Improved naming in patients with Broca’s aphasia with tDCS

Abstract

Background:

Language impairment (aphasia) is a common neurological deficit following stroke. For patients with chronic aphasia (6 months after the stroke), language improvements with speech therapy (ST) are often limited. Transcranial direct current stimulation (tDCS) is a promising approach to complement language recovery, but interindividual variability in treatment response is common after tDCS, suggesting a possible relationship between tDCS and type of linguistic impairment (aphasia type).

Methods:

This current study is a subgroup analysis of a randomized controlled phase II futility design clinical trial on tDCS in chronic post-stroke aphasia. All participants received ST coupled with tDCS (n=31) vs. sham tDCS (n=39). Confrontation naming was tested at baseline, and 1-, 4-, and 24-weeks post-treatment.

Results:

Broca’s aphasia was associated with maximal adjunctive benefit of tDCS, with an average improvement of 10 additional named items with tDCS+ST compared with ST alone at 4 weeks post-treatment. In comparison, tDCS was not associated with significant benefits in other aphasia types F(1)=4.23, p=0.04. Among participants with Broca’s aphasia, preservation of the perilesional posterior inferior temporal cortex was associated with higher treatment benefit (R=0.35, p=0.03).

Conclusions:

These results indicate that adjuvant tDCS can enhance ST to treat naming in Broca’s aphasia, and this may guide intervention approaches in future studies.

Introduction

Aphasia is a common, disabling neurological impairment post-stroke¹. Twenty percent of stroke survivors exhibit long-lasting language impairments beyond 6 months after the initial cerebrovascular event². Aphasia persisting beyond 6 months post-stroke (chronic aphasia) is strongly associated with depression, decreased quality of life, and social isolation^3–5. Speech therapy (ST) is the standard treatment for chronic post-stroke aphasia but is often associated with unsatisfactory and limited outcomes⁶. For this reason, there is a pressing need for new adjuvant approaches that can enhance the therapeutic benefits of ST.

Direct brain stimulation through transcranial direct current stimulation (tDCS) has recently emerged as a promising non-invasive and practical approach to complement ST^7–11. In a recent phase II prospective randomized controlled trial entitled Transcranial Direct Current Stimulation vs Sham Stimulation to Treat Aphasia After Stroke: A Randomized Clinical Trial (ClinicalTrials.gov Identifier: NCT01686373)⁷, we demonstrated that adjunct tDCS to treat chronic aphasia was associated with improvements in speech production (naming objects) at a sufficiently high level to conclude that future definitive clinical trials of tDCS in chronic aphasia would be warranted and not futile.

In this trial, tDCS was coupled with a form of computerized auditory-visual ST designed to treat naming impairments and recruit residual cortical areas in the left frontal lobe normally associated with speech production¹². The Anode-tDCS electrode was placed over the area in the left peri-lesional temporal lobe cortex with peak individualized pre-treatment naming fMRI activation to target cortical regions engaged in language processing and naming¹³ and to avoid the stroke lesion and current dispersion in the stroke cavity¹⁴. All aphasia types were included in the clinical trial, but we observed significant variability in therapeutical responses, suggesting a specific benefit of this tDCS approach for individuals with Broca’s aphasia and naming impairments. Here, we evaluated whether there was an association between adjuvant tDCS effects and aphasia type, and whether tDCS benefits were maximal for individuals with Broca’s aphasia.

Participants and Methods

This is a subgroup analysis of data from 70 individuals who had complete neuroimaging and long-term post-treatment behavioral assessments from the clinical trial tDCS vs Sham Stimulation to Treat Aphasia After Stroke (NCT01686373)⁷. The Institutional Review Board at the University of South Carolina and Medical University of South Carolina approved all procedures, including informed consent. A comprehensive description of the enrollment criteria, clinical trial design and methods are provided in Fridriksson et al., 2018⁷. In brief, 74 participants English-speaking individuals between 25 and 80 years of age with chronic (> 6 months) aphasia were randomized to two separate aphasia treatment arms: auditory-visual ST coupled with sham-tDCS vs. auditory-visual ST coupled with anodal-tDCS. Randomization was balanced based on enrollment site, age, aphasia type, and baseline aphasia severity. All participants underwent 15 treatment sessions lasting 45 minutes each over 3 consecutive weeks. tDCS was delivered using a Phoresor II PM 850 - Iomed Inc system with 1 mA of A-tDCS stimulation between two 5 × 5 cm saline-soaked sponges (electrodes). The cathode electrode was placed on the right supraorbital frontal scalp region. The anode tDCS electrode was placed over the residual temporal lobe cortical area, which demonstrated the highest personalized fMRI activation during two confrontation naming fMRI studies performed on two consecutive days at baseline. Sham-tDCS was performed by turning on the stimulation for 30 seconds, followed by a gradual reduction and ultimately cessation of the current within the subsequent 15 seconds. Each participant and clinician were asked to guess the stimulation type at the end of their treatment phase. Their guessing accuracy was 47.9% and 54.2%, respectively, indicating chance guessing and providing a confirmation of the adequate blinding procedure⁷. The trial’s primary outcome measure was the number of additional correctly named items 1 week after therapy on the 175 item Philadelphia Naming Test (PNT)¹⁵ and on trained items during ST.

