Bihemispheric stimulation over left and right inferior frontal region enhances recovery from apraxia of speech in chronic aphasia

Authors


  • Address from which the work originated: IRCCS Fondazione Santa Lucia, Via Ardeatina 306 00179, Rome, Italy.

Abstract

Several studies have already shown that transcranial direct current stimulation (tDCS) is a useful tool for enhancing recovery in aphasia. However, all tDCS studies have previously investigated the effects using unihemisperic stimulation. No reports to date have examined the role of bihemispheric tDCS on aphasia recovery. Here, eight aphasic persons with apraxia of speech underwent intensive language therapy in two different conditions: real bihemispheric anodic ipsilesional stimulation over the left Broca's area and cathodic contralesional stimulation over the right homologue of Broca's area, and a sham condition. In both conditions, patients underwent concurrent language therapy for their apraxia of speech. The language treatment lasted 10 days (Monday to Friday, then weekend off, then Monday to Friday). There was a 14-day intersession interval between the real and the sham conditions. In all patients, language measures were collected before (T0), at the end of (T10) and 1 week after the end of (F/U) treatment. Results showed that after simultaneous excitatory stimulation to the left frontal hemisphere and inhibitory stimulation to the right frontal hemisphere regions, patients exhibited a significant recovery not only in terms of better accuracy and speed in articulating the treated stimuli but also in other language tasks (picture description, noun and verb naming, word repetition, word reading) which persisted in the follow-up session. Taken together, these data suggest that bihemispheric anodic ipsilesional and cathodic contralesional stimulation in chronic aphasia patients may affect the treated function, resulting in a positive influence on different language tasks.

Introduction

Speech is probably one of the most complex and most intensively exercised motor skills of humans. In any language, the frequent use of always the same bundle of articulatory gestures participating in the construction of words transforms the recurring motor pattern into a stable, overlearned movement program represented onto the motor-cortical hard-disk that contains the human's phonetic lexicon. From there it can be accessed rapidly and safely whenever the words occur in an utterance (Levelt et al., 1999). Focal brain damage, such as a stroke in the left hemisphere, can cause a disorder in this alternation of movements, known as ‘apraxia of speech’. It is manifested as distortions of consonants and vowels that may be perceived as sound substitutions in the absence of reduced strength or tone of muscles and articulators controlling phonation (McNeil et al., 2000; Duffy, 2005). Since Paul Broca in 1865, the hypothesis has been advanced that damage to the left inferior frontal gyrus (IFG; Broca's area) might cause apraxia of speech disorders. Subsequent studies have suggested the involvement of the left anterior insula (Shuren, 1993; Dronkers, 1996; Donnan et al., 1997; Nestor et al., 2003), while others have confirmed that the most frequent area of damage in patients with apraxia of speech is Broca's region (Hillis et al., 2004).

Numerous treatments have been developed to remediate the apraxia speech disorder (Rosenbek et al., 1973; McNeil et al., 1997; Knock et al., 2000; Wambaugh, 2002). In recent years, new treatment approaches have emphasised the role of brain stimulation techniques, such as transcranial direct current stimulation (tDCS), as adjuvant strategies to speed up the language recovery process. Indeed, it has been demonstrated that aphasic patients exhibited greater recovery of word-retrieval deficits if the language treatment was coupled with repeated unihemispheric tDCS stimulation (Baker et al., 2010; Fiori et al., 2011; Flöel et al., 2011; Fridriksson et al., 2011; Kang et al., 2011; Monti et al., 2012; Marangolo et al., 2013). In a preliminary study, persistent beneficial effects were found in three chronic aphasic patients after 1 week of intensive language treatment for their apraxia of speech together with 20 min of anodic tDCS stimulation over the left Broca's area (Marangolo et al., 2011). Until now, the efficacy of bihemispheric tDCS stimulation has been mainly investigated in stroke motor recovery (Vines et al., 2008; Lindenberg et al., 2010; Lefebvre et al., 2012; Mordillo-Mateos et al., 2012). This was based on the assumption that upregulating excitability of intact portion of the ipsilesional motor cortex through anodic stimulation and downregulating excitability of the contralesional one through cathodic application should lead to the greatest recovery. Accordingly, bihemispheric tDCS and simultaneous physical and occupational therapy given over five consecutive sessions significantly improved motor function in a group of twenty chronic stroke patients when compared to the sham group (Lindenberg et al., 2010).

