The Precentral Insular Cortical Network for Speech Articulation

Abstract

Apraxia of speech is a motor disorder characterized by the impaired ability to coordinate the sequential articulatory movements necessary to produce speech. The critical cortical area(s) involved in speech apraxia remain controversial because many of the previously reported cases had additional aphasic impairments, preventing localization of the specific cortical circuit necessary for the somatomotor execution of speech. Four patients with “pure speech apraxia” (i.e., who had no aphasic and orofacial motor impairments) are reported here. The critical lesion in all four patients involved, in the left hemisphere, the precentral gyrus of the insula (gyrus brevis III) and, to a lesser extent, the nearby areas with which it is strongly connected: the adjacent subcentral opercular cortex (part of secondary somatosensory cortex) and the most inferior part of the central sulcus where the orofacial musculature is represented. There was no damage to rostrally adjacent Broca’s area in the inferior frontal gyrus. The present study demonstrates the critical circuit for the coordination of complex articulatory movements prior to and during the execution of the motor speech plans. Importantly, this specific cortical circuit is different from those that relate to the cognitive aspects of language production (e.g., Broca’s area on the inferior frontal gyrus).

Introduction

Apraxia of speech is an impairment in the ability to coordinate the sequential articulatory movements necessary to organize speech sounds to form syllables, words, phrases, and sentences (Wertz et al. 1984). Patients with speech apraxia, although aware of what they want to say and how it sounds, are not able to organize the particular articulatory movements necessary for speech. This deficit is not caused by impairment of muscle strength, tone, and range of motion (Darley et al. 1975). Patients exhibiting speech apraxia do not have language comprehension impairment and the production of nonspeech orofacial movements is possible (Darley et al. 1975). This disorder is often the result of brain damage from cerebrovascular accidents. The evidence regarding the critical brain damage for this syndrome remains controversial. Some evidence has suggested the involvement of Broca’s area (i.e., the left posterior inferior frontal gyrus, Hillis et al. 2004), the precentral gyrus (Itabashi et al. 2016), the temporo-parietal region (Square et al. 1997), and also subcortical lesions, including the left caudate nucleus and internal capsule (De Renzi et al. 1980; Papagno et al. 1993). Furthermore, there is also evidence that the superior part of the precentral gyrus of the insula in the left hemisphere may be critical for the coordination of speech articulation (Dronkers 1996; Nagao et al. 1999; Ogar et al. 2006; Baldo et al. 2011). In the above investigations, some of the cases exhibited pure apraxia of speech, but most had both aphasia and apraxia of speech. It was thus not possible to dissociate the contribution of the reported lesions to the apraxia of speech from the more general aphasic disorders. In the present study, we had the rare opportunity to investigate in detail the neuroanatomical locus of the lesions in four consecutive patients suffering from pure speech apraxia, that is, “without” orofacial motor impairments and without language comprehension defect.

Electrophysiological stimulation and functional neuroimaging studies in both the human and nonhuman primate brains have demonstrated that the most inferior part of the precentral motor region plays a major role in the production of orofacial movements (Penfield and Boldrey 1937; Penfield and Rasmussen 1950; Roux et al. 2018; Germann et al. 2020; Eichert, Papp et al. 2020; Eichert, Watkins et al. 2020). Indeed, the inferiormost part of the central sulcus is where the larynx, pharynx, velum, mandibulum, and tongue are represented. Experimental anatomical studies in the macaque monkey have provided a detailed understanding of the anatomical connectivity of these orofacial motor representations in the inferior precentral region (e.g., McGuinness et al. 1980; Simonyan and Jürgens 2002, 2005a, 2005b; Jürgens and Ehrenreich 2007; Morecraft et al. 2015).

