Abstract
Keywords
Introduction
Sensory input is required for maintaining normal cortical representations within the sensory and motor cortex, and motor performance as well as motor skills acquisition are dependent on somatosensory input.1-3 Not surprisingly, somatosensory deficits following a brain insult, such as a stroke, are associated with slower recovery of motor function.4,5 Research evidence from both animals and humans with no cortical impairments indicates that somatosensory inputs can induce plastic changes in the brain as they effect corticomotor excitability of the areas representing the stimulated body part that may outlast the stimulation period.6-11 It has therefore been postulated that augmenting somatosensory input may lead to long-term plasticity and enhanced motor recovery in patients with neurological impairments.3,12
Various modes of somatosensory input have been used as therapeutic interventions to enhance motor recovery following stroke, including needle electroacupuncture, 13 repetitive passive movement, 14 and mechanical vibration. 15 Transcutaneous electrical stimulation (TENS) is an additional modality that can provide sensory input by peripheral nerve stimulation via electrodes placed on the skin. Adjustment of the stimulation intensity, which is achieved primarily by manipulating pulse duration, pulse frequency, and/or current amplitude, determines whether the evoked responses are merely sensory or also include muscle contractions. 16
Although sensory TENS is frequently used in rehabilitation, primarily for pain modulation,17,18 there has been a growing interest in the past decade in its use as a means of enhancing motor recovery following a stroke.3,12 TENS stimulators are readily available, relatively cheap, and risk free, and unlike needle electroacupuncture, can be easily applied and controlled by most individuals, particularly if the stimulation remains at a pleasant sensory level. 16 Should it be found that somatosensory input via TENS can enhance motor recovery following a stroke, this noninvasive modality could be easily incorporated as part of the rehabilitation treatment regimen. Thus, the purpose of the present review is to determine the therapeutic effect that sensory TENS has on motor recovery following a stroke.
Methods
Literature Search Methods
A computerized search of articles published in English was conducted independently by the 2 authors. The following databases were searched from their inception to January 2010: PubMed, EMBASE, CINHAL, ISI Science Citation Index, Cochrane library, Cochrane Stroke Group Trials Register, Hooked on Evidence, and the Physiotherapy Evidence Database (PEDro). Articles appearing in these databases ahead of publication were also included. Reference sections of relevant review and research articles were used to identify additional pertinent articles. The following keywords were used: stroke/hemiplegia/hemiparesis/cerebrovascular accident; electric/electrical; stimulation/nerve stimulation; afferent/sensory/somatosensory. The full text of articles identified by the title and/or abstract as possibly relevant were retrieved and final decision on inclusion was made by both researchers.
Inclusion Criteria
Study design
Studies included were randomized or quasi-randomized controlled trials (quasi-RCTs) in which patients were treated with sensory TENS to enhance motor recovery following a cerebral vascular accident. We included studies in which participants were assigned to 1 of 2 or more treatment groups involving a comparison between the effects of stimulation and control or placebo conditions. Studies in which none of the treatment groups consisted of sensory stimulation alone were excluded as such a design could not isolate the effect of sensory stimulation.
Study population
Studies were included if the study participants were diagnosed with either ischemic or hemorrhagic stroke, with no limitation as to the amount of time elapsed since the insult, its location, or its size.
Intervention
The intervention of interest was peripheral electrical stimulation applied to the skin via external electrodes at a sensory or subthreshold level but not at a level inducing muscle contraction. Although electrical stimulation via skin electrodes at a sensory level is usually termed
Outcome measures
No limits were set as to outcome measures used to determine motor impairment and/or functional performance. However, we excluded studies that focused on the effect of sensory electrical stimulation on any of the following impairments without additional measures of motor recovery: dysphagia, neglect, pain, sensory perception, or spasticity.
Methodological Quality
The methodological quality of the included studies was graded independently by the 2 researchers using the PEDro Scale, which was developed by the Centre for Evidence Based Practice in Australia. 20 The scale was developed on the basis of expert consensus as to the determinants of the quality of RCT and has been shown to have good validity and reliability.21,22 It consists of 10 items leading to a maximal score of 10. Points are allocated for randomization (random and concealed allocation), data reporting (baseline similarity, number of participants reported and assessed, between-group statistical comparison of at least 1 key outcome measure, and point and variability estimates), blinding (participants, therapists, and evaluators), and intention to treat. Although reporting of eligibility criteria is also noted, it is not included in the total PEDro score. Studies were rated between excellent and poor on the basis of the PEDro score, as follows: 9 to 10, excellent; 6 to 8, good; 4 to 5, fair; <4, poor. Discrepancies between the graders were resolved by mutual agreement.
