Atomoxetine Enhances a Short-Term Model of Plasticity in Humans

Methods 

After interview to omit subjects with contraindications to the medications, 10 healthy subjects (5 women; age range, 21–69y; mean, 36.3y) signed written consent forms and attended their first sessions. Handedness was assessed by the Edinburgh Handedness Inventory. All procedures were approved by the Wake Forest University Health Sciences Institutional Review Board.

Participants were administered 40mg of atomoxetine (Strattera), 75mg of venlafaxine (Effexor), or placebo in a pseudorandomized, double-blinded crossover fashion across 3 visits. Randomization using a counterbalanced design was done by a study associate. Only this person knew the identity of the drug administered at each session, until unblinding after completion of all data collection and analyses. Subjects were assigned orders in which to receive the drugs, and each treatment was sealed in an opaque envelope. The primary investigator administered the drugs and TMS sessions. Envelopes were numbered according to session, along with subject number, and were opened on signing of informed consent at the beginning of the appropriate session. Subjects were required to allow a minimum of 1 week between their visits to permit washout of the drug previously administered. Drug administration occurred 90 minutes before the beginning of baseline testing (fig 1); this time was chosen to maximize the probability of peak plasma levels during training.20, 26 Blood pressure was measured immediately before drug administration and periodically until the completion of the session for that day. During the 90-minute waiting period, subjects remained seated and relaxed.

View Large Image Download to PowerPoint

Fig 1. Schema of study. Abbreviation: PO, per os.

After the 90-minute period, subjects underwent baseline TMS testing and motor training (see fig 1). Subjects received stimulation to the motor cortex contralateral to their dominant hand. TMS to the optimal position to produce isolated thumb movements of the dominant hand was delivered via a MagStim 200 Magnetic Stimulatora and a figure-of-8 shaped coil (wing diameter, 8.7cm) fixed in position. Coil stability was ensured by use of a head-coil holding kitb and temporary marks on the scalp. Subjects sat in a comfortable chair with their forearm supported in a molded arm cast. Four fingers were stabilized in a slight extension, whereas the thumb remained completely unrestrained. Electromyographic activity and motor evoked potentials (MEPs) were recorded through cloth electrocardiogram monitoring electrodesc placed over the belly of the extensor pollicis brevis (EPB) and flexor pollicis brevis (FPB) muscles. The electromyographic signal was amplified and band-pass filtered 30Hz to 1kHz using an isolated bioelectric amplifierd and fed into a laboratory computer for offline analysis. Thumb movements were recorded with a 3-dimensional accelerometer mounted on the phalanx of the thumbe and allowed identification of flexion, extension, abduction, and adduction movements. Accelerometer and electromyographic signals were digitized at 4000Hz.

After determining the resting motor threshold27 (minimum TMS stimulation intensity required to elicit at least 5/10 MEPs ≥50mV) and movement threshold (minimum stimulation intensity required to elicit reliable and consistent thumb movements), 60 stimuli were delivered at 0.1Hz (see fig 1). The baseline direction was defined as the predominant (mean) direction of TMS-evoked movements (fig 2).

View Large Image Download to PowerPoint

Fig 2. Illustration of the TMS paradigm modeling use-dependent plasticity. At baseline, TMS evoked movement in a consistent direction in subjects (ie extension, abduction), designated here by the black lines radiating from the accelerometer. Subjects then performed training movements in approximately the opposite direction (ie flexion, adduction) to baseline (black arrow). Posttraining, the direction of TMS-evoked thumb movements changed from the baseline direction to the trained direction (black lines falling into the training target zone, TTZ), close to a 180° change from baseline.