The purpose of our study was to investigate for the first time whether bihemispheric tDCS delivered over the IFG (in eight chronic aphasics) potentiated the recovery from apraxia of speech.

Materials and methods

Participants

Eight left-brain-damaged participants (four male and four female) were included in the study (see Fig. 1). Inclusion criteria were native Italian proficiency, pre-morbid right-handedness (based on the Edinburgh Handedness Questionnaire; Oldfield, 1971), a single left hemispheric stroke at least 6 months prior to the investigation, and no acute or chronic neurological symptoms requiring medication.

Figure 1.

Lesion descriptions for each aphasic patient. The figure shows the MRI acquisitions of all patients. Patients B.A., F.G., B.B., C.G. and C.M. underwent a 3.0-T scanner exam (Phillips Achieva Magnetom MR system) with acquisition of a T1 3-D sequence. Patients B.G., G.N. and N.R. underwent a 1.5-T scanner exam (Siemens Vision Magnetom MR system) with acquisition of a T1 3-D sequence for B.G. and N.R. and a clinical T1 sequence for G.N. (Top line) Sagittal and axial views of color-coded probability map of lesion overlap (range 1% dark purple to 100% white). Individual volume lesions were transformed into a standardised stereotaxic coordinate system using a computational semi-automatic procedure REGISTER; software provided by Brain Imaging Center, Montreal Neurological Institute, McGill University, Canada. B.G.'s lesion is localised in the left fronto-temporo-parieto-occipital cortex. At frontal level the damage involves the middle frontal gyrus and the IFG (Broca's area). At the temporo-parietal level the damage partially spared the temporal pole and involves the full extension of the superior temporal gyrus (Wernicke's area), the middle temporal gyrus, the angular and the supramarginal gyri and the inferior parietal lobule. At occipital level the damage involves the superior and the middle occipital gyri. The lesion also includes the insula. At sub-cortical level the damage partially involves the homolateral subcortical nuclei (putamen and globus pallidus) and the external and internal capsule. B.A.'s lesion is localised in the left fronto-temporo-parietal cortex. At frontal level the damage involves the middle frontal gyrus and the IFG (Broca's area). At the temporal–parietal level the damage includes the temporal pole, the full extent of the superior temporal gyrus (Wernicke's area), the middle and most of the inferior temporal gyrus and of the angular and supramarginal gyri. The lesion also includes the insula. At sub-cortical level the lesion involves the thalamus, the putamen and the globus pallidus, the caudate nucleus and the external and internal capsule. G.N.'s lesion is localised in the left fronto-parietal cortex. At frontal level the damage involves the IFG (Broca's area), the middle frontal gyrus and the pre-central gyrus. At the parietal level the damage includes the post-central gyrus and part of the inferior parietal lobule. The lesion also involves the insula. At subcortical level the lesion includes the thalamus, the putamen and the globus pallidus, the caudate nucleus and the external and internal capsule. In F.G. a cortical–subcortical atrophy involving the fronto-parietal cortex is present. The lesion is localised in the left insula. At the subcortical level the lesion involves the thalamus, the putamen and the globus pallidus, the caudate nucleus and the external and internal capsule. B.B.'s lesion is localised in the left fronto-temporo-parietal cortex. At frontal level the damage involves the IFG (Broca's area), the middle frontal gyrus and the pre-central gyrus. At temporal level the damage includes the superior temporal gyrus (Wernicke's area) and part of the inferior temporal gyrus. At parietal level the damage involves the post-central gyrus and the inferior parietal gyrus. The lesion also includes the insula. At the subcortical level the lesion partially involves the ipsilateral subcortical nuclei (putamen and globus pallidus) and the external and internal capsule. C.G.'s lesion is localised in the left fronto-temporo-parietal cortex. At the frontal level the damage involves the middle frontal gyrus, the IFG (Broca's area) and part of the lateral frontal pole and of the pre-central gyrus. At the temporal level the damage includes the temporal pole, the full extension of the superior temporal gyrus (Wernicke's area) and most of the middle temporal gyrus. At the parietal level the damage involves most of the post-central gyrus, of the inferior parietal lobule and of the angular and supramarginal gyri. The lesion also includes the insula. At the subcortical level the lesion includes the putamen and the globus pallidus, and the external and internal capsule. N.R.'s lesion is localised in the left fronto-temporo-parietal cortices. At the frontal level the damage laterally involves the IFG (Broca's area), the middle frontal gyrus, the superior frontal gyrus and the pre-central gyrus, and medially the medial frontal gyrus and the anterior cingulate gyrus. At the temporal level the lesion includes the pole as well as the superior (Wernicke's area) and the middle temporal gyrus. At the parietal level the damage involves the post-central gyrus and the inferior parietal lobule. The lesion also includes the insula. At the subcortical level the lesion includes the putamen and the globus pallidus, and the external and internal capsule. C.M.'s lesion is localised in the left fronto-temporo-parietal cortex. At the frontal level the damage involves the middle frontal gyrus, the IFG (Broca's area), part of the lateral frontal pole and the pre-central gyrus. At the temporal level the damage includes part of the temporal pole and the full extent of the superior temporal gyrus (Wernicke's area). At the parietal level the damage involves most of the post-central gyrus, the inferior parietal lobule and the angular and part of the supramarginal gyri. The lesion also includes the insula. At the subcortical level the lesion involves the putamen and the globus pallidus, the caudate nucleus and the external and the internal capsule.

