Original article

Changes in brain activation patterns after physiotherapy program:A preliminary randomized controlled trial study after Postural

Reconstruction1 and stretching programs

Changements dans les patterns d’activation cérébraux : une étude préliminaire randomiséecontrôlée après des programmes physiothérapiques de Reconstruction Posturale1 et de stretching

M. Nisand a,*, C. Callens b, V. Noblet c, N. Gaudreault d, P. Vautravers e,M.-E. Isner-Horobeti e, I.J. Namer f

a Institute of Postural Reconstruction1, 14, rue Wimpheling, 67000 Strasbourg, Franceb Institute of Physiotherapy, 67000 Strasbourg, France

c University of Strasbourg, CNRS, ICube, FMTS, 67037 Strasbourg, Franced École de réadaptation, université de Sherbrooke, Sherbrooke, Canada

e Institute of Rehabilitation, IUR Clémenceau, Strasbourg University, 45, boulevard Clémenceau, 67000 Strasbourg, Francef Department of Biophysics and Nuclear Medicine, University Hospitals of Strasbourg, 67200 Strasbourg, France

Received 13 April 2015; accepted 9 September 2015

Available online 24 October 2015

Abstract

Postural Reconstruction1 is a physiotherapy method that has been developed over the last two decades. Its main therapeutic objectives are todecrease pain levels, normalize joint alignment and posture and improve movement and function. The goal of this study was to substantiate theneuromuscular mechanism of action of Postural Reconstruction1 by evidencing pre- vs. post-intervention changes in brain activation patternsduring sustained ankle dorsiflexion. It concerns a single-centre, prospective, randomized, controlled, parallel-group trial. Sixteen healthy subjects(8 males and 8 females; age range: 20–23) were randomized into two groups: an interventional Postural Reconstruction1 group (n = 8) and acontrol stretching group (n = 8). Both groups performed 10 weekly sessions. The Postural Reconstruction1 sessions involved five manoeuvres andthe stretching sessions involved five different exercises. Brain activation patterns were measured via single-photon emission computedtomography. Each subject received two 1480 MBq doses (3 months apart). We performed a voxel-wise statistical analysis (using StatisticalParametric Mapping software) to detect changes in brain activation within each group.Results. – We observed statistically significant pre- vs. post-intervention changes in brain activation patterns in the Postural Reconstruction1

group (a neuromuscular approach), but also in the stretching group (viewed as a mechanical approach). There were no significant intergroupdifferences in the pre- or post-intervention brain activation patterns.Conclusions. – Our results suggest that the two different physiotherapy programmes have a neuromuscular mechanism of action and evidencedchanges in brain activation patterns in young, healthy adults (i.e. free of CNS lesions) during the performance of ankle movements.# 2015 Elsevier Masson SAS. All rights reserved.

Keywords: Neuroimaging; Neuroplasticity; Posture; Rehabilitation

Résumé

La Reconstruction Posturale1 est une physiothérapie qui a été développée au cours des deux dernières décennies. Ses objectifs sont la résolutiondes algies, la réaxation des segments, la normalisation de la posture, l’amélioration de la mobilité et de la fonction. Le but de cette étude était d’étayer lemode d’action neuromusculaire de la Reconstruction Posturale1 en mettant en évidence des modifications pré- vs post-intervention dans les patterns

Available online at

ScienceDirectwww.sciencedirect.com

Médecine Nucléaire 39 (2015) 502–513

* Corresponding author.E-mail address: m.nisand@free.fr (M. Nisand).

http://dx.doi.org/10.1016/j.mednuc.2015.09.0040928-1258/# 2015 Elsevier Masson SAS. All rights reserved.

d’activation cérébraux lors d’une dorsiflexion de la cheville. Il s’agit d’une étude prospective, monocentrique, randomisée contrôlée en deux groupesparallèles de sujets sains, 8 hommes et 8 femmes, (extrêmes : 20–23 ans) : un groupe expérimental de Reconstruction Posturale1 (n = 8) et un groupetémoin de stretching (n = 8). Dix sessions à un rythme hebdomadaire et 5 techniques pour chaque groupe. L’activation cérébrale a été mesurée partomographie d’émission monophotonique. Il a été administré à chaque sujet 2 � 1480 MBq à 3 mois d’intervalle. Nous avons effectué des analysesstatistiques en chaque voxel grâce au logiciel SPM pour détecter les changements d’activation cérébrale au sein de chaque groupe.Résultats. – Des changements statistiquement significatifs pré- vs post-intervention dans les patterns d’activation cérébrale ont été observés dansle groupe Reconstruction Posturale1 (approche neuromusculaire), mais aussi dans le groupe stretching (considéré comme une approchemécanique). Il n’y avait pas de différence significative entre les groupes.Discussion. – Nos résultats suggèrent que les deux programmes de physiothérapie ont un mode d’action neuromusculaire. Les changements ontété mis en évidence chez des jeunes adultes sains (non cérébro-lésés) lors d’un mouvement de la cheville.# 2015 Elsevier Masson SAS. Tous droits réservés.

Mots clés : Neuro-imagerie ; Neuroplasticité ; Posture ; Réhabilitation

M. Nisand et al. / Médecine Nucléaire 39 (2015) 502–513 503

1. Introduction

Postural Reconstruction1 (PR) is an innovative neuromus-cular physiotherapy paradigm that has been developed over thelast two decades. It is based on the hypothesis wherebyexcessive residual tension in large sets of overlapping musclesinvolved in human stance may change the joint moment andthus contribute to movement pattern disorders [1]. This can beviewed as analogous to building tension in the string of a bow;tension in the string causes bending and twisting of the bow (i.e.the spine and limbs).

Residual muscle tension (i.e. muscle tone) is not undervoluntary control. Hence, PR uses manoeuvres based onscientific studies of neurorehabilitation [2,3]. These manoeuvresimplement a specific therapeutic tool (referred to as normalizinginduction) to first increase the already excessive residual muscletension and then (in the same treatment session) fatigue it, with aview to normalizing spinal and/or peripheral joint alignmentdisorders [4,5]. Concomitant effects on pain, movementperformance and function are expected [6,7].