In addition, all studies were graded according to the therapy/prevention arm of the Oxford Centre for Evidence-Based Medicine levels of Evidence. 23 Accordingly, levels of evidence were classified as follows: 1a, systematic review of RCT; 1b, RCTs with narrow confidence interval; 1c, all-or-none case series; 2a, systematic review cohort studies; 2b, cohort study/low quality RCT; 2c, outcomes research; 3a, systematic review of case-controlled studies; 3b, case-controlled study; 4, case series, poor cohort case controlled; expert opinion.
Data Extraction and Analysis
Data from the retrieved studies were extracted independently by both researchers on the following topics: patient age, time elapsed since stroke, research design, treatment characteristics, and treatment effects. The various outcome measures were classified in accordance with the health dimensions addressed by the International Classification of Functioning Disability and Health.
24
MetaAnalyst software was used to perform meta-analysis where appropriate, sufficiently homogeneous data concerning performance were available.
25
When sufficient data were not reported to enable calculation of the effect size, the relevant author was contacted, and the necessary data were requested. Effect size (Hedges
Results
Selection of the Literature
The preliminary search in PubMed, using the different combinations of keywords and limited to the English language, yielded 455 titles. Only 15 of these were identified as relevant to the present review. Excluded titles were related to functional electric stimulation, electrical stimulation that induces muscle contraction, outcome measures not in the inclusion criteria, acupuncture treatments, needle or implanted stimulation, single-subject design, and other nonrelated titles. No additional titles were retrieved from the remaining databases. Characteristics of the retrieved studies are summarized in Table 1 and Table 2 (online).
Participant Characteristics, Outcome Measures, and Treatment Effects
Abbreviations: ExG, experimental group; CG, control group; SD, standard deviation; FU, follow-up; ICF, International Classification of Functioning, Disability and Health; EMG, electromyography; MVC, maximum voluntary contraction; Rx, treatment; UE, upper extremity; SEPs, somatosensory evoked potentials; MMAS, Modified Motor Assessment Scale; JTHFT, Jebsen-Taylor Hand Function test; CSS, Composite Spasticity Scale; TMS, transcranial magnetic stimulation; FIM, Functional Independence Measure; ARAT, Action Reach Arm Test.
Methodological Quality
Quality rating scores based on the PEDro classification as well as evidence level of each study are presented in Table 3 (online). The mean (standard deviation) PEDro score was 6.9 (1.2). Whereas only 1 of the studies was graded as excellent (score of 9), 26 the majority of the studies were graded as good, with scores ranging between 6 and 8.27-38 Only 2 studies were graded as fair (score of 5).39,40 Two items were scored positively in only 1 or 2 studies (blind therapist 26 and concealed allocation, respectively),32,36 whereas all the other items were scored positively in at least 10 studies. In terms of research design, 5 studies used a repeated-measures crossover design,27,29,31,33,37 4 studies allocated the participants to more than 2 groups,26,35,36,39 and 6 studies involved a 2-group comparison.28,30,32,34,38,40 With the exception of 1 study, 39 the number of participants per group was well balanced. Evidence level ranged between 1b and 3b, which is not surprising, because case studies were excluded a priori, and no expert opinion has been published on the topic. Evidence level of the majority of the studies was 2b or 3b, with only 4 articles involving rather large RCTs designated at evidence level 1b.26,35,36,38 Upper-extremity (UE) rehabilitation, whether in the acute30,32,40 or chronic27,29,31,33,37,39 stage was addressed by studies at evidence level 2b or 3b, with all studies indicating a positive effect on at least 1 outcome measure. Similarly, lower-extremity (LE) rehabilitation at the chronic stage was addressed by 2 studies at evidence level 1b35,36 and 2 at level 2b,28,34 indicating some positive outcomes. LE rehabilitation at the acute/subacute stage was addressed by only 2 studies at the 1b level26,38 with only 1 of them demonstrating a positive effect.