The resting electromyographic threshold was measured to permit use of a conditional triggering paradigm. The computer would thus trigger the TMS to fire only when the ambient electromyographic activity, and thus motor activity, was at or below the identified electromyographic activity threshold. This ensured constant cortical excitation because the suprathreshold motor activity of a subject would prevent the TMS from firing and elicit an “active” indicator icon onscreen. Auditory feedback continuously monitored the appropriate relaxation of the EPB and FPB during TMS. Additionally, a continuously updated graphical oscilloscope embedded in the LabView programf display provided visual feedback to the principal investigator, allowing artifacts and incongruities to be noted, further ensuring accuracy and correctness of data acquired.

After identification of the baseline movement direction, participants received instructions and were requested to make 5 movements to practice movement timing, speed, and direction. Participants then underwent 30 minutes of training. In blocks of 10 minutes, they performed voluntary brisk thumb movements in the direction opposite of their baseline (see Fig 1, Fig 2). During this time, minimum and maximum movement amplitudes were recorded. To describe the effects of training, we defined the training target zone (TTZ) as the window ±20° of the mean training direction (see fig 2).

To monitor the consistency of training kinematics across conditions, we measured the dispersion of training movement directions and the magnitude of the first peak acceleration of these movements. Using the accelerometer and LabView display, participants followed visual and auditory movement cues. The computer issued a tone at a rate of 1Hz, indicating that the participant should make 1 movement to the instructed direction at each tone. At each movement, the computer processed the data from the electromyographic and accelerometer recordings. If the amplitude was within the identified thresholds and the direction and timing were correct, then a green graphical indicator appeared onscreen by the respective recording displays; red signified incorrect. Thus, the direction and magnitude of each movement was monitored online, and participants were encouraged to perform accurately and consistently via visual and auditory computer feedback derived from the accelerometer and electromyographic signals. Of critical importance, constant participant attention was ensured through this paradigm.

After completion of each of the first 2 training epochs, the TMS-evoked movement direction was determined by delivering 10 pulses at 0.1Hz (see fig 1). Subjects were tested at 0.1Hz for 10 minutes after the final (third) training epoch (see Fig 1, Fig 2). During testing, we recorded MEPs in the hand muscles (EPB, FPB) mediating movements in the training (MEPagonist) and baseline (MEPantagonist) directions. The endpoint measure was the increase in the proportion of TMS-evoked movements into the TTZ after training.

Results 

Six subjects were right-handed, 3 were left-handed, and 1 was ambidextrous (tested on the right hand). One participant withdrew from the study because of venlafaxine-induced side effects and received no testing; another did not receive testing for the venlafaxine session because of side effects (table 1). Thus, data from 9 subjects were included for the group analyses. Effects of the drugs on blood pressure and heart rate were not significantly different from placebo (F=.589, P>.05) (fig3).

Table 1. Side Effects Experienced by Subjects

	Subject	Placebo	Venlafaxine	Atomoxetine
	1	Fatigue	None	Fatigue for 1h (4h after intake)
	2	None	Brief lightheadedness	Happiness for ≈3h
	3	None	Mild nausea, jittery, irritability, insomnia	None
	4	Fatigue	Severe nausea, vomiting, diarrhea	Fatigue
	5	None	None	None
	6	None	None	None
	7	None	Mild nausea	None
	8	None	Mild nausea, 1 episode of vomiting	None
	9	None	None	None
	10	None	Sweating, moderate nausea	NA

Abbreviation: NA, not applicable.

View Large Image Download to PowerPoint

Fig 3. Average vital signs for each condition across time. The effects of the drugs did not differ significantly from placebo.

At baseline, motor thresholds for muscles acting as agonist and antagonist to the training movements, movement thresholds, and amplitudes of MEPs did not differ across conditions (90min after drug intake, F=0.847, P>.05), pointing to the similar baseline excitability levels of the motor system during the 3 sessions (Table 2, Table 3). Training, however, led to significant changes in MEP ratio (posttraining MEP/baseline MEP, see table 3). Both placebo and atomoxetine administration resulted in an increase of MEPagonist after the training compared with baseline (P=.022 and P=.015, respectively). Venlafaxine showed no significant effect on MEPagonist after training compared with baseline (P>.05).