The normalizing induction uses an irradiation process that issimilar to motor overflow, i.e. a voluntary movement thattriggers distant, involuntary contractions of contralateralhomologous muscles or ipsilateral or contralateral non-homologous muscles [8,9]. In everyday implementations ofPR, the voluntary inducing movement is a complex motor taskthat is sustained for a relatively long period of time (10 to 30min) and requires full concentration by the subject. For a giveninducing motor task, the distant reaction is almost always thesame in a given subject but may vary from one subject toanother and may also affect the axial muscles.

The distant reactions are not related to central nervoussystem (CNS) damage or psychiatric disorders because theycan be triggered in healthy subjects [10]. The distant reactionscharacterized in the literature appear to have a central trigger(via cortical activation of motor overflow). However, asubcortical origin cannot yet be ruled out [11].

The progressive decrease in involuntary contractionsinduced by PR treatment suggests that the process involvesbrain plasticity. Hence, we hypothesized that PR’s mode ofaction might be related to changes in the spatial extent of brainactivation.

The goal of the present study was to investigate the effect ofPR on brain activation patterns during the performance of asimple, predefined motor task (sustained, active dorsiflexion ofthe right ankle) recorded with single-photon emissioncomputed tomography (SPECT). Indeed, SPECT has alreadybeen used to image functional brain activation patterns duringmovement, language and memory tasks [12–15].

To avoid methodological bias and the possible misinter-pretation of results, we sought to compare the putative changesin brain activation associated with PR (a neuromuscularapproach) with those associated with stretching (viewed as amechanical approach). According to many researchers [16–18],the effect of stretching is more likely to be due to theapplication of tensile stress to non-contractile, connectivetissues in and around the muscle than to inhibition of themuscle’s contractile element.

We therefore decided to perform a pilot study in young,healthy adults (i.e. free of CNS lesions).

2. Experimental procedure

We performed a single-centre, prospective, randomized(allocation ratio: 1:1), controlled, parallel-group, proof-of-concept study.

2.1. Participants

The study population comprised physiotherapy studentsfrom the Institut de Formation en Masso-Kinésithérapie(France). The inclusion criteria were as follows: age between18 and 35 (inclusive), negative pregnancy test (HCG Stat, fromRoche Diagnostics, Meylan, France, who has a sensitivity of1 mU/ml and able to detect a pregnancy as early as 10 days afterfertilization), command of the French language and valid socialsecurity coverage.

The main exclusion criteria were as follows: a history ofneurological or psychiatric impairments, recent musculoskeletaltrauma, contraindications to neuroimaging (such as the presenceof a pacemaker, ferromagnetic surgical clips, ocular implants,intraocular/neural metallic foreign bodies or hypersensitivity

M. Nisand et al. / Médecine Nucléaire 39 (2015) 502–513504

to the radiotracer or one of its excipients), breastfeeding andparticipation in another clinical trial.

Twenty-three eligible students volunteered to participate.Sixteen of the latter were chosen at random (using a blockrandomization design, an allocation ratio of 1:1 and the randomnumber generator in R software) and constituted the studypopulation (8 males and 8 females).

Given that this was a pilot study, the lack of literature data onthe effect of intervention on our primary efficacy variableprevented us from calculating a precise sample size. Hence, thesample size was determined as a compromise between studyfeasibility, physiological variability and financial imperatives.However, the results of the present study should facilitatecalculation of the sample size required in future trials. Eachsubject served as his/her own control.

2.2. Group allocation

To avoid methodological bias and the possible misinter-pretation of results, we decided to randomize the 16 participantsto an interventional group (PR) and a control group (stretchingsessions). Hence, in a second randomization, 4 men and 4women were attributed to each of the two groups in a blockrandomization design (allocation ratio: 1:1) by using therandom number generator in R software. The principalinvestigator enrolled participants and assigned them tointerventions in the Service de Biophysique et de MédecineNucléaire (University Hospital, France). The study’s objectiveand procedures were approved by an independent ethicscommittee and complied with the tenets of the Declaration ofHelsinki (1964). All subjects provided their written, informedconsent before participation in any study procedures.

2.3. Data collection

For each subject, the study was performed in three phases.Firstly, a clinical examination provided data on weight, height,blood pressure and heart rate. At baseline, SPECT wasperformed at rest and during activation (i.e. during sustainedright ankle dorsiflexion). Secondly, the subjects participated intheir allotted weekly physical therapy sessions over a 10-weekperiod. Thirdly, a post-intervention SPECT was performed twoweeks after the last physical therapy session. The overall studyperiod (from first inclusion to the end of data collection for theprimary outcome criterion) was 6 months.

2.4. The primary outcome criterion

The primary outcome criterion was defined as a change inthe brain activation pattern observed with SPECT.

The SPECT examinations were performed by an experien-ced radiologist and in accordance with the nuclear medicinedepartment’s standard clinical protocol. The ‘‘at rest’’ and‘‘during activation’’ SPECT measurements were performed inthat order at both the pre- and post-intervention time points.

There were no adverse or unexpected events in either group.

2.4.1. Measurement at rest (the baseline measurement)The first ‘‘at rest’’ measurement enabled us to quantify

changes in the brain activation patterns attributable to the motortask by comparison with post-intervention brain images.Fifteen mCi (555 MBq) of ethyl cysteinate dimer (ECM)-Tc99m (Neurolite1, Bristol-Myers Squibb Medical Imaging,Rueil-Malmaison, France) was injected intravenously while theparticipant was lying still in the supine position. The tracerupdate time was 90 seconds. Twenty minutes later a dual-headgamma camera (E-CAM, Siemens Healthcare, Erlangen,Germany) was used to image the brain. The acquisitionparameters were as follows: collimator setting: fan-beam; totalturnover: 2 � 1808; number of projections: 32; duration of eachprojection: 50 seconds; acquisition mode: stepper; zoom factor:1; acquisition matrix: 1282; acquisition time: 27 min.