Participants
The retrieved studies included a total of 446 patients, with 263 of them treated by sensory stimulation. (It should be noted that as the study reported by Ng and Hui-Chan in 2009 36 was an extension of their earlier study, 35 only the patients reported in 2009 were tallied. The effect on gait velocity, which was reported in both studies, was considered only in the later study.) The number of patients per study receiving active treatment ranged between 8 and 51, and the number receiving sham or control treatment ranged between 8 and 29. However, in the majority of studies (9/15), the number of participants receiving sensory stimulation was <15.
In 1 study, where TENS was applied to the LE, patients were in the acute phase of recovery, 26 and in 1 study, which involved the UE, patients were both in the acute and subacute stages of recovery. 32 Three studies involving treatment of the UE and/or LE were conducted with patients in the subacute stage.30,38,40 However, the majority of studies involved patients in the chronic stage of recovery, and 6 of these studies applied the TENS to the UE27,29,31,33,37,39 and 4 to the LE.28,34-36
Intervention
No adverse effects were reported following any of the interventions. The method of stimulation varied to a great extent between studies in terms of location, pulse characteristics, intensity, and duration of the stimulation.
Location
In 1 study, TENS was applied to both the LE and the UE, 40 whereas in another study, it was applied to the UE in some of the patients and to the LE in others. 39 However, in 7 of the studies, the stimulation was applied solely to the UE27,29-33,37and in 6 studies solely to the LE.26,28,34-36,38
In 7 of the 9 studies involving the UE, the surface electrodes were positioned over the median nerve at the wrist.27,29-33,37 In 2 of these studies, ulnar nerve stimulation was also induced,27,32 and in another study, stimulation was applied to the median, ulnar, and radial nerves. 37 Only 1 UE study focused on stimulating the triceps brachii muscle, using electrodes to cover the muscle. 40 Unlike most studies that used relatively small electrodes, 1 study used a glove-like electrode to provide sensory stimulation to the entire hand and a sock-like electrode to likewise stimulate the entire foot. 39
As for the studies involving stimulation to the LE, placement of the electrodes was over the junction of the gastrocnemius muscle and Achilles tendon in 1 study, 28 over the common peroneal nerve in 3 studies,34,38,40 and over 4 acupuncture points, related to LE function following traditional Chinese medicine, in 3 studies.26,35,36
Pulse characteristics
Pulse shape characteristics were not reported in 8 of the studies26,27,29-33,37 and varied in the remaining studies. Pulse duration ranged from 0.125 ms to 1 ms, with 7 of the studies employing the 1-ms-long duration26,29-33,37 and the other 8 studies using durations shorter than 0.240 ms.26,28,34-36,38-40 Pulse frequency ranged between 10 and 100 Hz. The most often used frequency was 10 Hz, delivered in 1-Hz pulse trains (5 pulses per train),27,29-33,37 and the next most often used frequency was 100 Hz.26,34-36 All studies used continuous stimulation, with the exception of the study by Yavuzer et al, 38 which used a 10-s on, 10-s off duty cycle.
Intensity of stimulation
Although current intensity (in mA) correlates with the level of sensation perceived by the patient, it cannot be used alone as an absolute determinant of perceived sensation because factors such as skin conductivity, location, and size of the electrode will also affect the level of sensation. 16 Therefore, intensity was in most cases defined in terms of induced sensation, which varied between studies. In the studies by Peurala et al 39 and Conforto et al, 30 the current intensity was raised to a level just below sensory threshold. In 3 of the studies, the current intensity was double the level necessary to reach threshold perception.34-36 Sensation induced in the remaining studies ranged from mild paresthesias (or tingling) to strong paresthesias, with all studies maintaining stimulation below the pain threshold and below visible motor contraction. In 5 of these studies, EMG (electromyography) was used to ensure that no muscle contractions were involved.27,29-31,33
Duration of treatment
In 5 of the studies, all involving the UE, treatment was provided in a single 2-hour-long session.29,31-33,37 In another similar study involving the UE, the stimulation was provided in two 2-hour sessions, whereby the ulnar and median nerves were stimulated simultaneously in 1 session and alternatively in the other. 27 All other studies used multiple treatment sessions, with the duration of treatments ranging between 20 and 120 minutes, the frequency of sessions per week ranging between 2 and 6, and the number of treatment weeks ranging between 3 and 8. Thus, the total number of treatment hours of studies using multiple treatments ranged from 8 hours 28 to 24 hours, 28 with a mean of 15.5 hours.