The detailed exploration of the cortico-cortical connections of the inferior precentral motor region, the adjacent subcentral opercular region, and the rostral insula in the macaque monkey (Morecraft et al. 2015) has provided important information by demonstrating the direct links between these cortical regions. It has also shown that the connectivity of the anterior insula extends posteriorly to involve the adjacent inferior part of the postcentral gyrus, where the orofacial musculature is represented in the primary somatosensory cortex, and the opercular parietal region, where a part of the secondary somatosensory cortex lies (Eickhoff et al. 2006). These experimental anatomical facts have thus provided a basis on which to explore the role of the orofacial cortex around the most inferior part of the central sulcus, the adjacent subcentral opercular region (where the anterior part of the secondary somatosensory cortex lies), and the anterior dysgranular insula in speech praxis in the language-dominant hemisphere.

We were able to examine in detail four patients who presented with “pure” speech apraxia (i.e., speech apraxia without aphasic symptoms and orofacial motor impairments) and to obtain excellent documentation of their lesions via modern structural neuroimaging. Our aim was to carry out detailed language examination, including speech praxis, and examine their symptoms in relation to the structural cortical damage in an effort to throw light on the specific cortico-cortical circuit underlying speech praxis, that is, the production of action for the articulation of speech.

Μaterials and Methods

Patients

Informed consent was obtained according to the Declaration of Helsinki from each patient or from their closest relative using the procedures approved by the local ethics committee. Participants were consecutively selected from the corpus of patients of the Auxilum Vitae Volterra Rehabilitation Hospital, from 2009 to 2018. None of the patients had impaired level of consciousness or muteness and there was no history of dementia, hearing loss, or visual loss. The patients were tested 40–90 days after the stroke.

Patient 1 (P1). She was in her 60th year of age, right-handed person, with 17 years of education. She was admitted to the rehabilitation hospital 36 days after a left hemisphere ischemic stroke of the supra-Sylvian branch of the left middle cerebral artery. Nonspeech orofacial movements were normal, as assessed by an apraxia test (Spinnler and Tognoni 1987) that requires the patient to produce distinct orofacial movements requested by the examiner via verbal instructions or requests to imitate the examiner’s movements. This patient had normal verbal comprehension and was able to produce a few comprehensible words that were relatively correctly articulated (see Table 1 and Materials and Methods section).

Patient 2 (P2). She was a right-handed person in her 80th year of age, with 5 years of education. She was admitted to the rehabilitation hospital 4 days after a left hemisphere hemorrhage in the territory of the middle cerebral artery. Nonspeech orofacial movements were normal, and she had normal verbal comprehension and was able to produce a few comprehensible words (see Table 1 and Materials and Methods section).

Patient 3 (P3). He was a right-handed person in his 70th year of age, with 13 years of education. He was admitted to the rehabilitation hospital 25 days after a left hemisphere ischaemia of the supra-Sylvian branch of the left middle cerebral artery. Nonspeech orofacial movements were normal, and he had normal verbal comprehension. He could produce and repeat some simple spoken words that were relatively correctly articulated (see Table 1 and Materials and Methods section).

Patient 4 (P4). He was a right-handed man in his 70th year of age, with 8 years of education. He was admitted to the rehabilitation hospital 5 days after a left hemisphere ischaemia of the supra-Sylvian branch of the left middle cerebral artery. Nonspeech orofacial movements were normal, and he had normal verbal comprehension and severe difficulties in producing comprehensible words (see Table 1 and Materials and Methods section).

Orofacial Praxis and Language Evaluation

Each patient was evaluated 2–3 days before the discharge from the rehabilitation hospital for language and general praxis ability with the following tests: the “Esame del Linguaggio II” (Ciurli et al. 1996), the Token Test (Spinnler and Tognoni 1987), the Oral Apraxia Test for nonspeech orofacial movements, and the Ideomotor Apraxia Test (Spinnler and Tognoni 1987).

The Esame del Linguaggio II is a general neuropsychological battery of tests, which evaluates various aspects of language performance: spontaneous speech, which requires the patient to describe an event or a complex figure; naming ability, which requires the naming of pictures by using nouns or verbs and also the verbal production of automatic series (i.e., the months, the days of the week, and counting forward the numbers from “1” to “10”); verbal and written comprehension is assessed by the identification of the correct pictures following the presentation by the examiner of spoken or written words and, for the sentences, the patient is required to align correctly a group of particular words; repetition requires the subject to listen and repeat words, nonwords, and sentences; and reading requires the patient to read aloud written words, nonwords, and sentences.

