Less is more: neural mechanisms underlying anomia treatment in chronic aphasic patients

Abstract

See Thompson and Woollams (doi:10.1093/brain/awx264) for a scientific commentary on this article. Previous research with aphasic patients has shown that picture naming can be facilitated by concurrent phonemic cueing [e.g. initial phoneme(s) of the word that the patient is trying to retrieve], both as an immediate word retrieval technique, and when practiced repeatedly over time as a long-term anomia treatment. Here, to investigate the neural mechanisms supporting word retrieval, we adopted—for the first time—a functional magnetic resonance imaging task using the same naming procedure as it occurs during the anomia treatment process. Before and directly after a 6-week anomia treatment programme, 18 chronic aphasic stroke patients completed our functional magnetic resonance imaging protocol—a picture naming task aided by three different types of phonemic cues (whole words, initial phonemes, final phonemes) and a noise-control condition. Patients completed a naming task based on the training materials, and a more general comprehensive battery of language tests both before and after the anomia treatment, to determine the effectiveness and specificity of the therapy. Our results demonstrate that the anomia treatment was effective and specific to speech production, significantly improving both patients’ naming accuracy and reaction time immediately post-treatment (unstandardized effect size: 29% and 17%, respectively; Cohen’s d: 3.45 and 1.83). Longer term gains in naming were maintained 3 months later. Functional imaging results showed that both immediate and long-term facilitation of naming involved a largely overlapping bilateral frontal network including the right anterior insula, inferior frontal and dorsal anterior cingulate cortices, and the left premotor cortex. These areas were associated with a neural priming effect (i.e. reduced blood oxygen level-dependent signal) during both immediate (phonemically-cued versus control-cue conditions), and long-term facilitation of naming (i.e. treated versus untreated items). Of note is that different brain regions were sensitive to different phonemic cue types. Processing of whole word cues was associated with increased activity in the right angular gyrus; whereas partial word cues (initial and final phonemes) recruited the left supplementary motor area, and right anterior insula, inferior frontal cortex, and basal ganglia. The recruitment of multiple and bilateral areas may help explain why phonemic cueing is such a successful behavioural facilitation tool for anomia treatment. Our results have important implications for optimizing current anomia treatment approaches, developing new treatments, and improving speech outcome for aphasic patients.

Keywords: anomia treatment; aphasia; phonemic cueing; picture naming; word retrieval.

Figure 1

Lesions overlap in our sample of patients. Colour range indicates the amount of overlap expressed as number of patients (colour bar). Numbers on the top represent z MNI coordinates of brain sections, displayed in neurological convention (i.e. L is L, R is R).

Figure 2

Study design (A) and functional MRI experimental protocol (B). fMRI = functional MRI; SOA = stimulus-onset asynchrony.

Figure 3

Behavioural results of Experiment 1 (A–B) and Experiment 2 (C–D). Dispersions represent standard errors of the mean (SEM). Significance of post hoc comparisons: *P ≤ 0.05; **P ≤ 0.01; ***P ≤ 0.005; ****P ≤ 0.001; n.s. = non-significant. For Experiment 1, ANOVAs reported are run across T1 and T2 only, for consistency with all other analyses (results at T3 are reported to show performance at follow-up). Note that (C) illustrates that the differences between untreated and ‘to-be-treated’ items at T1 were either non-significant or counter the predicted direction (i.e. UNT > TRE#), whereas at T2 all differences were significant and in line with the predicted direction (TRE > UNT). % = percentage of correct responses; F = final; I = initial; N = noise; TRE = treated items (at T1, TRE# = ‘to-be-treated' items); UNT = untreated items; W = word.

Figure 4

Functional MRI results. (A) Reductions in BOLD response related to immediate facilitation of naming. (B) Reductions in BOLD response related to long-term facilitation of naming. (C) Increased BOLD response for treated (as compared to untreated) items. (D) Activations related to partial (i.e. initial and final phonemes) cues processing. (E) Activations related to whole word cues processing. Results are displayed at P < 0.05 (FWE-corr.), except in (B), where they are displayed at P < 0.001 (unc.) for cluster extent, without correction at cluster-level (to allow for small-volume correction). Red and yellow bars refer to the conditions compared (red > yellow, as in the title of each contrast). Blue bars are conditions not included in the contrasts. a.u. = arbitrary unit; AG = angular gyrus; AIC = anterior insular cortex; Control = control trials (i.e. noise); Cued = cued trials (i.e. word, initial, and final); dACC = dorsal anterior cingulate cortex; F = final; I = initial; IFC = inferior frontal cortex; N = noise; Partial = partial cues (i.e. initial and final phonemes); PCN = precuneus; PMC = premotor cortex; SMA = supplementary motor area; TRE = treated items (TRE# = ‘to-be-treated' items); UNT = untreated items; W = word; Words = whole word cues; x y z = MNI coordinates of brain sections. Sections are displayed in neurological convention.

Figure 5

Correlation between behaviour and brain activity in the right anterior insular cortex (A) and in the right inferior frontal cortex (B). Plots show the relationship between naming efficiency (computed as the difference between mean RT: untreated-treated items at T2) and corresponding changes in BOLD response (i.e. untreated-treated) extracted during the noise-control condition. A greater improvement in naming efficiency (delta RT) is associated with greater changes in BOLD response (delta BOLD). R AIC = right anterior insular cortex; R IFC = right inferior frontal cortex; a.u. = arbitrary unit.