Control group treatment
In the majority of studies, the control groups were led to believe that they were receiving treatment, but the current intensity was actually maintained close to or at zero.26,27,29,32,34-36,38-40 However, in 4 studies, the control group received sensory stimulation at levels similar to those used in some of the studies as active treatment. Thus, stimulation of the control group ranged from the subsensory threshold level 29 to minimal perception, 32 to the subparesthesia level, 31 and to strong suprasensory stimulation. 30 Whereas most studies included only 1 control group receiving placebo treatment, in 1 study, the control group received no treatment. 33 In 1 study, control treatment was provided by stimulating the LE, whereas the intervention focus was actually on the UE function. 37 Additionally, 4 of the studies used 2 control groups, with one receiving placebo stimulation and the other no treatment.26,35-37
Assessment
Whereas the participants in all studies were assessed both pretreatment and posttreatment, follow-up assessment was conducted in only 6 of the studies,26,27,29,30,35,36 with follow-up time ranging from 24 hours 27 to 3 months. 30 A wide variety of outcome measures was used to examine treatment effect. In reference to the model of the International Classification of Functioning, none of the studies examined the effect on aspects of the person’s participation, whereas measures of body structure/function and activity were used in similar frequency.
The body structure/function most frequently addressed was spasticity, with 2 of the studies using EMG28,34 and 5 studies using a variety of clinical tools to assess it. 26 The next most frequently addressed body function was force, with 3 studies examining UE force30-32 and 3 examining LE force.34,35,39 The activity most often addressed was gait velocity.26,28,35,36,38,39 However, different studies measured velocity through a variety of activities, such as a 10-m walk, a 6-minute walk, or the Timed Up and Go test. The UE activity measure most commonly used was the Jebsen-Taylor Hand Function Test.27,29,30,37
Treatment Effect
A summary of treatment effects reported in the various studies is presented in Table 1 (last column). Effect sizes and 95% confidence interval of the primary outcome measures, and the pooled effects computed when data were available for more than 1 study are presented in Table 4 (online). The
Body structure and function domain
Only 1 study examined the effect of TENS on UE movement kinematics during reaching. Although the results showed a significant effect on wrist and finger velocity (
Pinch force was found to be favorably affected by TENS in 231,32 out of 3 studies. 30 However, the effect on pinch force could be pooled from only 2 of these studies,30,32 indicating no significant effect of TENS on this measure. Although the effect of TENS on LE force production was measured for both dorsiflexion and plantar flexion, significant effects were reported only for dorsiflexion. Pooling the results of 3 studies, the meta-analysis indicates a significant favorable effect of TENS on the dorsiflexion force.26,34,35 Finally, only 1 study used a global impairment scale (Brunnstrom Scale) to examine the effect of TENS on overall impairment level, indicating no significant effect. 38
Activity domain
The 10-m gait velocity test was the measure most frequently used to assess gait function.28,36,38,39 Whereas 3 of these studies indicated a significant improvement in the experimental group treated with sensory TENS, the effect size analysis and meta-analysis of 3 of these studies indicate no significant effect on this measure.28,36,38 Similar results were obtained regarding walking endurance, as measured by the 6-minute walk test 36 and regarding days to independent gait. 26 In contrast, the pooled effect on the Timed Up and Go test from 2 studies shows a significant result in favor of the experimental group.26,36
The Jebsen-Taylor Hand Function Test (JTHFT) was the most frequently used measure to assess UE function, with 4 studies using this measure.27,29,30,37 Whereas all studies reported a significant effect on the JTHFT, the data from 2 of these studies were not available to determine the effect size.29,37 The pooled effect of the 2 remaining studies, though not significant, indicates a positive tendency.27,30
The effect of TENS on finger and hand tapping frequency was evaluated in 1 study, with the results indicating a significant effect on finger tapping but no effect on hand tapping. 33 However, data were not available to determine the effect size. The Action Research Arm Test, which was used in 1 study to determine the effect on UE overall function, indicated no effect. 32 Finally, the effect of TENS on the general activity level was assessed in 3 studies.30,39,40 Different outcome measures were used in each study and thus could not be pooled. Although a significant treatment effect was reported for both the Barthel Index 40 and the Modified Motor Assessment Scale, 39 analysis indicates a significant effect size only for the Barthel Index. 40
Discussion
In all, 15 randomized or quasi-RCTs examining the effects of sensory TENS on motor recovery following a stroke were identified in the literature. All but 1 study, 38 which addressed LE function in the acute phase of rehabilitation reported statistically significant effects of sensory TENS on at least 1 aspect of motor performance (see Table 1). Yet because of the small number of participants per study and the large variability between them, the effect size analysis indicates a nonsignificant effect for most variables Moreover, the great variability between studies in terms of the outcome measures used as well as the insufficient data reported in some of the studies limited our ability to perform a meta-analysis. Thus, data for only 5 outcome measures could be pooled from 2 to 3 studies per outcome measure. The meta-analysis failed to show a favorable effect on gait velocity and pinch force. However, it indicates a significant effect of sensory stimulation to the LE on functional balance, as assessed by the Timed Up and Go test, and on force production of ankle dorsiflexion. Additionally, the pooling of data of only 2 of the 4 studies that demonstrated a significant effect on the JTHFT strongly suggests a positive impact of TENS on hand function as well.
