Prefrontal cortex rTMS enhances action naming in progressive non-fluent aphasia

Prefrontal cortex rTMS enhances action naming in progressive non-fluent aphasia M. Cotellia, R. Manentia, A. Albericib, M. Brambillaa, M. Cosseddub, O. Zanettia, A. Miozzob, A. Padovanib, C. Miniussia,c and B. Borronib aIRCCS Centro San Giovanni di Dio Fatebenefratelli, Brescia; bCentre for Aging Brain and Neurodegenerative Disorders, Neurology unit, University of Brescia, Brescia; and cDepartment of Biomedical Sciences and Biotechnologies, National Neuroscience Institute, University of Brescia, Brescia, Italy Keywords: language, non-invasive brain stimu- lation, progressive non- fluent aphasia, semantic dementia, transcranial magnetic stimulation Received 9 August 2011 Accepted 9 February 2012. Background and purpose: Progressive non-fluent aphasia (PNFA) is a neurodegen- erative disorder that is characterized by non-fluent speech with naming impairment and grammatical errors. It has been recently demonstrated that repetitive transcranial magnetic stimulation (rTMS) over the dorsolateral prefrontal cortex (DLPFC) im- proves action naming in healthy subjects and in subjects with Alzheimer�s disease. Purpose: To investigate whether the modulation of DLPFC circuits by rTMS mod- ifies naming performance in patients with PNFA. Methods: Ten patients with a diagnosis of PNFA were enrolled. High-frequency rTMS was applied to the left and right DLPFC and the sham (i.e. placebo) condition during object and action naming. A subgroup of patients with semantic dementia was enrolled as a comparison group. Results: A repeated-measure ANOVA with stimulus site (sham, left and right rTMS) showed significant effects. Action-naming performances during stimulation of both the left and right DLPFC were better than during placebo stimulation. No facilitating effect of rTMS to the DLPFC on object naming was observed. In patients with a diagnosis of semantic dementia, no effect of stimulation was reported. Conclusions: Our study demonstrated that rTMS improved action naming in subjects with PNFA, possibly due to the modulation of DLPFC pathways and a facilitation effect on lexical retrieval processes. Future studies on the potential of a rehabilitative protocol using rTMS applied to the DLPFC in this orphan disorder are required. Introduction Humans are highly dependent on language in their day- to-day functioning. As a result, language disorders are associated with substantial disability [1]. Progressive non-fluent aphasia (PNFA) is a neuro- degenerative condition that presents in the presenium and belongs to the primary progressive aphasia spec- trum that includes semantic dementia (SD) and log- openic progressive aphasia [2]. PNFA is characterized by a progressive effortful, non-fluent speech with grammatical errors and omissions [1,3] and naming impairment, with greater difficulty in naming actions than in naming objects [4]. Speech worsens gradually, and patients eventually become mute. The anatomical localization of PNFA is represented by focal anterior peri-Sylvian atrophy that involves the inferior, opercular and insular portions of the left frontal lobe and the left dorsolateral prefrontal cortex (DLPFC) [2,5]. Progressive non-fluent aphasia is considered an or- phan disorder because no evidence-based treatments are currently available to improve language perfor- mances or delay disease progression. It has been recently demonstrated that repetitive transcranial magnetic stimulation (rTMS) is effective in modulating the excitability of the DLPFC circuits and in facilitating naming [6–10]. In particular online, high- frequency rTMS administered at appropriate time intervals reduces vocal reaction times (vRTs) for picture naming in healthy individuals [7,11] and improves the number of correct responses in patients with Alzhei- mer�s disease [8,9]. Although the neuropsychological mechanisms responsible for rTMS-induced facilitation are still unclear, it has been postulated that rTMS may Correspondence: M. Cotelli, IRCCS Centro San Giovanni di Dio, Fatebenefratelli Via Pilastroni, 425125 Brescia, Italy (tel.: +0039 0303501593; fax: 0039-0303533513; e-mail: mcotelli@fatebe- nefratelli.it). 