2.4.2. Measurement during activationWhile in the supine position, the participant actively

dorsiflexed his/her right ankle to the fullest extent possible.The following instructions were given: ‘‘Flex your right foottowards your shin as far as possible and hold that position. Do notmove your left foot. The rest of your body must be relaxed’’. Next,25 mCi (925 MBq) of ECM-Tc99m was injected 15 seconds afterthe start of the motor task. The ankle dorsiflexion had to bemaintained for 90 seconds, to allow time for tracer uptake andbinding. Electromyographic recording was used to control for thepresence of involuntary co-contractions, which otherwise mighthave resulted in the activation of brain areas not involved in ankledorsiflexion. A keypoint stimulator (Medtronic-Dantec Medical,Royal Portbury, UK), dispersive electrodes (18 cm in length and20 mm in width-Medtronic, catalogue number 9013S07231) andfour surface recording electrodes (15 cm in length and 20 mm inwidth-Medtronic, catalogue number 901350211) were used forEMG recording. The electrodes were placed on the centre of thequadriceps and on the centres of the paravertebral muscles oneither side of the third lumbar vertebrae. A camcorder with a timerwas used to film the measurement session.

After the 20 min required for elimination of the tracer fromthe salivary glands, the gamma camera acquired a second set ofimages during a 27-minute period. The same measurementprotocol was used post-intervention for measuring brainactivation patterns at rest and during sustained ankle dorsiflexion.The volume of brain areas activated in ankle dorsiflexion (thesensorimotor cortex, the premotor cortex, the supplementarymotor area (SMA), the basal ganglia and the cerebellum) wasquantified and assessed in a statistical analysis.

In total, each subject received two 1480 MBq doses (3months apart). These doses were approved by the FrenchNational Agency for Healthcare Product Safety (Paris, France).

2.5. The physical therapy interventions

2.5.1. The interventionnal programme: PosturalReconstruction1

The PR group participated in 10 weekly, 30-minuteindividual sessions (all led by the same practitioner). In thePR paradigm, the treatment tool is referred to as ‘‘normalizing

M. Nisand et al. / Médecine Nucléaire 39 (2015) 502–513 505

induction’’, i.e. an irradiating, voluntary movement designed toinduce involuntary contractions (referred to as ‘‘evokedresponses’’) [4,5]. Inductions are complex motor tasksperformed with the greatest available range of motion. Notall full-range movements provoke a measurable evokedresponse; hence, some inductions must include certain combi-nations of movements (e.g. shoulder abduction and medialrotation) or avoid them (e.g. the dissociation of movement of thelittle toe and foot abduction). In some cases, the evoked responsewill occur if a conflicting factor (such as stiffness of the opposingmuscles) interferes with the normal range of movement or themovement’s trajectory. The inductions are selected aftermorphological examination of the patient in all three spatialplanes [19]. Clinically, the evoked responses translate intotransient aggravation of the pre-existing, localized, morpholo-gical impairment. The induction is maintained until there is adecrease in or disappearance of the evoked response. In clinicalterms, this decrease corresponds to a reduction of themorphological feature that was previously aggravated. Theinductions are performed with a specific breathing pattern and afocus on exhalation. Five inductions were used during the PRsessions. Two are described here (Fig. 1):

Fig. 1. a: dorsiflexion of the toes of the left foot against resistance, while in the sittingthe figure) on the submental muscles, the right and left paravertebral muscles oveabduction combined with internal rotation. Motor irradiation is enhanced by combinarm (with the greatest possible range of motion). The electrodes were placed (from

muscles and the right and left paraspinal muscles over T12-L1.a : dorsiflexion des orteils du pied gauche contre résistance en position assise. Les élles muscles paravertébraux droits et gauches en regard de C4 et le chef claviculaire dL’irradiation motrice est provoquée par la combinaison de l’abduction et de la roélectrodes ont été placées (du haut vers le bas) sur les muscles rhomboïdes droit

� a combination of shoulder abduction and internal rotation[20,21] and;� dorsiflexion of the toes of the left foot against resistance while

in the sitting position [22]. It is important to note that neitherof the two manoeuvres required right ankle dorsiflexion.

Dorsiflexion of the toes of the left foot against resistancewhile in the sitting position causes an involuntary, uncontrol-lable forward movement of the head. Electromyograms(EMGs) recorded in preliminary experiments revealed moder-ate activation of the right C4 paravertebral muscles and intenseactivation of the ipsilateral sternocleidomastoid muscle.

The combination of shoulder abduction with internalrotation causes uncontrollable aggravation of the ipsilateralthoracic curve, which is through to reflect accentuation ofspinal convexity. In preliminary experiments, the EMGsevidenced involuntary contractions of the contralateral para-spinal muscles (T12-L1).

2.5.2. The control programme: stretchingThe stretching group participated in 10 weekly, 30-minute

group sessions (all led by the same practitioner). In each

position. The electrodes were placed (from the top to the bottom of the screen inr C4 and the clavicular head of the sternocleidomastoid muscle; b: shouldering active, assisted, medial rotation of the shoulder with active abduction of thethe top to the bottom of the screen in the figure) on the right and left rhomboid

ectrodes ont été placées (du haut vers le bas) sur les muscles sous-mentonniers,u muscle sterno-cléïdo-mastoïdien ; b : abduction du bras avec rotation médiale.tation médiale du bras, réalisée dans la plus grande amplitude possible. Leset gauche et sur les paravertébraux en regard de T12-L1.

M. Nisand et al. / Médecine Nucléaire 39 (2015) 502–513506

session, each stretching posture was performed five timesalternately on each body side, in line with Taylor’s recommen-dations [18,23]. The stretching postures were intended tostretch the hamstrings, the triceps surae, the adductors, theiliopsoas and, lastly, the posterior muscles of the trunk and legs(Fig. 2). Passive stretching was chosen for the control groupbecause of its action on the elastic component of musclestructure [24,25]. The stretch was performed slowly, in order toavoid activation of the myotatic (stretch) reflex. Each stretchwas held for 10 seconds [26,27]. There was a 2-minute restperiod between each stretch.