Table 2. Average TMS Thresholds for Each Condition

	Condition	rMT Agonist	rMT Antagonist	MovT	% rMT
	Placebo	44.7±6.30	43.9±6.41	50.4±5.23	114.4±5.62
	Venlafaxine	43.6±6.05	43.1±5.74	49.8±6.23	114.9±5.39
	Atomoxetine	42.2±6.72	42.1±7.01	48.3±5.72	115.4±7.06

NOTE. Values are mean ± standard error (SE). Units for resting motor threshold (rMT) and movement threshold (MovT) are percentage of maximal stimulator output. Percentage rMT is the average percentage increase for movement threshold over the mean resting motor threshold. There were no significant differences between groups.

Table 3. MEP Measurements at Baseline and 30 Minutes Posttraining for Each Condition

	Condition	Baseline (mV)		Posttraining (mV)
		Agonist	Antagonist	Agonist	Antagonist
	Placebo	0.65±0.22	0.75±0.55	1.17±0.68⁎	0.80±0.34
	Venlafaxine	0.53±0.19	0.71±0.25	0.54±0.13	0.69±0.23
	Atomoxetine	0.68±0.33	0.96±0.80	1.20±1.01⁎	0.94±0.57

NOTE. Values are mean ± SE.

⁎ Significant comparison to baseline: P=.011 for placebo condition and P=.008 for atomoxetine condition posttraining compared with baseline MEPagonist.

All 3 groups had comparable angular dispersion (F=1.522, P>.05) and first peak acceleration of movements during training (F=2.029, P>.05) (table 4). Thus, drugs and placebo did not affect differently the quality of the training movements. Baseline movement directions across subjects were varied; however, it has previously been shown that the actual individual baseline directions observed across subjects are irrelevant to the validity of the model and results.4 Furthermore, the drugs were not seen to affect the baseline TMS-evoked direction. Analyses showed significant effects of intervention (placebo, venlafaxine, atomoxetine) on increase in the proportion of TMS-evoked movements into the TTZ (our endpoint measure, F=13.527, P<.001) (fig 4). Significant differences between drugs existed after 20 minutes (atomoxetine vs venlafaxine, P<.001) and 30 minutes (atomoxetine vs placebo, P<.001; atomoxetine vs venlafaxine, P<.001) of training (see fig 4). Venlafaxine did not have a significant effect on the proportion of movements into the TTZ compared with placebo (P>.05). The changes induced by the drugs were similar across subjects, as revealed by individual endpoint (TTZ) data after 30 minutes of training (fig 5).

Table 4. Training Kinematics for Each Condition

	Condition	Peak Acceleration (m/s2)	Angular Dispersion (length of unit vector)
	Placebo	4.503±0.438	0.918±0.056
	Venlafaxine	4.859±0.575	0.924±0.045
	Atomoxetine	4.501±0.602	0.959±0.043

NOTE. Values are mean ± SE. There were no significant differences between groups.

View Large Image Download to PowerPoint

Fig 4. Increase in proportion of movements in TTZ, organized by total elapsed training time (in minutes) and drug administered. Significant comparisons are denoted by group (P, placebo; V, venlafaxine; A, atomoxetine) and total elapsed training at time of measurement (10, 20, or 30min), on the higher group of the comparison. *P<.001.

View Large Image Download to PowerPoint

Fig 5. Individual subject scores (increase in proportion of movements in TTZ) across conditions after 30 minutes of training. Subjects represented by different symbols.