In this present study, to better access treatment generalizability, we focused on treatment benefits on untrained items only, i.e., the PNT, since the ST did not include the same items as the PNT. We performed an analysis of variance (ANOVA) with the dependent variable set as the number of additionally named items on the PNT post- vs. pre-therapy and 2 between-subjects factors: 1) Broca’s aphasia vs. other aphasia types, and 2) A-tDCS vs. S-tDCS. We also examined the relationship between structural brain integrity and treatment benefits. We focused on the posterior temporal inferior cortex since this is an area directly involved with naming^{16 17} and represented the location with the highest average tDCS anodal stimulation in the perilesional temporal area across all participants⁷. Cortical integrity was determined using structural neuroimaging. Chronic post-stroke lesions were manually outlined on T1 weighted MRI, normalized to standard MNI space using an enantiomorphic approach¹⁸ with unified segmentation-normalization¹⁹, which allows for the transformation of the T1-weighted images onto standard space, with the resultant spatial transformation matrix being applied to the stroke lesion mask. To avoid type II statistical errors due to the small sample size, region of interest (ROI) analyses were conducted instead of voxel-wise. The posterior inferior temporal gyrus ROI from the (John’s Hopkins University (JHU) brain Atlas^{20 21} was chosen as the best representation of the posterior temporal inferior cortex (this ROI corresponds to Brodmann Areas 20 and 37), and the percentage of damage to the posterior inferior temporal gyrus ROI was calculated for each individual. We then performed a partial correlation to test the relationship between ROI damage and recovery defined as additional named items in the PNT after therapy, controlling for lesion size.

Results

There were no statistically significant differences in age, baseline aphasia severity, aphasia type, or years of education between the active and sham tDCS groups⁷. There were also no significant differences in aphasia severity between treatment groups for the participants with Broca’s aphasia (T=−0.3, p=0.76). Overall (n=70) A-tDCS was associated with an increase over S-tDCS in correctly named PNT items of 3.32±2.45 items at 1 week, 4.70±3.0 at 4 weeks, and 5.49±3.38 at 24 weeks post-therapy. Conversely, among individuals with Broca’s aphasia (n=38), these differences were 6.75±3.42 (10.52 25.89%) items at 1 week, 10.23±3.95 (23.27 42.29) at 4 weeks, and 10.49±4.25 at 24 weeks. Among all other individuals besides those with Broca’s aphasia, these differences were −0.79±3.43 items at 1 week, −1.85±4.44 at 4 weeks, and −0.38±5.64 at 24 weeks, with ST alone yielding a higher benefit in this group.

At 4-weeks post-therapy, there was a significant interaction between aphasia type and tDCS (mixed effects ANOVA with 2 between-subjects factors: 1) Broca’s aphasia vs. other aphasia types, and 2) A-tDCS vs S-tDCS) F(1)=4.2258, p=0.0437. These results are shown in Figure 1A. The percentage of improvements and Cohen’s D effect sizes are demonstrated in Table 1. Interestingly, the numbers-needed-to-treat (NNT) for A-tDCS vs S-tDCS are shown in Figure 1B. Considering that a 9-item improvement in the PNT is a clinically meaningful difference, the NNT for A-tDCS was 1.9 for A-tDCS vs 4.2 for S-tDCS. A partial correlation, controlling for lesion size, demonstrated that individuals with more preservation of the posterior inferior temporal gyrus had higher treatment benefit (p=0.03, one-tailed) (Figure 1C).

Figure 1.

Figure 1A – Top row: lesion overlaps for each patient group, with the colorbars indicating the number patients with lesions at each voxel. The error bars correspond to the standard error of the mean (SEM - standard deviation of the sampling distribution). Bottom row: the subplots demonstrate treatment outcomes (gains in the number of correct named items in the PNT) at each time point post-treatment (1-, 4- and 24- weeks) for the corresponding patient groups.

