Oxford Textbook of Neurorehabilitation

Enhancement of neuroplasticity by drug therapy

 

Pharmacological modulation of stimulation-induced LTP in motor cortex: animal studies In the rat motor cortex, long-term potentiation (LTP) usually requires activation of the N-methyl-D-aspartate (NMDA) receptor (NMDAR) [1–3]. Local disinhibition of rat motor cortex by iontophoretic application of bicuculline, an antagonist of the gamma-aminobutyric acid (GABA) type A receptor (GABAAR), results in unmasking of latent horizontal intracortical connections and rapid changes in representational organization [4] . This seminal paper, therefore, identified GABAA-ergic inhibitory cortical circuits as of crucial importance in regulating synaptic plasticity and map reorganization. Accordingly, LTP is facilitated if the rat motor cortex is locally disinhibited by bicuculline [2, 3, 5]. Similarly, LTP is facilitated in the disinhibited non-lesional surround tissue of experimentally induced focal infarction in rat sensorimotor cortex [6]. In contrast, diazepam, a benzodiazepine, and positive allosteric modulator at the GABAAR, prevents the induction of LTP in rat motor cortex [7]. Neuromodulating transmitters such as dopa-mine (DA), norepinephrine (NE), acetylcholine (ACh), and serotonin (5-hydroxytryptamine, 5-HT) can exert significant effects on neocortical LTP (for a review, see [8]), but studies that have ad-dressed pharmacological modulation of LTP in motor cortex are still limited. The dopamine D1 receptor antagonist SCH23390 and the dopamine D2 receptor antagonist raclopride reduce LTP induction in rat primary motor cortex, indicating an important supportive role of dopaminergic neurotransmission in synaptic plasticity [9]. Similarly, pharmacological blockade of muscarinic receptors by atropine prevented the induction of LTP and rather favoured the induction of long-term depression (LTD) by the same stimulation protocol [10].

 

Pharmacological modulation of stimulation-induced LTP in motor cortex: human studies In human motor cortex, LTP-like plasticity can be induced by various non-invasive brain stimulation (NIBS) protocols (for reviews, see [11–20]). LTP-like plasticity is typically expressed by long-term increase in the amplitude of the motor evoked potential (MEP), elicited by single-pulse transcranial magnetic stimulation (TMS) of the motor cortex, before and after the NIBS induction protocol, and recorded by surface electromyography (EMG), typically from a hand muscle contralateral to the stimulated motor cortex [21, 22]. The term ‘LTP-like’ has been coined [23] because the increase in MEP amplitude exhibits close similarity compared to LTP as defined and studied at the cellular/molecular level, such as cooperativity, input-specificity, associativity, duration (>30 min), and dependence on NMDA receptor activation, but investigation is necessarily indirect at the systems level [14, 24]. LTP-like plasticity can be induced with several NIBS protocols and, although the detailed physiological mechanisms may be different between protocols, the effects of pharmacological modulation on LTP-like plasticity will be reviewed here from the perspective of classes of drugs with particular modes of action. All studies included in this review have been obtained in healthy subjects, and pharmacological modulation of LTP-like plasticity was typically assessed in placebo-controlled study designs by testing a single (oral) dose of study drug.

 

1. GABAergic disinhibition: no anti-GABAergic drugs are avail-able for human use due to their adverse pro-convulsive effects. However, GABAergic disinhibition can be experimentally induced by transient limb ischaemic nerve block [25, 26]. This GABAergic disinhibition permits LTP-like plasticity induction to occur by low-frequency (0.1 Hz) repetitive TMS (rTMS) [27, 28] a protocol that does not produce overt MEP change when given alone [29].

 

2. GABAergic inhibition: Neurotransmission through the GABAAR is enhanced by benzodiazepines, allosteric positive modulators of the GABAAR. The permissive effect of GABAergic disinhibition on low-frequency rTMS induced LTP-like plasticity can be prevented by the benzodiazepine lorazepam [28]. Lorazepam reduces LTP-like plasticity induced by high-frequency (5 Hz) rTMS [30]. The benzodiazepine diazepam and the GABA re-uptake inhibitor tiagabine [31], the GABAB receptor (GABABR) agonist baclofen [32], and activation of extrasynaptic GABAARs by low-dose ethanol [33] reduce LTP-like plasticity induced by paired associative transcranial magnetic stimulation (PAS). An additional subthreshold conditioning pulse 2 ms prior to the PAS test pulse, that would produce GABAA-ergic intracortical inhibition [34, 35], is also capable of blocking PAS-induce LTP-like plasticity [36]. Lorazepam attenuates LTP-like plasticity induced by anodal transcranial direct current stimulation (tDCS) in the first 10 min after stimulation [37], but MEP amplitude in-creases at later time points, an as of yet unexplained observation.