Discussion 

As hypothesized, a single oral dose of the selective norepinephrine reuptake inhibitor atomoxetine in healthy subjects elicited a significant increase in use-dependent plasticity, although the dual selective serotonin and norepinephrine reuptake inhibitor venlafaxine did not. Evidence thus far indicates a critical role for norepinephrine in mediating learning and memory28, 29, 30, 31 as well as functional recovery after brain injury.9, 32 Sympathomimetic drugs facilitate recovery from brain injury in rats.12, 33 Conversely, some drugs, such as α1-adrenergic antagonists, can impair or even reinstate motor deficits if given during a critical period after brain injury in animals.34 To extend these observations to humans, Goldstein35 showed that a cohort of stroke patients taking drugs known to impair plastic changes had worse clinical recovery after stroke than did a similar group not receiving those drugs. Clinical studies with amphetamine have sought to explore the beneficial effects of the nonselective noradrenergic drug amphetamine on plasticity in recovery. However, as noted in Long and Young,7 the timing of amphetamine administration with therapy has been a recurring problem. Animal models have shown the importance of a tight coordination of drug intake and physical therapy.

Atomoxetine is a highly selective inhibitor of the presynaptic norepinephrine transporter, with minimal affinity for cholinergic, serotoninergic, and dopaminergic receptors.16, 17 Atomoxetine is favorable for use in poststroke motor recovery not only for its specificity for norepinephrine but also for its longer half-life ensuring sustained effects during motor training after stroke. In addition, it has fewer cardiovascular side effects than amphetamine. Atomoxetine appears to be as effective as amphetamine in enhancing plasticity in this model. Finally, atomoxetine significantly increases c-fos expression in the prefrontal cortex; however, this effect may be related to its impact on extracellular dopamine levels.17

Venlafaxine is a phenylethylamine compound and a representative of a new class of antidepressants that selectively inhibit the presynaptic reuptake of serotonin and norepinephrine. Saletu et al20 showed improvements in attention, fine-motor activity, and reaction time under venlafaxine in a dose-dependent manner. The absence of an effect with venlafaxine in the present study may in part be because of the dose-response curve. Although no work has yet been done in humans, as seen in Koch et al,21 the time course of the effect of venlafaxine varies by dosage in animals. Thus, the 75-mg dose used in our study may have been insufficient to elicit a sufficiently prolonged increase in norepinephrine; furthermore, even at its peak activity, a higher dose may have been necessary because venlafaxine seems to predominantly affect serotonin at lower doses because of its 30-fold higher affinity for serotonin over norepinephrine.21 Alternatively, it may be that the time course of its TMS-detectable effects is unrelated to the time course of its plasma levels, and thus peak activity may have occurred outside of the training epoch.36

It also is possible that the greater affinity of venlafaxine for serotonin mediated the lack of increase of use-dependent plasticity in this study. TMS studies in humans treated with serotonergic drugs have yielded complex results, showing increases,37, 38 decreases,39 and both38 increases and decreases in motor cortical excitability. Pleger et al37 found that a single dose of fluoxetine increased the sum of MEP amplitudes and enlarged the motor map, as measured by TMS in a model of cortical excitability. Conversely, Robol et al39 showed that a single dose of citalopram increased motor threshold, silent period, and intracortical inhibition, consistent with findings with zolmitriptan.40 Ilic et al38 showed that a single dose of sertraline increased the steepness of the input-output curve but depressed paired-pulse facilitation. Functional magnetic resonance imaging studies found an increase in cortical motor activity after administration of selective serotonergic drugs correlating with improved motor function.41, 42 Our results show that venlafaxine suppressed the increase in agonist-muscle MEP; because motor training has been shown to transiently increase MEPs in the involved muscles, venlafaxine appears to have reduced training-induced changes in motor cortical excitability.43

The lack of increase of use-dependent plasticity under the effects of venlafaxine could also be related to the fact that several subjects in our study presented side effects, mostly nausea. This could have caused decreased attention during motor training; however, this is unlikely because the analysis of training kinematics showed the speed and consistency of trained movements under venlafaxine to be similar to those under the effects of placebo and atomoxetine.

Conclusions 

Our findings show the efficacy of atomoxetine to enhance use-dependent plasticity, and the specificity of this drug for norepinephrine transporters supports the hypothesis of the fundamental role of norepinephrine in mediating these plastic changes in healthy subjects. Such drugs may represent a more pharmacologically selective means of enhancing outcome after stroke.