2.6. Image processing

To perform a voxel-wise statistical analysis, images have tobe registered in a common space. To this end, all four imagesacquired for each patient were firstly rigidly aligned and thenaveraged. This average image was used to estimate an affinetransformation and thus map all four individual images onto theSPECT template provided in the Statistical ParametricMapping package (SPM – Wellcome Department of ImagingNeuroscience, London, UK; http://www.fil.ion.ucl.ac.uk/spm).Lastly, the intensities of the warped images were normalized(by estimating the scaling factor that minimized the sum ofsquared intensity differences with the SPECT template) andthen smoothed with a Gaussian kernel (full width at half-maximum: 12 mm). All processing steps were performed with

Fig. 2. The five passive stretching postures, for stretching the hamstrings (a), the trictrunk and lower limbs (e).Les 5 techniques d’étirement passif des muscles ischio-jambiers (a), des triceps suraudes membres inférieurs (e).

Medipy software (University, France; https://piiv.u-strasbg.fr/traitement-images/medipy).

2.7. Statistical analysis

All statistical analyses were performed using the SPM8software package (running on Matlab 7.10; MathWorks,Natick, MA, USA). Descriptive statistics were used tocharacterize the group samples. In activation studies, weconsidered the difference in intensity between images recordedduring activation and images recorded at rest; this differentialimage is referred to as the activation map. The pre-interventionbrain activation patterns (but not post-intervention activationpatterns) were obtained by performing a one-sample t-test on theactivation maps of all individuals (i.e. the two groups’ maps werepooled). Pre-intervention vs. post-intervention differences withineach group were obtained by performing a one-sample t-test onthe differences between the pre-intervention activation map andthe post-intervention activation map for each individual. Lastly,intergroup comparisons of pre- vs. post-intervention differencesin the activation maps were performed by applying a two-samplet-test. All statistical tests were performed in the absence of anycovariates, although global normalisation with SPM’s analysis ofcovariance routine was used to account for changes in intensityacross subjects. With a cluster spatial extent of 5 voxels, thethreshold for statistical significance of P (uncorrected) was setto < 0.001 for the analysis of pre-intervention brain activation

eps surae (b), the adductors (c), the iliopsoas (d) and the posterior muscles of the

x (b), des adducteurs (c), des iliopsoas (d) et des muscles postérieurs du tronc et

M. Nisand et al. / Médecine Nucléaire 39 (2015) 502–513 507

patterns and < 0.005 for the analysis of pre- vs. post-interventiondifferences in brain activation patterns.

This threshold enables the deletion of irrelevant, isolatedclusters with fewer than 5 voxels and thus provides a cleanermap. The choice of a value of 5 is somewhat arbitrary (we couldhave chosen 10) but this is the default setting in xjView (thesoftware we used to visualize the results).

Statistical maps were analysed with xjView software (http://www.alivelearn.net/xjview8/), which enabled us to identify thebrain regions associated with the detected clusters.

Only the statistician was blinded to the group assignments.He received two datasets but did not know which datasetcorresponded to which group.

3. Results

Fig. 3 shows the disposition of the study population:enrolment, allocation, follow-up and analysis.

The study participants’ individual characteristics aresummarized in Table 1. The data were not distributed normallyand so are reported as the median [1st–3rd interquartile range].There were no significant differences between the two studygroups in terms of age (range: 20–23), weight, height or restingheart rate (P > 0.05 in Mann-Whitney U tests).

Fig. 3. Disposition of the study population.Représentation schématique de la population de l’étude.

Electromyographic recordings of the quadriceps andparaspinal muscles and video recording confirmed the absenceof involuntary movements during the SPECT acquisitions.

3.1. Pre-intervention brain activation patterns duringsustained right ankle dorsiflexion

Certain brain areas were activated before any intervention(n = 16; one-sample t-tests without any covariates; statisticalthreshold, P < 0.001 (uncorrected); cluster size > 5 voxels).They were variously located in the left hemisphere (theposterior limb of the left internal capsule, the lentiform nucleusand the thalamus; the left temporo-occipital junction; the leftsub-gyral frontal lobe and the cingulate gyrus; the left superiorfrontal gyrus (Brodmann area (BA) 6 and the SMA) and theprecentral gyrus) and the right hemisphere (the right postcentralgyrus and the parietal inferior lobe) (Fig. 4).

3.2. Pre- vs. post-intervention differences in the PosturalReconstruction1 group

Differences were observed in the left hemisphere, the righthemisphere and the cerebellum (n = 8; one-sample t-testswithout any covariates; statistical threshold, P < 0.005

Table 1The clinical characteristics of the subjects in each group.Caracteristiques cliniques des sujets dans chaque groupe.

Patient number Gender(M/F)

Age(years)

Weight(kg)

Height(cm)

Systolic/diastolicblood pressure (mmHg)

Heart rate(beats per minute)

PR group

006 M 22 70.0 184 134/85 91001 M 21 75.3 185 142/85 55007 M 23 68.7 173 117/63 77017 M 21 71.0 190 130/85 85009 F 21 63.7 167 104/80 80011 F 21 56.0 170 120/80 72015 F 21 52.0 162 109/80 93014 F 21 67.0 169 114/78 75

Stretching group

005 M 21 66.1 175 102/69 77004 M 20 67.1 165 121/75 81002 M 22 88.4 184 136/72 52008 M 22 78.5 187 133/66 64010 F 21 50.1 158 121/80 87013 F 21 56.0 151 119/89 80012 F 20 58.6 165 110/77 54016 F 21 58.0 169 116/80 98

