain stimulation, DC stimulation for several minutes, at 1mA intensity with an electrode area of 35 cm², was considered safe by Nitsche and Paulus (2000,2001). How Nitsche and Paulus (2000)—according to the safety limits proposed by Agnew and McCreery (1987)—considered their protocol safe is unclear: 5 min of stimulation at 1 mA through one electrode with an area of 35 cm² implies a charge density of about 8500 µC/cm², well above the limit of 40 µC/cm².” (Priori, 2003) From this a reader might be led to think that the studies conducted by our group were not safe. We assume that Priori’s statement is due to a simple misunderstanding which we aim to clear up here. Because the safety issue is of central importance for the future application of tDCS in humans, it seems necessary to go into a detailed discussion and a reanalysis of the existing literature. Additionally we will propose a standard for safe DC stimulation according to currently available criteria.
Generally, it has to be clarified which mechanisms of electrical stimulation could cause neuronal or brain tissue damage. These are described in detail in the article of Agnew and McCreery (1987) and represent the crucial information which demonstrates that the tDCS protocols used by our group should be regarded as safe.First, general mechanisms of current-induced tissue damage, which are not restricted to suprathreshold pulsed stimulation protocols, but are also applicable to weak continuous DC stimulation, need to be considered.According to Agnew and McCreery (1987), important possible features of electrical brain stimulation, which could lead to brain damage, include electrochemically produced toxic brain products and (metallic) electrod dissolution products caused by the electrode–tissue interface.
Clearly, as Agnew and McCreery (1987) point out,these factors are not important in the case of transcranial stimulation, because stimulation electrodes and brain tissue do not come into direct contact. In order to minimize chemical processes at the electrode–skin interface, special
electrodes (see below) should be used. According to Agnew and McCreery (1987), another possible way in which the skin could be damaged at the interface would be heat development under the electrodes. This has been shown not to occur under the tDCS protocols we use (Nitsche and Paulus, 2000).
Second, the electrical stimulation could cause tissue damage by inducing neuronal hyperactivity and brain tissue heating (Agnew and McCreery, 1987). Damaging effects due to cortical hyperactivity refer to the effect of high-frequency suprathreshold stimulation over hours(Agnew et al., 1983). However, tDCS using our protocols induces only moderate changes in cortical excitability(TMS-elicited, muscle-evoked, potential amplitude changes are about 40%, with baseline-values of 1 mV,and a maximum elicitable amplitude of about 5 mV)(Nitsche and Paulus, 2000, 2001; Nitsche et al., 2003).Moreover, the effects of tDCS are subthreshold with regard to eliciting action potentials in neurons at resting membrane potential. Thus a damaging effect by neuronal hyperactivity seems improbable. Damage from the heating of neuronal tissue can be ruled out, in view of the fact that this was not even the case directly under the electrodes(Nitsche and Paulus, 2000) and that the critical current density or total charge entering the brain will only be about 50% of that directly under the electrode on the skin (Rush and Driscoll, 1968).
However, the question arises whether safety parameters for stimulation can be derived from currently available data.For repetitive suprathreshold electrical stimulation, factors that have been tested experimentally to determine the safety limits of stimulation are:
1. current density (stimulation strength (A)/electrode size(cm²)),
2. total charge (stimulation strength (A)/electrode size(cm²) ｘ total stimulation duration (pulse duration ｘnumber of pulses) (s) [C/cm²]) (in contrast to the original physical formula, here charge refers to the stimulated area),
3. charge per phase (stimulation strength (A) ｘ duration of a single pulse (µs) ＝ µC), and
4. charge density (stimulation strength (A)/electrode size(cm²) ｘ duration of a single pulse (µs) [µC/cm²])(Agnew and McCreery, 1987; McCreery et al., 1990; Yuen et al., 1981).