 

3. NMDAR: Dextromethorphan is a non-competitive partial antagonist at the NMDAR that results in blockade of PAS-induced [38–39], theta burst stimulation (TBS)-induced [40] and anodal tDCS-induced [41] LTP-like plasticity. Similarly, the NMDAR antagonist memantine blocks LTP-like plasticity induced by TBS [42]. In contrast, D-cycloserine, a partial NMDAR agonist pro-longs LTP-like plasticity induced by anodal tDCS [43]. 4. Blockers of voltage-gated sodium (Na+) and calcium (Ca2+ chanels: The voltage-gated Na+ and Ca2+) chan-channel-blocking anti-convulsant lamotrigine prevents LTP-like plasticity induced by low-frequency rTMS in the context of transient limb ischaemic nerve block [28]. Similarly, lamotrigine or carbamazepine, an-other voltage-gated Na+ and Ca2+ channel-blocking anticonvulsant, reduce PAS-induced [31, 44] and anodal tDCS-induced [41] LTP-like plasticity. Nimodipine, an L-type voltage-gated Ca2+ channel blocker abolishes TBS-induced [40] and PAS-induced [39] LTP-like plasticity, while the T-type voltage-gated Ca2+ channel blocker ethosuximide also abolishes TBS-induced LTP-like plasticity but turns PAS-induced LTP-like into LTD-like plasticity [39].

 

5. Dopamine: The effects of modulators of the dopaminergic system are complex. The DA precursor levodopa enhances PAS-induced [45] and quadripulse stimulation (QPS)-induced [46] LTP-like plasticity but switches the LTP-like effect induced by anodal tDCS to an LTD-like effect [45, 47]. The enhancement of PAS-induced LTP-like plasticity by levodopa shows an inverted U-shaped dose dependency [48]. The DA D2 receptor agonists cabergoline and pramipexole have no effect on PAS-induced [49] and QPS-induced [46] LTP-like plasticity, respectively, while the D2 receptor antagonist haloperidol suppresses PAS-induced LTP-like plasticity [49]. Ropinirole, a D2/D3 receptor agonist, demonstrates an inverted U-shaped dose-dependent suppression of PAS-and anodal tDCS-induced LTP-like plasticity at low and high doses but no difference to placebo at intermediate doses [50]. The selective D2 receptor agonist bromocriptine suppresses PAS-and anodal tDCS-induced LTP-like plasticity [51]. The selective D2 receptor antagonist sulpiride blocks LTP-like plasticity induced by TBS [52] and anodal tDCS [53] but has no significant effect on LTP-like plasticity induced by PAS [54]. Addition of the D1/D2 receptor agonist pergolide does not prevent the suppression of anodal tDCS-induced LTP-like plasticity by sulpiride, underscoring the significance of D2 receptors in regulating LTP-like plasticity in human motor cortex [53]. In summary, agonists versus antagonists of the D2 receptor enhance or suppress, respectively, LTP-like plasticity in human motor cortex, but the results are not fully consistent and may depend non-linearly on dose, the balance of neurotransmission through D1 vs. D2 receptors, and the stimulation protocol to in-duce LTP-like plasticity.

 

6. Norepinephrine: Methylphenidate, an indirect NE agonist, has no effect on PAS-induced LTP-like plasticity, whereas the NE antagonist and alpha-1-receptor antagonist prazosin suppresses it [49]. The NE reuptake inhibitors amphetamine and reboxetine enhance and prolong anodal tDCS-induced LTP-like plasticity [55, 56], whereas the β-adrenergic antagonist propranolol suppresses it [55].

 

7. Acetylcholine: The ACh esterase inhibitor tacrine has no effect on PAS-induced LTP-like plasticity, whereas the muscarinic (M1) receptor antagonist biperiden reduces it [49]. Rivastigmine, an-other ACh esterase inhibitor, increases PAS-induced LTP-like plasticity but, paradoxically, reduces anodal tDCS-induced LTP-like plasticity [57]. Similarly, nicotine results (in non-smokers) in enhancement of LTP-like effects induced by PAS [58] or TBS [59], but diminishes anodal tDCS-induced LTP-like plasticity [58, 60]. The α4β2-nicotinic receptor partial agonist varenicline does not modulate LTP-like plasticity induced by PAS but sup-presses it dose-dependently when induced by anodal tDCS [61].