Ethics statement

The data analysed in the current study conformed with The Code of Ethics of the World Medical Association (Declaration of Helsinki) printed in the British Medical Journal (18 July 1964) and were collected in accordance with the Institutional Review Board of the IRCCS Fondazione Santa Lucia, Rome, Italy. Our named Institutional Review Board specifically approved this study with the understanding and written consent of each subject.

Clinical data

Each patient had nonfluent speech. Subjects were not able to produce any words in spontaneous speech. Their language production was limited to a few syllables due to their apraxia speech disorder. Severe articulatory groping and distortions of phonemes were present in naming, repetition and reading tasks of twenty simple syllables (e.g. PA, MO, FU) and words [e.g. luna (moon), pipa (pipe)] of a standardised test for the evaluation of articulation (Fanzago test, 1983). To deeply investigate the aphasics’ language performance, each subject was also administered a standardised language test (Esame del Linguaggio II; Ciurli et al., 1996). The test included a picture description task, oral and written noun- and verb-naming tasks [= 20 for noun naming, i.e. topo (mouse); = 10 for verb naming, i.e. correre (to run), dormire (to sleep)], word repetition, reading and writing under dictation [= 20, i.e. letto (bed), tavolo (table)]. The test also comprised an auditory picture–word matching task (= 20) and a simple commands comprehension task [= 20, i.e. alzi la mano sinistra (raise your left hand), apra il libro (open the book)]. In all oral tasks, due to their apraxia speech disorder most patients did not produce any response (see Table 1). Some phonological errors were also present [i.e. topo (mouse) popo]. Noun and verb written naming and word writing under dictation were severely impaired. Errors were mostly omissions of the whole word. Auditory comprehension abilities were adequate for words and simple commands in the language test (Esame del Linguaggio II; Ciurli et al., 1996) while patients still had difficulties in a more complex auditory comprehension task (Token test cut-off 29/36; De Renzi & Faglioni, 1978; see Table 1). To evaluate nonverbal oral motor skills, the Oral Apraxia test (De Renzi et al., 1966) was administered. None of the patients showed apraxic disturbances.