It is important to note that current density is independent of stimulation duration and total charge reflects the product of current density and stimulation duration for a whole stimulation session, whereas charge per phase and charge density refer to only one pulse of a train of high-frequency suprathreshold stimuli applied over hours. Also, it is essential to realize that the safety limits stated for charge per phase and charge density apply only if repetitive highfrequency stimulation is given for several hours. This is the very reason why charge density and charge per phase are not applicable to tDCS, because in tDCS only one (continuous) stimulus is given in a whole session. Unfortunately, the only safety limit Priori refers to is charge density. In the opinion of McCreery (personal communication) the appropriate parameter for deriving safety limits for tDCS should be current density. As shown by McCreery et al. (1990),current densities below 25 mA/cm² do not induce brain tissue damage even by applying high-frequency stimulation over several hours. In our protocols, we stimulate with a maximum current density of 0.02857 mA/cm² which is a thousand fold lower than this limit. Because duration of stimulation is an important additional factor in causing tissue damage, total charge is the other important parameter for tDCS safety criteria. Tissue damage has been detected at a minimum total charge of 216 C/cm² (Yuen et al., 1981).Our protocols so far encompass maximum total charges of 0.022 C/cm² and, again, are far below these thresholds. This demonstrates that our study protocols are not beyond the safety limits described by Agnew and McCreery’s group,if the possible mechanisms of electrically induced tissue damage and the applicability of safety limits are considered.
Although these safety criteria derived from the studies of Agnew and McCreery’s group should generally be applicable to tDCS, due to technical differences between the stimulation protocols tested in their studies and tDCS,comparability may be restricted with regard to the exact values of the safety limits. We propose that unless more data are available than at present, current density and total charge should not be extended much beyond the limits of protocols used by our group. Additional studies have now been performed for these protocols, which produced no evidence that these may be harmful: they show that tDCS under our protocols does not cause heating effects under the electrodes (Nitsche and Paulus, 2000), does not elevate serum neurone-specific enolase level (Nitsche and Paulus,2001;Nitsche et al., 2003) (a sensitive marker of neuronal damage (Steinhoff et al., 1999) and does not result in changes of diffusion-weighted or contrast-enhanced MRI,or pathological EEG changes (unpublished observations).
Additionally, the accomplished excitability changes of about 40% compared to baseline should not result in neuronal damages due to hyperactivity, and the restricted duration of the effects do not induce stable (in terms of days or weeks) functional or structural cortical modifications, which could be undesirable in healthy subjects. This protocol has been tested in about 500 subjects in our laboratory so far without any side-effects, apart from a slight tingling sensation under the electrode during the first seconds of stimulation or the sensation of a short light flash if the stimulation was switched on or off abruptly.With regard to the latter point, we now prefer a wedgeshaped on and off-current switch. As it seems that current densities above 0.02857 mA/cm² (which refers to 1 mA/35cm²) could be painful (unpublished observations), we recommend that this value should not be exceeded. Also,electrode montages that could result in brainstem or heart nerve stimulation can be dangerous and should be avoided.
After stimulating the brainstem, Lippold and Redfearn (1964) described one case of disturbed breathing, speech arrest and psychosis, and it cannot be ruled out completely that a current flow could modulate rhythmogenesis of the heart. Thus, according to currently available knowledge,not only the cortical stimulation electrode, but also the remote one should be positioned so as to avoid current flow through the brainstem. The stimulation device should guarantee a constant current density, since current density and not voltage is the relevant parameter for inducing neuronal damage (Agnew and McCreery, 1987) and a constant voltage device could result in unwanted changes of current density if resistance is unstable. To minimize chemical reactions at the electrode–skin interface, tDCS should be performed with non-metallic, conductive rubber electrodes, covered completely by saline-soaked sponges(Nitsche and Paulus 2000). These sponges are then the only material in direct contact with the skin and chemical reactions should be minimized. If the stimulation is applied above foramina, currents could be focused and the effective electrode size diminished (Agnew and McCreery, 1987; Rush and Driscoll, 1968). Consequently,this should be avoided. Stimulation durations which are likely to result in excitability changes of more than an hour should be applied cautiously in healthy subjects,since excitability changes lasting for such a long time consolidate and stabilize (Abraham et al.,1993), and could be dysfunctional. For the same reason, long-term excitability changes should not be induced more than once a week, since repetitive daily stimulation in animals results in excitability changes that are stable for weeks or even months (Weiss et al., 1998).