The “Token Test” evaluates verbal comprehension. The subject is faced with 24 tokens of two different sizes (large or small) having two shapes (circle or square) and five different colors (red, black, white, yellow, and green) and is required to touch the appropriate tokens according to the verbal commands of the examiner (e.g., item 28: using only the large size tokens: “Put the red square away from the yellow one”).

The “Orofacial Apraxia Test” evaluates the production of orofacial movements required by verbal or imitation requests by the examiner. Thus, this test examines the ability of the patient to produce and also imitate orofacial acts that are not related to the production of speech (e.g., send a kiss or move the tongue to the left side).

The “Ideomotor Apraxia Test” evaluates the patient’s ability to perform voluntary meaningful actions of the hand and arm based on verbal instructions by the examiner and/or imitation requests by the examiner (e.g., imitate the military salute made by the examiner).

All the patients had a clinical follow-up after 1 year for general health evaluation.

Neuroimaging Data Acquisition

Imaging data for the patients were acquired with magnetic resonance imaging (MRI) or computed tomography (CT). For patient 1, T2w FLAIR MRI was acquired using a SIEMENS Symphony 1.5 T scanner (spin-echo inverse recovery sequence, TR = 10 000 ms, TE = 120 ms, TI = 2500 ms, 120° flip angle, 248 × 256 image matrix with 0.898 × 0.898 mm in-plane resolution and 20 6.5-mm thick slices. For patients 2–4, the brain images were acquired using a GE Light-Speed VCT 64 with a 512 × 512 image matrix with slightly different imaging parameters. Patient 2 had 36 5.338-mm thick slices with 0.469 × 0.469 mm pixels. Patient 3 had 30 5.00-mm thick slices with 0.488 × 0.488 mm pixels. Patient 4 had 29 4.15-mm thick slices with 0.488 × 0.488 mm pixels.

The morphological complexity of the fronto-opercular cortex and its proximity to insular cortex make proper identification and isolation of the loci of activation in this region difficult. The use of group averaging methods in human neuroimaging might contribute to this problem (Amiez et al. 2016). For this reason, we carried out both an average group analysis and analysis for each individual patient.

Neuroimaging Processing

All image volumes were transformed and resampled into a common brain-based coordinate system for comparison. The interactive graphics software program Register (http://www.bic.mni.mcgill.ca/ServicesSoftwareVisualization/Register) was used to identify homologous anatomical landmarks within the subject data and the MNI_ICBM152 unbiased nonlinear stereotaxic T1w average brain (Fonov et al. 2009, 2011). Using axial, sagittal, and coronal views, the user can navigate throughout the volumes to select at least five landmarks to define the transformation from the native patient image to MNI_ICBM152 space. The transformation is used to resample each dataset onto a 193 × 229 × 193 voxel grid with 1 mm³ spacing to facilitate comparison between the subjects.

Individual lesions were then drawn manually on the re-oriented brain normalized in Montreal Neurological Institute (MNI) standard stereotaxic space using Display http://www.bic.mni.mcgill.ca/software/Display /Display.html). This program permits labeling of the regions of voxels on each slice of the MRI volume and allows for the simultaneous visualization of the movement of the cursor on the screen within the sagittal, axial, and coronal planes of the MRI. Selection of any area can be performed by using the Display mouse-brush that marks the voxels by coloring. This coloring procedure that is accompanied by the simultaneous 3D view of the CT or MRI planes allows an unambiguous identification of the lesion landmarks. The axial view is the default view, but reference to the coronal and sagittal views can also be made when identifying the boundaries of the brain lesion.

Anatomical Analysis

Descriptive statistical lesion mapping was run by averaging the labeled voxels in each patient to generate the lesion probability maps. The boundary of each individual patient and of the patients’ group average lesion was identified using the Atlas of the Morphology of the Human Cerebral Cortex on the Average MNI Brain (Petrides 2019).

Brain Diagram

The schematic drawing of the left hemisphere presented in Figure 3 has been adapted from Economo and Koskinas (1925), but the cortical areas are named according to the Brodmann (1909) nomenclature that is more familiar to the neuroscience community.