1404 � 2012 The Author(s) European Journal of Neurology � 2012 EFNS European Journal of Neurology 2012, 19: 1404–1412 doi:10.1111/j.1468-1331.2012.03699.x promote novel activity patterns within the affected functional brain networks [12]. Repetitive transcranial magnetic stimulation is thought to induce long-lasting changes in cortical excitability, depending on a number of variables, such as the frequency of stimulation, stimulus intensity, site of stimulation and number of applications. One of these parameters, the frequency of stimulation, is widely thought to be a critical determinant in the modification of the cortical response. Both high (>5 Hz) and low- frequency (£1 Hz) rTMS have been employed, with the former mainly having an excitatory effect and the latter mainly having an inhibitory effect [13]. Deficits in action naming are the core feature in PNFA, which selectively involves DLPFC networks; therefore, it might be predicted that high-frequency rTMS of the DLPFC may be of help in efforts to im- prove action naming. In this study, we used high-frequency rTMS to investigate whether the modulation of activity of the DLPFC could modify the naming performance in PNFA. To test the specificity of the rTMS effect on language impairment in patients with PNFA, a group of patients with SD was used as a comparison group. Materials and methods Subjects Ten patients who fulfilled the current clinical criteria for PNFA [14–16] were recruited from the Centre for Ageing Brain and Neurodegenerative Disorders at the University of Brescia and from IRCCS Centro San Giovanni di Dio Fatebenefratelli, Brescia, Italy. Additionally, four patients with a diagnosis of SD were enrolled as a comparison group. Stringent exclusion criteria were applied as follows: (i) cere- brovascular disorders, hydrocephalus and intra-cra- nial mass, documented by MRI; (ii) a history of traumatic brain injury or another neurological dis- ease; (iii) significant medical problems (e.g. poorly controlled diabetes or hypertension or cancer within the past 5 years); (iv) major depressive disorder, bipolar disorder, schizophrenia, substance use disor- der or mental retardation according to the criteria of the DSM-IV; and (v) implanted metal objects or a history of seizures or any contraindication for rTMS [17]. The inclusion criteria were as follows: (i) only pa- tients with mild to moderate language impairment (Aachener Aphasie Test subtests with no severe impairment) entered the study; and (ii) patients had to be observed for at least 1 year after enrolment, and the diagnosis had to be confirmed. Of the ten patients with PNFA who were included in the study, eight were women and two were men. The mean (±standard deviation) age of the patients with PNFA was 69.1 (±9.3) years, and the mean age at onset was 66.8 (±9.1) years. A positive family history for dementia was recorded in 40% of cases. On average, patients had received 8.1 years (±4.1) of formal edu- cation. Two patients of 10 carried the PGRN Thr272fs mutation. Four patients with SD (one woman and three men) were considered as the comparison group to test the specificity of rTMS on language disturbances in patients with primary progressive aphasia. For patients with SD, the mean age was 68.2 (±10.1) years, and the mean age at onset was 66.2 years (±12.7). These two groups did not differ significantly with regard to age [t(12) = 0.19, P > 0.05] or education [t(12) = 0.06,P > 0.05]. PNFAwasdiagnosedwhen the first symptom was an isolated disorder of expressive lan- guage, whilst other aspects of cognition and daily living functions were relatively well preserved. PNFA was characterized by a reduction in the rate of speech with apraxia of speech, speech sound errors and agrammatism and relativelywell-preserved single-word comprehension. Semantic dementia was defined by a prominent sin- gle-word comprehension disorder (e.g. an impaired understanding of word meaning and/or object identity) and difficulty with confrontation naming. The diagnostic assessment involved a review of the full medical history, a semi-structured neurological exami- nation, a neuropsychological evaluation and a brain MRI study. All subjects underwent a brain MRI scan performed at 1.5 T Siemens (Symphony, Siemens, Erlangen, Germany) along with a 99 mTc-ECD SPEscan. Patients were administered an intravenous injection of 1110 MBq 99mTc-ECD (ethylcysteinate di- mer, Neurolite, Bristol-Myers Squibb Pharma) whilst resting and were imaged using a dual-head rotating gamma camera (VG Millenium GE) fitted with a low- energy, high-resolution collimator, 30 min after the intravenous injection of 99mTc-ECD, as previously de- scribed (GEGeneral Eletric Company, Easton Turnpike Fairfield, CT, USA). Statistical Parametric Mapping (SPM2; Welcome Department of Cognitive Neurology, University Col- lege, London) and Matlab 6.1 (Mathworks Inc., Sher- born, MA, USA) were used for image pre-processing. Images were spatially normalized to a reference ste- reotactic template [Montreal Neurological Institute (MNI)] and smoothed by a Gaussian kernel of 8 · 8 · 8 mm FWHM. Single-photon emission computed tomography (SPECT) data analysis was performed by researchers who were blinded to the clinical data. rTMS & PNFA 1405 � 2012 The Author(s) European Journal of Neurology � 2012 EFNS European Journal of Neurology Genetic sequencing of Microtubule-Associated Pro- tein Tau and Progranulin was also performed. The work was conducted in accordance with local clinical re- search regulations and conformed to the Helsinki Declaration. The study was approved by the local ethics committee, and