We are aware that an extension of the after-effects, most probably inducible by a further prolongation of stimulation duration, is the prerequisite for possible clinical applications of this technique. However, in our opinion additional systematic safety studies must be performed before stimulation duration can be extended. These studies are currently being performed in our laboratory.If all of these preconditions are met, which in total constitute a regimen for effective, but safe tDCS, there is no reason to suspect that tDCS could be harmful.
Acknowledgements
We thank Christine Crozier for improving the English style of this manuscript.
References
Abraham WC, Mason SE, Demmer J, Williams JM, Richardson CL, Tate WP, et al. Correlations between immediate early gene induction and the persistence of long-term potentiation. Neuroscience 1993;56:717–27.Agnew WF, McCreery DB. Considerations for safety in the use of extracranial stimulation for motor evoked potentials. Neurosurgery 1987;20:143–7.
Agnew WF, Yuen TG, McCreery DB. Morphologic changes after prolonged electrical stimulation of the cat’s cortex at defined charge densities. Exp Neurol 1983;79:397–411.
Antal A, Nitsche MA, Paulus W. External modulation of visual perception in humans. NeuroReport 2001;12:3553–5.Antal A, Kincses ZT, Nitsche MA, Paulus W. Manipulation of phosphene thresholds by transcranial direct current stimulation in man. Exp Brain Res 2003;150:375–8.
Lippold OCJ, Redfearn JWT. Mental changes resulting from the passage of small direct currents through the human brain. Br J Psychiatry 1964;110:768–72.
McCreery DB, Agnew WF, Yuen TG, Bullara L. Charge density and charge per phase as cofactors in neural injury induced by electrical stimulation.IEEE Trans Biomed Eng 1990;37:996–1001.Nitsche MA, Paulus W. Excitability changes induced in the human motor cortex by weak transcranial direct current stimulation. J Physiol 2000;527:633–9.
Nitsche MA, Paulus W. Sustained excitability elevations induced by transcranial DC motor cortex stimulation in humans. Neurology 2001;57:1899–901.
Nitsche MA, Nitsche MS, Klein CC, Tergau F, Rothwell J, Paulus W. Level of action of cathodal DC polarisation induced inhibition of the human motor cortex. Clin Neurophys 2003;114:6004.
Priori A. Brain polarization in humans: a reappraisal of an old tool for prolonged non-invasive modulation of brain excitability. Clin Neurophysiol 2003;14:889–95.
Rush S, Driscoll DA. Current distribution in the brain from surface electrodes. Anaesth Analg Curr Res 1968;47:717–23.
Steinhoff BJ, Tumani H, Otto M, Mursch K, Wiltfang J, Herrendorf G, et al.Cisternal S100 protein and neuron-specific enolase are elevated and site-specific markers in intractable temporal lobe epilepsy. Epilepsy Res 1999;35:75–82.
Weiss SR, Eidsath A, Li XL, Heynen T, Post RM. Quenching revisited: low level direct current inhibits amygdala-kindled seizures. Exp Neurol 1998;154:185–92.
Yuen TGH, Agnew WF, Bullara LA, Skip Jaques BS, McCreery DB.Histological evaluation of neural damage from electrical stimulation:considerations for the selection of parameters for clinical application.Neurosurgery 1981;9:292–9.
Michael A. Nitsche, David Liebetanz, Nicolas Lang,Andrea Antal, Frithjof Tergau, Walter Paulus* University of Goettingen,Department of Clinical Neurophysiology,Robert-Koch-Strasse 40, 37075 Goettingen, Germany * Corresponding author.
Tel.: t49-551-3966-50;
fax: t49-551-3981-26.
E-mail address: w.paulus@med.uni-goettingen.de
doi:10.1016/S1388-2457(03)00235-9