Results

All four patients presented here were right-handed adults: Three of them had an ischemic and the fourth had a hemorrhagic stroke affecting the left middle cerebral artery territory as determined by the CT or MRI scans. At the time of testing (40–90 days after the stroke), none of the patients had aphasia and they could all produce normally nonspeech-related orofacial movements. Their verbal production did not present “conduit d’approche” and it was characterized by speech sound errors approximating a target word (e.g., “peruto” or “papalo” instead of “saluto” [greeting]; or “maccio” or “giuggo or ciucciuccia or macciura or marigio” instead of “mangiare” [to eat]). Note also that the patients did not have reading difficulties (e.g., matching written words to the correct figures). Thus, all four patients exhibited pure speech apraxia.

Orofacial Praxis and Language Scores

All four patients had speech apraxia, that is, an impairment in the coordination of complex articulatory movements for speech. Note, however, that none of the patients had nonspeech orofacial apraxia, that is, they could all imitate and produce, upon verbal request, movements of the oral and facial musculature, as assessed by the Orofacial Apraxia Test (Spinnler and Tognoni 1987). Thus, the apraxia was “specific to the production of speech.” Note also that all four patients had a normal speech comprehension score and could understand written language, except for P3, who had some difficulties in comprehending written language. The results of the test scores are presented in Table 1.

The clinical follow-up approximately 1 year after the stroke demonstrated that all four patients had essentially the same speech apraxia that had been observed at least 40 days after the stroke.

Neuroanatomical Findings

The brain images for each patient are presented in Figure 1. Visual inspection of the brain scans of the individual patients demonstrated that they all had a lesion located in the precentral gyrus of the insula (gyrus brevis III) and, to a lesser extent, the adjacent subcentral opercular cortex and the most inferior part of the central sulcus (i.e., primary motor cortex in the anterior bank and somatosensory cortex in the posterior bank). The orofacial musculature is represented in the inferior part of the central sulcus. Possible damage to part of the claustrum and of the extreme capsule that lies next to the insula cannot be excluded. P1, P2, and P4 also had a partial lesion of the inferior postcentral gyrus where somatosensory representations of the orofacial musculature are found. None of the patients had a lesion of the pars triangularis (area 45) of the inferior frontal gyrus, and only patient P3 had minor damage to the pars opercularis (area 44) of the inferior frontal gyrus (see Fig. 1). In addition, the subcortical structures of the caudate nucleus and the anterior branch of the internal capsule were not damaged by the lesions.

The probability map of the lesions resulting from the group average (Fig. 2) demonstrates a peak of lesion overlap located in the precentral gyrus of the insula (gyrus brevis III, MNI stereotaxic coordinates: X: −38, Y: −3, and Z: 14; Petrides 2019). There was also minor involvement of the most inferior part of the central sulcus (MNI coordinates: X: −57, Y: −3, and Z: 31; Petrides 2019) and of the subcentral opercular cortex (MNI coordinates: X: −57, Y: −1, and Z: 9; Petrides 2019). The location of the lesion in the gyrus brevis III (precentral gyrus) of the insula is comparable to that reported previously as being commonly damaged in the patients with apraxia of speech (Dronkers 1996; Nagao et al. 1999; Ogar et al. 2006; Baldo et al. 2011).