M. Nisand et al. / Médecine Nucléaire 39 (2015) 502–513508

uncorrected; cluster size > 5 voxels): the left lingual andinferior occipital gyrus; the left inferior temporal gyrus andfusiform gyrus; the left middle temporal gyrus; the left anteriorcingulate and orbitofrontal gyrus; the left superior temporalgyrus; the left parahippocampal gyrus; the left inferior frontalgyrus; the left lentiform nucleus, internal capsule and thalamus;the left inferior parietal lobule and supramarginal gyrus; the leftprecuneus; the left inferior parietal lobule, Brodmann’s area(BAs) 40 and 7; the left precentral gyrus = 33 voxels (peakMontreal Neurological Institute (MNI) coordinate: 30 �6 48; t-statistics peak intensity: 5.07); the left precentral gyrus = 216voxels (peak MNI coordinate: 34 �28 60; t-statistics peakintensity: 5.03); the left SMA; the right superior temporalgyrus; the inferior orbitofrontal gyrus; the right inferioroccipital gyrus; the right caudate and lentiform nucleus; theright inferior parietal lobule and supramarginal gyrus; the rightsuperior and middle frontal gyri; the right cingulate gyrus andprecuneus = 472 voxels; the left cerebellum = 27 voxels (peakMNI coordinate: 54 �52 �42; t-statistics peak intensity: 4.5);the left cerebellum = 47 voxels (peak MNI coordinate: 14 �64�32; t-statistics peak intensity: 4.37); the left cerebellum andvermis; the right cerebellum = 235 voxels (peak MNI coor-dinate: �28 �38 �32; t-statistics peak intensity: 11.69); theright cerebellum = 7 voxels (peak MNI coordinate: �2 �52�40; t-statistics peak intensity: 4.07); the right cerebel-lum = 198 voxels (peak MNI coordinate: �20 �96 �24;t-statistics peak intensity: 5.98) (Fig. 5).

3.3. Pre- vs. post-intervention differences in the stretchinggroup

Differences were observed in the left hemisphere, the righthemisphere and the cerebellum (n = 8; one-sample t-testswithout any covariates; statistical threshold, P < 0.005

uncorrected; cluster size > 5 voxels): the left lingual andinferior occipital gyrus; the left lentiform nucleus; the leftmiddle temporal gyrus; the left temporo-occipital junction; theleft middle frontal gyrus; the left inferior frontal gyrus; the leftmiddle frontal gyrus; the left middle frontal gyrus, BA 6 andprecentral gyrus; the left thalamus; the left precuneus; the rightmiddle temporal gyrus; the right superior temporal gyrus; theright medial part of superior frontal gyrus; the right medialfrontal gyrus; the right superior frontal gyrus; the right superiorand middle frontal gyri, BA 6; the right SMA; the rightprecentral gyrus and opercula; the right precentral gyrus; theleft pre- and postcentral gyri; the right inferior parietal lobule;the right postcentral gyrus; the right caudate nucleus; thevermis; the left cerebellum; the right cerebellum = 24 voxels(peak MNI coordinate: �38 �60 �48; t-statistics peakintensity: 5.39); the right cerebellum = 20 voxels (peak MNIcoordinate: �28 �54 �42; t-statistics peak intensity: 9.07)(Fig. 6).

4. Discusssion

4.1. Pre-intervention brain activation patterns

The results for brain area activation during active ankledorsiflexion movement prior to the intervention agree with theliterature data [28,29]. To the best of our knowledge, only onestudy has explored brain area activation during sustained ankledorsiflexion [30]. The activated areas included the bilateralparacentral lobule and the precuneus (which are mainly locatedin the left hemisphere), BAs 1, 2, 4 and 7, the bilateral lingualgyrus (which is mainly located in the left hemisphere), BAs 17and 18, the right cuneus, BAs 18 and 19, the right inferioroccipital gyrus (adjacent to the lateral occipitotemporal gyrusand lingual gyrus) and the vermis.

Fig. 4. Brain activation during right ankle dorsiflexion before the intervention (n = 16; one-sample t-tests without any covariates; statistical threshold,P < 0.001 uncorrected; cluster size > 5 voxels).Activation cérébrale lors de la dorsiflexion de la cheville droite avant l’intervention (n = 16 ; t-test pour échantillon unique sans covariable ; seuil statistique,p < 0,001 non corrigée ; taille de cluster > 5 voxels).

M. Nisand et al. / Médecine Nucléaire 39 (2015) 502–513 509

4.2. Post-intervention brain activation patterns

In our study, changes in brain activation patterns were seenin both the PR group and the stretching group. One canspeculate that the interventions had effects on the afferent inputand then the efferent motor response when performing the task.Taken as a whole, our results raise questions about thephysiological mechanisms underlying these two types ofphysiotherapeutic intervention. Both the PR group and thecontrol group displayed significant pre- vs. post-interventiondifferences in activated brain areas. However, there were nosignificant intergroup differences in brain activation patternseither before or after the intervention.

To investigate the reproducibility of brain perfusion SPECTimaging over time, we also compared the pre- vs post-intervention baseline scans within each group and thecorresponding statistical map revealed no significant change

at the threshold P < 0.001, which is in accordance with severalalready published studies [31–33].

It is important to note that a lasting localized change inactivation patterns for a specific movement was obtained byphysical therapy interventions which did not involve the saidmovement at all. This experimental design was chosen in orderto rule out bias due to motor learning [34].

To the best of our knowledge, the present study is the first tohave recorded changes in brain activation patterns during anklemovement following the performance of rehabilitation sessionsin healthy subjects (without any practice or training in theexperimental motor task). Katiuscia Sacco et al. [35] used anfMRI motor imaging paradigm to evidence pre- vs. post-training differences in functional connectivity during ankledorsiflexion in normal subjects [35]. However, these changeswere observed after a week-long period of locomotor attentiontraining.

Fig. 5. Pre- vs. post-intervention differences in the Postural Reconstruction1 group: differences were observed in the left hemisphere, the right hemisphere and thecerebellum (n = 8; one-sample t-tests without any covariates; statistical threshold, P < 0.005 uncorrected; cluster size > 5 voxels).Différences pré- vs post-intervention dans le groupe Reconstruction Posturale1 : des différences ont été observées dans l’hémisphère gauche, l’hémisphère droit et lecervelet (n = 8 ; t-test pour échantillon unique sans covariable ; seuil statistique, p < 0,005 non corrigée ; taille de cluster > 5 voxels).