 

8. Serotonin: The selective 5-HT reuptake inhibitor citalopram enhances LTP-like plasticity induced by PAS [62] and anodal tDCS [63]. The overall picture of acute pharmacological effects on stimulation-induced LTP-like plasticity in human motor cortex is that anti-GABAergic manipulation and agonists of the neuromodulating neurotransmitter systems (DA, NE, ACh, and 5-HT) are usually enhancers (with few exceptions dependent on drug dose and stimulation protocol) while GABAergic drugs and antagonists of the neuromodulating neurotransmitter systems are suppressors. The currently available knowledge is summarized in Table 17.1. Pharmacological modulation of practice-dependent plasticity in motor cortex: animal studies

 

In animal models, repeated practice or motor skill learning can be associated with substantial representational plasticity of the trained motor cortex [64–66]. There exists a wealth of animal studies on pharmacological alteration of (motor) learning behavior in intact animals (for review see [67, 68]) and in animals after stroke lesion (for reviews see [68, 69]). It is beyond the scope of this chapter to provide a critical or even comprehensive review of this extensive research field. In addition, the problem with virtually all of these studies is that the observed behavioral effects were not sub-mitted to investigation of the underlying mechanisms at the level of practice-dependent representational plasticity in the motor cortex. Therefore, these studies provide no link between neurophysiological mechanism and behavioral effect, and therefore, interpretation is rather limited. In the following, single studies will be reviewed that were influential in the field, serve as important examples for modulating effects of drugs with different modes of action, and as primers for the human studies in the following subchapter.

 

GABAergic inhibition Infusion of muscimol, a GABAAR agonist, into the sensorimotor cortex of rats with surgical lesions in the ipsilateral anteromedial cortex significantly prolongs recovery from sensorimotor asymmetry when compared to animals with saline infusion [70]. Although the anteromedial cortex lesion creates a vulnerability to muscimol in the sensorimotor cortex, no detectable difference in the extent of cortical damage in this group accounts for the prolongation of behavioral asymmetry. These behavioral and anatomical data suggested for the first time that systemically delivered GABAA-ergic drugs might negatively interfere with restoration of function after cortical lesion. Cortical GABAergic signaling through GABAARs is divided into phasic synaptic and tonic extrasynaptic components. Tonic extrasynaptic GABAA-ergic inhibition is mediated primarily by α5-or δ-subunit-containing GABAARs and sets an excitability threshold for neurons [71, 72]. Pharmacological and genetic knockdown of α5-GABAARs enhance LTP and improve performance on learning and memory tasks [73]. Tonic extrasynaptic GABAA-ergic inhibition is enhanced in the perilesional tissue in a photothrombotic mice stroke model [74]. Reducing this excessive tonic inhibition by L655 708, a benzodiazepine inverse agonist of the α5-GABAAR, or by genetic reduction of the number of α5-or δ-subunit containing GABAARs has significant and sustained beneficial effects on motor recovery in this stroke model [74]. Specific antagonists of tonic extrasynaptic inhibition are not yet available for human use, but may constitute an interesting target for future drug development.

 

NMDAR D-Cycloserine, a partial NMDAR agonist enhances sensorimotor and cognitive recovery in rats when given 24 hours after 90 min of temporary medial cerebral artery occlusion (MCAO), as measured by functional magnetic resonance imaging (fMRI) and behavioral assessments 30 days after lesion [75]. Despite these favourable ef-fects, D-cycloserine, compared to saline injected control animals, does not affect final infarction size or secondary brain atrophy [75]. These data are of particular interest because it was long thought that excessive NMDAR activation in the acute stage after ischaemic stroke may contribute to delayed excitotoxic neuronal death [76]. However, all of the NMDAR antagonists studied so far have failed to show efficacy in large controlled clinical trials and, in some of the MCAO rat support a beneficial role for NMDAR stimulation during the recovery period after stroke, most likely caused by enhanced neuroplasticity rather than neuroprotection [75]. This may encourage testing of NMDAR agonists in clinical trials of human stroke rehabilitation.

 

Blockers of voltage-gated sodium (Na+ (Ca2+ ) channels The antiepileptic drug lamotrigine acts by stabilizing voltage-sensitive Na+ channels in a usage-dependent manner, preventing glutamate release and reversibly blocking excitatory neurotransmission. Therefore, lamotrigine was tested in the rat MCAO model by application of different doses or saline at the time of reperfusion [78]. In disagreement with a neuroprotective effect, lamotrigine does not demonstrate any effect on the total infarction volume, and several behavioral tests even show a disadvantage of the lamotrigine treated rats in sensorimotor recovery 7 days after infarction [78]. This is an important negative study, which suggests that blockade of voltage-gated Na+ detrimental for plasticity processes that support behavioral recovery and relearning.