The overall methodological quality of the retrieved studies, as evaluated using the PEDro classification, was good, with most items of the scale rated positively by the majority of the studies. However, some additional important features of the research methodology, which are not covered by the PEDro classification, also affect the ability to reach definitive conclusions regarding the effectiveness of sensory TENS. The 2 most important factors in this regard are the number of participants per study and the time span of the follow-up assessment. Whereas the number of participants per treatment group was between 8 and 51, the majority of studies, and in particular those with the more robust research methodology, involved a relatively small number of participants per treatment arm. Thus, the evidence level of most studies ranged between 2b and 3b. This resulted in insufficient power necessary to demonstrate effectiveness in at least some of the studies. Moreover, the small number of participants per study did not allow for their stratification by the severity of involvement or by lesion site—factors that have been shown to be significant determinants of the effectiveness of various treatments poststroke.41,42
Only 6 studies included a follow-up assessment that ranged from 24 hours 27 to 3 months.26,29,30,35,36 Whereas 5 of them indicate that the various gains achieved by sensory TENS were maintained in the follow-up assessment, Conforto et al 30 reported that the significant improvement noted in hand function (JTHFT) after 4 weeks of sensory stimulation was not maintained after the cessation of treatment. It is not possible from the data reviewed here to determine the cause of these differences in results because of the varying stages of recovery between the sample populations as well as the different stimulation protocols used in the studies.
Nevertheless, it seems that the use of active practice in conjunction with sensory stimulation is an important determining feature. In 1 study with a positive long-term effect, which used a crossover design and involved only a single session of sensory TENS, gains in JTHFT were maintained for 30 days only when motor training was combined with active stimulation (and not placebo treatment) in the second of the 2 sessions. 29 In other studies, Ng and Hui-Chan35,36 reported greater effects of sensory stimulation on gait velocity, the Timed Up and Go test, and ankle force production when active stimulation was combined with motor treatment as compared with using sensory TENS alone. However, given the small number of studies examining long-term effects and their contradictory results, further research is necessary to determine how best to facilitate the consolidation of the contribution of sensory stimulation to enduring cortical reorganization and to gains in motor performance.
Although all the reviewed studies used TENS to enhance peripheral sensory input, there was a high degree of variability between the studies in terms of the location of the electrodes, the duration of the treatments, and the stimulation characteristics. Thus, for example, whereas the majority of studies placed the electrodes over either specific muscles or nerves relevant to the impairment or activity assessed, others located them over acupuncture points.26,35,36 Generally, the application of electrodes over the peripheral nerves will induce widespread sensation along the nerve’s distribution and corresponding cortical activation. Yet some of the studies that placed the electrodes over the muscle belly used large electrodes (eg, a glove) that covered the entire distribution of more than 1 peripheral nerve. 39 As stimulation directly over the peripheral nerve or over the muscle belly generates very similar sensations, which are mediated by peripheral sensory nerve endings, a similar physiological mechanism may be expected for both techniques. Furthermore, studies examining the effect of pain modulation via sensory TENS have shown similar effects when TENS application was over the painful area versus over acupuncture points along the same peripheral nerve 43 or even over extrasegmental points (ie, remote acupuncture points). 44 Nevertheless, future studies are necessary to determine whether the physiological mechanism determining the effectiveness of sensory TENS on motor recovery is dependent on electrode location and size.