M. Nisand et al. / Médecine Nucléaire 39 (2015) 502–513510

We were not surprised to observe changes in brain activationpatterns following PR because this technique is based on aneuromuscular paradigm [36,37].

Our results indicate that changes in activation patternsoccurred in the stretching group. In retrospective, this is not sosurprising [38–40]. One can hypothesize that stretchingmodifies afferent pathways – most probably by increasingthe discomfort threshold rather than changing viscoelasticproperties per se [24] and thus the efferent motor response. Ifintervention can modify afferent inputs from the mechanore-ceptors in both contractile and non-contractile components ofmuscle, this would explain why brain activation patterns can bere-organized by stretching.

Our findings may have some important implications forrehabilitation in practice. It is generally accepted that stretchinghas only moderate effects on muscle length [24], musclestiffness [24], the risk of sports injuries [41], and delayed-onset

muscle soreness [42]. Nevertheless, with the above-mentionedtherapeutic objectives in mind, many therapists and coachesremain committed to stretching exercises. This conservatismsuggests that patients and athletes do gain some sort of benefit.Could this benefit be based on a physiological response relatedto an induced change in brain activation patterns? Similarly, arethere links between the normalization of joint alignment,posture restoration, the improved function observed in PR andthe activation pattern modification evidenced here?

4.3. Study limitations

The study design must be considered in the light of PR’spreliminary state of advancement, since researchers are still inthe early stages of using valid, reliable instruments to documentthe method’s putative effects.

Fig. 6. Pre- vs. post-intervention differences in the stretching group. Differences were observed in the left hemisphere, the right hemisphere and the cerebellum(n = 8; one-sample t-tests without any covariates; statistical threshold, P < 0.005 uncorrected; cluster size > 5 voxels).Différences pré- vs post-intervention dans le groupe stretching. Des différences ont été observées dans l’hémisphère gauche, l’hémisphère droit et le cervelet (n = 8 ;t-test pour échantillon unique sans covariable ; seuil statistique, p < 0,005 non corrigée ; taille de cluster > 5 voxels).

M. Nisand et al. / Médecine Nucléaire 39 (2015) 502–513 511

In PR, a trained practitioner has to choose and perform themost appropriate induction manoeuvres. PR is thereforeincompatible with the self-care principle that is oftenencouraged in physical therapy. In contrast, stretching enablesself-care. Hence, in an attempt to reduce potentiallyconfounding procedural differences between the two inter-ventions, the stretching sessions were led by a practitioner.

Given that this was a preliminary study, only healthysubjects were assessed. This was made possible by the fact thatPR can also be applied to asymptomatic individuals. This is alsotrue for stretching, which explains in part why we choose thattechnique for the control group. The study sample wascomposed of young adult volunteers (range: 20–23). In view ofthis narrow age range, our results cannot be extrapolated topopulations of healthy older adults or populations ofsymptomatic individuals.

Lastly, it is possible that the relatively short treatment period(10 weeks) was not long enough to produce statisticallysignificant differences in activation patterns when comparingthe two interventions. The absence of a statistically significantdifference between PR and stretching might also be due to thesmall sample size.

This explains why assessing pre- vs. post-interventionchanges within each group is of value and perhaps why we didnot observe a significant intergroup difference.

We tried to show the brain’s plasticity under the influence of aphysiotherapy technique, rather than a ‘‘snap-shot’’ of ankledorsiflexion. The two-week interval after the last physiotherapysession enabled the persistence of effects to be assessed. The riskof methodological bias due to this time interval is very low becauseif a specific activity had biased the results, it would have had tobe performed by all 16 subjects during the two-week period.

M. Nisand et al. / Médecine Nucléaire 39 (2015) 502–513512

The ‘‘evoked’’ motor task studied here (ankle dorsiflexion)was relatively simple, when compared with the normalizinginduction motor tasks applied in everyday PR practice. Eventhough the amplitudes of the inducing movements in PR arephysiologically normal, these movements are never sponta-neously performed in everyday life. In the present study, theinducing motor task did not involve the right ankle. The choiceof the right ankle was arbitrary and did not take account of theparticipant’s handedness. We chose ankle dorsiflexion overplantar flexion because the former movement (but not the latter)puts tension on the posterior muscle system. Hence, in mostcases, the range of movement in plantar flexion is greater thanthat of dorsiflexion. Plantar flexion is also easier to perform. Inour everyday practice, we find that dorsiflexion is more difficultfor patients because it is restricted by posterior muscle tension.Hence, dorsiflexion is more likely to be influenced by a shortcourse of PR and any changes in activation patterns should bemore visible. In the literature, neuroimaging studies ofactivation patterns for ankle plantarflexion-dorsiflexion havegenerally involved dynamic and often repetitive movements[43–46]. However, the study cited above [30] (which used echoplanar MRI to identify the brain areas activated duringsustained ankle active dorsiflexion) also analyzed ankleplantarflexion. The results showed that although ankleplantarflexion excited fewer cortical areas than ankle dorsifle-xion did, it excited more subcortical areas. In view of theserecent data, it might have been better to explore plantarflexionbecause the latter appears to involve subcortical structures to agreater extent.

The present study constituted a pilot neurological investiga-tion of upright, bipedal stance. Only contractions of the ankle,dorsiflexor and plantar flexor muscles are needed to maintainthe body’s centre of gravity within the support base [47].These muscles are directly involved in postural activity.Furthermore, the dorsiflexors are also involved in the gaitcycle (in heel-on and throughout the swing phase). Researchhas shown that ankle dorsiflexion produces much the samebrain activation pattern as gait does [48–50]. The physio-therapy exercises performed in the present study changed theactivation patterns for ankle dorsiflexion, even when themovement was not performed. Accordingly, one can hypo-thesize that upright stance and gait might also be affected byPR and stretching.

In recent years, the outcomes obtained by certain rehabilita-tion interventions have been explained by brain plasticity [51–

54]. However, most of the disorders studied and interventionsused have been related to the field of neurorehabilitation. In thepresent study, we demonstrated that musculoskeletal rehabilita-tion procedures (PR and stretching) are associated with changesin brain activation patterns. This is a noteworthy finding formusculoskeletal rehabilitation, which supposedly involves softtissue plasticity.