 

Dopamine Elimination of dopaminergic terminals in rat primary motor cortex by intracortical injection of 6-hydroxydopamine in conjunction with desipramine to protect noradrenergic terminals results in impairment of motor skill learning (food pellet retrieval with the contralateral forepaw) [9] . This deficit is not observed when destruction of dopaminergic terminals is initiated at a time when the motor skill is already achieved through training. In addition, the learning deficit can be rescued by local infusion of levodopa [9]. A similar learning deficit can also be obtained by pharmacological treatment with specific antagonists at the dopamine D1 receptor (SCH02339) or D2 receptor (sulpiride, raclopride) when given early into the training period [9]. In the rat transient MCAO model, treatment with levodopa significantly and dose-dependently improves recovery of sensorimotor function as assessed by rotating pole test, a strongly corroborate the concept of recovery enhancing actions of levodopa treatment after ischaemic stroke.

 

Norepinephrine A milestone publication in pharmacological enhancement of sensorimotor recovery was the investigation of the effects of the NE reuptake inhibitor d-amphetamine on sensorimotor recovery. Rats subjected to unilateral ablation of the motor cortex and placed on a narrow beam display transient contralateral paresis. An immediate and enduring acceleration of recovery is produced by a single dose of d-amphetamine given 24 hours after injury. This effect is blocked by the dopamine D2 receptor antagonist haloperidol or by restraining the animals for 8 hours beginning immediately after amphetamine administration [80]. The dramatic effect of d-amphetamine vs. saline on walking beam performance in stroke rats can be appreciated in N Video 17.1. Many subsequent studies in embolic and thrombotic stroke models in rats confirmed this original finding of a recovery promoting effect of d-amphetamine, in particular in conjunction with post-stroke motor skill training (e.g. [81–84]) and revealed that these effects on recovery are associated with increased structural plasticity in the contralesional [85, 86] and ipsilesional hemispheres [87].

 

Acetylcholine ACh is important for practice-dependent motor cortical plasticity in rats because lesions of the basal forebrain cholinergic system disrupt practice-dependent cortical map reorganization as assessed by intracortical microstimulation mapping, and at the same time, im-pair forepaw motor skill acquisition [88]. The potential of cholinergic drugs to enhance recovery in animal models of stroke has not been tested. Conversely, treatment with scopolamine, a muscarinic receptor antagonist, reinstates sensorimotor deficits in recovered rats after photothrombotic stroke [89].

 

Serotonin Little evidence exists for the efficacy of serotonin reuptake inhibitors to enhance practice-dependent recovery in animal models of stroke. Fluoxetine has no beneficial effect on sensorimotor recovery in rats with focal ischaemic lesions in motor cortex [90, 91]. Given this lack of preclinical evidence for a recovery enhancing role of serotonin in animal models of stroke, it is interesting to note that the first successful prospective Phase IIb randomized clinical trial (FLAME trial) investigated the effects of fluoxetine (20 mg/day) versus placebo on recovery of paretic arm/hand function in patients after ischaemic stroke [92] (for details, see the section ‘Impact of pharmacological modulation on neurorehabilitation of stroke’). This adds to the well-known notion that animal models in stroke often face translational roadblocks that prevent prediction of successful interventions in clinical stroke trials [93].

 

Pharmacological modulation of practice-dependent plasticity in motor cortex: human studies LTP is one important mechanism involved in motor learning. The strongest supporting evidence comes from interference experiments in rats: successful motor skill learning suppresses the subsequent induction of LTP in the training motor cortex when compared to LTP in an untrained motor cortex [94–97]. The same homeostatic interference between motor learning and subsequent induction of LTP-like plasticity is found in the intact human motor cortex [23, 98, 99]. From this tight interdependence of LTP and motor learning it is reasonable to assume that pharmacological modulation of LTP and motor learning are similar [100]. This paragraph will focus on the pharmacological modulation of learning of repetitive simple movements, a form of training that is particularly effective in motor rehabilitation after stroke [101, 102].

 

 

The following practice protocols will be reviewed systematically, as they have been studied most extensively with respect to pharmacological modulating effects: Protocol A: Practice of repetitive ballistic simple finger or arm movements results in an increase in corticospinal excitability of the trained movement representation as indexed by increase in MEP amplitude in the training muscle and an increase in the maximum peak acceleration of the trained movement [103, 104]. Virtual lesion experiments show that 1 Hz rTMS of the training motor cortex disrupts this form of practice-dependent learning, indicating that the primary motor cortex is essentially involved in this learning process [104].

 

 

Protocol B (Fig. 17.1A): Focal TMS of just suprathreshold intensity applied to the hand area of motor cortex results, in many subjects, in thumb movements consistently into one direction [105]. These subjects then train ballistic voluntary thumb movements into the opposite direction, typically for 30 min at a rate of 1 Hz. During and after training, practice-induced plasticity is assessed by the shift of TMS-induced thumb movements into the training direction [105].