Table 1. Sociodemographic and clinical data of the eight nonfluent aphasic subjects
Part.SexAgeEd. levelTime post-onsetPicture descrip.Noun namingVerb namingWord compr.Word rep.Word readSentence compr.Token test
  1. For each language task, the percentage of correct responses are reported (Esame del Linguaggio II, cut-off 100%; Ciurli et al., 1996; Token test, cut-off 29/36, De Renzi & Faglioni, 1978). Ed. level, Educational level; Picture descrip., Picture descriptions; Compr., comprehension, Rep., repetition; read, reading; y, years; mo, months.

B.G.M6082 y 6 mo00010000608/36
B.A.F62136 y 2 mo050100350759/36
G.N.F6081 y 2 mo0001000010010/36
F.G.M6854 y 8 mo000100603510010/36
B.B.F37186 mo0253010030257011/36
C.G.M49183 y047010012458012/36
N.R.M531310 mo00010070259013/36
C.M.F51136 mo015010010209014/36

Procedure

tDCS

Transcranial direct current stimulation was applied using a battery-driven Eldith (NeuroConn GmbH, Germany) Programmable Direct Current Stimulator with a pair of surface-soaked sponge electrodes (5 × 7 cm). Real stimulation consisted of 20 min of 2 mA direct current with the anode placed over the ipsilesional and the cathode over the contralesional IFG (F5 and F7 of the extended International 10–20 system for EEG electrode placement). For sham stimulation, the same electrode positions were used. The current was ramped up to 2 mA and slowly decreased over 30 s to ensure the typical initial tingling sensation (Gandiga et al., 2006). In both conditions, patients underwent concurrent speech therapy for their apraxia speech disorder. The language treatment was performed in ten daily sessions (Monday to Friday, then weekend off, then Monday to Friday). There was 14-day intersession interval between the real and the sham conditions. The order of conditions was randomised across subjects (see Fig. 2). Both the patient and the clinician were blinded with respect to the administration of tDCS. At the end of each condition, subjects were asked if they were aware of which condition (real or sham) they were in. None of the subjects was able to ascertain differences in intensity of sensation between the two conditions.

Figure 2.

Overview of study design.

Language treatment

Patients were administered all the standardised language tests at the beginning (baseline; T0) and at the end (T10) of each treatment condition, and 1 week after T10 (follow-up; F/U).

Before the treatment, 126 stimuli (syllables, words and sentences) were auditorily presented, one at a time, through an audiotape for three consecutive days. The participants had to repeat each stimulus within 20 s. We identified the stimuli the patients could not correctly produce or always omitted. As all subjects fail to correctly produce all the presented stimuli, the whole lists were considered. For each subject, the selected stimuli were subdivided into two lists of 63 stimuli. Each list included 28 syllables (e.g. PA, MO, CA, FU), 25 bysyllabic words [CV consonant–vowel, e.g. luna (moon), CVCCV, e.g. palla (ball)] and 10 S-V-O simple sentences (e.g. la donna fa la foto (the woman takes a picture)] were used.

According to the International Phonetic Alphabet (IPA, 1999), syllables included different places (e.g. plosive, nasal, fricative) and manners of articulation (e.g. bilabial, dental, velar). The two lists of words were matched for frequency and length.

Each list was randomly assigned to one of the two stimulation conditions (real vs. sham). In each condition, the order of presentation of stimuli was randomised across the treatment sessions.

The therapy method was similar for all patients. For each condition, the whole list of stimuli was presented during each session. The clinician and the patient were seated face-to-face so that the patient could watch the articulatory movements of the clinician as she spoke. The clinician presented one stimulus at a time and for each stimulus the treatment involved the use of four different steps which would progressively induce the patient to correctly reproduce it.

Step 1: The clinician auditorily presented the whole stimulus and asked the patient to repeat it. If the patient correctly repeated the stimulus, the clinician would present another stimulus but if he or she made errors the clinician would move on to the next step.