To the best of our knowledge, the present study is the first tohave recorded changes in brain activation patterns during anklemovement following the performance of rehabilitation sessionsin healthy subjects (without any practice or training in theexperimental motor task).

Future studies of the effects of these interventions on brainplasticity should be performed with larger samples, older adultsand longer treatment periods. Given that the results of thispreliminary study concerned asymptomatic participants, fur-ther work should assess patients with muscle and jointdisorders. Investigation of the effect on PR on brain activationpatterns while standing is currently being considered. Our studyresults not only argue in favour of the suggested neuromuscularmechanism of action of PR but also raise questions about themechanism of action of other kinds of rehabilitation.

Disclosure of interest

The authors declare that they have no competing interest.

Acknowledgments

We are grateful to Professors Christine Tranchant andChristian Marescaux (Faculty of Medicine, University ofStrasbourg) for assistance with the clinical trial and criticalreview of study proposal, respectively.

Funding: The study was funded by Strasbourg UniversityHospital (material support), the University of Strasbourg,France (financial support) and the Physiotherapie SchuleOrtenau, Germany (financial support).

References

[1] Sahrmann SA. Diagnosis and treatment of movement impairment syn-dromes. St. Louis: MO Mosby; 2002 [pp. 1–3, 42].

[2] Bobath B. Motor development, its effect on general development, andapplication to the treatment of cerebral palsy. Physiotherapy1971;57:526–32.

[3] Sherrington CS. The integrative action of the nervous system. In: Harper,Row, editors. Brunnstrom’s movement therapy in hemiplegia: a neuro-physiological approach Philadelphia. 5th ed., 1970. p. 165–7.

[4] Nisand M. La Reconstruction Posturale, un autre regard sur les algiesrachidiennes. In: Sauramps médical, editor. Rachis lombaire et thérapiesmanuelles. 2006. p. 155–69 [Montpellier].

[5] Nisand M. Le travail à distance : explication et illustration d’un principe debase de la Reconstruction Posturale à travers deux exemples de patholo-gies courantes. Kinesither Rev 2009;96:23–7.

[6] Destieux C, Gaudreault N, Isner-Horobetti M-E, et al. Use of PosturalReconstruction1 physiotherapy to treat an adolescent with asymmetricbilateral genu varum and idiopathic scoliosis. Ann Phys Rehabil Med2013;56:312–26. http://dx.doi.org/10.1016/j.rehab.2013.02.004.

[7] Engel O, Nisand M, Callens C. Étude de l’impact de la reconstructionposturale sur la scoliose idiopathique de l’adulte : à propos de sept cas.Kinesither Rev 2008;83:21–8.

[8] Carson RG. Neural pathways mediating bilateral interactions between theupper limbs.[Review] Brain Res Brain Res Rev 2005;49:641–62.

[9] Cernacek J. Contralateral motor irradiation cerebral dominance. Its chan-ges in hemiparesis. Arch Neurol 1961;4:165–72.

[10] Addamo PK, Farrow M, Bradshaw JL, et al. Characterizing the develop-mental profile of effort-induced motor overflow across a timed trial. Am JPsychol 2013;126:227–34.

[11] Hoy KE, Fitzgerald PB, Bradshaw JL, et al. Investigating the corticalorigins of motor overflow. Brain Res Brain Res Rev 2004;46:315–27.

[12] Borbely K, Nagy D, Toth M, et al. Speech activation SPECT and plasticityof language networks. J Neuroradiol 2005;32:345–7.

M. Nisand et al. / Médecine Nucléaire 39 (2015) 502–513 513

[13] Demonet JF, Celsis P, Agniel A, et al. Activation of regional cerebral bloodflow by a memorization task in early Parkinson’s disease patients andnormal subjects. J Cereb Blood Flow Metab 1994;14:431–8.

[14] Ebmeier KP, Murray CL, Dougall NJ, et al. Unilateral voluntary handmovement and regional cerebral uptake of technetium-99m-exametazimein human control subjects. J Nucl Med 1992;33:1623–7.

[15] Kononen M, Kuikka JT, Husso-Saastamoinen M, et al. Increased perfusionin motor areas after constraint-induced movement therapy in chronicstroke: a single-photon emission computerised tomography study. J CerebBlood Flow Metab 2005;25:1668–74.

[16] Magnusson SP, Simonsen EB, Aagaard P, et al. Biomechanical responsesto repeated stretches in human hamstring muscle in vivo. Am J Sports Med1996;24:622–8.

[17] Magnusson SP, Simonsen EB, Aagaard P, et al. A mechanism for alteredflexibility in human skeletal muscle. J Physiol 1996;497:291–8.

[18] Taylor DC, Dalton Jr JD, Seaber AV, et al. Viscoelastic properties ofmuscle-tendon units. The biomechanical effects of stretching. Am J SportsMed 1990;18:300–9.

[19] Nisand M. Bilan morphologique en Reconstruction Posturale : une autregrille de lecture de la scoliose idiopathique. Kinesither Rev 2009;9:25–32.

[20] Destieux C, Nisand M. Geste pratique : l’abduction en rotation médiale dumembre supérieur en Reconstruction Posturale. Kinesither Sci 2010;513:35–41.

[21] Nisand M. Reconstruction posturale1 : la manoeuvre d’abduction-rota-tion médiale du bras. Kinesither Scient 2011;527:39–46.

[22] Destieux C. La manoeuvre d’extension résistée des orteils en Reconstruc-tion Posturale. Mains Libres 2011;5:185–9, www.mainslibres.ch.

[23] Magnusson SP, Simonsen EB, Aagaard P, et al. Mechanical and physicalresponses to stretching with and without preisometric contraction inhuman skeletal muscle. Arch Phys Med Rehabil 1996;77:373–8.