 

 

This is an extremely elegant experimental protocol because the amount of learning is directly expressed by an electrophysiological measure of motor cortical plasticity. Protocol C: When subjects are requested to perform brisk movements of two different representations of one body side (either hand and leg [106], or hand and shoulder [107]) as synchronously as possible, motor learning occurs by improving synchronicity of the movement of the two trained motor representations, as can be assessed by the contraction onset delay of the two muscles in the electromyogram. The associated motor cortical plasticity is defined as the magnitude of the center of gravity shifts of the two trained motor representations as assessed by MEP mapping towards each other [106, 107]. A critical comparison of the three protocols of practice-dependent plasticity leads to the conclusion that PROTOCOL B [105] bears several advantages over the other ones: it translates practice-dependent plasticity directly into an electrophysiological measure—that is, the shift in direction of the TMS induced by thumb movement, that is closely related of the physiology of voluntary movement because the earliest signal emanating from motor cortex at voluntary movement onset encodes the direction of movement [108].

 

 

Furthermore, PROTOCOL B does not use MEP amplitude as an outcome measure for motor plasticity assessment. This is an advantage because MEP amplitude is rather indirectly linked to practice-dependent plasticity [109, 110]. Finally, all studies under PROTOCOL C report dissociation with effective pharmaco-logical modulation of motor cortical representational plasticity, but lacking pharmacological effects on the improvement of motor performance. This suggests that motor cortical plasticity is not a sufficient prerequisite for motor learning. It may well be a prerequisite for motor memory formation and lasting improvement in motor performance, but this has not been tested in any of the PROTOCOL C studies. The following paragraph summarizes the pharmaco-logical effects on practice-dependent motor learning, grouped ac-cording to pharmacological modes of action as already used in the section ‘Pharmacological modulation of stimulation-induced LTP in motor cortex: human studies’.

 

1. GABAergic disinhibition: If subjects practice repeated ballistic elbow movements (PROTOCOL A) during transient forearm ischaemic nerve block—i.e. in the context of a disinhibited motor cortex [26]—then the increase in MEP amplitude and peak acceleration of the trained movement are enhanced com-pared to when the same training is performed in the absence of disinhibition [111]. Anodal tDCS decreases GABA content in the training motor cortex as measured with MRI spectroscopy, and the amount of GABA decrease directly correlates with the amount of improvement in reaction times in a visually instructed finger-sequence learning task [112], further corroborating the notion that GABA plays a fundamental role in regulating the ex-tent of motor learning.

 

2. GABAergic inhibition: Application of lorazepam prior to practice abolishes motor learning in PROTOCOL A [111, 113], PROTOCOL B [114] and PROTOCOL C [107]. Similarly, diazepam and the GABABR agonist baclofen disrupt practice-dependent motor learning and plasticity in a visuomotor skill acquisition task requiring to follow a force trajectory by adjusting voluntary ankle dorsiflexor torque [115], and lorazepam disrupts skill acquisition of arm movements in a force field [116]. On the other hand, zolpidem, a positive modulator selectively at the alpha-1 subunit bearing subtype of the GABAAR does not disrupt practice-dependent motor learning in PROTOCOL A, suggesting that other GABAAR subtypes such as the alpha-2 subunit bearing subtype mediate the GABAA-ergic disruptive effect on motor learning [113]. 3. NMDAR: Pretreatment with the NMDAR antagonist dextromethorphan suppresses practice-dependent plasticity in PROTOCOL B [105] and skill acquisition of arm movements in a force field [116]. Similarly, the NMDAR antagonists amantadine [107] and memantine [117] reduced practice-dependent plasticity in PROTOCOL C. 4. Blockers of voltage-gated sodium (Na+) and calcium (Ca2+ ) channels: Lamotrigine has no significant disruptive effect on motor learning in PROTOCOL B [114] or learning arm movement trajectories in a force field [116].

 

5. Dopamine: Levodopa [109] and the D2 receptor agonist cabergoline [110] (Fig. 17.1B) enhance motor learning in PROTOCOL B, while the D2 receptor antagonist haloperidol decreases it [110] (Fig. 17.1B). Another study tested the impact of five genetic polymorphisms with established effects on dopa-mine neurotransmission on the effects of levodopa on practice-dependent motor learning and representational motor cortical plasticity in a marble navigation task, a skilled motor task that places intensive demands on the first dorsal interosseus muscle [118]. Levodopa results in enhancement in practice-dependent motor learning and motor cortex map enlargement of the trained motor representation in those individuals with polymorphisms associated with low dopamine neurotransmission, whereas levodopa is detrimental when compared to placebo in those individuals with polymorphisms associated with high dopamine neurotransmission [118]. These data are very important as they suggest that genetic variation in the dopamine system influences learning and its modulation by levodopa.