Step 2: The clinician auditorily presented the stimulus with a pause between syllables, prolonged the vowel sound, exaggerated the articulatory gestures and asked the patient to do the same.

Step 3: As in step 2, the clinician auditorily presented the stimulus, again with a pause between syllables, prolonged the vowel sound, exaggerated the articulatory gestures and asked the patient to do the same.

Step 4: The clinician auditorily presented one syllable at a time, prolonged the vowel sound, exaggerated the articulatory gestures and asked the patient to do the same.

If the patient was not able to articulate the stimulus in the first step, the clinician would move on to the next step and so on up to the last step. Any time the patient was able to reproduce the articulatory gestures facilitated by the clinician, he or she would be asked to repeat the whole stimulus without the clinician's help and only if he or she succeeded in doing so again was the response was considered correct. If the patient was not able to articulate the stimulus in the last step, the response was considered an error. The clinician manually recorded the response type on a separate sheet.

Data analysis

Statistical evaluations were performed using spss 13.0 software. Repeated-measures anovas (3 × 2) were run for syllables, words and sentences, with separate analyses for response accuracy and vocal reaction times (RTs). As RTs for syllables were already short at T0, for this group of stimuli RTs were not collected. For each analysis, two within-subject factors were included: Time (T0 vs. T10 vs. F/U) and Condition (real stimulation vs. sham). Interaction was explored using the Scheffé post hoc test.

For each stimulus, vocal RT was measured from the onset of the participant's response to the end of the stimulus production using Free Audio Editor 6.9.1 software.

Results

Accuracy data

Syllables

The analysis showed a significant effect of Time [Baseline (T0) vs. End of treatment (T10) vs. Follow-up (F/U), F2,14 = 31.76, = 0.000] and Condition (Real Stimulation vs. Sham, F1,7 = 16.76, = 0.005). The interaction of Time × Condition was also significant (F2,14 = 4.50, = 0.031). The Scheffé post hoc test revealed that, while no significant differences emerged in the mean percentage of correct syllables between the two conditions at T0 (differences between Real Stimulation and Sham, 2%; = 1), the mean percentage accuracy was significantly greater in the real stimulation than in the sham condition, both at T10 (differences between Real Stimulation vs. Sham at T10, 27%; = 0.027) and at F/U (differences between Real Stimulation vs. Sham at F/U, 24%; = 0.041). No significant differences emerged in the mean percentage accuracy between T0 and T10 for the sham condition (difference between T0 and T10, 12%; = 0.603; see Fig. 3).

Figure 3.

Mean percentage of correct syllables, words and sentences repeated at T0, T10 and F/U, for the real and sham stimulations respectively.

Words

The analysis showed a significant effect of Time (T0 vs. T10 vs. F/U; F2,14 = 38.93, = 0.000) and Condition (Real Stimulation vs. Sham; F1,7 = 7.88, = 0.026). The interaction of Time × Condition was also significant (F2,14 = 4.46, = 0.032). The Scheffé post hoc test revealed that, while no significant differences emerged in the mean percentage of correct words between the two conditions at T0 (differences between Real Stimulation and Sham, 7%; = 0.541), the mean percentage accuracy was significantly greater in the real stimulation than in the sham condition both at T10 (differences between Real Stimulation and Sham at T10, 22%; = 0.000) and at F/U (differences between Real Stimulation and Sham at F/U, 13%; = 0.004; see Fig. 3).

Sentences

The analysis showed a significant effect of Time (T0 vs. T10 vs. F/U; F2,14 = 15.11, = 0.000) and Condition (Real Stimulation vs. Sham; F1,7 = 6.76, = 0.035). The interaction Time × Condition was also significant (F2,14 = 6.33, = 0.011). The Scheffé post hoc test revealed that, while no significant differences emerged in the mean percentage of correct sentences between the two conditions at T0 (differences between Real Stimulation and Sham, 0%, = 1), the mean percentage accuracy was significantly greater in the real stimulation than in the sham condition both at T10 (differences between Real Stimulation and Sham at T10, 30%; = 0.009) and at F/U (differences between Real Stimulation and Sham at F/U, 19%; = 0.041). No significant differences emerged in the mean percentage of accuracy between T0 and T10 for the sham condition (differences between T0 and T10, 11%; = 0.641; see Fig. 3).