Figure 1B – The box and whisker plots demonstrate the PNT gains at 4 weeks post-treatment for participants with Broca’s aphasia. This is an expanded view of the data in the center error bar plot in 1A. Based on the PNT gains at 4 weeks post-treatment, the line graph demonstrates the number-needed-to-treat (NNT) (y axis) across a range of PNT gains (x axis) for A-tDCS vs S-tDCS. Note that for a PNT= 9 points, the NNT is 1.9 for A-tDCS (speech therapy with tDCS) and 4.2 for S-tDCS (speech therapy alone).

Figure 1C – The scatter plot demonstrates that a lower percentage of lesion (higher preservation) of the Posterior Inferior Temporal Gyrus (PSIG) was associated with higher treatment-related naming improvement among patients who received A-tDCS, controlling for total lesion size. The size of each data point in the scatter plot is proportional to the total lesion volume. The PSIG region of interest is shown in purple, and it is also identified by the red crosshair. The lesion overlay across all patients is shown in “jet” color scheme and the cortical projection of the electrode centers-of-mass across all participants is shown in blue in the lower brain rendering. At a group level, the PSIG was located in the perilesional temporal region with high anode electrode coverage.

Table 1 –

This table provides a summary of the PNT gains of A-tDCS over S-tDCS at 4 weeks post-treatment.

Category	Groups		PNT gains of A-tDCS over S-tDCS at 4 weeks post-treatment
	A-tDCS	S-tDCS	PNT items*	Percentage improvement compared with baseline	Proportion of maximal gains**	Cohen’s D* (95% CI)
All participants	n=31	n=39	4.70±3.0	6.00±4.07	3.00±1.71	0.37 (−0.10, 0.84)
Broca’s aphasia	n=17	n=21	10.23±3.95	11.00 4.19	6.00 2.26	0.83 (0.17, 1.48)
All except Broca’s aphasia	n=14	n=18	−1.85±4.44	−0.00 7.25	−1.00 2.54	−0.15 (−0.83, 0.54)

^*The statistical analyses described in the text, and the Cohen’s D calculations were performed based on PNT items for consistency with the original tDCS clinical trial. The PNT item improvement was chosen instead of proportion of maximal gains since to avoid numerical confounders given the high correlation between proportion of maximal gains and baseline impairments. Likewise, PNT item improvement was chosen instead of percentage of improvement compared with baseline to avoid comparisons between dissimilar levels of improvement between individuals with different baselines (e.g., a 17% improvement for a participant with a PNT = 150 at baseline indicates complete recovery, whereas the same level of improvement for a participant with a PNT = 50 at baseline is less impactful).

^**Proportion of maximal gain is calculated as improvement divided by total number of PNT items (175) minus baseline impairment.

Discussion

Chronic aphasia is not a stagnant condition, with significant improvements observed even years after stroke. The modern neurobiological framework of aphasia recovery suggests that improvements in the chronic period are related to the engagement of residual peri-lesional language areas or multimodal brain regions that compensate for the loss of language networks²². This process may be enhanced by external brain stimulation with tDCS, which may augment brain plasticity during treatment¹⁰.

The exact mechanisms of tDCS are not yet fully understood, and much of the existing literature is related to experimental models²³, underscoring key limitations in our understanding of the neurobiology of brain stimulation in humans. From animal studies, tDCS appears to be associated with physiological changes that improve neuroplasticity by increasing brain-derived neurotrophic factor (BDNF)²⁴, facilitating action potential generation²⁵, reducing Gamma-Aminobutyric Acid (GABA) response, and enhancing dendritic remodeling and dendritic density²⁵.

As demonstrated here, adjuvant tDCS improved recovery for individuals Broca’s aphasia, which could be considered an indirect marker of neuroplasticity. A-tDCS was targeted to the peri-lesional temporal cortex with the goal of engaging temporal-frontal networks involved in naming¹³. Indeed, better outcomes were associated with the preservation of the inferior temporal cortex, a region relevant for naming in the perilesional temporal region with high anode electrode coverage. This suggests that anodal effects in the preserved functionally relevant cortex is likely important, and this topic could be examined by direct studies assessing tDCS modeled electric fields in the future. It is possible that the tDCS anodal stimulation used in this study is better suited Broca’s aphasia irrespective of the form of ST, and these are also questions for future studies. Nonetheless, this study confirmed that there is a clear interaction between tDCS coupled with auditory-visual ST for Broca’s aphasia.

There is increasing agreement in the literature that tDCS coupled with speech therapy is a promising approach to improve language outcomes in chronic post-stroke. Additional definitive level I evidence studies are needed for optimal clinical translation. Our post-hoc analyses provide supporting evidence that maximal adjuvant benefit of tDCS coupled with auditory-visual ST can be achieved among individuals with Broca’s aphasia, and this information may guide future definitive treatment studies.