Discussion

We evaluated in detail the location of the brain lesions in four patients who presented with speech apraxia without exhibiting any signs of aphasia or impairment in the production of orofacial nonspeech movements. Thus, these patients exhibited the syndrome of pure speech apraxia. Both the brain images of the individual patients (Fig. 1) and also the probability map of the overlap of the lesions (Fig. 2) obtained by averaging the individual brain lesions in MNI stereotaxic space indicate that the critical lesion in all four patients involved the precentral gyrus of the insula (gyrus brevis III), which is composed of dysgranular cortex and, to a lesser extent, the nearby areas with which it is strongly inter-connected: parts of the adjacent subcentral opercular cortex (secondary somatosensory cortex) and the most inferior part of the central sulcus region where the orofacial musculature is represented. There was no damage to the rostrally adjacent Broca region (i.e., the pars opercularis, area 44, and the pars triangularis, area 45) on the inferior frontal gyrus, nor in the superior temporal gyrus and adjacent superior temporal sulcus (Wernicke’s region), which when damaged produce aphasia. The neuropsychological evaluation of all four patients was performed after the sub-acute phase of the brain damage (at least 40 days after the stroke) and, approximately 1 year later, all of the patients were still exhibiting pure speech apraxia without aphasia. It is important to emphasize here that none of the patients had a verbal comprehension deficit (i.e., aphasic symptoms) and none had a problem in producing nonspeech orofacial acts upon request. It is, therefore, clear that the impairment was restricted to the domain of speech praxis, that is, all four patients exhibited pure speech apraxia.

Because it had been suggested that speech apraxia may involve what is referred to as Broca’s area, that is, the left posterior inferior frontal gyrus (Hillis et al. 2004), we examined carefully this region for any infarct. In our cohort of four patients with pure speech apraxia, there was no lesion invading the pars triangularis (area 45) or the pars opercularis (area 44) of the inferior frontal gyrus, except for patient 3, who had minor damage to the pars opercularis (Fig. 1). The sparing of the posterior inferior frontal gyrus (Broca’s area) is consistent with the demonstration that none of the patients exhibited aphasic symptoms. The deficit was clearly pure apraxia of speech. The observations of Hillis et al. (2004) were made on patients in the acute stage of brain damage, when phenomena such as diaschisis could explain widespread damage. By contrast, our evaluations were performed after the acute stage (at least 40 days after the ictal event).

Note that a voxel-based lesion-symptom mapping study investigating patients who demonstrated speech apraxia during the acute stage (within 5–10 days from stroke) pointed out that the damage was focussed in the left precentral gyrus, but in the lesion overlap maps for all the patients suffering from pure speech apraxia, there was also damage in the left insula but not in the posterior inferior frontal gyrus (Broca’s area) (Fig. 1, panel B; Itabashi et al. 2016). In our patients, the diagnosis of pure speech apraxia was carried out at least 40 days after the brain damage and was confirmed by the clinical evaluation at follow-up approximately 1 year later. A possible lesion of the white matter of the extreme capsule and of the claustrum that lie next to the insula cannot be excluded in the investigated patients. Note, however, that the connections of the anterior insula with the subcentral opercular region (part of the second somatosensory cortex) and the region around the inferior central sulcus (precentral and postcentral banks) would involve axons coursing via this part of the extreme capsule, that is, white matter just below the insular cortex. The temporo-frontal extreme capsule fasciculus also provides anterior temporal input to the anterior insula and the lateral frontal lobe (Petrides and Pandya 1988; Petrides 2014). It is thus an important pathway enabling the anterior insula to receive input from the anterior superior temporal gyrus (auditory) and the middle temporal gyrus (multisensory semantic) in the language hemisphere. The anterior insula, in turn, is connected with the central-to-parietal opercular region and the orofacial somatomotor zone on the precentral and postcentral gyri (see Morecraft et al. 2015). Note, also, that none of the main structures of the basal ganglia and, in particular, the caudate nuclei were involved and that there was also no involvement of the anterior branch of the internal capsule as reported in some earlier investigations (De Renzi et al. 1980; Kertesz et al. 1984; Papagno et al. 1993). As to the possibility of temporo-parietal lesions (Square et al. 1997), in the present cases, there was minor damage to the inferior part of the postcentral gyrus where the orofacial region is represented.

Earlier work investigating patients in the chronic stage who had suffered from speech apraxia provided evidence that the superior precentral gyrus of the insula (gyrus brevis III) in the language-dominant left hemisphere might be critical for the coordination of speech articulation, but it could not provide unambiguous evidence because many of the cases had accompanying aphasic symptoms (Dronkers 1996; Ogar et al. 2006; Baldo et al. 2011). The anatomical findings from the four cases with speech apraxia without aphasia reported here provide strong support to the hypothesis that the superior part of the precentral gyrus of the insula (gyrus brevis III) and its connections with the nearby subcentral opercular cortex (part of the secondary somatosensory region) and the most inferior central sulcus (where the orofacial musculature is represented) constitute the critical neural circuit for speech praxis (Fig. 3).