informed consent was obtained from all participants prior to the beginning of the experiment. All of the included subjects were right-handed. Baseline cognitive assessment included screening tests for dementia (MMSE; Frontotemporal Lobar Degen- eration modified Clinical Dementia Rating Scale, FTLD-modified CDR [18–20]) and neuropsychological tests for non-verbal reasoning (Raven Coloured Pro- gressive Matrices), verbal fluency (phonemic and semantic), long-term memory (Story recall; Rey- Osterrieth Complex Figure, Recall), constructional and visuospatial abilities (Rey-Osterrieth Complex Figure, Copy) and attention and executive functions (Trial- Making Tests A and B). All the tests were administered and scored according to standard procedures [21]. The results of the baseline cognitive assessment are reported in Table 1. In addition, language functions were for- mally assessed with the full Italian version of the Aachener Aphasie Test (AAT). The neuropsychological data indicated that patients with PNFA showed impairment performance in all the assessed functions and preserved verbal short-term memory. Patients with SD obtained low scores on the semantic fluency test, long-term memory test and attention task. Language information about the patients with PNFA and SD is summarized in Table 2, along with the patients� scores on the four tasks (repetition, naming, writing and comprehension) of the AAT [22]. Formal speech evaluation revealed marked deficits in repetition, writing, naming and comprehension in patients with PNFA, whereas the SD revealed difficulties with repe- tition, naming and comprehension, and a preservation of the ability to create a written record of dictation. Stimuli The stimuli used in the action- and object-naming tasks were taken from the Center for Research in Language- International Picture Naming Project corpus CRL- IPNP [23]. These items have been tested and normalized in healthy and patient populations across seven differ- ent international sites and languages. We used 84 items (42 actions and 42 objects) selected from a previous experiment in healthy ageing subjects [11]. None of the action stimuli included in the task was associated with the objects selected. The nouns and verbs corresponding to the set of objects and actions used were matched for target-word frequency and length. The frequency, length of the target word, visual complexity and imageability of the pictures were mat- ched and counterbalanced between the experimental blocks. The items were divided into three blocks that were designed to represent the three stimulation con- ditions (left DLPFC, right DLPFC and placebo stim- ulation). The frequencies and lengths of the target words were counterbalanced in the experimental blocks. Table 1 Neuropsychological assessment in patients with progressive non-fluent aphasia (PNFA) and semantic dementia (SD) PNFA (n = 10) SD (n = 4) Cut-offs Screening for dementia MMSE 18.0 (3.7) 22.9 (5.3) >24 FTLD-modified CDR 4.0 (2.3) 3.0 (0.6) Non-verbal reasoning Raven Coloured Progressive Matrices 17.8 (8.8) 26.3 (7.1) >17.5 Memory Short Story, recall 7.2 (2.8) 6.8 (3.0) >7.5 Rey-Osterrieth Complex Figure, Recall 8.9 (3.9) 8.0 (7.6) >9.46 Digit Span 4.0 (0.3) 4.2 (0.3) >3.75 Language Fluency, phonemic 12.6 (6.4) 16.2 (11.4) >16 Fluency, semantic 21.2 (6.4) 14.2 (9.9) >24 Constructional and visuospatial abilities Rey-Osterrieth Complex Figure, Copy 13.2 (8.9) 31.3 (1.8) >28.87 Executive functions Trail-Making Test A 257.5 (194.2) 95.0 (93.8) <93.0 Trail-Making Test B 418.8 (28.0) 274.0 (99.1) <282.0 Results corrected for age and schooling. Cut-off scores referred to Italian normative data. MMSE, Mini-Mental State Examination. Standard deviation between brackets. Bold data refer to pathological scores. Table 2 Aachener Aphasie Test (AAT) subtests in patients with pro- gressive non-fluent aphasia (PNFA) and semantic dementia (SD) AAT subtests Mean scores ± standard deviation Cut-offs Token test (errors) PNFA 21.0/50 ± 12.0 <7 SD 18.6/50 ± 9.0 Repetition PNFA 124.9/150 ± 10.9 >142 SD 133.0/150 ± 6.5 Writing PNFA 57.0/90 ± 18.1 >81 SD 83.6/90 ± 5.8 Naming PNFA 86.5/120 ± 19.0 >104 SD 73.3/120 ± 8.0 Comprehension PNFA 93.4/120 ± 6.9 >108 SD 91.0/120 ± 14.7 Bold data refer to pathological scores. 