We ran further analyses by adding the order of conditions (real stimulation vs. sham) as fixed factor. The order of condition was not significant for the syllables, the words or the sentences (respectively, F1,6 = 0.56, = 0.483, F1,6 = 2.42, = 0.171 and F1,6 = 2.59, = 0.159).

Vocal reaction times data

Words

The analysis showed a significant effect of Time (T0 vs. T10 vs. F/U, F2,14 = 18.75, = 0.000) and of Condition (Real Stimulation vs. Sham, F1,7 = 6.1, = 0.043). The interaction Time × Condition was also significant (F2,14 = 4.27, = 0.036). The Scheffé post hoc test revealed that, while no significant differences emerged in the mean vocal reaction times between the two conditions at T0 (differences between Real Stimulation and Sham, 306 ms; = 0.984), the mean vocal reaction times were significantly faster in the real stimulation than in the sham condition, both at T10 (differences between Real Stimulation and Sham at T10, 2003 ms; = 0.013) and at F/U (differences between Real Stimulation and Sham at F/U, 1524 ms; = 0.042). No significant differences emerged in the mean vocal reaction times between T0 and T10 for the sham condition (differences between T0 and T10, 747 ms; = 0.599; see Fig. 4).

Figure 4.

Mean vocal reaction times for words and sentences at T0, T10 and F/U, for the real and sham stimulations respectively.

Sentences

The analysis showed a significant effect of Time (T0 vs. T10 vs. F/U; F2,14 = 15.11, = 0.000) and Condition (Real Stimulation vs. Sham; F1,7 = 6.38, = 0.040). The interaction of Time × Condition was also significant (F2,14 = 6.77, = 0.009). The Scheffé post hoc test revealed that, while no significant differences emerged in the mean vocal reaction time between the two conditions at T0 (differences between Real Stimulation and Sham, 135 ms; = 1), the mean vocal reaction times were significantly faster in the real stimulation condition than in the sham condition both at T10 (differences between Real Stimulation and Sham at T10, 5191 ms; = 0.006) and at F/U (differences between Real Stimulation and Sham at F/U, 3764 ms; = 0.048). No significant differences emerged in the mean vocal reaction times between T0 and T10 for the sham condition (differences between T0 and T10, 2594 ms; = 0.304; see Fig. 4).

We ran further analyses by adding the order of conditions (real stimulation vs. sham) as fixed factor. Neither for the words nor for the sentences was the order of condition significant (respectively, F1,6 = 4.59, = 0.076 and F1,6 = 1.32, = 0.294).

Discussion

The aim of the present study was to investigate whether bihemispheric frontal stimulation would enhance language recovery and, in particular, language articulation, in a group of left chronic aphasic persons. Bihemispheric tDCS may potentiate the effects of anodic stimulation to the left lesioned hemisphere (Fiori et al., 2011; Marangolo et al., 2011, 2013) through additional modulation of interhemispheric interactions via cathodic stimulation to the homologue contralesional area (Jung et al., 2011; Kang et al., 2011; You et al., 2011). Indeed, only after the real stimulation condition, articulatory errors significantly decreased and all patients were faster in repeating the stimuli compared to the sham condition. Most importantly, significant changes after therapy persisted at F/U and generalised to other tasks. Accordingly, most of the patients showed a significant improvement in different oral language tasks (picture description, noun and verb naming, word repetition and reading) administered before and after the treatment, an improvement which was still present 1 week after the therapy (see Table 2). This improvement revealed that the language treatment resulted in a positive effect on the production of stimuli not only treated but also belonging to other tasks. Indeed, after tDCS stimulation most patients were able to correctly produce the whole word and they showed a reduction in phonological errors, the reduction being due to improvement in speech praxis. This is consistent with previous transcranial direct current stimulation–tDCS literature showing longer-term changes (at 1 month or more) in word retrieval and other language measures (Naeser et al., 2010, 2011; Marangolo et al., 2011, 2013).