Using transcranial magnetic stimulation, it was recently demonstrated in nonimpaired individuals that 3 weeks of daily sensory TENS to the abductor pollicis brevis resulted in a global widespread effect on the motor map representations of the hand and forearm muscles that extended beyond the stimulated median nerve distribution. 11 This finding indicates that the effect of stimulation may extend beyond the specific stimulated muscle. However, future research is necessary to determine the optimal site of stimulation, particularly as rehabilitation goals usually involve the acquisition of functional abilities that require the coordinated activation of multiple muscle groups.
The differences in the stimulation intensity used in different studies and their contradictory effects are particularly striking. In a recent study conducted by Conforto et al, 30 which included patients in the subacute stage following a stroke, it was demonstrated that repeated sessions of TENS delivered at a subthreshold intensity over a period of 1 month enhanced hand function (as measured by the JTHFT), whereas higher intensity (ie, suprasensory) TENS did not affect performance. These results are in contrast to previous single-session studies of patients in both the acute 32 and chronic stages of recovery,27,33,37 which used similar stimulation parameters in terms of site of application as well as pulse frequency and duration. Those studies demonstrated better hand function when TENS was delivered at a higher stimulation intensity rather than at a subthreshold level. In fact, some studies used a subthreshold intensity as their control treatment. 34 At this stage, it is impossible to determine the factors responsible for these contradictory results and what role current intensity plays. Indeed, it remains to be seen whether optimal current intensity is determined by the number of stimulation sessions (ie, single session vs multiple sessions), by the amount of time elapsed since the insult, or by other confounding factors influencing responsiveness to different intensities.
In a series of experiments with healthy participants, it was demonstrated that reorganization of the human motor cortex in response to sensory TENS is dependent, among other factors, on the frequency of stimulation. Stimulation at 1 to 5 Hz was shown to increase cortical excitation, whereas stimulation at 10 to 40 Hz was found to decrease it. 45 A study of healthy participants also demonstrated that the pattern of pulse frequency may play a role in spinally mediated plasticity. Comparing the effect on reciprocal inhibition, it was shown that 10 pulses at 100 Hz delivered intermittently every 1.5 s was more effective than continuous stimulation delivered at a high frequency (approximately 360 Hz). 46 This may suggest that stimulation is more beneficial if delivered at a physiological-like rate rather than continuously.
In the present review, both types of frequency delivery were used. The majority of studies of the UE used 5 pulse trains delivered once a second,27,30-33,37 whereas the remaining studies used continuous stimulation delivered at frequencies ranging between 20 and 100 Hz. As most studies demonstrated some positive effects, it is impossible to draw any conclusions here about the relative contribution of stimulation frequency to enhanced motor recovery. Yet it is interesting to note that the only study that reported no advantage to sensory TENS used a completely different mode of stimulation delivery, with a continuous pulse frequency (at 35 Hz) that was delivered in a 10-s on, 10-s off duty cycle. 38 Whether this is a crucial factor determining the lack of observed effect in this study cannot be resolved on the basis of this review.
The present review indicates that sensory TENS may enhance motor recovery following a stroke, particularly when used in combination with active training. However, as the majority of the reviewed studies involved patients in the chronic stage of recovery, it has yet to be determined whether sensory TENS would be equally beneficial if used earlier during the rehabilitation process. It is also unclear as to what role location of the insult and severity of the deficits play in the responsiveness to sensory stimulation. Furthermore, given the great variability between studies in terms of stimulation parameters, further research is necessary to determine the effect of factors such as electrode location and stimulation parameters on long-term functional recovery of patients in both the acute and chronic stages of rehabilitation. Nevertheless, given the positive results reported by the majority of the reviewed studies, combined with the fact that TENS is relatively risk free and easy to implement, this modality when combined with active training seems to have potential as a noninvasive intervention that can enhance the benefits of customary rehabilitation treatments.
Footnotes
The author(s) declared no potential conflicts of interest with respect to the authorship and/or publication of this article.
The author(s) received no financial support for the research and/or authorship of this article.
References
Supplementary Material
Please find the following supplemental material available below.
For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.
For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.