[24] Gajdosik RL. Passive extensibility of skeletal muscle: review of the literaturewith clinical implications. Clin Biomech (Bristol Avon) 2001;16:87–101.

[25] Proske U, Morgan DL. Do cross-bridges contribute to the tension duringstretch of passive muscle? J Muscle Res Cell Motil 1999;20:433–42.

[26] Guissard N, Duchateau J, Hainaut K. Muscle stretching and motoneuronexcitability. Eur J Appl Physiol Occup Physiol 1988;58:47–52.

[27] Esnault M, Viel E. Stretching : étirements par chaînes musculairesillustrées, . 2e éd., Paris: Masson; 2002: 51–2.

[28] Kapreli E, Athanasopoulos S, Papathanasiou M, et al. A lateralization ofbrain activity during lower limb joints movement. An fMRI study.Neuroimage 2006;32:1709–21.

[29] Ciccarelli O, Toosy AT, Marsden JF, et al. Identifying brain regions forintegrative sensorimotor processing with ankle movements. Exp Brain Res2005;166:31–42.

[30] Jiang T, Wu W, Wang X, et al. Activation of brain areas following ankledorsiflexion versus plantar flexion functional magnetic resonance imagingverification. Neural Regen Res 2012;7:501–5. http://dx.doi.org/10.3969/j.issn.1673-5374.2012.07.004.

[31] Prohovnik I, Hakansson K, Risberg J. Observations on the functionalsignificance of regional blood flow in ‘‘resting’’ normal subjects. Neuro-psychologia 1981;18:203–17.

[32] Matsuda H, Higashi S, Kinuya K, et al. SPECT evaluation of brainperfusion reserve by the acetazolamide test using Tc-99m HMPAO. ClinNucl Med 1991;16:572–9.

[33] Holm S, Madsen PL, Sperling B, et al. Use of 99mTc-bicisate in activationstudies by split-dose technique. J Cereb Blood Flow Metabol 1994;14:S115–20.

[34] Shumway-Cook A, Woollacott MH. Motor control. Translating researchinto clinical practice. 3rd ed., Philadelphia: Lippincott Williams &Wilkins; 2007 : 21–45.

[35] Katiuscia Sacco K, Cauda F, D’agata F, et al. Reorganization andenhanced functional connectivity of motor areas in repetitive anklemovements after training in locomotor attention. Brain Res2009;1297:124–34. http://dx.doi.org/10.1016/j.brainres.2009.08.049.

[36] Davidoff RA. Skeletal muscle tone and the misunderstood stretch reflex.Neurology 1992;42:951–63.

[37] Paillard J. Tonus postures et mouvements. In: Kayser C, editor. Physio-logie, 3. Paris: Flammarion; 1976. p. 521–728.

[38] Lattanzio PJ, Petrella RJ. Knee proprioception: a review of mechanisms,measurements, and implications of muscular fatigue. Orthopedics1998;21:463–70.

[39] Solomonow M, Krogsgaard M. Sensorimotor control of knee stability. Areview. Scand J Med Sci Sports 2001;11:64–80.

[40] Valeriani M, Restuccia D, Di Lazzaro V, et al. Central nervous systemmodifications in patients with lesion of the anterior cruciate ligament ofthe knee. Brain 1996;119:1751–62.

[41] Weldon SM, Hill RH. The efficacy of stretching for prevention ofexercise-related injury: a systematic review of the literature. Man Ther2003;8:141–50.

[42] Herbert RD, de Noronha M, Kamper SJ. Stretching to prevent or reducemuscle soreness after exercise. Cochrane Database Syst Rev 2011;7:9[16–7].

[43] MacIntosh BJ, Mraz R, Baker N, et al. Optimizing the experimental designfor ankle dorsiflexion fMRI. Neuroimage 2004;22:1619–27.

[44] Trinastic JP, Kautz SA, McGregor K, et al. An fMRI study of thedifferences in brain activity during active ankle dorsiflexion and planta-rflexion. Brain Imaging Behav 2010;4:121–31. http://dx.doi.org/10.1007/s11682-010-9091-2.

[45] Christensen MS, Lundbye-Jensen J, Petersen N, et al. Watching your footmove an fMRI study of visuomotor interactions during foot movement.Cereb Cortex 2007;17:1906–17.

[46] Francis S, Lin X, Aboushoushah S, et al. fMRI analysis of active, passiveand electrically stimulated ankle dorsiflexion. Neuroimage 2009;44:469–79.

[47] Joseph J. Nightingale. Electromyography of muscles of posture: legmuscles in males. J Physiol 1954;126:81–5.

[48] Miyai I, Tanabe HC, Sase I, et al. Cortical mapping of gait in humans: anear-infrared spectroscopic topography study. NeuroImage 2001;14:1186–92.

[49] Sahyoun C, Floyer-Lea A, Johansen-Berg H, et al. Towards an under-standing of gait control: brain activation during the anticipation, prepara-tion and execution of foot movements. Neuroimage 2004;21:568–75.

[50] Dobkin BH, Firestine A, West M, et al. Ankle dorsiflexion as an fMRIparadigm to assay motor control for walking during rehabilitation. Neu-roimage 2004;23:370–81.

[51] Beales DJ, O’Sullivan PB, Briffa NK. Motor control patterns during anactive straight leg raise in chronic pelvic girdle pain subjects. Spine2009;34:861–70. http://dx.doi.org/10.1097/BRS.0b013e318198d212.

[52] Louw A, Diener I, Butler DS, et al. The effect of neuroscience educationon pain, disability, anxiety, and stress in chronic musculoskeletal pain.Arch Phys Med Rehabil 2011;92:2041–56. http://dx.doi.org/10.1016/j.apmr.2011.07.198.

[53] Hodges PW, Tucker K. Moving dierently in pain: a new theory to explainthe adaptation to pain. Pain 2011;152:S90–8. http://dx.doi.org/10.1016/j.pain.2010.10.020.

[54] Moseley GL. Pain, brain imaging and physiotherapy opportunity isknocking. Man Ther 2008;13:475–7. http://dx.doi.org/10.1016/j.math.2008.10.001.