Table 2. Mean percentage of correct responses in the different language tasks (Esame del Linguaggio II; Ciurli et al., 1996) for real stimulation and sham conditions
SCondPictur description (%)Verb naming (%)Noun naming (%)Word repetition (%)Word reading (%)
T0T1F/UT0T1F/UT0T1F/UT0T1F/UT0sT1F/U
  1. The order of conditions was randomised across subjects; cut-off score 100%. S, Subjects; Cond, Condition; *< 0.05; **< 0.01; ***< 0.001 (χ2 test).

Real first
B.G.
Real000000010***10020***20000
Sham0000002055202525000
B.A.
Real010***10050***50545***453555**55025***25
Sham1000504242452525553535500
G.N.
Real010***10000025***25020***20035***35
Sham101010000253030202222354040
F.G.
Real010***10010***0515*156080**80374040
Sham000000055606060353737
Sham first
B.B.
Sham055303030253535303535253535
Real550***503060***603562***623572***723565***65
C.G.
Sham000010***10475252125555455555
Real010***101045***455270*705582***825585***85
N.R.
Sham000055035***35708080252525
Real020***20525***253550***5080100***1002540*40
C.M.
Sham020***20030***301550***501027**272055***55
Real2060***603080***805075***752765***655580***80

As far as we know, this is the first study which has investigated the effects of bihemispheric stimulation on the recovery of language. As stated in the Introduction, several studies have already stressed the importance of associating specific language training with anodic unihemispheric tDCS stimulation over the perilesional language areas (Baker et al., 2010; Fiori et al., 2011; Fridriksson et al., 2011; Marangolo et al., 2013). This was based on the assumption that, in chronic patients, language recovery may be associated with the reactivation of left-hemispheric perilesional structures (Warburton et al., 1999; Saur et al., 2006; Winhuisen et al., 2007). Although it is often assumed that the right homologue of Broca's area takes over the function of the left if it is infarcted, the evidence for this is slender. Recent studies have stressed the importance of the left Broca's area or adjacent tissue in the natural recovery from post-stroke aphasia (Saur et al., 2006, 2008). Coherently with this assumption, some studies have also shown that the suppression of the right homologue language areas through repetitive transcranial magnetic stimulation (Naeser et al., 2005, 2010, 2011) or unihemispheric cathodic tDCS (Jung et al., 2011; Kang et al., 2011; You et al., 2011), reducing the inhibition on the ipsilesional cortex exerted by the unaffected hemisphere via the transcallosal pathway, determines significant changes in language recovery.

In line with this hypothesis, we examined the influence of simultaneous anodic ipsilesional and cathodic contralesional stimulation over the inferior frontal gyri (the left Broca's area and the right homologue of Broca's area) on apraxia of speech. Indeed, this bihemispheric stimulation over the inferior frontal gyri resulted in improvement of language. As most of our patients had left hemisphere cortical damage, it could be the case that bihemispheric stimulation engaged the remaining left cortico-subcortical hemispheric network, via interhemispheric white-matter pathways, leading to better recovery.

In conclusion, our data showed for the first time that bihemispheric stimulation is a useful tool for the treatment of apraxia of speech in chronic stroke aphasic persons. Further studies are needed to examine whether a bihemispheric stimulation technique might be more efficacious than unihemispheric stimulation in the recovery of language.

Competing interests

All authors declare that they have no significant competing interests that might have influenced the performance or presentation of the work described in this manuscript.

Funding

None.

Abbreviation
F/U

follow-up (1 week after the end of treatment)

IFG

inferior frontal gyrus

RT

reaction time

T0

baseline (before treatment)

T10

end of treatment

tDCS

transcranial direct current stimulation

Ancillary