 

6. Norepinephrine: The indirect NE agonists d-amphetamine [119, 120] and methylphenidate [110] (Fig. 17.1C) increase motor learning in PROTOCOL B, the selective NE reuptake inhibitor reboxetine enhances motor learning in PROTOCOL A [121], the selective NE reuptake inhibitor atomoxetine enhances motor learning in Protocol B [122], and d-amphetamine is an enhancer of motor learning in PROTOCOL C [123]. In contrast, the NE antagonist prazosin leads to suppression in PROTOCOL B [110, 124] (Fig. 17.1C) and the β-adrenergic blocker propranolol shows a non-significant trend towards sup-pression [124]. Of note, the beneficial effects of increasing NE neurotransmission on motor learning cannot be generalized to more complex practice tasks, such as finger-sequence learning, where reboxetine does not demonstrate a learning-enhancing effect [125].

 

7. Acetylcholine: The ACh esterase inhibitor tacrine enhances motor learning in PROTOCOL B [110] (Fig. 17.1D), while the muscarinic receptor antagonists biperiden and scopolamine result in suppression of motor learning in PROTOCOL B [110, 126] (Fig. 17.1D), but no effect of scopolamine in learning an arm movement trajectory in a force field [116].

 

8. Serotonin: Fluoxetine, a specific serotonin reuptake inhibitor enhances practice-dependent motor cortical representational plasticity [127] but not motor learning in PROTOCOL B [128] or PROTOCOL C [127]. The specific serotonin reuptake inhibitor paroxetine improves performance gain in the 9-hole peg test [129]. The acute pharmacological effects on practice-dependent plasticity in human motor cortex are summarized in Table 17.2. Of note, the pattern of effects is very similar to the acute pharmacological effects on LTP-like plasticity induced by NIBS supporting the view of over-lapping mechanisms of LTP and learning in motor cortex. Impact of pharmacological modulation on neurorehabilitation of stroke

 

The large body of preclinical studies in animal models of stroke on pharmacological modulation of recovery contrasts with the paucity of controlled studies in humans on pharmacotherapy for recovery after ischaemic stroke [130]. Almost all human stroke studies are based on either retrospective analyses, case reports, or controlled clinical trials with small numbers of patients (typically less than 50 patients). Retrospective studies are the only way to obtain information in cases of suspected detrimental drugs because prospective controlled clinical trials would be unethical. For instance, stroke patients were retrospectively divided into a ‘detrimental group’ and a ‘neutral group’ depending on whether or not they had received one or more drugs within the first 28 days after stroke that were identified to impair stroke recovery in animal models, namely the antihypertensives clonidine and prazosin, neuroleptics, and other dopamine receptor antagonists, benzodiazepines, and the anticonvulsants phenytoin and phenobarbital. Stepwise regression analyses incorporating other potential prognostic factors indicated that drug group independently influenced both the degree of upper extremity motor impairment (as measured by the Toronto stroke scale) and independence in activities of daily living (as measured by the Barthel index) 84 days after stroke.

 

These data are consistent with the detrimental effects of certain drugs on recovery in laboratory animals and suggest that similar effects may occur in humans [131]. It is another critical shortcoming of the available clinical studies that only very rarely have mechanisms of pharmacological modulation of recovery been explored. In one study [132], treatment with a single oral dose of 100 mg of levodopa enhances practice-dependent motor cortical plasticity in chronic subcortical stroke patients, as assessed with the described (section ‘Pharmacological modulation of practice-dependent plasticity in motor cortex: Human Studies’) practice PROTOCOL B [105].

 

This enhancement is similar to the one observed when levodopa is administered to aged healthy subjects [109]. In another study [133], a single oral dose of 20 mg of the selective serotonin reuptake inhibitor fluoxetine enhances ipsilesional activation of the sensorimotor cortex in lacunar stroke patients during movements of the paretic hand, as assessed with fMRI, and this enhancement correlates with the improvement in performance of finger-tapping and dynamometer tests. These two studies are extremely important because, for the first time, they go beyond the purely clinical evaluation of motor outcome and move pharmacological modulation of practice-dependent plasticity in healthy subjects (see section on ‘Pharmacological modulation of practice-dependent plasticity in motor cortex: human studies’) to the clinical stage. Exploration of pharmacological modulation of practice-dependent plasticity by electrophysiological and neuroimaging techniques may advance our knowledge on the mechanisms that enhance recovery of function after stroke lesion and, potentially, allow testing and predicting the responsiveness to a particular pharmacotherapy in individual patients. The following detailed review on the effects of pharmacological modulation of practice-dependent recovery from stroke will focus on motor recovery, as practice-dependent plasticity in the motor domain has been the focus in the preceding sections, and motor impairment is the most prevalent disability after stroke [134]. Even after having completed standard motor rehabilitation in 50–60% of stroke patients at least some degree of motor impairment will persist [135–138]. Pharmacological modulation of recovery of other major disabilities after stroke, in particular aphasia, neglect, and other cognitive deficits, has been surveyed in other authoritative reviews [139, 140].