The representations of the orofacial musculature lie in the inferior part of the central sulcus. In the most inferior part lie the oral motor representations (larynx, pharynx, velum, mandibulum, tongue, etc.) on the anterior bank of the sulcus and the nearby precentral gyrus (Germann et al. 2020; Eichert, Papp et al. 2020; Eichert, Watkins et al. 2020) and the somatosensory representations on the posterior bank of the central sulcus (Zlatkina et al. 2016). The inferior motor region representing the orofacial musculature (i.e., primary motor cortical area 4) is connected with: 1) the adjacent posterior bank of the central sulcus where the orofacial components of the primary somatosensory cortex are represented, 2) the adjacent subcentral opercular cortex where part of the secondary somatosensory cortical areas OP3 and OP4 lie (Eickhoff et al. 2006), and 3) the anterior dysgranular insula (Kertesz et al. 1984; Morecraft et al. 2015). The neuroanatomical cortico-cortical connections of this specific region, namely, the precentral gyrus of the insula (gyrus brevis III) with the inferior central sulcal region, would suggest that, in the language-dominant hemisphere, will underlie the articulation of words, and damage “restricted” to this circuit would produce pure speech apraxia, as was the case in the four patients reported here.

It is of interest to note that cytoarchitectonic studies have suggested that the inferior part of the motor region (representing the orofacial musculature) stems from the anterior insular cortex and the adjacent dorsal Sylvian opercular region (subcentral gyrus; Morecraft et al. 2015; Pandya et al. 2015). This local cortico-cortical circuit involving the dysgranular insula, the adjacent subcentral opercular region (part of the secondary somatosensory cortex) and the most inferior parts of the central sulcal region (see Fig. 3) has been shown by the results of the present investigation to be the critical one for the phonic articulation of verbal information (i.e., speech coordination of the sequential articulatory movements) in the language-dominant hemisphere. This local orofacial articulation circuit has connections with Broca’s region on the inferior frontal gyrus (Frey et al. 2014) that is planning the speech utterance via the precentral premotor cortex (area 6), and lesions of this circuit together with damage to Broca’s region will yield speech apraxia together with other language-related impairments. However, note that the highly focal lesions restricted to the speech articulation circuit in the present four patients led to pure speech apraxia without aphasia. Thus, the present study demonstrated a clear dissociation of the cortico-cortical circuit for speech praxis from other language production circuits, damage of which leads to aphasia.

Note that the four patients with speech apraxia reported here had no problem in producing nonspeech-related orofacial movements. This observation can be explained by the fact that the orofacial representation in the motor cortex is bilateral (Kuypers 1958; Jenny and Saper 1987) and also the fact that there are other representations of the orofacial region in the cortex, namely, the supplementary motor cortex and the cingulate motor areas M3 and M4 (Morecraft et al. 2004). The latter orofacial representations have direct subcortical projections to the facial bulbar nucleus (Morecraft et al. 2004).

In speech production models, speech apraxia occurs after the cognitive formulation of the speech utterance (failure of which would lead to aphasic errors) and reflects an impairment in the coordination of complex articulatory movements before the execution of the motor plan. Note that speech apraxia must be distinguished from dysarthria due to muscle weakness and loss of control (Darley et al. 1975). Here, it is shown that the critical circuit for speech praxis involves the dysgranular motor cortex of the precentral gyrus of the insula (gyrus brevis III) and its connections with adjacent areas where the representations of the orofacial musculature are found: the inferior central sulcal region as well as the subcentral opercular cortex.

Funding

M.P. was funded by CIHR Foundation Grant (FDN-143212).

Notes

We thank Professor Costanza Papagno for her comments. We also thank the speech therapists of the Auxilium Vitae Volterra, Elena Baroncini, Elisabetta Cigni, Susan Di Sacco, and Monica Orlandini for their contributions to data collection Conflict of Interest: None declared.

References