1406 M. Cotelli et al. � 2012 The Author(s) European Journal of Neurology � 2012 EFNS European Journal of Neurology The visual complexity and imageability of the pictures were also matched between blocks. Ten additional ob- jects and actions were used for a practice block (five actions and five objects). Procedure Subjects sat in front of a 17-inch monitor that was controlled by a personal computer running Presenta- tion software (http://www.neurobs.com). The trial structure is illustrated in Fig. 1. After a frame that indicated the category of the stimulus to the subject (�ACTION� or �OBJECT�), a warning sound 50 ms in duration was presented at the onset of a centrally lo- cated fixation cross that was present for 1000 ms. After disappearance of the fixation cross, the stimulus was presented and remained on the screen for 1000 ms. A blank screen then followed for a time period varying from 4000 to 5000 ms. The subject�s task was to name, as rapidly as possible, the stimuli that appeared on the computer screen. Vocal responses were recorded and digitized at 44.1 kHz, using the program GoldWave (V. 5.12; http://www.goldwave.com). The responses were then analysed offline for accuracy and vRTs. The vRT analysis was performed only on correct responses that were <2 standard deviations from the mean RT (3.3% of the responses were eliminated). In the case of uncertain initial vocalization, the start of the response was considered at the beginning of the correct complete response. For each stimulus, we calculated the mean response accuracy percentage and the mean vRTs. The three stimulation sites (the left and right DLPFC and placebo stimulation) and the block orders were counterbalanced across subjects. For the placebo con- trol condition, a 3-cm-thick piece of plywood was ap- plied to the coil [24] so that no magnetic fields reached the cortex. For the placebo condition, the junction of the two-coil wings was placed over the vertex (CZ in the EEG 10/20 international system) using the same pro- cedure as is used for the real rTMS. For left and right DLPFC, the Talairach coordinates of the cortical sites underlying the coil were estimated for each subject using the SofTaxic Evolution Navigator system (V. 2.0; http://www.emsmedical.net). This system was used to identify the stimulation site on the scalp above the DLPFC (Talairach coordinates X = ±35, Y = 24, Z = 48, middle frontal), as in previous studies [8,9,11]. To stimulate the DLPFC, we used a 70-mm figure-eight cooled coil and placed the junction of the two-coil wings above the target point. rTMS was delivered for 500 ms from the onset of the visual stimulus using a frequency of 20 Hz. The stimulation intensity used during the experiment was set at 90% of each subject�s resting motor threshold. These parameters are consis- tent with the safety recommendations for rTMS [17], and none of the subjects showed or reported any side- effects of stimulation. The coil position and the specific brain areas of hypoperfusion for representative patients with PNFA and SD are reported in Fig. 2. The applied procedure was exactly the same as was used in a recent study on healthy older adults [11] that showed that the naming latency for actions was short- ened after stimulation of the left and right DLPFC compared with application of the sham stimulation (actions: left 963 ± 20, right 976 ± 40, sham 1078 ± 36). Stimulation was not observed to have any effect on the accuracy of naming in this healthy group. Inter- estingly, the older adults included in this previously published report and the patients with PNFA tested in the present work did not differ significantly with regard to age [t(22) = 0.26, P > 0.05]. Results We analysed both accuracy and vRTs using a repeated- measure ANOVA for each patient group (PNFA or SD) as the between-subject factor and site (sham, left and right) and stimulus type (actions or objects) as within- subject factors. Post hoc analyses (Fisher�s least signif- icant difference, LSD test) were performed. The results are expressed as the mean ± standard deviation or percentage, as indicated. Statistical significance was set at P £ 0.05. Figure 1 Schematic representation of the essential steps of the study design (see Methods for details). rTMS & PNFA 1407 � 2012 The Author(s) European Journal of Neurology � 2012 EFNS European Journal of Neurology Accuracy The effect of rTMS of the DLPFC on object and action naming was analysed. In PNFA subjects, the repeated- measure ANOVA with site (placebo, left and right rTMS) as a factor demonstrated significant effects of stimula- tion [F(2, 18) = 3.66, P = 0.04]. Figure 3 shows the mean naming accuracy scores in each of the stimulation conditions for objects and actions. Action-naming performance during left (mean = 48.88 ± 6.3, P = 0.036) and right (49.57 ± 6.8, P = 0.027) DLPFC stimulation was enhanced in comparison with that observed during placebo stimu- lation (38.15 ± 6.9). Conversely, object-naming per- formance did not differ significantly between the conditions (left, 86.68 ± 2.7; right, 79.14 ± 5.8; pla- cebo stimulation, 78.57 ± 7.8). The single-subject scores showed that 50% of pa- tients with PNFA demonstrated bilateral effects in ac- tion-naming improvement, but 30% of the patients reported a selective right DLPFC effect, and 20% showed an improvement only after stimulation of the left DLPFC. The difference in individual scores for action naming during left and right DLPFC stimula- tion, as compared with placebo stimulation, are reported in Fig. 4. In patients with SD, no facili