 

GABAergic disinhibition One trial tested the effects of GABAergic disinhibition as induced by selective upper brachial plexus anaesthesia of the paretic arm in seven chronic subcortical stroke patients on practice-dependent improvement of the kinematics of ballistic pincer-grip movements [141]. Out of a performance plateau that was reached through a 1-month long training, this intervention results in significant improvement of peak acceleration and strength of the trained movement including some activities of daily living. This improvement is associated with an increase in the MEP amplitude in practice hand muscles [141].

 

Dopamine A single dose of 100 mg/d of levodopa given over 3 weeks and combined with conventional physiotherapy improves motor recovery when compared to placebo in ischaemic stroke patients, an effect that is maintained 3 weeks after the end of levodopa treatment [142]. This study was influential as it was one of the first prospective randomized double-blind placebo-controlled pharmacological trials in stroke rehabilitation, but also criticized because the levodopa vs. placebo groups were not balanced for age and stroke hemisphere, and other factors of importance for determining stroke outcome such as stroke location and stroke size were not even mentioned. Other more recent levodopa trials are listed in Table 17.3. The findings are inconsistent, so that the efficacy of levodopa in enhancing stroke rehabilitation is currently not clear. Moreover, a Val-to-Met substitution in codon 108/158 of the catechol-O-methyltransferase (COMT) decreases the dopamine break-down enzymatic activity 3–4-fold, increases the base level of dopamine in the central nervous system, and is associated with a disadvantage for recovery of motor function after stroke [143]. The interaction of this COMT polymorphism with levodopa therapy during stroke rehabilitation has not been investigated yet.

 

Norepinephrine Based on the early evidence on the recovery enhancing effects of d-amphetamine in the rat [80], the first controlled randomized clinical trial ever to enhance recovery in stroke patients tested the effects of d-amphetamine [144]. This prospective randomized double-blind placebo-controlled pilot study in acute stroke patients (less than 10 days after stroke) showed that a single dose of 10 mg of d-amphetamine coupled with physiotherapy enhances motor recovery as measured by the Fugl-Meyer scale more than placebo. However, it is not clear to which extent spontaneous recovery differences in this small sample of acute stroke patients contributed to the results. Furthermore, this initial finding could not be replicated in several other trials (see Table 17.3) and d-amphetamine has potentially serious adverse effects (in particular blood pressure elevation, in-creased mortality) so that d-amphetamine is no longer considered as a primary choice in pharmacotherapy of stroke rehabilitation [145]. Novel selective NE reuptake inhibitors such as reboxetine may have a superior safety profile. One particularly interesting study showed that a single oral dose of 6 mg of reboxetine significantly increases maximum grip power and index finger-tapping speed of the paretic hand of subacute or chronic stroke patients [146]. This enhanced motor performance is associated with a re-duction of cortical hyperactivity towards physiological levels, especially in the ipsilesional ventral premotor cortex and supplementary motor area. Connectivity analyses revealed that in stroke patients neural coupling of ipsilesional primary motor cortex with ventral premotor cortex and supplementary motor area is significantly reduced compared with healthy controls and that reboxetine treatment normalizes this deficient connectivity [146]. These findings underscore the capability of modern neuroimaging to further our understanding of the mechanisms that mediate pharmacological enhancement of function at the systems level of cortical networks.

 

Acetylcholine No randomized clinical trials investigating the effects of ACh esterase inhibitors on stroke motor recovery have been published, as of May 2018.

 

Serotonin The largest multicentre randomized clinical trial published to date to explore pharmacological enhancement of stroke rehabilitation is the Fluoxetine for Motor Recovery after Acute Ischaemic Stroke (FLAME) trial [92]. The trial showed that early (within the first 5–10 days after stroke) initiation of a daily dose of 20 mg of fluoxetine in combination with physiotherapy improves recovery, measured by the Fugl-Meyer scale, 90 days after stroke compared to placebo. Although this is an impressive and clinically relevant result, the mechanisms underlying this recovery enhancing effect remained unexplored. The fluoxetine group also shows a significantly lower occurrence of depression at 90 days compared to the placebo group, which may have contributed (non-specifically) to the enhanced motor recovery, even though the superiority of fluoxetine over placebo as measured by the Fugl-Meyer scale was independent of the effect of fluoxetine on depression [92]. The finding that fluoxetine improves the rate of favourable outcomes measured by the modified Rankin scale [92] has important public health implications as this rate improvement is similar to those achieved with thrombolytic therapy [147]. Unlike thrombolytic therapy, however, the usefulness of which is limited by having to give it within 4.5 hours of stroke onset [148], a selective serotonin reuptake inhibitor (SSRI) could be given to a much larger cohort of stroke patients, as there is no (known) critical time window.

 

Limitations and outlook Pharmacological enhancement of neuroplasticity and neurorehabilitation success is still an emergent field, despite its longstanding tradition in preclinical research. Evidence-based guidelines are not yet available and recent systematic reviews conclude that more evidence by adequately powered large-scale randomized controlled clinical trials is warranted before more definite recommendations can be given as the presently available evidence is promising but largely based on small-scale proof-of-principle studies [130, 166, 167]. Also, the optimal interval after stroke onset to start pharmacotherapy, the frequency and dose of medication, the optimal duration of medication and rehabilitation, and the in-tensity and components of concomitant skills practice, remain important features for the design of randomized clinical trials [168]. On the other hand, a recent Phase IIb randomized double-blind placebo-controlled parallel group clinical trial with 118 stroke patients demonstrated clinically relevant efficacy of early treatment with the SSRI fluoxetine in enhancing stroke recovery [92]. The rate of improvement of disability as measured with the modified Rankin scale 90 days post-stroke was comparable to the benefit obtained in thrombolysis trials for acute stroke treatment [147]. The effect size on the Fugl-Meyer Motor Score 90 days post-stroke in the fluoxetine trial, expressed by Cohen’s d was 0.76, indicating a strong effect [83].

 

This opens up the realistic perspective that early pharmacotherapy for enhancing stroke recovery becomes a standard practice with clinically meaningful effects in the management of subacute stroke patients soon. However, there are still important gaps of knowledge that need to be filled by further research. One important issue relates to the complexity of stroke pathobiology: ‘therapy must acknowledge the ‘Janus-faced’ nature of many stroke targets and must identify endogenous neuroprotective and repair mechanisms’ [93]. That is, excitability decreasing pharmacotherapy (e.g. by anticonvulsants) in the acute phase of stroke may be neuroprotective but this positive effect may be overridden by detrimental effects on neuroplasticity and recovery [131, 159]. In addition, recovery-modulating drugs may have potentially serious adverse effects that limit their wide ap-plication, such as blood pressure elevation and increased mortality by d-amphetamine [145]. Furthermore, it is not clear to what extent effects of pharmacological modulation on ischaemic stroke recovery reviewed in this chapter can be generalized to other causes of acquired lesions of the central nervous system, such as traumatic brain injury or spinal cord injury. While it appears that there is large consistency for traumatic brain injury and ischaemic stroke (for reviews see [140, 169–171]), there are also examples of divergence: For instance, patients with incomplete spinal cord injury do not benefit from adding levodopa to training of gait [172] while add-on treatment with levodopa improves motor recovery patients after ischaemic stroke [132, 142]. Clonidine, an α2-adrenergic receptor agonist, and therefore an inhibitor of presynaptic NE release, impairs recovery of function in an animal model of focal traumatic brain injury [173], and was identified as a ‘detrimental drug’ in motor recovery after ischaemic stroke in humans [131], while clonidine has demonstrated beneficial effects in combination with locomotor training in patients with severe but incomplete spinal cord injury (for review see [174]). The reasons for these discrepancies are only partly understood. Likely, they are related to the specific functions of neuromodulating neurotransmitters in neuronal circuits of cerebral cortex versus spinal cord. For instance, the presynaptic inhibition of NE release in spinal cord circuitry by clonidine can lead to relief from pain and spasticity, and induce concomitant changes towards normalization of the cyclic EMG locomotor activation pattern in incomplete spinal cord injury patients [174], while the same mechanism of presynaptic inhibition of NE release in cerebral cortex has detrimental effects on motor recovery after ischaemic or traumatic brain lesions, most likely by its negative interference with cortical LTP [175]. Another issue relates to the neurobiological mechanisms under-lying enhancement of neuroplasticity and recovery. This chapter demonstrates that animal models are generally very useful for identification of mechanisms that may be of translational value for clinical applications. However, this is not a straight avenue, and the chapter closes with the urgent recommendation that future clinical trials should include measurements of network reorganization, using fMRI, electroencephalography (EEG), magnetoencephalography (MEG), or TMS (for laudable examples, see [132, 133, 146]), in addition to clinical endpoints, in order to improve our understanding of the mechanisms underlying recovery enhancement and to open opportunities for applying more specific interventions to residual neuronal networks in individual stroke patients.

 

Close window by clicking in grey area >